Patients and Caregivers as Developers of Medical Devices: An Empirical Study on User Innovation in the Healthcare Sector (Forschungs-/Entwicklungs-/Innovations-Management) 3658320400, 9783658320409

Citation preview

Forschungs-/ Entwicklungs- / Innovations-Management

Moritz Göldner

Patients and Caregivers as Developers of Medical Devices An Empirical Study on User Innovation in the Healthcare Sector

Forschungs-/Entwicklungs/Innovations-Management Series Editors Hans Dietmar Bürgel (em.), Stuttgart, Germany Diana Grosse, Freiberg, Germany Cornelius Herstatt, Hamburg, Germany Hans Koller, Hamburg, Germany Christian Lüthje, Hamburg, Germany Martin G. Möhrle, Bremen, Germany

Die Reihe stellt aus integrierter Sicht von Betriebswirtschaft und Technik Arbeitsergebnisse auf den Gebieten Forschung, Entwicklung und Innovation vor. Die einzelnen Beiträge sollen dem wissenschaftlichen Fortschritt dienen und die Forderungen der Praxis auf Umsetzbarkeit erfüllen. Professor Dr. Hans Dietmar Bürgel (em.), Universität Stuttgart Professorin Dr. Diana Grosse vorm. de Pay, Technische Universität Bergakademie Freiberg Professor Dr. Cornelius Herstatt, Technische Universität Hamburg Professor Dr. Hans Koller, Helmut-Schmidt-Universität Hamburg Professor Dr. Christian Lüthje, Technische Universität Hamburg Professor Dr. Martin G. Möhrle, Universität Bremen

More information about this series at http://www.springer.com/series/12195

Moritz Göldner

Patients and Caregivers as Developers of Medical Devices An Empirical Study on User Innovation in the Healthcare Sector

Moritz Göldner Hamburg, Germany Moritz Göldner Institute for Technology and Innovation Management Hamburg University of Technology Hamburg, Germany Dissertation Technische Universit¨at Hamburg/2020

Forschungs-/Entwicklungs-/Innovations-Management ISBN 978-3-658-32040-9 ISBN 978-3-658-32041-6 (eBook) https://doi.org/10.1007/978-3-658-32041-6 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Responsible Editor: Carina Reibold This Springer Gabler imprint is published by the registered company Springer Fachmedien Wiesbaden GmbH part of Springer Nature. The registered company address is: Abraham-Lincoln-Str. 46, 65189 Wiesbaden, Germany

Foreword

Healthcare professionals have long been recognized as a valuable source of innovation in the healthcare sector. Yet, the roles of patients and caregivers in innovation of medical devices has so far seen little attention in academia and industry. Also from our own experience we know that patients are increasingly evolving from passive consumers to knowledgeable and critical recipients of healthcare products and services. The spread of the Internet plays an important role in this process, but does not fully explain this phenomenon comprehensively. In the context of his dissertation, Moritz Göldner is investigating the largely unexplored phenomenon of innovative patients and their caregivers who develop and even market self-developed medical devices. His work is closely connected to the rich body of research on so-called Lead Users. In this research stream the prevalence of patients and caregivers as developers of medical devices has been repeatedly reported, but not yet explicitly empirically investigated. To date, studies on large-scale and real-world data on the development and diffusion of patient-driven innovations have been lacking. His dissertation focuses on three research questions: 1) How do patients and caregivers contribute to innovation in the healthcare sector? 2) What are the characteristics of patients and caregivers as innovators and how do they differ from other innovators in the healthcare sector? 3) What are reasons for patients and caregivers diffusing their innovations to others and thus become user entrepreneurs? To answer his research questions, Moritz Göldner conducts two complementary studies. In his first study he uses a mixed-method approach to evaluate quantitative data consisting of two datasets on analytical app data of more than 1,100 medical smartphone apps each. Based on these results he analyzes qualitative data from 16 interviews with developers of medical apps. In the second

v

vi

Foreword

study, he examines the commercialization activities of patients and caregivers based on 14 cases where a tangible medical device was developed and successfully brought into market. In this study, he conducted in-depth interviews and analyzed secondary data such as patents and online sources. As a result, he has made what I consider to be at least three significant contributions: First, he finds significant evidence for innovative patients and their caregivers who are developers of (digital and tangible) medical devices. Second, he shows that user-developed medical apps are rated significantly better than applications that are innovated by non-user-developers such as manufacturers. A third finding is that innovating patients and caregivers who offer their tangible medical device on the market do not maximize their profits, but try to market their solutions at reasonable, cost-covering prices in order to make them available to as many other patients as possible. By doing so, those patients and caregivers actively contribute to improving the quality of life of many patients. In this way, Moritz Göldner builds a bridge to the emerging research stream on social innovation, which has been essentially decoupled from Lead User research so far. Finally, he develops proposals to important questions of current healthcare system, such as how affordable solutions with high customer value can be generated. The high quality of his results in combination with the application of sophisticated scientific methods as well as the sound interpretation and presentation of the results confirm the unique research approach chosen by Moritz Göldner. For me, the essential contribution of his work lies in the very well-founded discussion and extension of theory as well as his aspirations to disseminate a phenomenon of high theoretical and practical relevance. Here, Moritz Göldner is accomplishing pioneering work and makes a clearly recognizable scientific contribution. It has been a privilege and my pleasure to supervise this dissertation which has been evaluated by all examiners with “summa cum laude”. Hamburg August 2020

Univ. Prof. Dr. Dr. h.c. Cornelius Herstatt

Acknowledgments

My research is based on the idea that some patients and caregivers develop medical devices for their own unmet medical needs—mainly because there is no suitable solution available. During my dissertation project, I had the opportunity to talk to more than hundred patients, caregivers and healthcare professionals and to listen to their incredible journeys. It was fascinating to see how dedicated they were in solving their own and ultimately also others’ health-related problems and how engaged they were in disseminating their innovation. This dissertation would not have seen the light without all these people who were willing to share their stories and experiences—I am very grateful for their time and their support. I would also like to acknowledge those who have supported me during my PhD trajectory at the Institute of Technology and Innovation Management at Hamburg University of Technology (TUHH). First and foremost, I am very grateful to my supervisor, Prof. Dr. Dr. h.c. Cornelius Herstatt. Thank you for all your ideas, stories and feedback during my time at the TIM-institute. You gave me the freedom and confidence to pursue my own research projects and always provided valuable support when needed—on academic and non-academic issues. I have very much enjoyed the time we spent together in Hamburg, but also on conferences and consulting projects around the globe. I am also thankful to my second supervisor, Prof. Pedro Oliveira, PhD from Copenhagen Business School. I very much enjoyed discussing my research with you on many occasions. Your enthusiasm for the phenomenon of patient innovation and your engagement in offering a platform for innovative patients has been a tremendous motivation for me. I would also like to thank Prof. Dr. Christian Lüthje from TUHH for chairing my semi-virtual PhD-defense in times of the COVID-19-pandemic.

vii

viii

Acknowledgments

Further, my gratitude goes to Prof. Eric von Hippel from Massachusetts Institute of Technology who was advising me on my PhD project on a regular basis and repeatedly challenged my approaches. Thank you for the continuous, valuable exchange and the opportunity to present my work on several Innovation Lab meetings at MIT Sloan School. I would also like to thank the deputy director of the TIM-institute, Dr. Stephan Buse for his fostering and the great experiences I gained in our joint projects. I am thankful to all my fellow colleagues at the institute for exchanging knowledge and experiences. I have always appreciated the positive and friendly atmosphere at the institute. My special thanks go to Dr. Thorsten Pieper and Viktoria Drabe for the great time we had together. Thank you for your continuous support, the continuous exchange, countless quips, and so many beautiful hours we spent together in the office and elsewhere. Thank you, Niels Hackius, for always being around to talk about anything and everything and for being my coffee-mate in these years. I would also like to thank Dr. Holger Borchers and Heinrich Schwarz, PhD. I very much enjoyed working together with you on many occasions during industryand teaching-related projects and I am very happy to see that our collaboration will continue in the future. Finally, my greatest thanks go to my parents for their encouragement and continuous support and to my life companion Dr. Franziska Bomba who has stood by me during all stages of this journey. Thank you so much for your love and your tremendous support. Hamburg August 2020

Moritz Göldner

Contents

Part I

Focus and Scope of this Thesis

1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Research Problem and Relevance . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 The Structure of This Dissertation . . . . . . . . . . . . . . . . . . . . . . . .

3 4 5 6

2

Conceptual Foundations: The Phenomenon of Patients and Caregivers as User Innovators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 User Innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1 Heterogeneity of Needs . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.2 Stickiness of Information . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Free Innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Lead User Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 User Entrepreneurship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 The Market Success of User Innovations . . . . . . . . . . . . . . . . . . . 2.6 User Innovation in Healthcare . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9 9 12 12 14 16 17 19 21

Research Gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

27

3

Part II 4 5

Are Patients and Caregivers the Better Innovators? The Case of Medical Smartphone Applications

Introduction: User Innovation and Medical Smartphone Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31

Theoretical Background and Hypothesis Development . . . . . . . . . . . 5.1 Empirical Field: e-Health, m-Health, Digital Health . . . . . . . . .

35 35

ix

x

Contents

5.2 5.3

6

7

8

Medical Smartphone Apps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hypothesis Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.1 Free Revealing of User-Developed Medical Apps . . . . 5.3.2 The Early Development of Medical Apps . . . . . . . . . . 5.3.3 Developer Type and the Quality of Medical Apps . . . 5.3.4 Developer Type and the Download Numbers of Medical Apps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.5 Developer Type and Medical App Revenues . . . . . . . . 5.3.6 Summary of Hypotheses: Research Model for Mediated Regression Analysis . . . . . . . . . . . . . . . . .

38 41 42 43 43

Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1 Analytical App Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.1 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.2 Operationalization of the Measures . . . . . . . . . . . . . . . . 6.1.3 Data Preparation and Assumptions for Regression Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.4 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Qualitative Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.1 Sampling and Data Collection . . . . . . . . . . . . . . . . . . . . 6.2.2 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

49 50 50 54

45 46 46

57 63 64 64 65

Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1 Findings Concerning the Analytical App Data . . . . . . . . . . . . . . 7.1.1 Descriptive Analysis: Dataset 1 (2014) . . . . . . . . . . . . . 7.1.2 Descriptive Analysis: Dataset 2 (2018) . . . . . . . . . . . . . 7.1.3 The Emergence of Medical Apps: A Comparison Between Datasets 1 (2014) and 2 (2018) . . . . . . . . . . . 7.1.4 Regression Analysis of Dataset 2 (2018) . . . . . . . . . . . 7.1.5 Summary of the Findings: Analytical App Data . . . . . 7.2 Findings of Qualitative Data on Medical App Developers . . . . 7.2.1 Triggers for Innovative Endeavors . . . . . . . . . . . . . . . . . 7.2.2 Product Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2.3 Commercialization and Outcomes . . . . . . . . . . . . . . . . . 7.2.4 A Summary of the Findings: Qualitative Data on Medical App Developers . . . . . . . . . . . . . . . . . . . . . .

67 67 67 70

101

Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.1 Summary of the Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Patients and Caregivers are Not (All) Free Innovators . . . . . . .

103 103 104

77 80 94 96 96 99 99

Contents

8.3 8.4 8.5 8.6 9

xi

The Early Development of Medical Smartphone Apps was Triggered by User-Developers . . . . . . . . . . . . . . . . . . . . . . . . Apps Developed by Patients and Caregivers are of Higher Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Company-Developed Apps are the Most Frequently Downloaded Apps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Patients and Caregivers Develop Financially Successful Medical Apps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Preliminary Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.1 Theoretical Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 Managerial Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3 Limitations and Further Research . . . . . . . . . . . . . . . . . . . . . . . . .

Part III

105 107 108 109 111 111 113 114

User Entrepreneurs for Social Innovation – The Case of Patients and Caregivers as Developers of Tangible Medical Devices

10 Introduction: User Entrepreneurs for Social Innovation . . . . . . . . .

119

11 Theoretical Background and Research Questions . . . . . . . . . . . . . . . 11.1 Opportunity Recognition and Exploitation in Entrepreneurship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Opportunity Recognition and Exploitation by Patients and Their Caregivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3 Social Innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

123

12 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

129

13 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.1 Descriptive Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2 Unmet Medical Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3 Opportunity Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3.1 Ideation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3.2 Prototype Development . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3.3 Intellectual Property . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4 Opportunity Exploitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4.1 Product Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4.2 Regulatory Approval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4.3 Production and Distribution . . . . . . . . . . . . . . . . . . . . . . .

133 133 136 137 137 137 139 140 140 141 142

123 124 125 126

xii

Contents

13.5 Market Launch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

144

14 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

147

15 Preliminary Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.1 Theoretical Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.2 Managerial Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.3 Limitations and Further Research . . . . . . . . . . . . . . . . . . . . . . . . .

151 151 153 154

Part IV

Integration of Findings, Implications, and Conclusion

16 Summary of Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

157

17 Implications and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.1 Theoretical Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.2 Managerial Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.3 Implications for Health Policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

161 161 164 165 166

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

169

List of Figures

Figure 1.1 Figure 2.1 Figure 5.1 Figure 5.2 Figure 5.3 Figure 5.4 Figure 5.5 Figure 5.6 Figure Figure Figure Figure

6.1 6.2 6.3 6.4

Figure 6.5 Figure 6.6 Figure 7.1 Figure 7.2 Figure 7.3

The structure of this dissertation . . . . . . . . . . . . . . . . . . . . . . The interrelation between the free innovation paradigm and the user innovation paradigm . . . . . . . . . . . . . The evolution of e-health, m-health and digital health . . . . Global smartphone shipments between 2009 and 2017 . . . . Downloads of medical apps in the Apple AppStore worldwide between 2014 and 2017 . . . . . . . . . . . . . . . . . . . . Revenues generated by smartphone apps per segment in Germany between 2008 and 2018 . . . . . . . . . . . . . . . . . . . The mediated regression model without the control variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The mediated regression model, including the control variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Histogram: Number of downloads . . . . . . . . . . . . . . . . . . . . . Histogram: Revenue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Histogram: Number of ratings . . . . . . . . . . . . . . . . . . . . . . . . The normality of residuals for exogenous variable revenue: Histogram (left) and normal P-P plot (right) . . . . . Scatterplot for exogenous variable revenue: standardized predicted value vs. residuals . . . . . . . . . . . . . . . Conceptual depiction of a simple mediation model . . . . . . . The download estimations Q2 2018 for paid and free apps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Number of downloads vs. revenue . . . . . . . . . . . . . . . . . . . . . The mediation analysis: A visualization of indirect effects 1 and 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6 14 36 37 39 40 47 47 59 60 60 61 62 64 74 77 85

xiii

xiv

Figure 7.4 Figure 7.5 Figure 7.6 Figure 8.1 Figure 14.1 Figure 17.1

List of Figures

Model overview: The mediated regression analysis . . . . . . . Simplified model for mediated regression analysis of truly free apps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Revenue-targeting apps: Apps with in-app-purchases and without in-app purchases . . . . . . . . . . . . . . . . . . . . . . . . . The ratio of user-developed apps vs. non-user developed apps that are newly released every year . . . . . . . Process overview of opportunity recognition and exploitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Interrelationships between user innovation, free innovation, and social innovation theory . . . . . . . . . . . . . . . .

86 89 93 107 147 163

List of Tables

Table 6.1 Table 6.2 Table 7.1 Table 7.2 Table 7.3 Table 7.4 Table 7.5 Table 7.6 Table 7.7 Table 7.8 Table 7.9 Table 7.10 Table 7.11

The descriptive statistics of variables used in empirical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The analysis of multicollinearity . . . . . . . . . . . . . . . . . . . . . . . The number of developers and the number of developed apps: The 2014 data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . App pricing: The 2014 data . . . . . . . . . . . . . . . . . . . . . . . . . . . Market tenure (days): The mean, median, S.D., and p-values: The 2014 data . . . . . . . . . . . . . . . . . . . . . . . . . . Ratings: The mean, median, S.D., and p-values: The 2014 data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The number of developers and the number of developed apps: The 2018 data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . App pricing: The 2018 data . . . . . . . . . . . . . . . . . . . . . . . . . . . Market tenure (days): The mean, median, S.D., and p-values: The 2018 data . . . . . . . . . . . . . . . . . . . . . . . . . . Ratings: The mean, median, S.D., and p-values: The 2018 data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The download estimations for Q2 2018 (paid apps): The mean, S.D., median, and p-values . . . . . . . . . . . . . . . . . . The download estimations for Q2 2018 (free apps): The mean, S.D., median, and p-values . . . . . . . . . . . . . . . . . . The revenue estimations for Q2 2018 (apps with in-app purchases) in U.S. $: The mean, S.D., median, and p-values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

58 63 68 69 69 70 70 71 72 72 73 73

75

xv

xvi

List of Tables

Table 7.12

Table Table Table Table Table Table

7.13 7.14 7.15 7.16 7.17 7.18

Table 7.19 Table 7.20 Table 7.21

Table 7.22 Table 7.23 Table 7.24

Table 7.25 Table 7.26

Table 7.27 Table 7.28 Table 7.29 Table 13.1

The revenue estimations for Q2 2018 (apps without in-app purchases) in U.S. $ : The mean, S.D., median, and p-values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Revenue-targeting apps and truly free apps . . . . . . . . . . . . . . A comparison of the datasets: Developed apps . . . . . . . . . . . A comparison of datasets: App parameters . . . . . . . . . . . . . . . The release year of the apps: The 2014 data . . . . . . . . . . . . . The release year of the apps: The 2018 data . . . . . . . . . . . . . The correlation of variables used in the statistical analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The model coefficients for the mediated regression analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The mediation analysis: Indirect effects . . . . . . . . . . . . . . . . . The model coefficients for the mediated regression analysis, including categorical values for the exogenous variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The mediation analysis: The indirect effects, including categorical values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The model coefficients for the mediated regression analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The model coefficients for mediated regression analysis (truly free apps), including categorical values for the exogenous variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . The mediation analysis: Indirect effects, including categorical values for truly free apps . . . . . . . . . . . . . . . . . . . . The model coefficients for the mediated regression analysis (revenue-targeting apps), including categorical values for the exogenous variable . . . . . . . . . . . . . . . . . . . . . . The mediation analysis: Indirect effects, including categorical values for revenue-targeting apps . . . . . . . . . . . . . A summary of the analytical app data analysis (an overview of the hypotheses) . . . . . . . . . . . . . . . . . . . . . . . A qualitative study on medical smartphone apps—characteristics of interviewees . . . . . . . . . . . . . . . . . . . A qualitative study on user entrepreneurs for social innovation—characteristics of interviewees . . . . . . . . . . . . . .

76 76 78 78 79 80 82 84 85

87 88 90

90 91

92 94 95 97 134

Part I Focus and Scope of this Thesis

1

Introduction

Contents 1.1 1.2 1.3

Research Problem and Relevance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Structure of This Dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4 5 6

In 1991, a woman from Frankfurt (Germany) developed leukemia. A bone marrow transplant from a matching donor would have been her only hope of survival. Her husband immediately began to search for a suitable donor. At the time, only 3,000 people in Germany were registered as potential stem cell donors. As time was running out, the woman’s husband started an initiative to increase the number of donors and to identify a match for his wife. A well-known bone marrow transplant specialist soon joined the team. Within a few months, more than 20,000 new people were registered and, more importantly, a suitable match for her was identified. However, it was too late. The woman died some months after the transplant from the consequences of her leukemia. While things ended badly for her, her husband promised her that he would find a suitable stem cell donor for every single blood cancer patient. Today, Germany’s bone marrow donor file (DKMS, Deutsche Knochenmarkspenderdatei gGmbH) is by far the largest donor file in the world. As at the end of 2017, it is a multinational non-profit company with more than 7.8 million registered potential donors in its database and has provided almost 60,000 stem cell donations.1

Although the founding of DKMS involved a tragic incident, it underlines the massive impact individuals with unmet medical needs may have if they set out to improve their own or their relatives’ health outcomes. Yet, DKMS is not perceived 1 Source:

Rutt and Müller-Eschenbach (2013), https://mediacenter.dkms.de/wp-content/upl oads/2018/10/DKMS-Annual-Report-2017.pdf, accessed January 15, 2019. © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_1

3

4

1

Introduction

as a user-developed innovation in the healthcare sector, but as a national authority that works for the public good. DKMS’s emergence sheds some light to the phenomenon of user-directed innovation. The early work of Eric von Hippel (1976) on users who innovate around their needs has sparked a new research stream on innovation sources. His research heralded the end of the fixation on manufacturer-developed innovation and has opened several interlaced research avenues toward user-developed innovation. This dissertation investigates one of these avenues in breadth and depth to better understand the meaning and relevance of patients and caregivers2 as developers of medical devices around their own unmet medical needs.

1.1

Research Problem and Relevance

Users have been proven to be a key source of innovation. While manufacturers expect to benefit from selling a product or a service, users expect to benefit from using their innovations (von Hippel 2009). A sound body of literature on user innovation has emerged over the past decades, indicating that users are the source of many important innovations in diverse industries such as healthcare, sport, banking, scientific instruments, and the humanitarian sector (Goeldner et al. 2017; Oliveira and von Hippel 2011; Baldwin et al. 2006; Lüthje et al. 2005; Riggs and von Hippel 1994). In the healthcare sector, healthcare professionals have long been recognized as a valuable source of innovation for the development of medical devices. Both companies and scholars have found substantial evidence that involving them can lead to the successful development of new products (Chatterji et al. 2008; Lettl et al. 2006). Healthcare professionals who have already developed medical devices (Lüthje and Herstatt 2004) and procedures (Hinsch et al. 2014) for their own needs or who have discovered new off-label usages for drugs (von Hippel et al. 2016) are particularly valuable. While the contributions to innovation of healthcare professionals and medical device manufacturers as providers of medical devices and services have been described intensively, the roles of patients and caregivers in innovation in the healthcare sector have received little attention in academia and industry. Current research indicates that the role of patients in healthcare is changing from that of a

2I

use caregivers as an umbrella term for relatives or close friends who care for a loved one; it does not encompass professional care by healthcare professionals.

1.2 Research Questions

5

passive consumer of healthcare to a knowledgeable and critical recipient of healthcare products and services (DeMonaco et al. 2019; Oliveira et al. 2015; Pols 2014). This development is leveraged by two major trends: first, the prevalence of digital health services such as the available online health-related information resources, patient-centered online communities, and mobile health services that give patients access to information and control over their data (Amann et al. 2016; Dwivedi et al. 2016); second, the dramatic increase in chronic diseases (such as diabetes and hypertension), which require patients to frequently think about a disease and eventually implement behavioral changes to their daily activities (Goodman et al. 2013). Significant unmet medical needs, better access to medical information, and the more active role of patients and their caregivers may accelerate innovation in the healthcare sector; this offers great potential for patients, scholars, and managers. In this thesis, I investigate innovative patients and caregivers who develop medical devices for their own unmet medical needs. To date, there has been very limited evidence on the prevalence of patients and caregivers as developers of medical devices for their own needs, and the few studies have focused on case studies (DeMonaco et al. 2019; Habicht et al. 2013) and on innovative services and techniques instead of tangible medical devices (Oliveira et al. 2015). Thus, there is a lack of large-scale, real-world data on the development and diffusion of patient-directed innovation in the healthcare sector.

1.2

Research Questions

I have outlined the importance of user innovation and have given an initial indication that patients and caregivers have been an underestimated source of innovation in healthcare. Although user innovation scholars have yielded a significant body of research on innovative end-users in various industries and on intermediary users in healthcare, few scholars have studied this phenomenon in detail. Building on this gap, this dissertation studies the phenomenon of patients and caregivers as developers of medical devices beyond the current status of research, in width and depth. By studying an extreme user innovation case, my goal is to draw conclusions that are relevant to innovation scholars and practitioners beyond the healthcare sector (Eisenhardt 1989). The following research questions will be examined in the course of this dissertation:

6

1

Introduction

• How do patients and caregivers contribute to innovation in the healthcare sector? • What are the characteristics of patients and caregivers as innovators and how do they differ from other innovators in the healthcare sector? • What are reasons for patients and caregivers diffusing their innovations to others and thus become user entrepreneurs? Thus, this thesis exploratively examines how and why patients and caregivers contribute to the development of medical devices.

1.3

The Structure of This Dissertation

As depicted in Figure 1.1, this dissertation has four parts. In part I, I review the conceptual foundations of the phenomenon of patients and caregivers as developers of medical devices in a literature review. I then identify and explain the gaps in theory I identified during the literature review.

Part I

Part II

Part III

Part IV

Focus and Scope: Conceptual Foundaons Research Quesons

Study 1: Paents and caregiver as developer of medical smartphone applicaons

Study 2: Paents and caregivers as developers of tangible medical devices

Integraon of Findings Implicaons Contribuons

Literature

Analycal app data & interview data

Interview data & secondary data

Figure 1.1 The structure of this dissertation

In part II, I present my first study on patients and caregivers as developers of medical smartphone applications. Using a mixed-method approach (Johnson and Onwuegbuzie 2004), I analyze both quantitative data using two datasets on more than 1,100 medical smartphone apps each and qualitative data from 16 interviews with developers of medical smartphone apps. The quantitative data is used to develop an empirical model for a differentiated analysis of user innovation’s impact on the medical smartphone applications market. In part III, I further explore the commercialization activities of patients and caregivers by analyzing

1.3 The Structure of This Dissertation

7

14 case studies of patients and caregivers who successfully brought a tangible medical device to the market. For this study, I conducted interviews and used secondary data such as patents, reports, and online resources on the innovations. Finally, in part IV, I integrate the findings from both studies and the literature review to derive implications that are valid for the healthcare sector and beyond.

2

Conceptual Foundations: The Phenomenon of Patients and Caregivers as User Innovators

Contents 2.1

2.2 2.3 2.4 2.5 2.6

2.1

User Innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1 Heterogeneity of Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.2 Stickiness of Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Free Innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lead User Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . User Entrepreneurship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Market Success of User Innovations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . User Innovation in Healthcare . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9 12 12 14 16 17 19 21

User Innovation

The phenomenon of innovative users is rooted in the seminal work of Eric von Hippel, who began to challenge the traditional manufacturer-centric view of innovation. In an early article, he analyzed the dominant role of users in the scientific instrument innovation process (von Hippel 1976). At the time, this notion was not at all popular, and both scholars and managers strongly questioned his research. Yet, over time, a significant body of empirical research on the user-centered view on innovation has emerged, confirming that users are not solely recipients of manufacturer-developed innovations, but innovate according to their needs (Baldwin and von Hippel 2011).

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_2

9

10

2

Conceptual Foundations: The Phenomenon of Patients …

I follow von Hippel’s (2009, p. 30) definition, differentiating between users and manufacturers: Users […] are firms or individual consumers that expect to benefit from using a product or a service. In contrast, manufacturers expect to benefit from selling a product or a service.

Users in this sense are individuals or single user firms who develop a product or service for their own needs. For instance: Some years back, the German postal service provider (Deutsche Post DHL group) tried to convince some major German car manufacturers to provide it with a functional and inexpensive electric vehicle for its postal services within cities. Yet, all the companies DHL contacted were unwilling or unable to provide it with such a car. Out of necessity, DHL started collaborating with a small university spinoff and subsequently developed a car according to DHL’s needs (Kampker et al. 2016). As at early 2019, DHL was using more than 9,000 of the StreetScooter. Further, owing to the high market demand, it has also begun to sell this e-vehicle to other companies.1 Thus, DHL is a user firm that has developed a tool it needed to better provide its regular services. Coincidentally, it has recently become the fifth-largest manufacturer of e-vehicles in Germany, a status it never intended to reach in the first place.2 Another example: In the 1970 s, a woman from Sweden with poliomyelitis began to develop a wheeled walker—known as rollator—because her musculoskeletal system was becoming increasingly restricted by the disease (Kohlbacher et al. 2015). At the time, only walkers without wheels were available. She aimed to make her innovation public to as many people as possible and therefore did not apply for a patent. Bogers and colleagues (2010) defined two user subtypes for a more granular differentiation: Intermediate users are users such as firms or healthcare professionals who use equipment, components, or techniques from manufacturers to produce goods and services (see example of DHL StreetScooter). On the other hand, there are end-users who are typically individual end-customers or a community of end-users (Bogers et al. 2010), such as the inventor of the wheeled walker. While there are many examples from the sport and leisure industry (Baldwin et al. 2006; Lüthje et al. 2005), the inventor of the wheeled walker is notable

1 Source: https://discover.dhl.com/business/business-ethics/future-of-electric-vehicles, accessed on February 2, 2019. 2 Source: https://de.statista.com/statistik/daten/studie/578385/umfrage/beantragte-umwelt boni-fuer-elektroautos-in-deutschland-nach-herstellern, accessed February 2, 2019.

2.1 User Innovation

11

example from the healthcare sector. Since these two user subtypes are particularly prevalent in healthcare, I will elaborate further on this in section 2.6. There are many studies of user innovation in diverse empirical fields, such as software (Lakhani and von Hippel 2003; Franke and von Hippel 2003), plumbing accessories (Herstatt and von Hippel 1992), scientific instruments (Riggs and von Hippel 1994), sport (Tietze et al. 2015; Baldwin et al. 2006; Franke et al. 2006; Lüthje et al. 2005), and the humanitarian sector (Kruse et al. 2019; Cooper et al. 2017). Besides the abovementioned product-based innovations, there is evidence of service innovation by users in the areas of banking (Oliveira and von Hippel 2011), innovation of procedures (Hinsch et al. 2014), and process innovation (de Jong and von Hippel 2009; von Hippel and Tyre 1995). The healthcare sector has also been the subject of several studies, which I will outline in detail in section 2.6. Several national studies have been conducted to assess the potential of user innovation across nations: Von Hippel and colleagues (2011) found that the percentage of user innovators in the overall population aged 18 and above ranged from 3.7% in Japan to 5.2% in the U.S., and up to 6.1% in the UK. These numbers from representative surveys indicate that millions of people are innovating around their needs and that their innovations are a significant share of the countries’ overall research and development (R&D) spending. This study has been replicated in Canada, China, Finland, Russia, South Korea, Sweden, and the UAE, reaching values from 1.5% (South Korea, China) to 9.6% (Russia) of the general population (Jin et al. 2018; von Hippel 2017). Franke and colleagues (2015) even pointed out that these numbers are too low, suggesting that the de facto value is about four times higher. In sum, a sound body of empirical research supports the proposition that users are a key source of innovation. Owing to a significantly different sequence of activities during the development and, eventually, the diffusion of an innovation, Gambardella, Raasch, and von Hippel (2017) argued that the emergence of user innovation is a general paradigm shift from manufacturer-centered innovation to innovation by individual users, in which users increasingly take responsibility during the process. Previous studies have identified two important factors that influence the costs and benefits of manufacturer innovation and user innovation: the heterogeneity of needs and the stickiness of information (Sánchez-González et al. 2009; Lüthje and Herstatt 2004; Franke and von Hippel 2003; von Hippel and Katz 2002). I will now elaborate on both.

12

2.1.1

2

Conceptual Foundations: The Phenomenon of Patients …

Heterogeneity of Needs

Sánchez-González and colleagues (2009, p. 1594) defined heterogeneity of needs as follows: “Heterogeneity of needs in a group can be defined as the degree to which the needs of i individuals can be satisfied with j standard products which optimally meet the needs of those individuals.” Thus, this is high if many standard products are needed to satisfy the needs of a group of individuals (j ≈ i), and low when few products are needed to fulfill the same group’s needs (j < < i) (Franke and von Hippel 2003). This has important implications for manufacturers and users: Even if manufacturers fragment a given market with a high heterogeneity according to the users’ needs, many users would still be dissatisfied, because manufacturers cannot come up with an individual product for every customer (Lüthje and Herstatt 2004). Although the concept of mass customization is a meaningful way to reduce needs heterogeneity in a market by delivering individually tailored products on a large scale (Salvador et al. 2009; Zipkin 2001), there is evidence that mass customization has limits and cannot be successfully transferred across industries (Zipkin 2001). Thus, if a user needs a non-average product, there are three options, according to Franke and von Hippel (2003): First, the need remains unserved, leaving the user dissatisfied. Second, the user identifies a custom supplier that individually creates the product according to the needs. Third, users may serve themselves by crafting their own product or by modifying an existing product (Franke and von Hippel 2003). Thus, heterogeneity of needs and the associated user dissatisfaction is a major driver of user innovation, particularly in fields in which companies have little incentive to innovate owing to small market sizes and lower expected revenues—for instance in the case of rare diseases. A strategy for manufacturers to overcome the burden of needs heterogeneity is the development of toolkits, i.e. to outsource key need-related innovation tasks to users (von Hippel and Katz 2002). Franke and von Hippel’s (2003) study in the field of web server software revealed that users who were able to modify their own software were much more satisfied than non-innovating users. Thus, although there are ways for manufacturers to address heterogeneity of needs by using mass customization or toolkits, it remains a major source of innovation for users who are dissatisfied with existing solutions (von Hippel 2005).

2.1.2

Stickiness of Information

Generally, an innovation requires two knowledge types: information about the problem (need knowledge) and about how to solve it (solution knowledge)

2.1 User Innovation

13

(Schweisfurth and Herstatt 2016; von Hippel 1998). Different user groups have different knowledge bases—most end-users have need knowledge and lack solution knowledge, while manufacturers have solution knowledge and need to absorb external need knowledge if they aim to better understand their customers’ needs (Block et al. 2016; von Hippel and Katz 2002). In their study on user firms, Adams and colleagues (2013) argued that intermediate users are key players in innovation owing to their unique knowledge base. If need knowledge and solution knowledge are not situated in the same entity, they must be brought together—physically or virtually—in a single locus (von Hippel 1994). Since such knowledge is often costly to acquire, transfer, and use in a new location, it is sticky information (von Hippel 1994). If the information on the need knowledge is sticky, other patterns of innovation will be expected as if the information on the solution is sticky. Thus, depending on the stickiness of the need knowledge or the solution knowledge, innovation is more likely to happen among manufacturers or among users (Ogawa 1998). If the solution knowledge’s transaction cost is low—for instance if only general programming skills are required to develop a smartphone app—it is very likely that users who have these skills and who are aware of an unmet need will develop a solution they need. On the other hand, if solution knowledge has high transaction costs—for instance when developing a bio-engineered drug that is supposed to treat cancer—this information is very sticky to transfer and will most likely not be adopted by cancer patients. The same is true for the need knowledge’s stickiness (von Hippel 1994). Further, a recipient of information must have the skills and prior knowledge to be able to understand the information and incorporate it into the development activities. This relates to the concept of absorptive capacity (Cohen and Levinthal 1990), which incorporates a manufacturer’s ability to recognize the value of new, external information, assimilate it, and apply it. Schweisfurth and Raasch (2018) found that manufacturers who want to incorporate sticky knowledge from users require need absorptive capacity, which is distinct from solution absorptive capacity. Thus, sticky information is hard to transfer. As noted, manufacturers are incentivized by selling their products or services. Users are incentivized by using the self-developed products or services. Yet, there is research on a more diverse characteristics set that innovative users possess.

14

2

2.2

Conceptual Foundations: The Phenomenon of Patients …

Free Innovation

An alternative theoretical lens on user innovation is consumer innovation or free innovation. Von Hippel (2017, p. 1) coined the term: Free innovation are defined as a functionally novel product, service, or process that: Was developed by consumers at private cost during their unpaid discretionary time and

Monetary or process-related benefits

Benefits from use

Is not protected by its developers, and so is potentially acquirable by anyone without payment – for free. No compensated transactions take place in the development or in the diffusion of free innovations.

+ User innovation

Manufacturer innovation Paid development and/or no free revealing

Free user innovation

Free innovation

Unpaid development and free revealing

Figure 2.1 The interrelation between the free innovation paradigm and the user innovation paradigm

2.2 Free Innovation

15

While the scope of user innovation theory is about the needs of an individual or a firm when developing an innovation, free innovators are noncommercial entities, i.e. consumers (von Hippel 2017). There are similarities and differences between the two theoretical lenses: Free innovation takes place at a household level, while user innovation may happen during work and during free time at home. Since the national studies that I mentioned in section 2.1 were conducted on a household level (Jin et al. 2018; von Hippel 2017; de Jong et al. 2015), I assume that most of the innovations reported in these studies were cases of free innovation. Figure 2.1 illustrates the differences between the two concepts: In the lower left area, there are manufacturers who develop the innovation for sale in their paid time. In the upper left area, user innovators (that may be individual users or user firms) develop innovations for own use, but do not reveal them for free (for instance if a user firm sells the innovation afterwards) (Adams et al. 2013) or people who develop innovations for own use during work time (Oliveira and von Hippel 2011). In the upper right area, there are individuals who developed an innovation in their unpaid discretionary time who benefited from the innovation and who revealed it for free. The several studies conducted in the household sector are a good example (Jin et al. 2018; de Jong et al. 2015). Further, there are individuals who developed an innovation in their unpaid discretionary time and reveal the innovation for free, but who did not benefit from the innovation (lower right area). Building on the dichotomous differentiation between utilitarian user motives and pecuniary remuneration of manufacturers, Raasch and von Hippel (2013) found that there were also users who developed innovations for the joy of innovation. This phenomenon was observed first with hackers who developed software for own use as well as for the joy and learning during the process (von Krogh et al. 2012; Lakhani and Wolf 2005). According to Raasch and von Hippel (2013), enjoyment from creating an innovation, altruism, and learning from creating an innovation are important process-related motivations. Building on these initial results, Stock and colleagues (2015) confirmed that some users have utilitarian user motives, while others have hedonic user motives or a combination of both. Their main findings were that innovations developed by innovators with a utilitarian purpose have greater utility, while users who develop for the fun of developing an innovation build solutions of greater novelty. Thus, considering these studies on free innovation and on participators, one would expect to detect all three user groups—manufacturers, users, and participators—in a marketplace. Although the theoretical lens of free innovation offers a set of exciting new research avenues, my research focuses on the unmet medical needs of patients and caregivers in relation to their diseases. Thus, I mainly follow

16

2

Conceptual Foundations: The Phenomenon of Patients …

the user innovation research stream, since it has a stronger emphasis on the own needs of individuals for developing products and services.

2.3

Lead User Theory

In 1986, von Hippel established the concept of leading-edge users—so-called lead users. In his definition (von Hippel 1986, p. 796), lead users of a novel product, process, or service have two characteristics: Lead users face needs that will be general in a marketplace – but face them months or years before the bulk of that marketplace encounters them, and – Lead users are positioned to benefit significantly by obtaining a solution to those needs.

These two characteristics were later defined as trend leadership and high expected benefit from the innovative solution (Hienerth and Lettl 2017). The rationale behind trend leadership is that customer needs tend to evolve over time along underlying trends in the marketplace (Schreier and Prügl 2008). Persons at the leading edge of such a trend will encounter new needs much earlier than regular users. The high expected benefit from an innovative solution builds on lead users’ motivations: by innovating, they solve their own problems, particularly if manufacturers are not (yet) able or willing to do so (Lüthje and Herstatt 2004). When the lead user concept was established, the main aim was to differentiate between innovative and non-innovative users. Further, scholars have begun to measure individuals’ extent of lead userness (Schreier and Prügl 2008). Thus, measurement of the construct is not dichotomous, i.e. the extent of lead userness is pivotal for measurement. Further, the two characteristics of lead users are independent dimensions that must be assessed separately (Franke et al. 2006). Hienerth and Lettl’s (2017) review summarized additional features of lead users: lead users are domain-specific and trend-specific, i.e. a lead user in one domain is not necessarily a lead user in a very different domain. Further, the lead user construct is not a trait, but may emerge or vanish within a user over time. In short, integrating lead users into a manufacturer’s R&D activities is a meaningful way to integrate sticky knowledge on needs and on solutions located outside a firm’s boundaries (von Hippel 1986). The lead user method is a four-step process to identify such user innovators (Lüthje and Herstatt 2004). Its main purpose is to surface product or service concepts for companies, which later further develop these concepts. According to

2.4 User Entrepreneurship

17

Lüthje and Herstatt (2004), its steps are: 1) the start of the lead user process, 2) the identification of needs and trends, 3) the identification of lead users, and 4) the concept design. The lead user method has drawn significant attention from scholars and managers (Hienerth and Lettl 2017; Schreier and Prügl 2008; Lüthje and Herstatt 2004; von Hippel 1986). While scholars have opened several research avenues to study lead users and their innovations from various perspectives (Hienerth and Lettl 2017; Bogers et al. 2010), practitioners have emphasized the significant contributions of lead users to highly innovative and commercially attractive products (Franke et al. 2006; Herstatt and von Hippel 1992). From a company’s perspective, there is evidence that the timeframe for co-development among lead users is twice as fast as and occurs at half the cost of other methods to identify new product concepts (Herstatt and von Hippel 1992). While lead users and manufacturers are mainly seen as organizationally distinct entities, there is evidence that employees in a company also have lead user characteristics in relation to the company’s products or services (Schweisfurth and Herstatt 2016). These so-called embedded lead users are more customer-oriented, more active in internal boundary spanning, and undertake more innovation effort than regular employees (Schweisfurth and Raasch 2014).

2.4

User Entrepreneurship

The predominant motivation of users to develop an innovation is own use of this product or service. Thus, users capture limited economic benefit beyond own use (Shah and Tripsas 2016). Still, there may be reasons—such as positive feedback from peers—to further develop and bring an innovation to the market. Most user innovators require professional help to commercialize and distribute their innovations at some point, particularly if the innovation is not diffused freely peer-to-peer, but commercially (Gambardella et al. 2017). In this case, the innovation is often licensed to an established producer already present in the market (de Jong et al. 2015). This pattern has been observed particularly with healthcare professionals who often collaborate with established medical device manufacturers (Chatterji et al. 2008). In one of his early studies, von Hippel (1976) found that 80% of scientific instruments in his sample were invented, prototyped, and fieldtested by users and subsequently commercialized by established manufacturers. Yet, there is empirical evidence that some user innovators become entrepreneurs and bring their product or service to the market themselves. If this is successful, they eventually get an economic benefit from their innovation. However, this was

18

2

Conceptual Foundations: The Phenomenon of Patients …

not the initial motivation for the innovation, but rather the outcome of a journey triggered by own needs. Shah and Tripsas (2007) extended user innovation theory by coining the term user entrepreneur for user innovators who become entrepreneurs with a self-developed product or service. They found that the development, adaptation, and initial diffusion of an idea very often occurs before a formal evaluation as a business opportunity is performed. Having a prototype of the innovation, public interaction and community interaction supplies an innovator with feedback that ultimately triggers the entrepreneurial process (Shah and Tripsas 2007). They postulated that enjoyment during the development and commercialization process as well as low opportunity costs are primary drivers of user entrepreneurship. Haefliger, Jäger, and von Krogh (2010) revealed that, in some cases, user entrepreneurs developed products based on assets they acquired in one industry and then used in another industry. Shah and Tripsas (2016) examined how users and manufacturers estimated the potential financial returns of their innovations in relation to their unique profit thresholds. Apparently, the estimates of users and manufacturers differ owing to information asymmetries, complementary assets, and interpretations of available information. Companies are estimated to have higher profit thresholds than individual users and usually have a portfolio of different projects that are constantly evaluated and prioritized. Shah and Tripsas found that factors such as open product design, modular product architecture, an industry in an early lifecycle, and little government regulation positively impact on user entrepreneurship (Shah and Tripsas 2016). Crowdfunding platforms have been proven to foster user entrepreneurship, since they help to overcome financial burdens during the development phase (Brem et al. 2017). The career decision to become an entrepreneur is generally influenced by the pursuit to maximize own utility, as an individual combination of financial rewards estimated by being employed or being entrepreneur, as well as other incentives such as working conditions (i.e. decision-making control, risk exposure, work effort required) and other factors (Douglas and Shepherd 2000). Thus, as it was proven that user entrepreneurs don’t necessarily maximize their pecuniary remuneration (Shah and Tripsas 2016), since they receive sufficient nonpecuniary utility from being self-employed and working on a topic they are passionate about. In short, user entrepreneurs substitute love for money (Shah and Tripsas 2016; Podolny and Scott Morton 2002). There is evidence of user entrepreneurship across industries, for instance in the development of animation movies (Haefliger et al. 2010), stereo systems (Langlois and Robertson 1992), medical devices (Lettl et al. 2006), juvenile products (Shah and Tripsas 2007), and sporting goods (Shah and Tripsas 2016; Baldwin et al. 2006; Franke and Shah 2003). A study by Shah et al. (2012) using longitudinal

2.5 The Market Success of User Innovations

19

data of 4,928 firms founded in 2004 in the U.S. revealed that 10.7% of all startups and 46.6% of innovative startups that survived the first five years were established by users. In sum, user entrepreneurship is a significant driver of innovation across industries that is driven by users who seek to maximize their utility by combining nonpecuniary utilities and pecuniary remuneration. Yet, user entrepreneurship is just one particular case of diffusion of user innovation. Returning to Figure 2.1, user entrepreneurs are initially located with user innovators and then move to the lower left side of the matrix, since they also have a monetary benefit from revealing their innovation. Often companies incorporate need knowledge from innovative users, further develop them, and market them as their product on the market. I will explore how user innovations are diffused through several channels and how user innovation’s success can be assessed.

2.5

The Market Success of User Innovations

The literature on user innovation has always emphasized the fairly selfish motivation of user innovators—users innovate because they obtain benefits for themselves from an innovation (von Hippel 2009). Yet, there are also attempts to understand how and why user innovators diffuse their innovations: Harhoff and colleagues (2003) showed that users often freely reveal their innovations, since they benefit from using an innovation rather than from selling it. The free revelation of user innovations has been perceived in several industries, such as sporting equipment (Franke and Shah 2003), library information systems (Morrison et al. 2000), semiconductor production (Harhoff et al. 2003), the household sector (de Jong et al. 2018; von Hippel 2017), and open-source software (Füller et al. 2013; Balka et al. 2010; von Krogh and von Hippel 2006). Not only user innovators, but also manufacturers may have reasons to free reveal some—carefully selected—innovations (Henkel 2006). Free innovation (von Hippel 2017) is revealed freely, yet the diffusion of freely revealed innovation is often hard, since there are few rewards for an innovator to do so (de Jong et al. 2018). Online communities and platforms are a way for users to disseminate and further develop innovations with peers (Hienerth and Lettl 2011; Franke et al. 2008)—even if a community is hosted by a manufacturer (Jeppesen and Frederiksen 2006). No matter how an innovation is diffused, to date, very little is known about user-innovated products’ ability to compete with producer-innovated products in

20

2

Conceptual Foundations: The Phenomenon of Patients …

real-world settings. One reason for this predicament is the lack of reliable quantitative data in real-world settings. Thus, few scholars approach this question via the various approaches I will now outline: Riggs and von Hippel (1994) found, in their retrospective analysis of 64 innovations regarding scientific instruments, that users tend to develop new functional capabilities, while manufacturers tend to improve the convenience or reliability of existing devices. Both user types equally contributed to improvements of the scientific instruments’ sensitivity, resolution, or accuracy. Further, the users’ innovations had higher scientific importance and were developed earlier, while manufacturer-developed innovations had higher commercial importance and were developed later. A study in the area of extreme sport, in which expert ratings were obtained to assess innovations’ commercial attractiveness, revealed that innovators with high lead userness developed commercially more attractive innovations than innovators with lower lead userness (Franke et al. 2006). A study by Lilien and colleagues (2002) revealed that products that were surfaced via the lead user method (n = 5 vs. n = 42 manufacturer-developed products) and that were subsequently incorporated into a manufacturer’s R&D activities were superior concerning novelty and strategic importance, and had a significantly higher estimated sales volume within year five after market launch. In an experimental setting, user-generated ideas (n = 52) were rated as superior concerning novelty and customer benefits and lower concerning feasibility than ideas developed within a company (n = 51) (Poetz and Schreier 2012). Kristensson et al. (2004) found similar results: ideas developed by ordinary users in an experiment were superior concerning originality and value compared to ideas developed by professional developers. A recent study by Hamdi-Kidar et al. (2019) found that lead users developed superior product concepts than other user groups, no matter if they are developed in a collective setting or an individual setting. Further, users have been proven to be capable of screening ideas in conformance to professional experts (Magnusson et al. 2016). Yet, these experiments are borderline cases between user innovation and crowdsourcing3 and do not provide reliable information on how user innovations perform in real-world settings. One of the few studies to show the market performance of user-generated designs was by Nishikawa and colleagues (2013): User-generated designs (n = 6) were significantly more successful than professional-generated designs (n = 37) along financial parameters such as sales volume and profit margin. Yet, these 3 Crowdsourcing

is an online-based activity in which a crowd is invited to participate in undertaking a task posted by an individual or an organization. See Estellés-Arolas and González-Ladrón-De-Guevara (2012) for a comprehensive definition of crowdsourcing.

2.6 User Innovation in Healthcare

21

designs were further developed, manufactured, and sold by an established company. Building on this study, Nishikawa et al. (2017) designed a field experiment in which they labeled a consumer product in some shops as customer-ideated, while the same product was sold in other shops without mentioning this source of design. They found that this label increased the product’s de facto market performance by up to 20%. In contrast, Fuchs and colleagues (2013) found that labeling a luxury fashion brand collection as user-designed reduces the demand, most likely because such a luxury product is perceived to be lower in quality and fails to signal high status. All the above mentioned studies have limitations: they had low sample sizes, were based on experiments rather than real-world market data, the sample products were commercialized by a company or evaluations were based on expert assessments rather than on large-scale customer feedback. To date, there is a lack of large-scale, real-world data on innovations that have been commercialized by user innovators.

2.6

User Innovation in Healthcare

Research has provided robust arguments on user innovation’s prevalence in the healthcare sector. Scholars focused mostly on innovative endeavors by healthcare professionals. Shaw’s (1985) early work provided initial evidence on the role of interaction between medical device manufacturers and healthcare professionals. He stressed that multiple and continuous interactions with the user throughout the innovation process positively impacted on the innovation outcome. Benefits of involving users in medical device development such as increased access to user needs, improvements in designs, and an increase in device functionality, usability, and quality were identified by Shah and Robinson (2007). They also listed barriers such as resource constraints that hinder user involvement. In general, several studies stressed the positive impact of involving healthcare professionals in medical device R&D (Kesselheim et al. 2014), particularly for the development of new products or services with a high degree of innovativeness (so-called radical innovation) (Chatterji and Fabrizio 2012; Lettl et al. 2008; Lettl et al. 2006). As healthcare professionals may develop solutions according to their unmet needs, some possess lead user characteristics in relation to their discipline (Lüthje and Herstatt 2004). A study evaluation of U.S. patent data by Chatterji and colleagues (2008) revealed that physicians contribute to medical device innovation and account for almost 20% of approximately 26,000 medical device patents filed

22

2

Conceptual Foundations: The Phenomenon of Patients …

over a period of six years. Further, they found that patents developed by physicians had more influence on subsequent inventive activity than non-physician patents. A study that analyzed investment by incumbents into physician-founded startups found that that established medical device manufacturers are much more likely to cite healthcare professional-founded companies’ patents and to incorporate them into new medical devices (Smith and Sfekas 2013). An analysis of another patent dataset revealed that highly specialized healthcare professionals develop innovations with higher technological impact than those developed by corporate innovators (Lettl et al. 2009). A study that analyzed a panel dataset of medical device companies found evidence that inventive collaborations with healthcare professionals enhanced corporate product innovation, particularly for generation of radical innovations (Chatterji and Fabrizio 2014). Despite these encouraging results, it has been proven that some manufacturers try to hamper user innovation in the medical device sector, particularly by implementing measures that prevent users from modifying and improving their products (Braun and Herstatt 2008). This pattern may be partly but not entirely explained by restrictive medical device regulations that hinder the modification of medical devices. Although drug development is complex, there is also evidence that healthcare professionals develop drugs jointly with manufacturers (Xu and Kesselheim 2014). Innovative behaviors on the part of healthcare professionals regarding offlabel use of drugs, i.e. the legal prescription of a medication in a different way than approved by the regulatory agencies (Wittich et al. 2012), was studied by DeMonaco and colleagues (2006). They found that 57% of new applications to regulatory agencies for existing drugs was identified by clinicians, who prescribed these drugs off-label. A follow-up study revealed that healthcare professionals successfully used off-label prescription of drugs to treat their own patients; however, they bemoan that clinicians have few incentives to diffuse the innovation to other healthcare professionals. The authors suggest addressing this market failure via public policy measures to improve patient care (von Hippel et al. 2016). Makerspaces are open communities for tinkering and innovating that have been proven to enhance user-generated innovation (Halbinger 2018). In a study of six newly created makerspaces in hospitals in Sweden, Svensson and Hartmann (2018) found that, already in the first year of operation, the potential returns from the innovations developed in the makerspaces by hospital employees exceeded the initial investment more than 10 times. Most of the innovations would not have been developed without the makerspace in the hospital. This sheds light on the relevance of incorporating not only physicians, but also other healthcare professionals such as nurses or psychologists who actively innovated according to their own and their patients’ needs (Svensson and Hartmann 2018). Thus, the

2.6 User Innovation in Healthcare

23

hospital (i.e. all those who work in it) are potential innovators (Thune and Mina 2016). However, innovation is not limited to product and service innovation: Hinsch and colleagues (2014) revealed that healthcare professionals developed new techniques that trigger the subsequent development of new medical devices by manufacturers and users, although the diffusion of techniques differs significantly from the diffusion of medical devices and therefore requires considerable interpersonal interaction. According to user entrepreneurship theory (Shah and Tripsas 2007), enjoyment during the development and commercialization process as well as low opportunity costs have been identified as the main drivers of user entrepreneurship. Thus, Shah and Tripsas noted that they would not expect innovative healthcare professionals to develop medical devices on their own, but would anticipate established firms or startups to take over the concept. As noted, several studies have proven this pattern (Chatterji and Fabrizio 2014; Chatterji et al. 2008; Lettl et al. 2006). There is only limited evidence that healthcare professionals become successful entrepreneurs (Smith and Sfekas 2013; Lettl et al. 2009). These ambiguous findings imply a difficult prediction on the de facto commercialization activities of medical devices developed by healthcare professionals. As shown in the previous paragraphs, most research into user innovation in the healthcare sector has focused on healthcare professionals as users. According to Bogers and colleagues (2010), healthcare professionals can be seen as intermediary users in the healthcare sector, since they mostly innovate along their patients’ needs and not to satisfy their own health-related needs. Yet, innovative endeavors of patients and their caregivers—the end-users of health-related products or services—have only occasionally been examined (DeMonaco et al. 2019; Oliveira et al. 2015; Habicht et al. 2013). In their review of hospitals as innovators in the healthcare system, Thune and Mina (2016, p. 1549) summarized this gap in the literature, stating that “the role of patients in medical innovation is often highlighted as important, but rarely explicitly investigated.” Thus, there is initial evidence that patients and their caregivers are a source of innovation in the healthcare sector: long since, the relationship between patient and their healthcare professional had a significant information asymmetry: healthcare professionals dominate medical encounters and patients are the passive recipients of medical information (Camerini et al. 2012; Budych et al. 2012; Hartzband and Groopman 2010). Three trends in healthcare are changing healthcare professionals’ dominance: The increase in chronic diseases, the desire to treat rare diseases, and the emergence of digital health services. I will outline the first two trends in this section, and the latter in section 5.1.

24

2

Conceptual Foundations: The Phenomenon of Patients …

There is a shift underway from acute to chronic noncommunicable diseases that require long-term commitment from healthcare professionals to their patients (Holman and Lorig 2004; Mascie-Taylor and Karim 2003). The increase of chronic noncommunicable diseases is reaching epidemic magnitudes worldwide (Daar et al. 2007). Yet, patients are acting increasingly proactively and are becoming experts in their health issues as chronic conditions are prevailing more and more (Hartzband and Groopman 2010). Further, patients actively seek to increase their quality of life4 and wellbeing, rather than to increase their life expectancy (Steptoe et al. 2015). As chronic diseases are associated with a decrease in health-related quality of life, particularly if experienced with comorbid diseases (Rothrock et al. 2010), patients have a strong incentive to increase their medical knowledge about their chronic disease. Rare diseases are defined as diseases that affect less than one in 2,000 people5 (Remuzzi and Garattini 2008). Yet, many disorders fit this definition: according to the WHO, there are more than 5,000 rare diseases (Schieppati et al. 2008), leading to a huge heterogeneity of needs (Sánchez-González et al. 2009) in the healthcare system worldwide that is hard to meet by pharmaceutical companies and medical device manufacturers. Further, the lack of disease-related knowledge among healthcare professionals, the emotional challenges that accompany a diagnosis of a rare (and often chronic) disease, and the frequently greater travel distances to the next expert are significant burdens to patients with a rare disease (Budych et al. 2012). Thus, based on the abovementioned information, one would expect to identify user innovation by patients and their caregivers, particularly in the case of chronic and rare diseases. There has been initial but limited scholarly attention on the phenomenon of patients and caregivers as innovators in the healthcare sector. Habicht et al. (2013) argued that patients have a strong incentive to innovate, since they generally expect to benefit from using self-developed solutions. DeMonaco and colleagues (2019) pointed to two examples of free innovation developed by patients. Oliveira and colleagues (2015) found, in their study on 500 patients with a rare condition, that 36% of the sample innovated in relation to their needs and that 8% developed

4 Quality

of life refers to an overall assessment of human experience that is applied across multiple disciplines, such as psychology, medicine, economics, environmental science, and sociology (Costanza et al. 2006). Related concepts include subjective wellbeing, wellness, happiness, and life satisfaction (Edwards-Schachter and Wallace 2017). 5 This definition is valid for the EU. In the U.S., a rare disease is defined as one that affects fewer than one in 1,250 people (Remuzzi and Garattini 2008).

2.6 User Innovation in Healthcare

25

solutions that were not yet know to medicine. Of the innovations, 10% was products and the remaining 90% was service innovations. Yet, the authors concluded that the diffusion of the innovations was partial, mainly directly to other patients known to the innovator (Oliveira et al. 2015). In their study on a web-based innovation platform for patients and caregivers, Bullinger and colleagues (2012) found that, of 53 innovations developed on the platform, 32% was software and apps, 30% tangible devices, 21% service innovations, and 17% media-related innovations. An analysis of 22 innovative contribution of patients with a chronic condition (patients with diabetes types 1 and 2) revealed that their innovations in the health information technologies field were highly original (Kanstrup et al. 2015). In sum, there is initial evidence that patients and caregivers develop medical devices for their own needs, yet there are many blind spots in this field. I will address some in the course of this thesis.

3

Research Gaps

This dissertation contains two complementary studies in order to answer the overall research questions, as mentioned in section 1.2: • How do patients and caregivers contribute to innovation in the healthcare sector? • What are the characteristics of patients and caregivers as innovators and how do they differ from other innovators in the healthcare sector? • What are reasons for patients and caregivers to diffuse their innovations to others and thus become user entrepreneurs? Thus, I exploratively examine how and why patients and caregivers contribute to the development of medical devices. Specifically, the literature review in chapter 2 has revealed a set of research gaps I address with these studies. I have outlined the prevalence of user innovation across industries, particularly in healthcare. There is broad empirical evidence that users, particularly those who provide health-related products and services (i.e. healthcare professionals), are a primary source of innovation. However, to date, very little is known about innovative patients, i.e. people who receive health-related treatments and products. This very large group of users has largely been excluded from innovation by scholars and companies. Yet, it seems promising and valuable to integrate patients and their relatives into companies’ product development: patients are highly intrinsically motivated to innovate, and often have complementary knowledge to that of healthcare professionals about certain aspects of their disease (Elberse et al. 2011; Wilson 1999).

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_3

27

28

3

Research Gaps

A second gap in the literature is the lack of large-scale, real-world market data in the realm of user innovation. As outlined in section 2.6, it is hard to assess the success of user-developed products, since most user-generated ideas remain in the prototype stage and are later (if at all) commercialized jointly with a manufacturer. To date, little is known about the de facto market behaviors of user-developed and commercialized products vs. manufacturer-developed and commercialized products (Ogawa and Piller 2006). A reason for this dilemma is the lack of reliable quantitative data in real-world settings. Some experimental studies have confirmed that user-generated ideas can compete or even outpace manufacturer-developed products in the marketplace (Nishikawa et al. 2013; Poetz and Schreier 2012), yet there is a lack of large-scale real-world data. In study 1, I address these two gaps by analyzing how and why patients and caregivers develop medical devices for their own needs. Using a mixed-method approach (Johnson and Onwuegbuzie 2004), I will analyze a large-scale, realworld dataset of analytical app data on medical smartphone apps developed by user-developers (such as patients and caregivers as well as healthcare professionals) and non-user-developers. I will then analyze qualitative data from 16 interviews with user-developers of medical smartphone apps. Building on this, study 2 addresses two additional gaps in the literature: First, I extend the user entrepreneurship literature by elaborating on the prevalence of innovative patients and caregivers who develop tangible medical devices—not software, as in study 1—according to their own or their caregivers’ needs. As previous studies emphasized that they would not expect users to develop medical devices owing to the high opportunity costs of tangible medical device development (Shah and Tripsas 2007), my study elaborates on the reasons why patients and caregivers develop such devices in spite of the associated barriers. Further, I provide data on how these users recognize and subsequently exploit the entrepreneurial opportunity for a social innovation associated with their health burden. This is particularly relevant, since the development outcome for patients is an increase in their own quality of life—a parameter that is becoming increasingly important in healthcare-related studies (Wikman et al. 2011). To address the two latter research gaps, I conducted 14 case studies on user entrepreneurs who developed and commercialized a tangible medical device.

Part II Are Patients and Caregivers the Better Innovators? The Case of Medical Smartphone Applications Partial results of the chapter, particularly selected parts of dataset 1 and the qualitative data, were published earlier by Goeldner and Herstatt (2016).

4

Introduction: User Innovation and Medical Smartphone Applications

Hello! […] I’m an iPhone application developer and a type 1 diabetic. I’m asking for £7.5k to help fund the development of an innovative new iPhone application that could help thousands of diabetics worldwide. With your help, I’d like to release it completely free on the App Store and possibly bring it to other devices in the future. In 2011 diabetes was estimated to have affected 366 million people worldwide. In the US, over 8.3% of the population are affected, and here in the UK it’s estimated that more than 1 in 20 people has diabetes. […] But here’s the good news: good diabetic control has been shown to severely reduce the risk of complications later in life. The more information we have to work with, the healthier we can be. […] Most of the existing diabetic applications on the App Store make it tedious and difficult to keep your journal up to date. After all, my disease takes up so much of my life already that I didn’t want to spend any more time than necessary recording data. So I built my own diabetic journal application, and I made it fast. More importantly, I made it intelligent. […] By using an innovative new technology called Smart Input, the application I’ve developed looks at your medication usage and uses that information to determine which medicine you’re mostly likely to be taking at any given point in time. By learning your habits, it’s able to pre-populate fields that other applications make you type out by hand or select from menus—cutting the time needed to enter information from minutes to seconds. It lets you monitor every aspect of your health, from tracking carbs and watching your diet to physical activities. It can even remind you to take your medication when you next leave the house. I’ve slaved over every aspect of the applications interface in order to make it as efficient as possible. Electronic supplementary material The online version of this chapter (https://doi.org/10.1007/978-3-658-32041-6_4) contains supplementary material, which is available to authorized users. © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_4

31

32

4

Introduction: User Innovation and Medical Smartphone Applications

The result is an application that I believe is more focused than any other diabetic management software out there. With your help, I want to prove that good design can really help to improve people’s lives. […]1

The trigger to develop a medical smartphone app often relates to a personal need developers have during their daily routines. I gained this insight in early 2013 when I read the abovementioned quote on a crowdfunding platform. This developer’s need to improve his diabetes management was so high that he decided to invest a significant amount of time to develop a medical smartphone app that suits his needs. He also wanted to share the app with peers for free and thus asked for money on a crowdfunding platform in order to make the design appealing to future app users. At the time this crowdfunding project was initiated, there was very limited evidence that patients could develop tools to improve their health outcomes. Yet, there was substantial evidence that users innovate along their needs if no suitable products are available (see part I of this dissertation), also in the healthcare sector. However, most studies focused on users who provided health-related products or services, such as healthcare professionals. The evidence on the innovative endeavors of patients (i.e. people who receive health-related treatments as well as products) was very limited (DeMonaco et al. 2019; Oliveira et al. 2015). Although first studies indicated that it is promising to integrate patients into the development of medical devices (Bullinger et al. 2012), scholars and companies have not managed to include patients and caregivers into their research activities. The second gap in the literature I address is the lack of large-scale, real-world market data on user innovations. It is complex to assess user-developed products’ success, since most user-generated ideas remain in the prototype stage and are later (if at all) commercialized together with a manufacturer. There is a lack of studies into the real-world market behaviors of user-developed and commercialized products vs. manufacturer-developed and commercialized products (Ogawa and Piller 2006). A reason for this dilemma is the lack of reliable quantitative data in real-world settings (Nishikawa et al. 2013).

1 Source: https://www.kickstarter.com/projects/nialg/the-diabetic-journal, accessed September 13, 2018.

4

Introduction: User Innovation and Medical Smartphone Applications

33

Thus, based on these two gaps in the literature, I developed the two research questions for study 1: • How do patients and caregivers contribute to innovation in the medical device sector? • How do user-developed medical devices compete with non-user-developed medical devices in a large-scale, real-world setting? These two questions serve as the exploratory framework for study 1, in which I draw on these two gaps in the literature, analyzing the innovative behaviors of patients and caregivers, their motivations, and their contributions to improving the quality of their own and ultimately of other patients’ therapy. I compare innovations of patients and caregivers with innovations of healthcare professionals and companies with two complementary datasets: First, I use publicly available data to detect and evaluate a large user innovations set. Second, I use interview data obtained from 16 interviews with innovative patients, caregivers, and healthcare professionals. As an empirical field, I selected the medical smartphone apps market, which allowed me to analyze innovations that are already diffused to the App Store and that are available to millions of potential customers. Apple’s App Store has been investigated by innovation scholars in various ways to draw conclusions regarding app development and beyond (Miric et al. 2019; Yin et al. 2014; West and Mace 2010). Customers download and rate medical apps with very limited knowledge about the origins (user innovator vs. non-user-developer) and therefore have unbiased opinions, which I will evaluate. Barriers that limit diffusion such as regulatory approval, market admission, high technical knowledge required for market access, manufacturing, and distribution are less prominent for medical apps than in the case of tangible medical devices. Regarding the analytical app data, I analyzed a dataset obtained in mid-2014 and another obtained in mid-2018. Both datasets comprise data from more than 1,100 medical smartphone apps from Germany, the UK, and the U.S. In this study, I make at least two major contributions: First, I found significant evidence for the phenomenon of patients and caregivers as developers of medical smartphone apps. The analysis of 1,265 apps developed by 912 developers yielded a relative number of user-developed apps of about 40% of the overall sample, while healthcare professionals accounted for about 29%, and patients and caregivers for about 11%. This large portion is of both user types is somewhat unexpected, since the medical device industry is dominated by incumbents that should be able to enter this new market.

34

4

Introduction: User Innovation and Medical Smartphone Applications

Second, I responded to the call for large-scale, real-world empirical data on the market performance assessment of user-developed products. In this study, I analyze a subset of 837 apps on the adoption of user-developed products in a marketplace without co-development by intermediaries. I find that user-developed apps are rated significantly higher than non-user-developed apps and that the revenue is highest for patient-developed apps. This underlines the important contribution of user-developed medical apps to the healthcare system. Study 1 is structured as follows: In chapter 5, I discuss the study’s theoretical background and derive the hypotheses for the empirical model. In chapter 6, I outline methodology and the dataset. In chapter 7, I present the empirical findings; in chapter 8, I discuss these findings; in chapter 9, I draw preliminary conclusions from the findings, study limitations, and avenues for further research.

5

Theoretical Background and Hypothesis Development

Contents 5.1 5.2 5.3

5.1

Empirical Field: e-Health, m-Health, Digital Health . . . . . . . . . . . . . . . . . . . . . . . . . Medical Smartphone Apps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hypothesis Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.1 Free Revealing of User-Developed Medical Apps . . . . . . . . . . . . . . . . . . . . 5.3.2 The Early Development of Medical Apps . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.3 Developer Type and the Quality of Medical Apps . . . . . . . . . . . . . . . . . . . . 5.3.4 Developer Type and the Download Numbers of Medical Apps . . . . . . . . 5.3.5 Developer Type and Medical App Revenues . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.6 Summary of Hypotheses: Research Model for Mediated Regression Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

35 38 41 42 43 43 45 46 46

Empirical Field: e-Health, m-Health, Digital Health

In a traditional patient-physician relationship, the patient is very dependent on the information obtained from a healthcare professional. Assuming that physicians dominate medical encounters, Bodkin and Miaoulis (2007) described physicians as gatekeepers of healthcare information. Several scholars identified the emergence of the Internet as a major factor in challenging these traditional roles (Hartzband and Groopman 2010; Mukherjee and McGinnis 2007). Electronic healthcare (e-health) services can be defined as “health services and information delivered or enhanced through the internet” (Eysenbach 2001). Subsequent studies found evidence that the Internet has become a significant source of health information (Hartzband and Groopman 2010; Camacho et al. 2010; Khechine et al. 2008; Bodkin and Miaoulis 2007; Mukherjee and McGinnis 2007). © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_5

35

36

5 Theoretical Background and Hypothesis Development

Since 1990

Since 2010

Since 2015 digital health m-health

e-health

Digitalizaon & the Internet

Mobile devices

Big data

Figure 5.1 The evolution of e-health, m-health and digital health (adapted from Meister et al. 2017)

Today, patients actively use the Internet to gain access to medical information and thus become informed patients who no longer only depend on their local healthcare providers (Meister et al. 2017; Budych et al. 2012; Camacho et al. 2010). For healthcare professionals, the Internet also serves as a significant source of information (Hartzband and Groopman 2010). Yet, the Internet—with the manifold information available online—poses risks to patients, their caregivers, and healthcare professionals: information overload and the likelihood of falsehoods that are easily and rapidly propagated on the Internet (Hartzband and Groopman 2010). Jacobs et al. (2017) found that only people who are quite young, welleducated, Internet-savvy, and who have a high socioeconomic status really benefit from health-related information available online. This must be considered when assessing the potential of digital health-related services for the whole population. The mobile health services (m-health) trend relates closely to e-health and telemedicine (see Figure 5.1). As stated by Istepanian et al. (2004, p. 405), m-health “represents the evolution of e-health systems from traditional desktop ‘telemedicine’ platforms to wireless and mobile configurations.” The continuously growing

5.1 Empirical Field: e-Health, m-Health, Digital Health

37

number of smartphones (Figure 5.2)1 has leveraged a wide range of new, appbased medical devices and services (Agu et al. 2013). This ecosystem allows nonprofessional developers to actively design solutions that correspond to their medical needs and to invite others to use these apps.

1,500

1,438

1,469

1,465

2015

2016

2017

1,302 1,019 1,000 725

495

500 305 174 0

2009

2010

2011

2012

2013

2014

Figure 5.2 Global smartphone shipments between 2009 and 2017 (in million units)

As indicated by Terry (2010), utilization fields for medical smartphone apps are very diverse, ranging from patient communication, point-of-care documentation, and disease management and diagnosis, to public health services and ambulance services. In 2015, about 48% of all available medical apps targeted chronic conditions (Research2Guidance 2015). Digital health has become increasingly popular after 2015. Additional to ehealth and m-health, it comprises the meaningful use of so-called smart devices such as wearables as well as the utilization of large and often real-time datasets (big data) (Meister et al. 2017). Ultimately, the convergence of these different data sources should lead to more personalized treatment of patients. Digital health is a multidisciplinary domain that involves healthcare professionals, computer scientists, engineers, public health workers, information scientists, and others working together closely (Kostkova 2015).

1 Source:

Statista: https://www.statista.com/statistics/271491/worldwide-shipments-of-sma rtphones-since-2009, accessed January 29, 2019.

38

5.2

5 Theoretical Background and Hypothesis Development

Medical Smartphone Apps

The introduction of the Apple iPhone in 2007 marked the start of a new era of mobile phones: the smartphone. Although there were a few phones with touchscreens and mobile Internet available at the time, the iPhone was the first phone to combine a large touchscreen, a web browser, a user interface made for touchscreens, and no physical keyboard in an intuitive and user-friendly way (West and Mace 2010). In July 2008, Apple launched the second edition of the iPhone and opened the Apple App Store, which offers complementary third-party software for free or for sale. Initially, it offered about 500 pieces of software or apps. Only a few of these apps were developed by Apple—most came from other developers who used the software development kit (SDK) provided by Apple. Apple’s major competitor is Google, with its Android platform, which is available to several mobile phone manufacturers. The Android Market, later renamed Google Play, was launched in October 2008 and offers a similar ecosystem for software developers and users (Basole and Karla 2012). Both app stores allow developers to make their apps available at a small annual cost to a huge audience. The store owner offers toolkits to its developers, facilitating user participation by creating an ecosystem for development (Franke and von Hippel 2003). Apple keeps 30% of all revenues and distributes 70% to developers. In 2018, more than two million apps were available in Apple’s App Store.2 While there are also several other stores available for smartphone apps, the Google Play Store and Apple App Store offer by far the largest number of available apps and have the largest customer bases (Research2Guidance 2017). The availability of complementary apps is increasing the value of these operating systems and thus increases hardware sales. Platform owners such as Apple or Google are taking measures to increase the ease-of-access and to lower the costs of developing for their platform (Miric et al. 2019). Thus, it may also be valuable for individual developers to join the platform, since barriers are particularly low. My study considered only Apple’s App Store, since it offered the highest number of medical smartphone apps at the time of the first data acquisition in 2014 (Pramann et al. 2014). Medical apps make up a fraction (< 2% of all available apps3 ) compared to larger groups such as games (> 24%). In this app store, a few popular apps account for the most downloads and revenue, while most apps are 2 Source:

https://www.statista.com/statistics/276623/number-of-apps-available-in-leadingapp-stores, accessed January 29, 2019. 3 Source: https://www.statista.com/statistics/270291/popular-categories-in-the-app-store, accessed January 29, 2019.

5.2 Medical Smartphone Apps

39

downloaded only a few hundred or thousand times per month (Garg and Telang 2013). Yet, online platforms such as Apple’s App Store offer opportunities for developers who develop apps for niche markets: owing to the much larger product selection than traditional retail channels, successful niche products lead to a longer tail in the sales distribution (Brynjolfsson et al. 2011). This long tail of developers is inviting not only professional developers and companies, but also hobbyists and amateurs to develop apps. While such small-scale innovations only account for limited individual revenue, the overall share of such innovation may represent a considerable fraction of the overall revenue generated on the platform (Miric et al. 2019). The worldwide download numbers of medical apps (Figure 5.3) have increased from about 2.3 billion in 2014 to 3.7 billion in 2017 (Research2Guidance 2017). This indicates that this sector is steadily growing over time and that medical smartphone apps are increasingly used.

Downloads (Billions) 4.0

3.7

3.5 3.0

3.0 2.5

3.2

2.3

2.0 1.5 1.0 0.5 0.0 2014

2015

2016

2017

Figure 5.3 Downloads of medical apps in the Apple AppStore worldwide between 2014 and 2017 (Research2Guidance 2017)

To date, there are no mandatory regulations for medical smartphone apps (Visser et al. 2013), although some of these apps can be seen as medical devices

40

5 Theoretical Background and Hypothesis Development

according to EU-Directive 93/42/EEC (2007). Thus, it is much easier to diffuse a medical smartphone app than a tangible medical device that requires clearance by a regulatory body. Approval for market launch is only required from the platform provider (i.e. Apple), not from the regulatory bodies, such as the Food and Drug Administration (FDA) in the U.S. or EU bodies. To date, very few apps are being voluntarily regulated (Shuren et al. 2018; Hussain et al. 2015; Kamerow 2013), mostly because they intend to be reimbursed by health insurances (Elenko et al. 2015). Regulatory agencies in Europe and the U.S. are currently evaluating ways to regulate smartphone apps so as to ensure user safety (Shuren et al. 2018). There are three general ways in which an app may generate revenue: app sales, in-app purchases, and in-app adverts. Particularly in-app purchases (i.e. apps with additional features that can be unlocked by paying a fee or by a recurring payment) have been proven to substantially increase revenue (Roma and Ragaglia 2016). In-app advertising is a way to get revenue from free apps.

Revenue (in € million) 1,600

1,575 1,456

Paid downloads 1,400

In-app purchases

1,200

In-app adversments

1,509

1,287

1,000

909

800 548

600 430 400

242 200 0

97 3

23

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018*

Figure 5.4 Revenues generated by smartphone apps per segment in Germany between 2008 and 2018 (*forecast for 2018, in e million)

In 2018 in Germany, in-app purchases were estimated to yield about 76% of all revenues, with in-app adverts accounting for 18% and app sales for 6%

5.3 Hypothesis Development

41

(Figure 5.4)4 . However, this data was collected across several app stores available in Germany and considered all available apps, not only medical apps. Nonetheless, it outlines the importance of in-app purchases in generating revenue via smartphone apps.

5.3

Hypothesis Development

In this study, I seek to contribute to a novel research stream about the innovativeness of patients and caregivers by analyzing the medical smartphone apps of patients, their caregivers, and healthcare professionals, compared to manufacturers’ apps. Thus, I seek to shed some light on this research field, which has been mainly driven by qualitative studies and thus small sample sizes. According to the review of user innovation literature (see section 2.1), I assume that there are several developer types in the medical smartphone app market. Regarding the non-user-developers, I assume that there will be professional medical device manufacturers and professional software developers active in the market for medical smartphone apps. Other players in the healthcare sector such as health insurances, pharmaceutical companies, and others can also be expected. Further, I assume that there are independent developers who don’t develop according to their own or their relatives’ needs, but for the joy of developing software and learning along the way (Miric et al. 2019). These developers are related to the phenomenon of participators, as described by Raasch and von Hippel (2013). Lakhani and Wolf (2005) found, in their study in the field of open-source software, that the strongest motivational driver of software developer was enjoyment-based intrinsic motivation, particularly how creative a developer feels when working on a project. Pecuniary remuneration is in many cases only a secondary motivational factor for these developers. Regarding the user-developers, there is comprehensive evidence from the literature that healthcare professionals (i.e. people who provide health-related products and services, such as physicians, surgeons, dentists, psychologists, pharmacists, midwives, nurses, and others) develop medical devices according to their own work-related needs and their patients’ unmet medical needs (Chatterji 2009; Lettl et al. 2008; Lüthje and Herstatt 2004). Although most of them have no technical knowledge (i.e. software development skills), I assume that individual healthcare professionals and associations of healthcare professionals know about 4 Source:

https://de.statista.com/statistik/daten/studie/801670/umfrage/umsatz-mit-mobilenapps-nach-segmenten-in-deutschland, accessed January 29, 2019.

42

5 Theoretical Background and Hypothesis Development

their and their patients’ unmet medical needs and are able to address these needs in a medical smartphone app, potentially with the help of a professional software developer. There is limited evidence that patients develop medical devices for their own unmet medical needs (Habicht et al. 2013), particularly in the case of chronic and rare diseases (Oliveira et al. 2015). Initial data from a web-based platform revealed that patients and their caregivers developed software to improve their health outcomes (Bullinger et al. 2012). Since both patients and caregivers may possess significant knowledge on the unmet medical needs of a patient, and as solution knowledge on developing an app is comparatively easy to obtain, I expect patients and their caregivers to develop medical devices for their own or their relatives’ needs. Considering these four developer types, I developed five hypotheses that I will test using the dataset and the resulting empirical model.

5.3.1

Free Revealing of User-Developed Medical Apps

Studies of the diffusion of user innovations have shown that users often freely reveal their innovations, since they benefit from using the innovation, as opposed to selling it (Harhoff et al. 2003). The pattern of free revealing of user innovations has been studied in several industries (Franke and Shah 2003; Harhoff et al. 2003; Morrison et al. 2000) and has been analyzed in-depth for open-source software (Füller et al. 2013; Balka et al. 2010; von Hippel and von Krogh 2006). In the case of open-source software, developers not only give away the software for free, but also publish the source code on the Internet (Tirole and Lerner 2002). Yet, there may be reasons for manufacturers to reveal their carefully selected innovations for free, for instance, owing to reputation in a community or for marketing reasons (Henkel 2006). In contrast, some user innovators may not want their innovations to be freely revealed, for instance when they seek to commercialize their idea and become a user entrepreneur (Bogers and West 2012; Shah and Tripsas 2007). Research into free innovation has recently emphasized that many end-users (e.g. patients and their caregivers) are not incentivized by monetary rewards and therefore reveal their innovation for free (von Hippel 2017). In the medical smartphone apps field, I assume that—similar to other examples in (open-source) software—patients and caregivers freely diffuse their innovations, compared to professional software developing firms or medical device manufacturers. On the other hand, healthcare professionals, who likely developed a medical app out of their professional experience, may not want to freely reveal their innovations. Independent developers develop for several purposes, for

5.3 Hypothesis Development

43

instance, the joy of developing an app, learning along the way, and for financial benefits. Thus, financial benefits are not their primary intention (Raasch and von Hippel 2013). This argumentation yields the following hypothesis: H1a: Patients and caregivers freely reveal their medical apps more often than professional software developers (companies). H1b: Patients and caregivers more often reveal their paid medical apps at a lower price than professional software developers (companies).

5.3.2

The Early Development of Medical Apps

Users innovate because they face unmet needs months or years before other users (Herstatt and von Hippel 1992). In a study about the development of scientific instruments, Riggs and von Hippel (1994) showed that early development of new scientific instruments (with high scientific importance) is driven by users, while manufacturers step in later and mainly develop devices of high commercial importance. Since patients, caregivers, and healthcare professionals are all users of medical devices, I assume that some also face needs earlier than the general market and therefore are the first to develop novel medical smartphone apps. Healthcare professionals have already been proven to be lead users and are able to develop medical devices earlier than manufacturers (Lettl et al. 2006; Lüthje and Herstatt 2004). The Apple App Store was opened to developers in mid2008. At the time, few professional software developers were developing mobile smartphone apps. Barriers for patients, caregivers, and healthcare professionals to diffuse their apps are very low—there is as yet no mandatory regulatory process for medical smartphone apps (Shuren et al. 2018). Thus, I hypothesize: H2: The early development of medical smartphone apps was triggered by userdevelopers.

5.3.3

Developer Type and the Quality of Medical Apps

It is hard to assess the quality of user-developed products, since most usergenerated ideas remain in a prototype stage and are later (if at all) commercialized together with a manufacturer. To date, little is known about the de facto market

44

5 Theoretical Background and Hypothesis Development

behaviors of user-developed and commercialized products vs. manufacturerdeveloped and commercialized ones (Ogawa and Piller 2006). Lilien et al. (2002) found that user-developed products outperformed products developed by internal developers, not only concerning their novelty, but also concerning their sales forecast. Similarly, Poetz and Schreier (2012) found that user-developed products had higher novelty and use value, while the feasibility was lower than products developed by a manufacturer’s internal development team. More generally, several studies found a positive correlation between usage experience and innovativeness (Schreier and Prügl 2008; Lüthje et al. 2005). Particularly in the healthcare sector, patients with long-lasting—often chronic—diseases have high usage experience in time-consuming daily routines. During these enduring treatments, patients, their caregivers, and healthcare professionals gain sticky knowledge about the disease and the treatment, which is costly to acquire and to transfer (von Hippel 1994). Previous research has shown that healthcare professionals use this sticky information to develop for instance off-label drug therapies (DeMonaco et al. 2006). Thus, I propose that patients and caregivers also have latent needs that are costly to transfer and that are hard to anticipate by someone who is not affected by or who does not care about someone who is affected by a disease. In section 2.1, I outlined that need knowledge and solution knowledge are needed for a successful innovation (Schweisfurth and Herstatt 2016). I postulate that many patients and caregivers possess significant need knowledge and that some of these—who have access to solution knowledge (i.e. software development skills)—are able to develop software according to their own needs or the needs of the person to be cared for. Access to this sticky knowledge on unmet medical needs leads to higher satisfaction of the client of an app who would then assess the app’s quality by rating it. Ratings have been proven to be a simple yet meaningful way of assessing a product’s quality, particularly for buying decisions (Chen 2017; Rozenkrants et al. 2017; Stoyanov et al. 2015; Kranz et al. 2013). Although there are several parameters of an app that influence its ratings (e.g. functionality, design, reliability, etc.), I assume that patients who use this app value it most if it best suits their unmet medical needs. Thus, I hypothesize: H3a: Developer type impacts on the quality (operationalized by ratings) of medical apps. H3b: Apps developed by patients and caregivers receive better ratings than apps developed by other developers.

5.3 Hypothesis Development

5.3.4

45

Developer Type and the Download Numbers of Medical Apps

As noted above, while first studies have assessed the quality of user-developed vs. non-user-developed products, there is very limited evidence on the de facto uses of user-developed products in a marketplace. As mentioned by Nishikawa et al. (2013), there are several reasons for this shortcoming: First, companies usually translate user-developed concepts into a marketable product. This process involves a decisions set that strongly impact on a product’s market success. Second, Poetz and Schreier (2012) found that the feasibility of user-developed products is lower than manufacturer-developed ones. Thus, further development and commercialization of user-developed products may be harder than further developing less novel manufacturer-developed products. Third, it is challenging to assess whether and how consumers’ responses (and thus their buying decisions) are aligned to managers’ perceptions of new products and what impacts an eventual mismatch of both perceptions has on market performance (Nishikawa et al. 2013). In the medical smartphone apps market, the abovementioned constraints are all not in place: users themselves create marketable products on their own—no manufacturer is needed to complete the app. The platform provider, the Apple App Store, only evaluates whether the software meets its general guidelines5 , but it obviously does not further develop the app. Thus, customers’ buying decisions are only influenced by the information on an app as it is displayed in the App Store. One of the very few available studies on consumer goods’ market performance revealed that user-generated products sold about twice as much as a manufacturerdeveloped products in the same category (Nishikawa et al. 2013). Based on these findings, and considering the abovementioned advantages of the App Store as an empirical field with a direct developer-customer interface, I hypothesize: H4a: Developer type impacts on medical apps’ download numbers. H4b: Apps developed by patients and caregivers achieve higher download numbers than company-developed apps.

5 Source:

2019

https://developer.apple.com/app-store/review/guidelines, accessed January 15,

46

5 Theoretical Background and Hypothesis Development

5.3.5

Developer Type and Medical App Revenues

The abovementioned variables, an app’s quality (operationalized by user ratings) and an app’s downloads, provide a good estimate of how users perceive an app. However, the bottom line for marketplace success is the revenue generated by an innovation. A study by Franke and colleagues (2006) in the field of extreme sport showed that a user innovation’s sales potential increases with this user’s lead userness. Svensson and Hartmann (2018) revealed that the potential revenues from innovations developed by healthcare professionals in a hospital-based makerspace already exceeded the required investment more than 10 times in the first year of operation. Lilien and colleagues (2002) found that user-developed products that were further developed by a manufacturer and subsequently commercialized yielded significantly higher estimated sales volume within year five after market launch than a solely manufacturer-developed product. The de facto sales revenue of a user-developed product was recorded by Nishikawa et al. (2013): sales revenues from a tangible, user-developed product that was further developed and commercialized by a manufacturer were about three times higher than those of a solely manufacturer-developed product. An app’s revenue is determined by price, downloads, and in-app purchases. Thus, I can only assume a mediating effect of an app’s developer type on revenue. In H1, I postulated that apps developed by patients and caregivers are freely revealed more often and that they are offered at a lower price than manufacturerdeveloped apps. Yet, I found that in-app purchases account for most smartphone app revenues (see section 5.2). Considering these two opposing tendencies, I hypothesize: H5a: Developer type has a mediating impact on medical apps’ revenues H5b: The developer type patients and caregivers has a significantly positively mediating impact on apps’ revenues, compared company-developed apps

5.3.6

Summary of Hypotheses: Research Model for Mediated Regression Analysis

I will test H1 and H2 using the descriptive analysis of the dataset, and H3, H4, and H5 using a mediated linear regression model. Besides the exogenous variable developer type and the endogenous variable revenue, I included the two moderating variables ratings and number of downloads in the model (see Figure 5.5).

5.3 Hypothesis Development

Exogenous variable

47

App variables

Revenue variables

Ratings

Downloads

Endogenous variable

H4

H3

Developer

H5

Revenue

Mediator variable

Figure 5.5 The mediated regression model without the control variables

Further, I incorporated four control variables into the model. The control variables are subdivided into app variables (number of ratings and market tenure) as well as the two revenue variables price and in-app purchases (see Figure 5.6). The interview data will be used to better understand and interpret the results in chapter 8.

Exogenous variable

Developer

Control variable

App variables

Revenue variables

Number of ratings

Price

Ratings

Downloads

Market tenure

In-app purchases

Mediator variable

Figure 5.6 The mediated regression model, including the control variables

Endogenous variable

Revenue

48

5 Theoretical Background and Hypothesis Development

As noted in section 5.2, an app’s revenue is only determined by number of downloads, price, and in-app purchases. Thus, the control variables number of ratings and market tenure only point to the two mediating variables, but not to the endogenous variable.

6

Methodology

Contents 6.1

6.2

Analytical App Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.1 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.2 Operationalization of the Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1.3 Data Preparation and Assumptions for Regression Analysis . . . . . . . . . . . 6.1.4 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Qualitative Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.1 Sampling and Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

50 50 54 57 63 64 64 65

The phenomenon of innovative patients and caregivers is still under-explored. I take a phenomenon-based approach (von Krogh et al. 2012) to derive implications for both the user innovation literature and health policy. The healthcare sector has been subject to several studies and has proven to be a field with high innovation potential, particularly for user innovation (Svensson and Hartmann 2018; Hinsch et al. 2014; Chatterji et al. 2008; Braun and Herstatt 2008; Lettl et al. 2006; Shaw 1985). For study 1, I obtained two complementary data sources: analytical data on medical smartphone apps and qualitative data on interviews with user innovators (patients, caregivers, and healthcare professionals) in the field of medical smartphone apps.

Electronic supplementary material The online version of this chapter (https://doi.org/10.1007/978-3-658-32041-6_6) contains supplementary material, which is available to authorized users.

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_6

49

50

6.1

6

Methodology

Analytical App Data

This study focuses only on apps of the Apple App Store, as at the time of first data acquisition (2014), considerably more medical apps were available for iOS than for Android (Pramann et al. 2014). In the meantime, the number of medical apps for Android has increased substantially. In 2017, the Apple App Store and the Google Play both had about 150,000 medical smartphone apps available (Research2Guidance 2017). Analytic app data such as user ratings, downloads, and app prices have been investigated in different medical disciplines, such as mental health (Bakker et al. 2018), obesity (Stevens et al. 2014), skin cancer diagnosis (Abbott and Smith 2018), orthopedics (Franko and Bhola 2011), hernia repair (Connor et al. 2013), and surgery (O’Neill and Brady 2012; Dala-Ali et al. 2011). I will now outline how I collected the analytical app data, how the variables were operationalized, how data aggregation and processing was performed, and how the data’s robustness was tested.

6.1.1

Data Collection

6.1.1.1 Dataset 1 (2014) I used the software WebSundew 4 Professional (Sundewsoft) to collect and extract data from five websites that provide publicly available analytical data about smartphone apps. I collected 30 parameters such as name, app ID, initial release date, and ratings, which have been used to measure app success (Lee and Raghu 2014; Garg and Telang 2013). The initial collection took place on May 26, 2014; data was collected for some parameters weekly and for others monthly until August 28, 2014. Although I measured data over about three months, I used only the most recent values obtained in August 2014 for further analysis in this thesis. All data was stored in a MySQL database (release 5.6.16) using an Apache web server (release 2.4.7) and was accessed using the web browser based software phpMyAdmin (release 4.1.6). The database was accessed using Microsoft Access 2010 and was evaluated using IBM SPSS Statistics 22 and Microsoft Excel 2010. First, I analyzed all apps in the top 1,000 medical apps in the German, UK, and U.S. Apple App Stores in the three categories free, paid, and grossing. Those three markets were selected owing to their good market conditions and high availability of medical apps (Research2Guidance 2017) as well as for reasons of language. Since some apps appeared in several markets and some in several categories, this first search yielded n = 4,550 apps. Further, I selected only apps with a true

6.1 Analytical App Data

51

medical purpose and that thus can be classified as a medical device according to a medical device regulation (cf. EU-Directive 93/42/EEC 2007). This led to 1,233 apps published by 878 developers. According to Armstrong et al. (1997), this classification was done again by an experienced innovation scholar and was later cross-checked. Applying Cohen’s κ (Cohen 1968) yielded an agreement of 89.2%. In a next step, I categorized the apps according to the developer type into four groups: I differentiated between user-developers, namely patients and caregivers, as well as healthcare professionals, and non-user-developers (i.e. professional developers, namely companies and independent developers). For each app, I did a brief Internet search about each developer’s group affiliation. This information was mainly obtained from the About us section of the developer’s website, the iTunes website, linkedin.com, twitter.com, websites with media releases, and other websites. If the app was developed in a team, it was classified as user-developed if at least one user was the initiator of the idea (i.e. they were a co-founder) and if the idea was developed out of the need experienced by that individual. If a healthcare professional and a software developer were both co-founders, I assumed that the need knowledge that triggered the innovation came from the healthcare professional and that the software developer contributed solution knowledge to the app development; thus, it was a user innovation by a healthcare professional. If there was only one software developer as a single founder and the team had a chief medical officer but not a co-founder, I classified the developer as a company. Independent developers are individuals who develop software for various indications (i.e. games, fitness, medical) and often also deliver software development as a service to others. During the classification, I found several indications that independent developers often develop apps for the fun of developing apps and not for maximizing profit from app revenues. Thus, they can be seen as participators (Raasch and von Hippel 2013), as outlined in section 2.2. In some cases, associations of healthcare professionals developed apps that were also classified under healthcare professionals. In the following, I will give examples of the classification. This app was classified as patients and caregivers, according to the app’s website: Sanovation started based on a personal experience: […] Our co-founder had been suffering from chronic pain for several years. In search of a suitable treatment to relieve his pain he went from one doctor to another, trying to figure out the cause for his pain and what to do about it. It was tiring, difficult, and above all frustrating. After hitting rock bottom, […] he developed a prototype to track his own medical condition. A physical therapist to whom he showed his pain drawings, finally found

52

6

Methodology

the cause for Daniel’s pain. As a result of this breakthrough, […] he was able to receive the correct treatment and slowly get his life back on track.1

The next developer is a healthcare professional who developed an app. Yet, he did this in his role as a patient, thus, I classified him as patient and caregiver: […] I am flattered that you discovered my app. I am both a patient and a physician. I am a cardiothoracic surgeon who had to stop operating due to multiple sclerosis. So I am now semi-retired and work as the director of the Cardiac Surgery ICU. Making apps is a hobby of mine. I am self-taught as far as programming and developing. The idea for this app came to me because I often would forget where I had given myself my last medication injection and thought it would be useful to have an app in which to keep a record of the sites. The original version was a simple 2D picture background. While experimenting with developing a game, I came across the 3D technology and decided to upgrade InjectionTracker to 3D. That’s how it evolved to the current version. I receive no monetary benefit for the app. I do it for fun and in hopes it might help others. […]2

This developer is a healthcare professional, as he indicated on a developer community: I am a doctor and amateur developer with a particular interest in coding for medical technology. I started my journey into code in 2015. Ampoule helps healthcare professionals track their personal drug inventory. Never let your stock run out or expire again!3

The following app was developed by a man for his wife, thus, I classified him under patient and caregiver: […] The app was developed out of necessity. My wife was diagnosed with an ailment that required her to take a myriad of medications – some of which caused her to have short term memory loss. After many attempts using off the shelf solutions – including alarm clocks – notes and ribbons, we looked to see if there wasn’t an app that had been created that reminded people to take their medication as prescribed – get to doctors’ appointments as scheduled and refill prescriptions in a timely fashion. There was nothing on the market like that... So we built it ourselves. […] We used a combination of agile development and common sense. We asked the 1 Source:

https://www.sanovation.com/story, accessed January 7, 2019. personal email received on August 1, 2014. 3 Source: https://stackoverflow.com/users/story/8289095, accessed February 18, 2019. 2 Source:

6.1 Analytical App Data

53

people using the app to tell us what they needed – we reviewed the feedback and developed accordingly. To date we have released over 50 iterations. Med Helper has been used over 10,000,000 times in over 150 countries worldwide. We have received thousands of suggestions and continue to get feedback daily which we are using to build our next phase. – Med Helper Cloud. […]4

I classified the following developer under independent developer, because he is a single innovator who developed the app in his free time just for the fun of developing it: I made an app called "Doctor Mole" […] which scans your moles and gives you feedback on the ABCDE metrics (Asymmetry, border color, diameter, evolution). I took the same engine from that app, and applied it to an ECG chart. I made the app for fun, I do app development in my spare time at home. I am not a healthcare professional. The motivation for the app was just to see how easy/difficult it would be to scan an ECG chart from a phone with the goal of improving it to the point where it can add value when trying to diagnose an ECG. […]5

For each user innovator, I have proof of their affiliation. This task was then completed again by an experienced innovation scholar separately so as to ensure rigorous results. Cohen’s κ for the classification of all four developer types yielded 78.7% (κ = 0.787). I then double-checked all classifications with no consensus, and decided on one developer type according to the two independently identified results. For developer groups with more than one person, I evaluated the app’s founder or idea originator. In 86 cases, the information could not be obtained from the Internet. I sent a clarification e-mail to these developers; 55 answered—a 63.9% return rate. To ensure being on the conservative side concerning this classification, I deleted all apps for which the group affiliation could not be identified unambiguously. This yielded 1,192 apps by 847 developers.

6.1.1.2 Dataset 2 (2018) The second dataset was established in mid-2018 in order to confirm the 2014 data and analyze the development of the Apple App Store and thus of the innovative developers over time. Instead of relying on several data sources, a comprehensive dataset on analytical app data provided by the company Priori Data6 was purchased.

4 Source:

personal email received on January 31, 2014. personal email received on January 31, 2014. 6 Source: www.prioridata.com, accessed July 5, 2018. 5 Source:

54

6

Methodology

The dataset comprised 18 parameters such as name, app ID, initial release date, download numbers, and revenue of the top 1,000 apps in the medical category for free and paid apps from Germany, the UK, and the U.S. in Q1 2018 and Q2 2018. As some apps appeared in several markets and some in several categories, this dataset comprised n = 4,154 individual apps. The web-based dataset was accessed using a web browser, downloaded, and evaluated using IBM SPSS Statistics 24 and Microsoft Excel 2016. As the ratings of the apps were not available in the 2018 dataset, this data was obtained using a self-developed script directly from the Apple App Store using Google Sheets. The data on ratings was then merged with the dataset. Further, I applied exactly the same procedure as with the 2014 dataset for classifying the medical apps that are medical devices according to the EU medical device regulation (cf. EU-Directive 93/42/EEC 2007). This yielded 1,310 apps developed by 957 developers. Again, this classification was done by an experienced innovation scholar and was later cross-checked. Applying Cohen’s κ (Cohen 1968) yielded an agreement of 92.1%. The categorization into the four developer groups, namely user-developers (patients and caregivers, as well as healthcare professionals) and non-user-developers (i.e. professional developers, namely companies and independent developers) followed the same guidelines as in the 2014 dataset. Again, this work was redone by an experienced innovation scholar separately. Cohen’s κ for the classification of all four developer types was 88.6% (κ = 0.886). In 63 cases, the information could not be obtained during the Internet search. I sent an e-mail to these developers; 18 answered—a 28.6% return rate. To ensure being on the conservative side concerning for the classification, I deleted all apps for which the group affiliation could not be identified correctly. This yielded 1,265 apps by 912 developers.

6.1.2

Operationalization of the Measures

I will now explain the measures that will be analyzed in the empirical model (including the independent variable, the dependent variable, mediating variables and control variables). Only the 2018 data was used for the empirical model owing to the higher data quality in this dataset.

6.1.2.1 Exogenous Variable The developer type is the independent variable in my model and was determined during data collection. In section 2.1, I outlined the importance of user innovation,

6.1 Analytical App Data

55

various user types, and their characteristics. According to the literature, I identified companies (i.e. manufacturers of medical apps) and independent developers (participators, i.e. individuals who enjoy developing apps and where economic benefit is of subordinate priority) as non-user-developers and healthcare professionals (who develop medical apps for their own or for their patients’ needs) and patients and caregivers (who develop medical apps for their own needs or for the needs of their relatives) as user-developers. Each app was associated with one developer type: company, independent developer (both non-user-developers), healthcare professional, patient and caregiver (both user-developers). As described in the previous chapter, this variable was identified manually for each of the 912 developers. Since the users of medical apps don’t know the developer type (this information is not directly available on the App Store), I assumed that decisions about downloading and rating the app are unbiased concerning this measure.

6.1.2.2 Endogenous Variable The primary variable for evaluating a smartphone app’s performance is an app’s revenue. Revenues can be generated by selling apps for a given price or by offering in-app purchases. Although both ways have been proven to successfully increase revenues in the App Store (Roma and Ragaglia 2016), in section 5.2 I elaborated on the large fraction of revenue generated by in-app purchases. Since revenue-related data is not published in the App Store, the data was estimated by the data provider. The revenue was estimated for Q2 2018 (April 1 to June 30, 2018) for the German, UK, and U.S. markets. The variable revenue was calculated by summing up the values from these three countries.

6.1.2.3 Mediating Variables Since the independent variable developer type had no direct effect on the dependent variable revenue (the revenue was only influenced by downloads, price, and in-app purchases), I assume that there is a mediating effect to be found in the data: App quality was operationalized using the app’s rating. Users are able to rate a smartphone app that was downloaded between one and five stars (one star = the worst rating, five stars = the best rating). User ratings are “peer-generated product evaluations posted on company or third party websites” (Mudambi and Schuff 2010, p. 186) and have been used by scholars to evaluate a smartphone app’s quality (Chen 2017; Rozenkrants et al. 2017; Stoyanov et al. 2015; Kranz et al. 2013). Although there is initial evidence that the validity of user ratings of physical products is limited (de Langhe et al. 2016), other studies have shown that customer satisfaction with a product relates to customer ratings (Engler et al.

56

6

Methodology

2015). Yet, for digital products, it is common sense to use customer ratings as an indicator to measure quality (Roma and Ragaglia 2016; Lee and Raghu 2014). Further, users can add a text-based review of the app. The ratings were the mean value of all ratings an app has received since it was published in the App Store. The variable rating was the weighted mean value of the rating of the German, UK, and U.S. markets, considering the number of ratings in each country. Second, app popularity was operationalized using the number of downloads. The more often an app is downloaded, the higher the number of potential users and thus the app’s revenue (Lee and Raghu 2014). Although some apps are just downloaded and subsequently deleted, this is the best indicator available to measure how many people actually use an app. This data is not available from the App Store, but was estimated by the data provider. The download numbers have been estimated for Q2 2018 (April 1 to June 30, 2018) for the German, UK, and U.S. markets. The variable downloads was calculated by summing up the values from these three countries.

6.1.2.4 Control Variables The number of ratings indicated how often an app has been rated by users. Engler et al. (2015) found that the number of ratings enhances a rating’s subjective weight. The date an app was first uploaded to the App Store was used to calculate the market tenure (Miric et al. 2019). The app’s price can be set by the developer in predefined price tiers (i.e. U.S. $0.99, U.S. $1.99, U.S. $2.99). This variable is stored in U.S. $ for all apps in the dataset, since the price tiers have fixed equivalents in other countries (EUR and GBP). As noted above, in-app purchases are another way to increase revenues: the basic version of the app is provided for free or at relatively low cost, and additional features can be unlocked by additional single or monthly payments. This is called the freemium business model (Liu et al. 2014). This variable is dichotomous, i.e. an app has or does not have in-app purchases. While the first two control variables relate to the app properties, the latter two relate to the revenue. I called the two variables number of ratings and market tenure app-related variables and the price and in-app purchases revenue-related variables.

6.1 Analytical App Data

6.1.3

57

Data Preparation and Assumptions for Regression Analysis

Both datasets (2014 and 2018 data) were stored in a database and analyzed similarly using IBM SPSS Statistics 24, the SPSS macro PROCESS developed by Hayes (2017), and Microsoft Excel 2016. As the data was obtained from an independent source and the developer classification was performed without knowledge on an app’s performance, I assume that there is no common method bias regarding the data (Podsakoff et al. 2003). Owing to the better data quality and the more recent data, I decided to use only the 2018 dataset for the regression analysis (see section 7.1.4). Thus, the data preparation in this chapter only considered the data used in the regression analysis. The datasets contained no missing values except ratings, because some apps had no ratings. These apps without ratings were excluded for the regression analysis. Thus, the dataset was reduced from 1,265 apps to 837 apps. Regression analysis requires the dataset to meet a set of assumptions. The data should be normally distributed, checked for homoscedasticity and multicollinearity, and for independence of error terms (Hayes 2017; Hair et al. 2014; Menard 2002). Both visual inspections and statistical tests can be applied to check these assumptions. To test the dataset for the abovementioned assumptions, I applied a simple linear regression analysis using SPSS. For this analysis, revenue served as the dependent variable and developer type, as well as the mediating variables ratings and number of downloads as well as the control variables were included as independent variables in three separate layers. For the mediated regression analysis applied later, a more sophisticated model was applied, using the SPSS macro PROCESS (Hayes 2017). However, to test the assumptions, a simple linear regression model was adequate. To reduce the nonnormality of the data, I used the transformed data for three variables in the simple regression analysis, as indicated below. A dummy variable was used for analyzing the four different developer types.

6.1.3.1 Analysis of Normality In general, exogenous and endogenous variables should be normally distributed in multivariate data analysis. A violation of this assumption may influence a regression analysis’ results. There are several ways to check for normality: visual inspection, analyzing skew and kurtosis of the data, or applying commonly used tests, i.e. the Kolmogorov-Smirnov test (Field 2009). Other scholars have argued that, for linear regression analysis, the normality of the residuals (i.e. the error

58

6

Methodology

terms) is more important than that of the exogenous and endogenous variables (Hair et al. 2014). Table 6.1 The descriptive statistics of variables used in empirical model Min.

Max.

Mean

S.D.

Skewness

Kurtosis

K-S test

Rating

1.00

5.00

3.699

0.959

−0.765

−0.130

0.000

No. of downloads

0

384,399

4,750

21,233

11.89

170.110

0.000

Revenue

0

648,601

6,166

38,037

11.25

151.370

0.000

No. of ratings

5

51,709

1,004

3,966

7.01

60.475

0.000

Price

0.00

94.99

1.9389

5.622

9.53

129.296

0.000

Market tenure

47

3,624

1,657

930.33

0.18

−1.037

0.000

No. of ratings (Box-Cox transformed data)

0.30

0.89

0.5445

0.156

0.248

−0.959

0.000

No. of downloads (Box-Cox transformed data)

1.00

1.05

1.026

0.008

0.855

0.969

0.000

Revenue (Box-Cox transformed data)

0.00

0.15

0.041

0.044

0.449

−1.214

0.000

Std. residuals (revenue)

−6.175

2.945

0.0

0.994

−0.386

−0.085

0.000

Notes: n = 837; 428 apps without ratings were excluded from this analysis. Std. error for skewness: 0.085. Std. error for kurtosis: 0.169. Lambda values for Box-Cox transformation: −0.2000 (number of ratings), 0.0039 (number of downloads), −0.01247 (revenue)

As presented in Table 6.1, the values for number of downloads, revenue, number of ratings, and price had high skewness and kurtosis. Thus, I decided to reduce these variables’ skewness and kurtosis by applying a data transformation. Data

6.1 Analytical App Data

59

transformations are commonly used to correct violations of the statistical assumptions that underlie the multivariate techniques and can also be used to improve the correlation between variables (Hair et al. 2014). I decided to follow the procedure of Box and Cox (1964), a parametric power transformation technique that seeks to reduce anomalies in the data such as nonnormality and heteroscedasticity (Sakia 1992). The Box-Cox transformation is based on a log transformation of the data. Thus, I could not transform the control variable price, as a significant number of apps are available for free, and log transformations don’t allow the data to take zero or negative values (Olivier and Norberg 2017). For the variables revenue and downloads, which also may reach the value zero, I shifted the dataset one increment to the right, as suggested by Olivier and Norberg (2017). As shown in Table 6.1, this shift is insignificant for revenue and downloads, but would have been a significant change for the control variable price. The data transformation must be kept in mind for the interpretation of the data. Since the lambda values for number of ratings and revenue were negative, I subsequently transformed these two variables by mirroring them in order to restore the initial direction. The visual inspection of Figures 6.1, 6.2 and 6.3, as well as the decrease of skewness and kurtosis, indicated that the data transformation improved the data for better usage in linear regression analysis. Nonetheless, the KolmogorovSmirnov test revealed nonsignificant results for all variables.

Figure 6.1 Histogram: Number of downloads (left: original data, right: Box-Cox transformed data)

60

6

Methodology

Figure 6.2 Histogram: Revenue (left: original data, right: Box-Cox transformed data)

Figure 6.3 Histogram: Number of ratings (left: original data, right: Box-Cox transformed data)

Similarly, the analysis of normality of residuals revealed that the KolmogorovSmirnov test was nonsignificant. Yet, a visual inspection of the normal probability plots (Figure 6.4) indicated that the deviation from normality seemed to be within an acceptable range. In sum, the calculations using the Kolmogorov-Smirnov test revealed that the variables in the dataset were nonnormally distributed. However, the assumption of normality is particularly true for smaller sample sizes (n < 50). For larger sample sizes (n > 200), the impact of nonnormality of data diminishes (Hair et al. 2014).

6.1 Analytical App Data

61

Figure 6.4 The normality of residuals for exogenous variable revenue: Histogram (left) and normal P-P plot (right)

My sample consisted of n = 837 apps and thus can be considered to be comparatively large. A study by Lumley and colleagues (2002) of healthcare-related data found that, for very large samples, linear regression is also valid for extremely nonnormal data. Thus, I assume that, owing to the large sample size and the Box-Cox transformation that was applied, the dataset can be utilized for further analysis.

6.1.3.2 Analysis of Homoscedasticity Another assumption for regression analysis is the presence of equal variances of the residuals. As stated by Hair et al. (2014), unequal variances are one of the most common violations of assumption. Homoscedasticity is present if the variance is equally distributed, i.e. if there are no triangle-shaped or diamond-shaped patterns (Hair et al. 2014). Visual inspection of the scatterplot (Figure 6.5) revealed that there were no major deviating patterns; however, clearly, some of the apps are released for free, while other apps are sold.

6.1.3.3 Analysis of Multicollinearity Next, the dataset was evaluated concerning multicollinearity, i.e. the extent to which the independent variables of the regression analysis correlate with one another. Two commonly used parameters to assess both pairwise and multiple variable collinearity are tolerance and its inverse, the variance inflation factor

62

6

Methodology

Figure 6.5 Scatterplot for exogenous variable revenue: standardized predicted value vs. residuals

(VIF) (Hair et al. 2014). Menard (2002) stated that values for tolerance below 0.2 are cause for concern and that values below 0.1 indicate serious correlation between variables. The analysis of the dataset (Table 6.2) revealed that all values for tolerance were far above 0.2, indicating no multicollinearity in the data.

6.1.3.4 Analysis of Independence of Residuals Another assumption for regression analysis is that the predicted values should be independent, i.e. the residuals should be uncorrelated for two randomly selected observations (Hair et al. 2014). To test for independence of error terms, I ran the Durbin-Waston test (Durbin and Waston 1951) in the simple linear regression model. As stated in the literature (Field 2009), the test should yield values between 1 and 3. The Durbin-Watson test revealed that the value for the simple linear regression was 1.665. This indicates that independence of error terms is not a concern for further analysis of the data.

6.1.3.5 Summary: Assumptions for Regression Analysis The examination of a simple regression analysis to check for assumptions for the mediated regression analysis done later revealed only minor violations of the

6.1 Analytical App Data

63

Table 6.2 The analysis of multicollinearity Tolerance

VIF

Developer type: Independent developers

0,845

1,184

Developer type: Healthcare professionals

0,767

1,303

Developer type: Patients and caregivers

0,828

1,208

Number of downloads

0,481

2,080

Rating

0,818

1,223

Price

0,897

1,115

Market tenure

0,715

1,399

In-app purchases

0,913

1,095

Number of ratings

0,441

2,267

Notes: For the category variable developer type, a dummy variable was created. The category company served as a baseline, against which the other categories were compared.

dataset. The analysis for normality revealed a nonnormally distributed dataset. I applied a Box-Cox transformation (Box and Cox 1964) to three variables to minimize the impact of nonnormal data on the results. Yet, there was sound evidence that, for larger datasets, regression analyses are pretty robust to minor violations of the normality assumption (Hair et al. 2014), particularly for health-related datasets (Lumley et al. 2002). Further analysis on homoscedasticity, multicollinearity as well as for independence of residuals revealed no violations of the assumptions required for regression analysis.

6.1.4

Data Analysis

H1a/b and H2 were tested using the descriptive data analysis, since they required an analysis of both datasets (2014 and 2018 data). H3a/b, H4a/b and H5a/b were tested using a mediated regression analysis with two mediators and four control variables using only the 2018 dataset. This analysis sought to test hypotheses on the relationship between the exogenous variable developer type, the endogenous variable revenue, and two mediating variables (app quality and downloads) and several control variables. I expected linear relationships between these variables and various interconnections between the mediating variables and the control variables. I considered using an analysis of covariance (ANCOVA), yet this approach would not have mirrored the entanglement of the data, particularly

64

6

Methodology

the interactions between the variables ratings, the number of downloads, and the revenue (Field 2009). Thus, a simple linear regression with ordinary least squares estimation including two mediating variables and a set of control variables was a suitable type of analysis (Hayes 2017). In general, mediation analysis is a statistical method to test if and how an effect of an exogenous variable on an endogenous variable can be partitioned into a direct and an indirect effect mediated by another variable (see Figure 6.6). I used the SPSS macro PROCESS7 (Hayes 2017) to model and to calculate the mediated regression analysis.

Mediator variable

Exogenous variable

Endogenous variable

Figure 6.6 Conceptual depiction of a simple mediation model (adapted from Hayes (2017))

The PROCESS macro is particularly suited to analyze a dichotomous or a three- or four-category ordinal exogenous variable (Hayes 2017). Since developer type—the exogenous variable in this study—is a four-category variable, this confirmed the decision for applying this tool. The syntax of the mediated regression model in PROCESS appears in Appendix 1.

6.2

Qualitative Data

6.2.1

Sampling and Data Collection

Regarding the qualitative data, I conducted 16 semi-structured interviews with developers from Germany, Austria, Switzerland, the U.S., and India to find out about the motivations for their innovative endeavors, their specific knowledge of 7 Source:

https://processmacro.org, accessed January 15, 2019.

6.2 Qualitative Data

65

and experience in the topic, supportive and required contextual factors from their perspectives, and their roles and activities in the innovation process. These themes were developed using prior research (Franke et al. 2006; Lüthje and Herstatt 2004; Franke and Shah 2003; von Hippel 1998) and resulted in an interview guideline for semi-structured interviews. Twenty-two user-developers from the 2014 dataset were contacted, and 15 were willing to be interviewed. Further, I interviewed a developer who released his app in early 2015. In order to increase the validity of the findings (Goffin et al. 2019; Gehman et al. 2018), I sought to reach app developers that varied concerning user group, professional experience, market tenure, and the apps’ business models. I conducted the first set of interviews between May and July 2014, and the second set in March 2015. During this timeframe, six patients, seven caregivers, and three healthcare professionals were interviewed. The developers were between 18 and 55 years old (average age: 39). I interviewed one female and 15 male developers. The interviews lasted between 23 and 49 minutes, with an average length of 35 minutes.

6.2.2

Data Analysis

All interviews were transcribed and coded using MAXQDA 11. For data reduction, a threefold approach was used: first-order analysis, second-order analysis, and aggregation (Gioia et al. 2013; Miles et al. 1994). I sought to ensure the results’ rigor: Concerning data triangulation, three different participant groups were interviewed: patients, caregivers, and healthcare professionals. Concerning investigator triangulation, the interviews were conducted together with a Master’s student or an experienced innovation scholar. The interview analysis was done independently by two scholars and was cross-checked (Gioia et al. 2013; Gibbert et al. 2008).

7

Findings

Contents 7.1

7.2

Findings Concerning the Analytical App Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 7.1.1 Descriptive Analysis: Dataset 1 (2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 7.1.2 Descriptive Analysis: Dataset 2 (2018) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 7.1.3 The Emergence of Medical Apps: A Comparison Between Datasets 1 (2014) and 2 (2018) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 7.1.4 Regression Analysis of Dataset 2 (2018) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 7.1.5 Summary of the Findings: Analytical App Data . . . . . . . . . . . . . . . . . . . . . 94 Findings of Qualitative Data on Medical App Developers . . . . . . . . . . . . . . . . . . 96 7.2.1 Triggers for Innovative Endeavors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 7.2.2 Product Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 7.2.3 Commercialization and Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 7.2.4 A Summary of the Findings: Qualitative Data on Medical App Developers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

This section contains the findings of the analysis of analytical app data and the qualitative data.

7.1

Findings Concerning the Analytical App Data

7.1.1

Descriptive Analysis: Dataset 1 (2014)

During this study, 1,192 apps developed by 847 developers were analyzed; 441 developers (52.1%) were non-user-developers: 336 companies and 105 independent developers. They accounted for 52.2% of all developed apps. The remaining 406 developers (47.9%) were users: 338 healthcare professionals, and 68 patients

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_7

67

68

7

Findings

and caregivers. As shown in Table 7.1, the number of developers and the number of developed apps correlated across the sample.

Table 7.1 The number of developers and the number of developed apps: The 2014 data Companies

Independent developers

Non-user-developers

Healthcare professionals

Patients and caregivers

User-developers

No. of developers

336 (0.397)

105 (0.124)

338 (0.399)

68 (0.080)

No. of developed apps

492 (0.411)

132 (0.111)

487 (0.409)

81 (0.069)

Notes: n = 1,192 apps developed by 847 developers

The analysis of the developers’ business models revealed that 57.9% of all companies released their apps for free, but only 51.5% of healthcare professionals and 42.0% of patients and caregivers (see Table 7.2). Mean prices for paid apps varied from U.S. $5.30 (independent developers) to U.S. $5.45 (companies) for non-user-developed apps and from U.S. $5.30 (healthcare professionals) to U.S. $9.01 (patients and caregivers) for user-developed apps. Looking at mean, standard deviation (S.D.), and median values for prices (Table 7.2) revealed that, for patient or caregiver-developed apps, there was high variance, with few very expensive apps and the majority of apps priced between U.S. $1.99 and U.S. $2.99. Some apps offered in-app purchases, i.e. the basic version of the app is provided for free or at low cost, and additional features can be unlocked by additional payments. This freemium business model (Baden-Fuller and Haefliger 2013) relates positively to increased downloads and app revenue (Liu et al. 2014). I found that 20.4% of all healthcare professionals and 18.5% of patients and caregivers offered in-app purchases, followed by companies (11.0%) and independent developers (8.2%). Mann-Whitney U-tests regarding the percentage of free apps revealed that companies offered significantly (p = 0.007) more often apps for free than patients and caregivers. There was no significant (p = 0.662) difference in price for paid apps between these two developer types. The market tenure of each app was calculated using the date on which the app was first uploaded to the App Store. As depicted in the analysis concerning the developer type showed that apps developed by healthcare professionals are on average in the market the longest (884 days or 2 years and 5 months), followed by companies (857 days), patients and caregivers (852 days), and independent developers (767 days). Apps developed by healthcare professionals or companies

7.1 Findings Concerning the Analytical App Data

69

Table 7.2 App pricing: The 2014 data % of free apps

Mean price of paid apps (U.S. $)

S.D. price of paid apps (U.S. $)

Median price of paid apps (U.S. $)

% of in-app purchases

0.579

5.45

10.24

2.99

0.110

2 Independent 0.427 developers

5.31

8.90

1.99

0.082

3 Healthcare professionals

0.515

5.30

9.40

2.99

0.204

4 Patients and 0.420 caregivers

9.01

23.79

2.99

0.185

1 Companies

Notes: n = 1,192 apps

were in the market on average significantly longer (p < 0.05) than apps developed by independent developers (see Table 7.3).

Table 7.3 Market tenure (days): The mean, median, S.D., and p-values: The 2014 data Mean

Median

S.D.

1

2

1 Companies

857

795

518

2 Independent developers

767

599

541

0.045*

3 Healthcare professionals

884

823

574

0.721

0.047*

4 Patients and caregivers

852

796

586

0.722

0.358

3

0.619

Notes: *p < 0.05. n = 1,192; I used Mann-Whitney U-tests, since the data were not normally distributed

Regarding the ratings, I found that apps developed by patients were rated best, at an average value of 3.97 stars (Table 7.4), followed by healthcare professionals (3.77 stars). Companies and independent developers received 3.49 stars and 3.12 stars, respectively. Companies had significantly lower ratings (p = 0.002) than patients and caregivers as well as healthcare professionals (p < 0.001). Independent developers had significantly lower ratings than healthcare professionals (p < 0.001) as well as patients and caregivers (p < 0.001). The overall sample contained 255,641 user ratings. In sum, I found that patient-developed apps were rated better and thus were of superior quality than non-user-developed apps, confirming H3. On the other hand, I found that patients did not reveal their innovations more often for free

70

7

Findings

Table 7.4 Ratings: The mean, median, S.D., and p-values: The 2014 data Mean

Median

S.D.

1 Companies

3,494

3,500

0.950

1

2

3

2 Independent developers

3,121

3,417

1.100

0.013*

3 Healthcare professionals

3,770

4,000

0.918

0.000***

0.000***

4 Patients and caregivers

3,972

4,162

0.620

0.002**

0.000***

0.384

Notes: *p < 0.05; **p < 0.01; ***p < 0.001. n = 830; 362 apps without ratings were excluded from this analysis. I used Mann-Whitney U-tests, since the data were not normally distributed. In total, 255,641 user ratings were analyzed

and did not offer paid apps at a lower price level; thus, H1 had to be rejected. Although all the hypotheses were evaluated using only the 2018 dataset, the analysis of the 2014 dataset in this section already delivered some first insights into the performance of user-developed vs. non-user-developed medical apps.

7.1.2

Descriptive Analysis: Dataset 2 (2018)

The 2018 dataset revealed a similar picture to the 2014 dataset. It comprised 1,265 apps developed by 912 developers. Of the developers, 516 (56.6%) were nonuser-developers: 363 companies and 153 independent developers. They accounted for 60.1% of all developed apps. Thus, 396 developers (43.4%) were users: 292 healthcare professionals as well as 104 patients and caregivers. As outlined in Table 7.5, the number of developers and the number of developed apps correlated across the sample.

Table 7.5 The number of developers and the number of developed apps: The 2018 data Companies

Independent developers

Non-user-developers

Healthcare professionals

Patients and caregivers

User-developers

No. of developers

363 (0.398)

153 (0.168)

292 (0.320)

104 (0.114)

No. of developed apps

537 (0.425)

224 (0.177)

370 (0.292)

134 (0.106)

Note: n = 1,265 apps developed by 912 developers

7.1 Findings Concerning the Analytical App Data

71

The percentage of free apps was highest for companies (71.1%), followed by patients and caregivers (55.2%), healthcare professionals (43.5%), and independent developers (30.8%). The mean price for paid apps varied between U.S. $7.56 for companies, U.S. $5.31 for healthcare professionals, U.S. $4.61 for patients and caregivers, and U.S. $2.94 for independent developers (see Table 7.6). Standard deviation of the price was highest for company-developed apps and lowest for apps developed by independent developers. The percentage of in-app purchases was highest for apps developed by patients and caregivers (25.4%) and lowest for healthcare professionals (11.1%). Table 7.6 App pricing: The 2018 data % of free apps

Mean price of paid apps (U.S. $)

S.D. price of paid apps (U.S. $)

Median price of paid apps (U.S. $)

% of in-app purchases

1 Companies

0.711

7.56

13.34

2.99

0.142

2 Independent developers

0.308

2.94

3.45

1.99

0.196

3 Healthcare professionals

0.435

5.31

8.57

2.99

0.111

4 Patients and caregivers

0.552

4.61

6.93

2.99

0.254

Notes: n = 1,265 apps

Similar to the 2014 data, Mann-Whitney U-tests regarding the percentage of free apps revealed that companies offered apps significantly more (p < 0.000) for free than patients and caregivers. Again, there was no significant difference (p = 0.270) in price for paid apps between companies and patients and caregivers. In H1a, I posited that patients and caregivers revealed their innovations freely more often than companies. I found no evidence for this hypothesis in the 2014 or in the 2018 dataset; thus, H1a had to be rejected. In H1b, I anticipated that patients and caregivers reveal their innovations more often at a lower price than companies. Again, I found no evidence for this hypothesis in the two datasets; thus, H1b had to be rejected. The analysis of market tenure concerning the developer type revealed that apps developed by healthcare professionals were on average in the market the longest (1,744 days or 4 years and 9.5 months), followed by patients and caregivers (1,595 days), companies (1,458 days), and independent developers (1,339 days). Apps developed by healthcare professionals had on average a significantly longer tenure

72

7

Findings

(p < 0.001) than apps developed by companies or independent developers, while apps developed by patients were in the market on average significantly longer (p = 0.004) than apps developed by independent developers (Table 7.7). Table 7.7 Market tenure (days): The mean, median, S.D., and p-values: The 2018 data Mean

Median

S.D.

1 Companies

1,458

1,320

889

1

2

2 Independent developers

1,339

1,159

867

0.086

3 Healthcare professionals

1,744

1,710

909

0.000***

0.000***

4 Patients and caregivers

1,595

1,536

868

0.088

0.004**

3

0.104

Notes: *p < 0.05; **p < 0.01; ***p < 0.001. n = 1,265; I used Mann-Whitney U-tests, since the data were not normally distributed

Regarding the user ratings, I found that (similar to the 2014 data) patients and caregivers received on average the highest ratings regarding mean (4.18 stars) and median value (4.47 stars). Second, healthcare professionals reached on average 3.88 stars, followed by independent developers (3.54 stars). The differences were highly significant (p < 0.001) between patients and caregivers and companies, and between healthcare professionals and companies (Table 7.8). Similar results were obtained from the comparison of independent developers to the two user-developer groups. Further, there was a significant difference between patients and caregivers and healthcare professionals (p = 0.015). The overall sample contained 841,167 user ratings. Table 7.8 Ratings: The mean, median, S.D., and p-values: The 2018 data Mean

Median

S.D.

1 Companies

3,502

3,500

0.950

1

2

2 Independent developers

3,538

3,928

1.100

0.385

3 Healthcare professionals

3,883

4,000

0.918

0.000***

0.008**

4 Patients and caregivers

4,184

4,474

0.620

0.000***

0.000***

3

0.015*

Notes: *p < 0.05; **p < 0.01; ***p < 0.001. n = 837; 428 apps without ratings were excluded from this analysis. I used Mann-Whitney U-tests, since the data were not normally distributed. In total, 841,167 ratings were analyzed

Next, I analyzed the download estimations regarding the four developer types. As downloads vary considerably between paid and free apps (Garg and Telang

7.1 Findings Concerning the Analytical App Data

73

2013), I differentiated between these two subgroups for analyzing download estimates. I found that mean values ranged from 180 downloads in Q2 2018 for independent developers (median = 106 downloads) to an average 576 downloads per quarter for patients and caregivers (median = 122 downloads). Healthcare professionals (mean = 242 downloads, median = 126 downloads) achieved significantly more downloads than independent developers (Table 7.9).

Table 7.9 The download estimations for Q2 2018 (paid apps): The mean, S.D., median, and p-values Mean

Median

S.D.

1

2

3

1 Companies

203

115

500

2 Independent developers

180

106

517

0.183

3 Healthcare professionals

242

126

419

0.535

0.029*

4 Patients and caregivers

576

122

3,020

0.922

0.358

0.463

Notes: *p < 0.05; **p < 0.01; ***p < 0.001. n = 579; I used Mann-Whitney U-tests, since the data were not normally distributed

The download numbers for free medical apps in Q2 2018 were about 20 times higher than for paid apps: patients and caregivers received on average 10,218 downloads (median = 1,538), followed by companies (mean = 5,974), independent developers (mean = 4,264), and healthcare professionals (mean = 4,019). As depicted in Table 7.10, patients and caregivers as well as companies had significantly more downloads than healthcare professionals (p = 0.010 and p = 0.008, respectively).

Table 7.10 The download estimations for Q2 2018 (free apps): The mean, S.D., median, and p-values Mean

Median

S.D.

1

2

1 Companies

5,974

1,200

26,141

2 Independent developers

4,264

866

9,659

0.306

3 Healthcare professionals

4,019

832

19,183

0.008**

0.482

4 Patients and caregivers

10,218

1,538

24,762

0.245

0.116

3

0.010*

Notes: *p < 0.05; **p < 0.01; ***p < 0.001. n = 686. I used Mann-Whitney U-tests, since the data were not normally distributed

74

7

Findings

As the mean values and the median values showed a huge difference, and as the S.D. values were also high, I calculated the skew of the download estimations: for paid apps, the skew was 17.75, for free apps, it was 10.83. This indicates that few apps dominated the mean download numbers: for paid apps, there were 144,547 downloads in total in Q2 2018. Of those, 107,741 downloads (75%) were associated with 20% of the apps. For free apps, there were 3,979,372 downloads in total in Q2 2018. Of those, 3,339,758 downloads (84%) were associated with 20% of the apps. Figure 7.1 depicts the download numbers associated with the top 300 paid and free apps, respectively. Smartphone app revenue is generated by app sales, in-app purchases, and inapp adverts. Since in-app adverts can’t be tracked by the platform provider, they couldn’t be considered for the revenue estimations. Thus, the revenue estimations were based on app sales and in-app purchases. In section 5.2 I showed that, in a sample of different app types, app sales and in-app purchases accounted for more than 80% of all revenues, with in-app purchases being responsible for the by far largest fraction of revenues. I assume that there is a similar pattern within medical apps. Thus, I analyzed the revenue estimations for apps with and apps without in-app purchases separately.

Downloads in Q2 2018 (log scale) 1,000,000 paid free 100,000

10,000

1,000

100

Figure 7.1 The download estimations Q2 2018 for paid and free apps (top 300 apps, log scale)

Regarding the apps with in-app purchases (see Table 7.11), I found that patients and caregivers had by far the highest value for revenue: the mean value was

7.1 Findings Concerning the Analytical App Data

75

U.S. $61,344 for Q2 2018 (median = U.S. $9,412). Ranked second, independent developer had a mean revenue of U.S. $24,821 (median = U.S. $499), followed by healthcare professionals who received on average U.S. $20,632 (median = U.S. $1,789). Companies had the lowest average revenue (mean = U.S. $10,492, median = U.S. $1,755). The mean values of patients and caregivers’ revenue are significantly higher than those of companies (p = 0.007), independent developers (p = 0.001), and healthcare professionals (p = 0.092). Table 7.11 The revenue estimations for Q2 2018 (apps with in-app purchases) in U.S. $: The mean, S.D., median, and p-values Mean

Median

S.D.

1

2

1 Companies

10,492

1,755

25,294

2 Independent developers

24,821

499

101,464

0.193

3 Healthcare professionals

20,632

1,789

43,011

0.494

0.109

4 Patients and caregivers

61,344

9,412

121,379

0.007***

0.001**

3

0.092**

Notes: *p < 0.05; **p < 0.01; ***p < 0.001. n = 195. I used Mann-Whitney U-tests, since the data were not normally distributed

The apps without in-app purchases (see Table 7.12) had on average about 2% of the revenues compared to apps with in-app purchases: patients and caregivers were ranked highest (mean = U.S. $1,140), followed by healthcare professionals (mean = U.S. $482), companies (mean = U.S. $322), and independent developers (mean = U.S. $195). Since those apps included a large set of free apps, the median value was U.S. $0.00 for three of the four developer groups. The mean revenue for apps without in-app purchases were significantly higher for patients and caregivers than for independent developers (p = 0.048) and companies (p < 0.001). The mean revenue for healthcare professionals was also significantly higher than for companies (p < 0.001). The abovementioned data indicated substantial differences between free and paid apps, as well as between apps with and without in-app purchases. In preparation for the regression analysis in section 7.1.4.4, I split the dataset of all apps into potentially revenue-targeting apps (i.e. apps that are not offered for free and/or apps with in-app purchases) and truly free apps (i.e. apps that are offered for free and without in-app purchases). Since the regression analysis included the rating as a mediating variable, only apps with at least one rating were included.

76

7

Findings

Table 7.12 The revenue estimations for Q2 2018 (apps without in-app purchases) in U.S. $ : The mean, S.D., median, and p-values Mean

Median

S.D. 2,456

1 Companies

322

0.00

2 Independent developers

195

18.50

3 Healthcare professionals

482

4 Patients and caregivers

1,140

1

2

365

0.00***

0.00

1,989

0.00***

0.314

0.00

8,443

0.00***

0.048*

3

0.239

Notes: *p < 0.05; **p < 0.01; ***p < 0.001. n = 1,070. I used Mann-Whitney U-tests, since the data were not normally distributed

The data from Table 7.13 is visualized in Figure 7.2 using the variables number of downloads and revenue: on this log scale graph, truly free apps are on the x-axis, while revenue-targeting apps are on an approximately linear curve.

Table 7.13 Revenue-targeting apps and truly free apps (upper left part) Without in-app purchases

With in-app purchases

Sum

Free apps

381 (0.455)

141 (0.168)

522 (0.624)

Companies

231 (0.276)

59 (0.070)

290 (0.346)

Independent developers

20 (0.024)

28 (0.033)

48 (0.057)

Healthcare professionals

96 (0.115)

24 (0.029)

120 (0.143)

Patients and caregivers

34 (0.041)

30 (0.036)

64 (0.076)

Paid apps

296 (0.354)

19 (0.023)

315 (0.376)

Companies

86 (0.103)

7 (0.008)

93 (0.111)

Independent developers

62 (0.074)

5 (0.006)

67 (0.080)

Healthcare professionals

108 (0.129)

6 (0.007)

114 (0.136)

Patients and caregiver

40 (0.048)

1 (0.001)

41 (0.049)

Sum

677 (0.809)

160 (0.191)

837 (1.000)

Notes: n = 837; 428 apps without ratings were excluded from this analysis

Thus, differentiating between revenue-targeting apps and truly free apps is important to further estimate the success of user-developed vs. non-userdeveloped medical apps.

7.1 Findings Concerning the Analytical App Data

77

Revenue 1,000,000 Companies Independent developers 100,000

Healthcare professionals Paents or caregivers

10,000

1,000

100

10

1 1

10

100

1,000

10,000

100,000 1,000,000 Downloads

Figure 7.2 Number of downloads vs. revenue (log scale)

7.1.3

The Emergence of Medical Apps: A Comparison Between Datasets 1 (2014) and 2 (2018)

In this chapter, the datasets I presented in the two preceding chapters were compared regarding the developer type and few other parameters available in both datasets. As the estimations for download numbers and revenue were only available in dataset 2 (2018), these data could not be compared. The number of developed apps is summarized in Table 7.14. The relative number of companies is almost the same (41.1% vs. 42.5%), while the number of apps developed by healthcare professionals decreased from 40.9% to 29.2%. Independent developers as well as patients and caregivers developed more apps in 2018 than in 2014 (11.1% vs. 17.7% and 6.9% vs. 10.6%, respectively). In 2018, there were 253 apps available that were already available in the 2014 dataset. The remaining 1,012 apps were added later to the App Store or were not among the top 1,000 in the three target countries Germany, the UK, or the U.S. in 2014. Surprisingly, the relative number of patient-developed apps available in both

78

7

Findings

Table 7.14 A comparison of the datasets: Developed apps 2014

2018

Apps in both datasets

n

%

n

%

n

% of 2014

Companies

492

0.411

537

0.425

90

0.183

Independent developers

132

0.111

224

0.177

17

0.129

Healthcare professionals

487

0.409

370

0.292

106

0.218

81

0.069

134

0.106

40

0.494

1,192

1

1,265

1

253

0.212

Patients and caregivers

datasets (49%) exceeded the relative number of healthcare professionals (22%), companies (18%), and independent developers (13%). The analysis of six app parameters revealed that the mean ratings were very similar between the two datasets (Table 7.15). The number of ratings per app and the market tenure (displayed in years) significantly increased over time (p < 0.001 in both cases). As the apps in 2018 were available for longer in the App Store than those of 2014 and the number of users had increased, the number of ratings increased. The average price of all paid apps has significantly decreased (p < 0.001) from U.S. $5.65 to U.S. $4.06, and the number of free apps significantly increased (p < 0.001) by 4 percentage points. The number of in-app purchases significantly increased (p < 0.000) by 4 percentage points.

Table 7.15 A comparison of datasets: App parameters (ratings, no. of ratings, price of paid apps, market tenure, % of free apps, and % of in-app purchases) 2014 mean

2018 S.D.

mean

p-value S.D.

Rating

2.51

1.89

2.45

1.91

0.399

No. of ratings

214

1,758

665

3,260

0.000***

Price of paid apps

5.65

11.28

4.06

8.44

0.000***

Market tenure

2.35

1.51

4.26

2.50

0.000***

% of free apps

0.50

–

0.54

–

0.000***

% of in-app purchases

0.11

–

0.15

–

0.004**

Notes: *p < 0.05; **p < 0.01; ***p < 0.001. I used Mann-Whitney U-tests, since the data were not normally distributed

7.1 Findings Concerning the Analytical App Data

79

Next, I analyzed the release year of all apps for both datasets, looking at the emergence of medical apps. The Apple App Store initially opened on July 10, 2008 for developers to upload and share their smartphone apps. The release data of each app is indicated in both datasets. However, as apps that were removed from the App Store are no longer visible, I could only analyze apps within the top 1,000 apps (and thus included in the datasets). The findings must be analyzed with this constraint in mind: Analyzing the release date of the apps (2014 dataset) in my sample revealed that healthcare professionals initially uploaded 22 apps, followed by five company-developed apps, two apps developed by patients and caregivers, and one by an independent developer (Table 7.16). Overall, 80% of apps released in 2008 were developed by users. In 2009, the user innovation ratio (UIR) shrunk to 50% and remained between 46% and 48% until 2013. The absolute number of released apps was fairly low in the early years owing to the low diffusion of smartphones at the time. The number of newly released apps rose in the next few years: in 2010, companies and independent developers accounted for 55% of all released apps. The compound annual growth rate (CAGR) of all newly released apps was 58%.

Table 7.16 The release year of the apps: The 2014 data 2008

2009

2010

2011

2012

2013

1 Companies

5

40

82

95

118

113

2 Independent developers

1

12

18

19

22

50

3 Healthcare professionals

22

39

74

87

113

113

4 Patients and caregivers

2

12

8

11

16

24

User innovation ratio

0.80

0.50

0.45

0.46

0.48

0.46

Notes: n = 1,096; 96 apps released in early 2014 were excluded from the analysis

Analyzing the second dataset (2018) revealed a similar picture: Still, users accounted for the majority of initially released apps in 2008 (Table 7.17). However, in the 2018 data, the number of apps released in 2008 decreased from 30 to 10. This indicates that particularly healthcare professionals have withdrawn their apps from the App Store or that these apps are no longer listed in the Top 1,000 in one of the three countries. As the ratio of user-developed apps was very similar between both datasets, the loss of apps seemed prevalent across all developer types.

80

7

Findings

Table 7.17 The release year of the apps: The 2018 data 2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

1C

2

21

31

42

54

55

66

92

75

72

2 ID

1

8

16

9

8

19

39

34

46

32

3 HP

6

18

37

37

42

51

45

48

45

32

4 P&C

1

6

9

13

13

19

14

20

22

12

UIR

0.70

0.45

0.49

0.50

0.47

0.49

0.36

0.35

0.36

0.30

Notes: n = 1,212; 53 apps that were released in early 2018 were excluded from the analysis

After 2013, the ratio of user-developed apps is decreasing from an average of 48% (2009 to 2012) to an average of 33% (2013 to 2017). Yet, there were on average 17 new apps released per year by patients and caregivers, and 43 by healthcare professionals. Companies released about 76 new apps per year between 2014 and 2017. In H2, I posited that the early development of medical smartphone apps was triggered by user-developers. Based on the findings from both datasets, the ratio of user-developed apps compared to non-user-developed apps was between 70% and 80% in first year of the App Store and subsequently fell a little below 50% in 2012 and finally to 30% in 2017. Thus, H2 is supported.

7.1.4

Regression Analysis of Dataset 2 (2018)

The regression analysis had four parts. First, and in preparation for the regression analysis, the correlation of the variables was analyzed. Second, the mediated regression analysis was conducted. Third, the mediated regression analysis was conducted again, considering the four individual categorical values for the exogenous variable developer type. Finally, to shed some light on another aspect of the app analysis, I subdivided the dataset into apps that seek to generate revenue and apps that do not.

7.1.4.1 Correlation analysis Table 7.18 provides the correlations between the variables used in the statistical analysis. First, the findings of the descriptive analysis are supported with respect to the developer type. There is a significant correlation between rating, number

7.1 Findings Concerning the Analytical App Data

81

of downloads, and revenue as well as between market tenure and the developer type. There is no significant correlation between price and the developer type. Second, the rating of an app was significantly negatively correlating with the number of ratings and with the presence of in-app purchases. This indicates that apps with a high number of ratings had on average lower ratings than apps with a lower number of ratings. Third, the download numbers of an app were significantly positively correlated with developer type, ratings, revenues, prices, and were significantly negatively correlated with in-app purchases. Thus, as stated in the literature, this means that better ratings increased the download numbers (Kübler et al. 2018). Further, the correlations indicated that downloads may mediate between developer type, ratings, and revenue. Fourth, app revenue correlates significantly positively with the two revenuerelated control variables price and in-app purchases as well as with the mediating variable number of downloads, the independent variable developer type, and the control variable market tenure. Finally, the correlations between the control variables and the mediating variables as well as the dependent variable indicated that it is beneficial to also control for these variables.

7.1.4.2 Mediated Regression Analysis The model coefficients for the mediated regression analysis are listed in Table 7.19. The variables number of ratings, number of downloads, and revenue were transformed owing to non-normality of the data (see section 6.1.3.1). Thus, the coefficients had to be analyzed with this transformation in mind. The variances of the mediating variables (R2 (ratings) = 16.9%, R2 (number of downloads) = 49.6%) and of the endogenous variable (R2 (revenue) = 40.0%) were explained to a very satisfactory amount (Field 2009). I found that developer type had, as hypothesized, a significant (p < 0.001) impact on an app’s ratings. Regarding the regression coefficient, I found that developer type had the second largest coefficient (coeff. = 0.223) after the transformed variable number of ratings (coeff. = 1.959). The positive significant regression coefficient of price indicated that more expensive apps had better ratings. The negative coefficient of the variable market tenure indicated a weak yet significant negative effect of older apps on their ratings. Regarding the downloads, I found that developer type had a significant negative (p = 0.019) impact on the number of downloads. Thus, H4a can be accepted. Ratings had a significantly (p = 0.003) negative impact on number of downloads. An analysis of the categorical values, as shown below, explain this pattern more

0.017

0.039

0.099**

Revenues

Prices

With IAP

Tenure

Notes: *p < 0.05; **p < 0.01

−0.016

0.198**

Downloads

No. rating

0.247**

0.088*

Rating

Developers

Developer

0.052

0.287**

0.116** 0.329**

−0.042 −0.292**

0.286** 0.299**

0.111***

Revenue

−0.102**

0.426**

Download

0.095**

0.037

0.051

−0.032

Rating

Table 7.18 The correlation of variables used in the statistical analysis

0.132**

0.203**

−0.054

Price

−0.243**

−0.050

With IAP

−0.178**

Tenure

No. rating

82 7 Findings

7.1 Findings Concerning the Analytical App Data

83

in some detail. Further, there are also some expectable results. Number of ratings had a significant positive (p < 0.001) impact, while price had a significant negative (p < 0.001) impact. Intuitively, apps that are offered for a higher price have lower download numbers than cheaper apps. The availability of in-app purchases had no significant effect on the two mediator variables. Regarding the endogenous variable revenue, I found that number of downloads related significantly negatively (p < 0.001) to revenue. This rather counter-intuitive result is further analyzed below. Price and in-app purchases both had, as expected, a significant positive impact (p < 0.001) on revenue. The about 25 times higher regression coefficient for in-app purchases indicated that the implementation of in-app purchases had a far greater effect on revenue than a change in price. Next, the mediating effect of ratings and number of downloads were analyzed. The indirect effects were determined using bootstrapping (Hayes 2017). I used 5,000 bootstrap samples to calculate the indirect effect on a 95% confidence interval. The two indirect effects are shown in Figure 7.3: Indirect effect 1 analyzed the mediating effect of developer type via both ratings and number of downloads on revenue. Indirect effect 2 analyzed the indirect effect of developer type on revenue only via number of downloads. The results of the mediation analysis are listed in Table 7.20. Indirect effect 1 (via only number of downloads) is about three times as strong as indirect effect 2 (via ratings and number of downloads). Thus, as the confidence interval touches zero (BootLLCI is the lower limit of the confidence interval and BootULCI is the upper limit of the confidence interval), the effect can be considered as nonsignificant of a 95% confidence interval (Hayes 2017). Thus, H5a, which proposed a mediating effect of developer type on revenue, had to be rejected. The results of this first mediation analysis are summarized in Figure 7.4. The continuous arrows indicate significant correlation between two variables, and the dotted arrows no significant correlation. The direction and the significance level of each significant effect were also indicated. To better understand the significantly negative effects of ratings on number of downloads and of number of downloads on revenue, I conducted the same analysis again including categorical values for the four developer types.

7.1.4.3 Mediated Regression Analysis, Including Categorical Values for the Exogenous Variable In this mediated regression analysis, I analyzed the four developer types in some detail. I used company-developed apps as a baseline and compared the other three developer groups separately to this baseline data. The data is provided in

0.0000

−0.0001

1.9589

0.0178

0.0168

2.3784

Market tenure

No. of ratings

Price

In-app purchases

Constant

0.0000

0.8336

0.0016

0.0000

0.0000

–

–

0.0000

p 0.0002

−0.0007

0.0010

R2 = 0.497 F(6, 830) = 136.482 p < 0.001

1.0162

0.0005

0.0000

−0.0002 0.0006

0.0014

0.0000

0.0338

0.0000

–

0.0002

−0.0004 –

S.E.

No. of downloads Coeff.

0.0000

0.2627

0.0000

0.0000

0.0000

–

0.0027

0.0188

p

0.1574

0.0030

0.0002

–

R2 = 0.400 F(3, 833) = 184.962 p < 0.001

0.6332

0.0615

0.0025

–

–

0.1533

−0.5937 –

–

–

S.E.

Revenue

–

–

Coeff.

–

–

p

0.0001

0.0000

0.0000

–

–

0.0001

Endogenous variable

Notes: n = 837; 428 apps without ratings were excluded from this analysis. The data of the variables number of ratings, number of downloads, and revenue was transformed

R2 = 0.164 F(5, 831) = 32.667 p < 0.001

0.1294

0.0801

0.0056

0.2077

–

–

No. of downloads

–

0.0275

–

0.2235

S.E.

Ratings

Ratings

Developer type

Coeff.

Mediator variables

Table 7.19 The model coefficients for the mediated regression analysis

84 7 Findings

7.1 Findings Concerning the Analytical App Data Exogenous variable

App variables

85 Revenue variables

Endogenous variable

Download

Revenue

Download

Revenue

Indirect effect 1 Developer

Ratings

Indirect effect 2

Developer

Ratings

Mediator variable

Figure 7.3 The mediation analysis: A visualization of indirect effects 1 and 2

Table 7.20 The mediation analysis: Indirect effects Effect

BootSE

BootLLCI

BootULCI

Significance

Indirect effect 1

0.0003

0.0002

0.0000

0.0006

no

Indirect effect 2

0.0001

0.0000

0.0000

0.0002

no

Notes: n = 837; 428 apps without ratings were excluded from this analysis. Indirect effects were determined using bootstrapping (5,000 bootstrap samples, 95% confidence interval)

Table 7.21. A slight increase of R2 can be observed, indicating that this model explains more of the variance of the two mediating variables. The results indicate that ratings of healthcare professionals as well as patients and caregivers were significantly higher (p < 0.001) than the ratings of companies, supporting H3b. The regression coefficients of the control variables changed slightly, but the significance levels remained stable for all of them. Regarding the number of downloads, both independent developers as well as patients and caregivers had a significantly negative impact (p < 0.001) on number of downloads compared to the baseline data (companies). Since there was no significant effect for healthcare professionals compared to the baseline concerning downloads, H4b is rejected.

86

7 Exogenous variable

App variables

Revenue variables

No. of ratings

Price

***

+*** +** Ratings -***

Control variable

Mediator variable

+*** -**

Endogenous variable

+***

-** Developer

Findings

-***

Download

-**

Revenue

+***

Market tenure

In-app purchases

R2 = 16.9% (Ratings)

R2 = 49.6% (Downloads)

+***

R2 = 40.0% (Revenue)

Figure 7.4 Model overview: The mediated regression analysis

The results for the endogenous variable revenue remained the same, since there was no direct effect from developer type on revenue. The analysis of the mediating effect (Table 7.22) revealed further details regarding the indirect effects of developer type on revenue. I found that indirect effect 1 (from developer type on revenue via ratings and number of downloads) was insignificant for any of the three user types compared to the baseline. However, the bootstrap analysis of indirect effect 2 (developer type on revenue via number of downloads) highlighted that patients and caregivers as well as independent developers had significant positive indirect (p < 0.05) effects on revenue. This confirmed that the developer type patients and caregivers mediated revenue positively via number of downloads, confirming H5b. While the abovementioned data on the mediated regression analysis yielded sufficient results to answer H3, H4, and H5, it remains unclear why a higher number of downloads resulted in lower app revenue. Thus, I further analyzed the data by splitting the dataset into two subsets.

7.1.4.4 Mediated Regression Analysis: Revenue-Targeting Apps vs. Truly Free apps In section 7.1.2, I elaborated on two app subtypes. First, revenue-targeting apps, i.e. apps where the developer intends to receive money from the app’s users. This category comprises apps that are not available for free as well as apps (free or

2.5800

In-app purchases

Constant

0.0056 0.1255

0.0805 0.0000

0.6384

0.0018

0.0000

0.0000

–

–

0.0000

0.0000 0.0002

−0.0008

0.0000 0.0010

= 0.519 F(8, 28) = 111.750 p < 0.001 R2

1.0160

0.0005

−0.0002 0.0010

0.0014

0.0000

0.0346

0.0000

–

0.0006

−0.0025 –

0.0005

0.0002

S.E. 0.0006

Coeff. −0.0032

p

0.0000

0.0585

0.0000

0.0000

0.0000

–

0.0004

0.0001

0.6167

0.0000

0.1574

0.0030

0.0002

–

= 0.400 F(3, 33) = 184.962 p < 0.001 R2

0.6332

0.0615

0.0025

–

–

0.1533

−0.5937 –

–

–

–

–

S.E.

–

–

–

–

Coeff.

Revenue

–

–

–

–

p

0.0001

0.0000

0.0000

-

–

0.0001

Endogenous variable

Notes: n = 837; 428 apps without ratings were excluded from this analysis. The data of the variables number of ratings, number of downloads, and revenue were transformed. The categorical values were measured against the baseline value (company)

R2

p 0.2714

Number of downloads

Mediator variables

= 0.170 F(7, 829) = 24.234 p < 0.001

0.0175 0.0379

Prices

2.0268

No. of ratings

0.2138

– 0.0000

–

–

0.0977

0.0749

−0.0002

No. of downloads

S.E. 0.0942

Market tenure

–

0.5717

Patients and caregivers

Ratings

0.5367

Healthcare professionals

Coeff. 0.1037

Independent developers

Ratings

Table 7.21 The model coefficients for the mediated regression analysis, including categorical values for the exogenous variable

7.1 Findings Concerning the Analytical App Data 87

88

7

Findings

Table 7.22 The mediation analysis: The indirect effects, including categorical values Effect

BootSE

BootLLCI

BootULCI

Significance

Independent developers

0.0000

0.0001

0.0000

0.0002

no

Healthcare professionals

0.0003

0.0001

0.0000

0.0005

no

Patients and caregivers

0.0003

0.0001

0.0000

0.0005

no

Independent developers

0.0019

0.0008

0.0002

0.0034

yes

Healthcare professionals

−0.0001

0.0003

−0.0007

0.0004

no

Patients and caregivers

0.0015

0.0007

0.0001

0.0028

yes

Indirect effect 1

Indirect effect 2

Notes: n = 837; 428 apps without ratings were excluded from this analysis. Indirect effects were determined using bootstrapping (5,000 bootstrap samples, 95% confidence interval). The categorical values were measured against the baseline value (company)

paid apps) that offer in-app purchases. Although some of these apps may not receive revenue (because the app is not downloaded or because the users don’t buy in-app purchases), the developer intended to receive revenue with this app. On the other hand, there are truly free apps, i.e. apps that are completely free (no paid apps and all without in-app purchases). As depicted in Table 7.1, there were 381 truly free apps and 456 revenue-targeting apps in the dataset. Since the developers of these 381 apps had no intention to generate revenue, I analyzed these apps separately. Thus, I assumed that their major aim was to generate a high number of downloads and thus reach a huge user set. This results in a simplified mediation regression analysis (see Figure 7.5) with only one mediator (ratings) and number of downloads as the endogenous variable. The revenue-related variables price and in-app purchases are constant and were therefore omitted. For the analysis of truly free apps vs. revenue-targeting apps, I first present the results of the mediated regression analysis without the categorical values (Table 7.23) and then with the categorical values (Table 7.24). Analogous to the previous section, company-developed apps were the baseline for this analysis. Developer type had a significant impact (p < 0.001) on ratings of a truly free medical app. Number of ratings had a significantly positive impact (p < 0.001) on ratings. Regarding number of downloads, which serve as the endogenous variable in this model, there were no significant effects concerning developer type. However, ratings had a significant negative impact (p = 0.022) on download numbers. This indicates that the apps that are downloaded most often are not necessarily

7.1 Findings Concerning the Analytical App Data

Exogenous variable

App variables

89

Endogenous variables

No. of ratings

Developer

Control variable

Ratings

Mediator variable

Download

Market tenure

Figure 7.5 Simplified model for mediated regression analysis of truly free apps

the apps that received the highest ratings. Number of ratings and market tenure both had a significant positive impact (p < 0.001) on downloads. This indicates that apps that are available for longer and that have received many ratings are more likely to be downloaded. The analysis of the categorical values for truly free apps (Table 7.24) revealed that user-developed apps, namely apps developed by healthcare professionals as well as patients and caregivers had significantly higher (p < 0.001) ratings than apps developed by companies. There was no significant difference between apps developed by independent developers and the baseline. The remaining effects were very similar to the effects mentioned in Table 7.23. The analysis of the mediating effects (indirect effect of developer type on number of downloads via ratings) revealed a significantly negative effect (p < 0.05) of user-developed apps (patients and caregivers as well as healthcare professionals) compared to the baseline (company-developed apps) (Table 7.25). Thus, counterintuitively, companies is the developer type that is most successful concerning truly free apps. I then analyzed the revenue-targeting apps using the same mediated regression analysis as in section 7.1.4.3. I found a significantly positive impact of healthcare professionals (p < 0.001) as well as patients and caregivers (p = 0.010) on ratings.

90

7

Findings

Table 7.23 The model coefficients for the mediated regression analysis (truly free apps)

Developer type

Mediator variable

Endogenous variable

Ratings

Number of downloads

Coeff.

S.E.

p

0.2671

0.0418

0.0000

−0.0002

0.0002

0.4184

–

–

–

−0.0006

0.0003

0.0223

Ratings

Coeff.

S.E.

p

−0.0001

0.0001

0.1185

0.0000

0.0000

0.0000

Number of ratings

1.0939

0,3031

0.0003

0.0294

0.0016

0.0000

Constant

2.6497

0.1926

0.0000

1.0187

0.0012

0.0000

Market tenure

R2 = 0.121

R2 = 0.493

F(5, 377) = 17.318 p < 0.001

F(6, 376) = 91.401 p < 0.001

Notes: n = 381; 428 apps without ratings and 456 revenue-targeting apps were excluded from this analysis. The data of the variables number of ratings, number of downloads, and revenue were transformed

Table 7.24 The model coefficients for mediated regression analysis (truly free apps), including categorical values for the exogenous variable Mediator variable

Endogenous variable

Ratings

Number of downloads

Coeff.

S.E.

p

Coeff.

S.E.

p

Independent developers

0.1757

0.2082

0.3991

−0.0015

0.0011

0.1654

Healthcare professionals

0.5792

0.1100

0.0000

0.0001

0.0006

0.8657

Patients and caregivers

0.7387

0.1640

0.0000

−0.0011

0.0009

0.1827

–

–

–

−0.0006

0.0003

0.0184

−0.0001

0.0001

0.1062

0.0000

0.0000

0.0000

Number of ratings

1.1226

0.3123

0.0004

0.0297

0.0016

0.0000

Constant

2.9058

0.1842

0.0000

1.0185

0.0012

0.0000

Ratings Market tenure

R2 = 0.123

F(5, 375) = 10.476 p < 0.001

R2 = 0.497

F(6, 374) = 61.660 p < 0.001

Notes: n = 381; 428 apps without ratings and 456 revenue-targeting apps were excluded from this analysis. The data of the variables number of ratings, number of downloads, and revenue were transformed. The categorical values are measured against the baseline value (company)

7.1 Findings Concerning the Analytical App Data

91

Table 7.25 The mediation analysis: Indirect effects, including categorical values for truly free apps Effect

BootSE

BootLLCI

BootULCI

Significance

Independent developers

−0.0001

0.0002

−0.0004

0.0002

no

Healthcare professionals

−0.0004

0.0002

−0.0007

−0.0001

yes

Patients and caregivers

−0.0005

0.0002

−0.0009

−0.0001

yes

Indirect effect 1

Notes: n = 381; 428 apps without ratings and 456 revenue-targeting apps were excluded from this analysis. Indirect effects were determined using bootstrapping (5,000 bootstrap samples, 95% confidence interval). The categorical values were measured against the baseline value (company)

However, the regression coefficients for ratings were about half the value of the regression coefficients for truly free apps. This indicates that the quality difference was much larger for truly free apps than for revenue-targeting apps. The negative coefficients of market tenure and in-app purchases indicate that apps that were newer on the App Store and apps without in-app purchases likely have better ratings. Number of ratings and price both had a significantly positive impact on the rating of a revenue-targeting app, indicating that apps with many reviews (no matter whether the review is good or bad) and more expensive apps are perceived as being of higher quality. Regarding the number of downloads for revenue-generating apps, I find that only apps developed by healthcare professionals had significantly more downloads (p = 0.0019) than apps developed by companies. There was no significant effect for patients and caregivers vs. companies or independent developers vs. companies. In contrast to truly free apps, there was no significant effect of ratings on downloads for revenue-targeting apps. App price did not influence the downloads, i.e. more expensive apps are also often downloaded. Market tenure, number of ratings, and in-app purchases positively influenced the number of downloads for revenue-targeting apps (Table 7.26). As expected, the revenue of revenue-targeting apps was significantly positively influenced by number of downloads and price—more downloads and a higher price result in more revenue. However, the presence of in-app purchases had a significant negative (p = 0.0002) impact on revenue. This puzzling result contradicts previous research (see section 5.2) and must be further elaborated on. A visual inspection of the dataset of revenue-targeting apps revealed that although all

3.1877

Number of ratings

2.4615

In-app purchases

Constant

0.0000

0.0092

0.0085

0.0000

0.0000

–

–

0.0100

0.0001

Coeff.

S.E.

0.0003

−0.0005

0.0013

0.0006

0.0000

0.0021

0.0000

= 0.583 F(8, 447) = 78.036 p < 0.001 R2

1.0092

0.0056

0.0000

0.0308

0.0000

–

0.0008

−0.0005 –

0.0006

0.0007

0.0020

0.0000

p

0.0000

0.0000

0.6317

0.0000

0.0000

–

0.1059

0.5134

0.0019

0.9495

= 0.574 F(3, 452) = 202.750 p < 0.001

0.1498

−3.2289 R2

0.0025

0.0001

–

–

0.1468

–

–

–

–

S.E.

−0.0094

0.0010

–

–

3.2284

–

–

–

–

Coeff.

Revenue

–

–

–

–

p

0.0000

0.0002

0.0000

–

–

0.0000

Endogenous variable

Notes: n = 456; 428 apps without ratings and 381 truly free apps were excluded from this analysis. The data of the variables number of ratings, number of downloads, and revenue were transformed. The categorical values are measured against the baseline value (company)

R2

p 0.3769

Number of downloads

Mediator variables

= 0.259 F(7, 448) = 24.234 p < 0.001

0.1707

0.0057 0.0974

0.0152 −0.2547

Price

0.2964

– 0.0000

– −0.0003

Number of downloads

–

0.1235

Market tenure

–

0.3196

Patients and caregivers

Ratings

0.4132

Healthcare professionals

0.1016

S.E. 0.1124

Coeff. −0.0994

Independent developers

Ratings

Table 7.26 The model coefficients for the mediated regression analysis (revenue-targeting apps), including categorical values for the exogenous variable

92 7 Findings

7.1 Findings Concerning the Analytical App Data

93

those developers aimed at generating revenue, not all of them succeeded in generating revenue (Figure 7.6). A substantial number of apps offered in-app purchases but generated zero revenue. There were 10 apps with a price but with zero downloads and thus zero revenue. There were 18 apps with partly considerable downloads and in-app purchases but no revenue. These apps failed to generate revenue, although the developer intended to do so. The total revenue generated in Q2 2018 was estimated to be about U.S. $5,161,000. The lower 50% of revenue-targeting apps accounted for less than 1% of this revenue.

Revenue 1,000,000 no in-app purchases in-app purchases

100,000 10,000 1,000 100 10 1 1

10

100

1,000

10,000

Downloads 100,000

Figure 7.6 Revenue-targeting apps: Apps with in-app-purchases and without in-app purchases (log scale)

Thus, although in-app purchases correlated positively with revenue for the overall model, I found a negative effect for revenue-targeting apps owing to the presence of a share of apps that generated no revenue, although this was intended by the developers. The analysis of the mediating effect for revenue-generating apps (Table 7.27) revealed that, for healthcare professionals, indirect effect 2 (developer type via

94

7

Findings

Table 7.27 The mediation analysis: Indirect effects, including categorical values for revenue-targeting apps Effect

BootSE

BootLLCI

BootULCI

Significance

Independent developers

0.0002

0.0002

−0.0003

0,0007

no

Healthcare professionals

−0.0006

0.0004

−0.0016

0.0001

no

Patients and caregivers

−0.0005

0.0004

−0.0013

0.0001

no

Independent developers

−0.0001

0.0023

−0.0047

0.0043

no

Healthcare professionals

0.0065

0.0021

0.0024

0.0108

yes

Patients and caregivers

−0.0016

0.0025

−0.0066

0.0033

no

Indirect effect 1

Indirect effect 2

Notes: n = 456; 428 apps without ratings and 381 truly free apps were excluded from this analysis. The categorical values were measured against the baseline value (company)

downloads on revenue) was positively significant (p < 0.05). Thus, only healthcare professionals developed revenue-generating apps that were financially more successful than those of companies.

7.1.5

Summary of the Findings: Analytical App Data

In section 7.1, I analyzed two large-scale, real-world datasets of analytical app data in order to draw conclusions on developer type’s impacts on app performance. Therefore, I clustered the app developers into four groups: companies and independent developers (both non-user-developers) as well as patients and their caregivers in one group and healthcare professionals in the last group (both user-developers). One dataset comprising 1,192 apps was gathered in mid2014, and the second dataset comprising 1,265 apps was gathered in mid-2018. Subsequently, both datasets were processed and analyzed similarly. A mediated regression analysis was applied to the 2018 dataset. Overall, I found that, in the field of medical smartphone apps, a significant number of apps was developed by user-developers (47.8% in the 2014 dataset and 39.9% in the 2018 dataset). Regarding H1a/b, I found no evidence that patients and caregivers revealed their medical apps more often freely or at a lower price than professional software developers (companies). Concerning H2, I found evidence that the development in the first years after the opening of the App Store of medical smartphone apps was mainly triggered by user-developers. Further,

7.1 Findings Concerning the Analytical App Data

95

H3a/b confirms that the developer type of an app had a significant influence on the ratings of an app. I found that apps developed by patients and caregivers had significantly better ratings than apps developed by other developers. Regarding H4a/b, I found that developer type impacted on medical apps’ download numbers; however, apps developed by patients and caregivers did not have higher download numbers than other developers. Regarding H5a/b, I could not confirm that developer type had a mediating impact on revenue of medical apps. However, when analyzing the categorical values only in the regression model, I found evidence that the developer type patients and caregivers had a significantly positively mediating impact on revenue of apps compared to company-developed apps. In sum, the analysis of the dataset 2 (2018) revealed support for H2, H3a, H3b, H4a, and H5b. I found no support for H1a, H1b, H4b, or H5a (see Table 7.28).

Table 7.28 A summary of the analytical app data analysis (an overview of the hypotheses) Significance Support for hypothesis H1a: Patients and caregivers freely reveal their medical apps more often than professional software developers (companies).

–

no

H1b: Patients and caregivers more often reveal their paid medical apps at a lower price than professional software developers (companies).

–

no

H2: The early development of medical smartphone apps was triggered by user-developers.

–

yes

H3a: Developer type impacts on the quality (operationalized by ratings) of medical apps.

p < 0.000

yes

H3b: Apps developed by patients and caregivers receive better ratings than apps developed by other developers.

p < 0.000

yes

H4a: Developer type impacts on medical apps’ download numbers.

p < 0.000

yes

H4b: Apps developed by patients and caregivers achieve higher download numbers than company-developed apps.

p < 0.000

no

H5a: Developer type has a mediating impact on medical apps’ revenue.

–

no

p < 0.05

yes

H5b: The developer type patients and caregivers has a significantly positively mediating impact on apps’ revenues, compared to company-developed apps

96

7

Findings

In the following chapter, a qualitative study using interview data with userdevelopers of medical apps will shed more light on these findings, particularly regarding the rather unexpected results concerning app pricing (H1a/b), the early development of the App Store (H2), and the superiority of user-developed apps regarding user ratings (H3a/b).

7.2

Findings of Qualitative Data on Medical App Developers

After gathering the first dataset on analytical data about user innovation in the field of medical smartphone apps, I conducted 16 semi-structured interviews with userdevelopers (six patients, seven caregivers, and three healthcare professionals1 ). This qualitative study is grouped into three main clusters and a concluding section, where remaining aspects will be presented and summarized. Some information on the interviewees and their apps are listed in Table 7.29. The sample developers were based in Germany, Austria, Switzerland, the U.S., and India. The aim of the interviews was to find complementary information to the analytical app dataset, particularly about the motivations for their innovative endeavors, their specific knowledge of and experience in the topic, supportive and required contextual factors from their perspectives, and their roles and activities in the innovation process.

7.2.1

Triggers for Innovative Endeavors

The initiations of the innovations differed, since patients were directly confronted with a disease, while caregivers only indirectly suffered from it. The sample patients showed strong personal motivations concerning their level of suffering, particularly those with chronic conditions such as diabetes and hypertension. Four of the interviewed patients were diabetics and received their diagnoses at a young age. They dealt with these health issues for many years and sought to help themselves in order to bring relief. Thus, their personal needs were at the origin of their innovative endeavors. In most cases, there was a daily demand for a solution. For instance, diabetics must document their food intake and must calculate insulin injections to regulate their blood glucose level. Hypertension patients must regularly monitor their blood pressure: 1 An

anesthetist, a general practitioner, and a physical therapist.

7.2 Findings of Qualitative Data on Medical App Developers

97

Table 7.29 A qualitative study on medical smartphone apps—characteristics of interviewees ID/role

Country of origin

Target disease

Patient 1

Germany

Diabetes

Patient 2

Germany

Hypertension

Patient 3

Germany

Diabetes

Patient 4

Austria

Diabetes

Patient 5

Germany

Multiple sclerosis (MS)

Patient 6

U.S.

Diabetes

Caregiver 1

Germany

Hypertension

Caregiver 2

Germany

Speaking aid

Caregiver 3

Germany

First aid

Caregiver 4

Germany

Poisoning treatment

Caregiver 5

Germany

Diabetes

Caregiver 6

India

Diabetes

Caregiver 7

Germany

Diabetes

Healthcare professional 1

Switzerland

Anesthesia management

Healthcare professional 2

Germany

Acupressure

Healthcare professional 3

Germany

Bruxism

Whenever I measured [my blood pressure] at work, I wrote an e-mail home in order to put the values into a table, but that was very annoying. (Patient 2)

The described health issues involve daily obligations that are very timeconsuming. A patient’s motivation to meet such obligations may vary over time. Thus, the need for a solution emerges, and becomes stronger every day. Caregivers’ innovative endeavors related to someone else’s personal needs. However, I found that curiosity and fun during development were two additional major motivational aspects that drove their development process. Initially, I had the intention to develop any kind of app; I just wanted to try it. Then my father was diagnosed with high blood pressure. […] This brought me to the field of hypertension management quickly. (Caregiver 1)

98

7

Findings

The triggers of the participating healthcare professionals were operational—they tried to simplify time-consuming tasks in their daily activities, such as dose calculation. Others aimed to inform patients about certain health topics so as to improve the medical encounter, which is hampered by time constraints during daily activities. In the early days of the App Store, users could choose between a few apps available online. However, the number of available apps increased rapidly. The interview data revealed that patients and caregivers who uploaded their app more recently already had a concept for the software in mind and then checked which existing solutions were available online. However, in all cases, the interviewees stated that they were dissatisfied with existing apps. This dissatisfaction—combined with their severe need—were the major triggers for their product development. Thus, all interviewees matched the conceptual description of lead users, as defined by von Hippel (von Hippel 1986). In section 7.1.3, I showed that the ratio of user-developed apps remained high between 2010 and 2015, although one could expect that some years after the founding of the App Store, there would be a sufficient number of apps available that suit most patients’ needs. However, the apps of the interviewees released after 2011 had some novelty compared to existing solutions. One developer connected existing PC software to a mobile app, developing the first cross-platform diabetes management app; another interviewee designed an app for children, opening a new market segment. One patient intended to make a very flexible app with many tracking possibilities to suit his needs, while one relative wanted to make a very streamlined and simple piece of software for his sister. My sister was diagnosed with diabetes type 1 some time ago. […] Then the idea came to make an app with a nicer, simpler design that is easier to use […]. There were a lot of apps where you could enter an incredible amount of data, but in the end you didn’t need it at all, you just want to get a rough overview on your data. (Caregiver 7)

Although the basic functions of all the sample apps were comparable to existing solutions, I found that every single new app had some novelty—for instance in terms of design, connectivity, target group, or by introducing gamification and thereby making diabetes monitoring into a game. Thus, all interviewees had good reasons to develop their own software, because there was no fitting solution available at the time.

7.2 Findings of Qualitative Data on Medical App Developers

7.2.2

99

Product Development

Most of the analyzed apps were developed privately, i.e. without external support. Only two healthcare professionals collaborated with a partner during the development process owing to a lack of programming skills. All other 14 interviewees had the skills and knowledge to develop a smartphone app; some were trained software architects, while others had a high affinity to software development and improved their skills during the development process. A patient who was not a trained software developer described his search process: Well, [my source of information was] basically a lot of googling. […] I had no professional help. The Internet is very rich. It is funny, developers are really free with sharing their ideas on blogs. (Patient 6)

Depending on an app’s scope and complexity, the development timeframe varied tremendously. It ranged from a few days to several months to finalize the app. The intensity expressed in hours per day or week also differed. At one extreme, a participant worked full-time in order to finish the development. The majority estimated the time they spent on developing the app to between 10 to 20 hours per week over several months. Of course this development process takes time and therefore costs you money. (Patient 6)

Owing to personal preferences, current knowledge, and limited resources, only a few apps were recreated for Android. Although the regulation of the app as a medical device demonstrates trustworthiness to its customers (Cortez et al. 2014), only two apps had been regulated. The majority did not pursue this option, since current regulatory frameworks bore too many administrative, financial, and procedural efforts.

7.2.3

Commercialization and Outcomes

Investigating the commercialization process, I found that some sample developers initially did not intend to upload their app to the App Store, but changed their decision during the course of private utilization. Mostly, these developers recognized their app’s usefulness; therefore, every participant uploaded the own app

100

7

Findings

to offer it to millions of smartphone users worldwide in order to not only increase the own, but also others’ quality of life. If it helps me, why should it not help the others as well? (Patient 3)

Most of the sample developers had set a price for downloading their app. However, owing to relatively low download rates, these fees were not a major revenue stream, but allowed them to compensate for their development costs (hardware, Apple developer account, etc.). I thought about the pricing for a long time—whether the app should be free or not. I invested several hundred hours of my free time—that’s why I didn’t really want to give it away for free. And then I decided to give it away for two, three euros or so. […] However, you don’t get incredibly rich… (Healthcare professional 1)

Yet, several interviewees stated that they did not want to give the app away for free, since the development took a lot of resources. This is in line with the findings from section 7.1, where I found that patient-developed apps are more often paid apps compared to company-developed apps. I didn’t want to do release the app for free because of course there is a lot of effort behind it. And it’s always good to have some extra money. (Caregiver 7)

While one interviewee approached a medical device manufacturer in order to jointly develop the medical app further, the manufacturer declined cooperation. When it comes to improving the situation of patients through innovations, then large medical device manufacturers are very resistant to such ideas. (Patient 3)

Only two interviewees founded a commercial startup. Although return on investment is relatively low in financial terms, most participants had lists of suggestions for future improvement and ideas on how to further develop their app. Generally, these suggestions mostly originated directly from users’ feedback. In addition, half of the participants had ideas for new apps they want to pursue in the near future; however, a prerequisite is that there are resources and no time constraints. Nonetheless, some participants will not invest any further effort into their app, owing to a lack of stimuli and time.

7.2 Findings of Qualitative Data on Medical App Developers

7.2.4

101

A Summary of the Findings: Qualitative Data on Medical App Developers

The investigated phenomenon can be described by considering the findings on triggers for innovative endeavors, the development process, commercialization, and outcomes. Before the participants started developing a medical app, several and varying triggers such as daily demand (i.e. blood pressure measurement or blood glucose measurement), the high suffering or the dissatisfaction with apps available on the market fostered the planning for and the final realization of the app. Subsequent to the creation of the app, participants experienced its usefulness via private use. Commercialization introduced the medical app to millions of App Store users worldwide. Although many apps for their disease were already available, users were dissatisfied and developed their own software, introducing some novelty into the market. Further, the data showed that certain attributes fostered innovative endeavors, for instance, a high affinity for smartphones may positively influence the desire to develop an app. The likeliness that someone with a high affinity for IT will develop an app seems much higher than someone without such affinity. Further, the participants expressed their personal attitudes toward their personal medical environment. One participant did not perceive healthcare professionals as valuable problem-solvers concerning his diabetes. Similarly, he perceived large medical device manufacturers as “resistant to innovation.” These attitudes strengthened his desire to build an app to satisfy his needs. The sample developers’ prior competencies and knowledge played a key role. Sound medical knowledge about the disease in question was present in all three user-developer groups. Healthcare professionals were well informed about their medical subject, but all patients were able to discuss their condition with healthcare professional at a high level. Often a general practitioner possesses less knowledge on diabetes than the patient himself. (Patient 4)

Those who developed a medical app by themselves had some prior programming knowledge. Most of these participants also worked in IT-related jobs and therefore possessed various competencies.

102

7

Findings

A long time ago, I did some research in the field of genetics and […] those huge datasets are only manageable by using IT. Therefore, I took some classes in bioinformatics and taught myself to use the programming language PERL. (Healthcare professional 1)

Two healthcare professionals instructed someone else to realize their idea. They represent an exception in the dataset. This external software developer is a critical prerequisite to realizing an app.

8

Discussion

Contents 8.1 8.2 8.3 8.4 8.5 8.6

8.1

Summary of the Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Patients and Caregivers are Not (All) Free Innovators . . . . . . . . . . . . . . . . . . . . . . The Early Development of Medical Smartphone Apps was Triggered by User-Developers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Apps Developed by Patients and Caregivers are of Higher Quality . . . . . . . . . . Company-Developed Apps are the Most Frequently Downloaded Apps . . . . . Patients and Caregivers Develop Financially Successful Medical Apps . . . . . .

103 104 105 107 108 109

Summary of the Findings

In this study, I analyzed different user groups in a large-scale, real-world setting. In the medical smartphone apps market, I identified four developer types: non-user-developers (companies and independent developers) and user-developers (patients and their caregivers as well as healthcare professionals). I evaluated two datasets on analytical app data: one containing data of 1,192 apps from mid-2014 and another with data of 1,265 apps from mid-2018. Both datasets were processed and analyzed similarly. The subsequently applied regression analysis was conducted only with the 2018 dataset. Further, a set of semi-structured interviews was conducted to qualitatively confirm the findings of the quantitative analysis. In the field of medical smartphone apps, the number of user-developed apps is as high as 47.8% (2014) and 39.9% (2018). The percentage of patient-developed apps has increased from 6.9% in 2014 to 10.6% in 2018. Although this value is not representative to the overall population in the three target countries (Germany,

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_8

103

104

8

Discussion

the UK, and the U.S.), it indicates that more and more patients are taking own actions to improve their health outcomes. Oliveira et al. (2015) found that 36% of all patients with a chronic condition or their caregivers innovated in relation to their needs. These findings are in line with a growing body of literature that indicates that patients are gaining increased medical expertise in relation to their disease (Pols 2014; Budych et al. 2012; Hartzler and Pratt 2011; Greenhalgh 2009). Regarding the two research questions I developed in chapter 4, my study confirms the results of extant literature (Oliveira et al. 2015; Habicht et al. 2013) on patients and caregivers as a key contributor group to innovation in the medical device sector using a large-scale, real-world dataset. The performance assessment and thus research question 2 will be answered in the course of this chapter, along with a discussion of the outcomes of the five hypotheses.

8.2

Patients and Caregivers are Not (All) Free Innovators

Based on user innovation theory and particularly based on the patterns of freely revealing user innovation, I hypothesized that patients and caregivers revealed their medical apps for free more often (H1a) and that their paid apps are sold at a lower price (H1b). Both hypotheses had to be rejected. In 2014, the percentage of free apps developed was lowest among patients and caregivers (42%) and highest among companies (58%). In 2018, independent developers had the lowest percentage of free apps (31%), followed by healthcare professionals (43%), patients and caregivers (55%), and companies (71%). Analyzing the mean and median values of the prices of paid apps, I found that independent developers offered their apps at the lowest price and patients and caregivers (2014 data) and companies (2018 data) at the highest price. These results are surprising and must be evaluated considering the background of app development. The interviews with user-developers revealed that they developed their medical apps independently according to their own, their relatives’, or their patients’ needs. On the other hand, a substantial number of company-developed apps were developed by large, multinational medical device manufacturers, pharmaceutical companies, or health insurances. A visual inspection of the titles of free company-developed apps revealed that many of these apps related to a tangible medical device or a drug (i.e. a communication tool for a continuous glucose monitoring or a reminder for insulin injections) or are distributed by health insurances who generate their revenues from recurring payments, bypassing the App Store. This finding on free apps by companies is in line with

8.3 The Early Development of Medical Smartphone Apps …

105

research into the beyond the pill strategy: Pharmaceutical companies currently face an increasingly competitive environment and patent expiries, and must deliver outcome-based evidence on their drugs more often (Wenzel et al. 2014; Herstatt and Raasch 2007). Thus, they began to offer value-adding services such as medical smartphone apps that come along with a drug or a medical device to improve medical compliance, to track health-related quality of life, to improve disease interception, or simply to gain a competitive advantage by offering a digital tool (Narayan et al. 2013). On the other hand, user-developers only have app sales and contingently in-app purchases to receive compensation for their development work. In the interviews, it was mentioned several times that a small pecuniary remuneration is important to them as recognition for their work. Another reason for the relatively high number of patients and caregivers who seek to receive revenue may be that the App Store offers a very convenient ecosystem for user entrepreneurs. The interviews revealed that the user-developers received positive feedback from peers, enjoyed the development process, and had low opportunity costs for the development. These three factors were identified by Shah and Tripsas (2007) as levering user entrepreneurship. A report about the economics of mobile apps revealed similar results: about 53% of all developers stated that their primary goal was to help others; likewise, 67% of developers received no or less than U.S. $10,000 in revenue from their sales per year (Research2Guidance 2015). In my study, I can also confirm the presence of so-called participators, i.e. individuals who profit from the process of developing a medical app (Raasch and von Hippel 2013). I identified a large set of independent developers who developed a medical device without a direct link to a person with a disease. Although not all offered their innovations for free, I found evidence in both datasets that independent developers sold their innovations at a relatively low price and hardly offer in-app purchases. In sum, I found that companies often freely reveal their medical apps, since many of them have alternative revenue streams compared to the three remaining developer groups. Nonetheless, there was evidence that about 50% of patients and caregivers revealed their medical apps for free.

8.3

The Early Development of Medical Smartphone Apps was Triggered by User-Developers

In H2, I proposed that, in the early days of the App Store, user-developers dominated development. In line with this hypothesis, the analysis of the emergence of medical smartphone apps revealed user-developers’ dominance in the early stage of the App Store. In the first six months after opening, about two-thirds of

106

8

Discussion

all apps were developed by healthcare professionals. The ratio of user-developed apps vs. non-user-developed apps was between 80% and 70% (in the 2014 and 2018 dataset, respectively) in 2008 and dropped in the following years to about 50%. From 2013 on, the ratio has settled between 35% and 30% (Figure 8.1). Although the relative number of users in the market is decreasing over time, the absolute numbers are increasing or are remaining constant. This is due to the market’s overall growth over time. Similarly, Riggs and von Hippel (1994) found that, for scientific instruments, the number of user-developed innovations was high at the outset and then decreased over time, while producer-developed innovations increased over time. Further, this pattern confirms the model of Utterback and Abernathy (1975), who stated that product innovation is needs-driven and more often takes place in earlier development stages and process innovation (that is technology-driven or cost-driven and that is associated more with companies) in later stages. As noted above, industries in early phases of a lifecycle have a higher likelihood of user entrepreneurs (Shah and Tripsas 2007). The more mature an industry, the more non-user-developers are present in the market. This pattern was clearly observable in both datasets. Another notable finding in this analysis was the high survivor rate of patients and caregivers in the two datasets. Each app has a unique app ID. Using this app ID, I could determine whether an app was in both the 2014 and the 2018 datasets. I found that the apps developed by patients and caregivers present in both datasets was by far the largest fraction (49%), followed by healthcare professionals (22%), companies (18%), and independent developers (13%). This indicates that apps developed by patients and caregivers remained in the top 1,000 longer in the three target countries (Germany, the UK, and the U.S.). Although I only relied on two data points (2014 and 2018), the apps in both datasets were likely available in the top 1,000 during most of the four years in between. It seems that companies and independent developers tend to develop new apps, while patients and caregivers tend to continually improve their apps. One reason for this pattern was revealed in the interviews: patients and caregivers use their apps daily and thus have a high incentive to continually improve them over time. This pattern is in line with research by Nishikawa and colleagues (2013), who found that user-ideated consumer products in the marketplace were more likely to survive the three-year study observation period than non-user-ideated products. Further, my results support the argument by Shah and colleagues (2012), who found that about 46% of innovative startups that survived a five-year period were user-funded.

8.4 Apps Developed by Patients and Caregivers are of Higher Quality

107

100% 90%

User innovation ratio

User innovation ratio

in 2008 = 70%

in 2017 = 30%

80% 70% 60% 50% 40% 30% 20% 10% 0% 2008

2009

Companies

2010

2011

2012

Independent developers

2013

2014

Healthcare professionals

2015

2016

2017

Patients and caregivers

Figure 8.1 The ratio of user-developed apps vs. non-user developed apps that are newly released every year (2018 dataset)

8.4

Apps Developed by Patients and Caregivers are of Higher Quality

Based on initial findings on the performance parameters of user-developed products (Nishikawa et al. 2013; Poetz and Schreier 2012; Lilien et al. 2002), I posited in H3a and H3b that the developer type impacts on a medical app’s quality (operationalized by user ratings) and that apps developed by patients and caregivers are of higher quality than those developed by companies. In total, I analyzed 255,641 user ratings in the 2014 dataset and 841,167 user ratings in the 2018 dataset. Five-star user ratings is generally a well-suited proxy for assessing product quality (Chen 2017; de Langhe et al. 2016). I found highly significant evidence for both subhypotheses in both datasets as well as in the regression analysis of dataset 2, confirming H3a and H3b. The regression analysis as well as the descriptive results revealed that, on average, user-developed apps have an about 0.50 star better rating—a substantial difference considering that the scale ranged from one to five. I suggest several reasons for this pattern. First, I assume that there is a high heterogeneity of needs (Franke and von Hippel 2003) in the medical smartphone app market, which fosters user innovation. Further, there is a large number of diseases for which data-driven m-health solutions may help to reduce individuals’ health burdens. The interview data revealed that only for a single disease such as diabetes, even

108

8

Discussion

if there are hundreds of medical apps, some patients and caregivers would still be dissatisfied and would develop an own app according to their needs. In this way, gamification, cloud-based solutions, the integration of wearables, or apps for specific target groups (i.e. children) have been introduced to the App Store. Yet, it remains unclear why company-developed apps didn’t reach the same quality level. A second explanation may be the difficulty of companies acquiring sticky need knowledge from patients, caregivers, and healthcare professionals. Although companies have the solution knowledge to develop an app from a technical perspective, some fail to incorporate the information on the needs and the associated functionalities required for a medical app (Schweisfurth and Raasch 2018; von Hippel 1994). From the interviews, I identified one case where a patient tried to cooperate with a medical device manufacturer and offered to jointly further develop an app, yet the company declined to work with the patient. A quantitative follow-up study with a large set of user-developers could confirm if this is a common pattern or an exceptional case. Interestingly, I found the quality difference between user-developed apps and company-developed apps to be smaller for revenue-targeting apps than for truly free apps. For truly free apps, the regression coefficients for healthcare professionals (coeff. = 0.579) as well as for patients and caregivers (coeff. = 0.739) were much larger than for revenue-targeting apps (coeff. = 0.413 for healthcare professionals and 0.320 for patients and caregivers). Although the difference between company-developed apps and user-developed apps was still highly significant (p < 0.001 in all four cases), the coefficients indicated that companies seem to put more effort into better meeting customer needs with revenue-targeting apps.

8.5

Company-Developed Apps are the Most Frequently Downloaded Apps

Regarding number of downloads, I hypothesized that developer type impacts on medical app download numbers (H4a) and that apps developed by patients and caregivers achieve higher download numbers than company-developed ones (H4b). While I found that developer type impacted on medical app download numbers (H4a was confirmed), the impact direction is opposite to what I expected; thus, H4b had to be rejected. Patients and caregivers as well as independent developers had significantly negative impacts (p < 0.001 for both) on downloads compared to company-developed apps. There was no significant difference for healthcare professionals vs. companies concerning number of downloads. According to the

8.6 Patients and Caregivers Develop Financially Successful Medical Apps

109

results for H3, I found that ratings (that were superior for user-developers) had a significant negative impact (p < 0.001) on download numbers. The medical smartphone app market has addressed various diseases. There is likely a massive imbalance between the demand and supply for individual diseases. Some diseases may be over-represented, while for other ailments, very few apps may be available. Thus, many users cannot rely only on (good) ratings, but must download the available apps. Further, price had a significantly negative impact on number of downloads. On the other hand, for some apps, there is a high competition; here, good ratings likely do make a difference. Further research into user-developed medical apps should consider the disease and the number of competitors for a disease in order to deliver a more fine-grained picture on this assumption. As noted, the developer type of a medical app is not visible in most cases. This information was manually added using extensive desk research. However, what is often seen are brand names of well-known medical device manufacturers, pharmaceutical companies, or health insurances. These brand names may more easily convince a customer to download an app instead of an unknown brand. Building on this, I found that for truly free apps, there is only a negative mediating effect of user-developed apps on number of downloads via ratings. Thus, the significantly better ratings of user-developers did not positively impact on their download numbers. On the other hand, number of ratings had a significant positive impact on downloads. This indicates that users evaluate several ratings carefully and then weigh their decision on whether to download an app—even if there are some negative ratings. Nishikawa and colleagues (2017) revealed that labeling a product customerideated resulted in a 20% increase in market performance. Transferring this insight to the field of medical apps, I would assume that if patients, caregivers, and healthcare professionals emphasize their initial reasons for development more, this may positively impact on download numbers.

8.6

Patients and Caregivers Develop Financially Successful Medical Apps

App revenue is determined by number of downloads, price, and in-app purchases. Thus, developer type only had an indirect effect on app revenue. I theorized that developer type has an indirect impact on medical app revenue (H5a) and that patients and caregivers have a significantly positive mediating impact on app revenue compared to company-developed apps (H5b). The mediated regression analysis

110

8

Discussion

revealed that there was no indirect effect of developer type on revenue; thus, H5a had to be rejected. However, considering the categorical values of developer type, the analysis highlighted that patients and caregivers had an indirect effect via number of downloads on revenue. Thus, apps developed by patients and caregivers were—as indicated already in the descriptive analysis—more financially successful at developing medical apps. This is in line with the scarce research into the financial performance of user-developed or user-ideated products (Shah and Tripsas 2016; Nishikawa et al. 2013; Lilien et al. 2002). The interview data has revealed only limited evidence concerning the financial performance of user-developed apps. Only one interviewee mentioned that they aimed to develop a business model with recurrent payments (in-app purchases) instead of single app sales. The remaining developer who did not offer their app for free just mentioned that they seek to get compensated for their efforts, but do not wish to earn money with the development. This is in line with research by Lettl and colleagues (2009) on heroes and hobbyists. The authors found that user innovators who apply low technological diversity and high specialization in their development are able to outperform corporate innovators and thus become heroes (i.e. technologically successful with their innovation), while other users often remain hobbyists (i.e. the innovation remains at a low technical level). Yet, the user-developers who began to integrate in-app purchases into their business model early may have been heroes who applied high specialization into their software. In sum, these results are the first to prove the financial viability of userdeveloped products in a large-scale, real-world setting and thus contribute to both theory and practice. I will now outline these contributions and limitations.

9

Preliminary Conclusions

Contents 9.1 9.2 9.3

Theoretical Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 Managerial Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 Limitations and Further Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

I will now derive theoretical and managerial implications from study 1, mentioning some limitations and possible ways to overcome them. In Part IV, I will integrate the findings from studies 1 and 2.

9.1

Theoretical Implications

This study has contributed to user innovation theory in several ways: First, I have extended user innovation theory by analyzing the hitherto under-researched phenomenon of patients and caregivers as user-developers, which has notable implications for research. The starting point for this research was the few patient-developed medical smartphone apps observed in the Apple App Store. In the course of this study, I found significant evidence of this phenomenon. In the medical smartphone apps market, patients and caregivers develop medical devices along with healthcare professionals, companies, and independent developers. The relative number of user-developed apps was about 40% of the overall sample, with healthcare professionals accounting for about 29% and patients and caregivers for about 11%. This large fraction is somewhat surprising, since the medical device market is long dominated by established firms that could (theoretically) very easily enter

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_9

111

112

9

Preliminary Conclusions

the market. A recent study that used the 2014 dataset of my analysis even revealed that both managers and scholars by far under-estimate the number of available user-developed apps (Bradonjic et al. 2019). The study supports the findings of previous research that healthcare professionals are a key source of innovation in the healthcare sector (Hinsch et al. 2014; Chatterji et al. 2008; Lettl et al. 2008; Lüthje and Herstatt 2004). In my study, healthcare professionals were the largest user group, accounting for about 29% of all developed apps. This is remarkable, since medical doctors generally don’t have the solution knowledge to develop software. I found that some trained themselves to write software, while others hired a third party to develop an app according to their concept or invited a technically skilled co-founder in a startup. The emerging research stream on free innovation focuses on the free revealing aspect of innovations developed in a user’s unpaid discretionary time. Although I expected differently, I found that only about 50% of apps developed by patients and caregivers were disseminated as free apps. A possible explanation may be that the App Store offers a platform to easily adopt user innovation globally— in contrast to the adoption of tangible products, which is often cumbersome and without direct benefits to an innovator (de Jong et al. 2018). I propose that this particular ecosystem of smartphone apps is a valuable empirical field that enables researchers to further study user innovation and innovation adoption in manifold ways. The ease of this platform may also be a reason for the unexpectedly high number of user-developed paid apps. My qualitative study revealed that because it is so easy for them to set a low price, many users decided to do so in order to get a small compensation for their efforts. This study confirmed that in markets that are about to emerge, users are at the forefront of development. Similar to previous studies (Riggs and von Hippel 1994), the initial development was triggered by user innovators. The ratio of newly released apps decreased from 70% in the first year to about 30% in year 10. In line with the scarce research into the survival rate of user innovations (Nishikawa et al. 2013; Shah et al. 2012), an analysis of the apps in both datasets (2014 and 2018) revealed that about 49% of the patient-developed apps from 2014 were still available in the top 1,000 apps in one of the target countries. The survival rate dropped to 22% for apps developed by healthcare professionals, 18% for company-developed apps, and 13% for apps developed by independent developers. Building on this promising result and the interview results, one could argue that patient-developed apps have been in the market longer, since the users actually use and improve the app over time, while other developer groups may not further develop the app and eventually release a completely new app into the market.

9.2 Managerial Implications

113

Finally, I have advanced user innovation theory by responding to the call for large-scale, real-world empirical data on market performance assessment of userdeveloped products (Nishikawa et al. 2013) that go beyond experimental data (Poetz and Schreier 2012). Drawing on a unique dataset of 837 apps, this study is to my best knowledge the first to analyze real-world data on the adoption of userdeveloped products in a marketplace without co-development by intermediaries. Analyzing a mediated regression analysis, I found that user-developed apps are rated significantly higher than non-user-developed apps and that revenue is highest for patient-developed apps. This is surprising, since I did not expect users to seek financial benefits to this extent. One explanation for this pattern may be that companies tend to not strive for financial success, since many see medical apps as complementary products to existing medical devices or pharmaceuticals, while for user-developers, the app is the single source of revenue. In sum, further research is needed if we are to better understand the interplays between user-developed products and non-user-developed products in the marketplace. This study provides important evidence that users should not be under-estimated as a source of innovation in the healthcare sector.

9.2

Managerial Implications

It is valuable for companies to integrate innovative users into their R&D process, since it helps to reduce information asymmetries and increases the efficiency of the innovation process (Henkel and von Hippel 2004). This study has shown that both user types—intermediate users such as healthcare professionals and endusers such as patients and their caregivers innovated successfully and with high quality. Commercial mobile app publishers and healthcare companies should take advantage of this and should consider including patients, caregivers, and healthcare professionals into their R&D process. Depending on the task and the required knowledge set, involving patients and caregivers may be superior to involving healthcare professionals. The significantly better performance of user-developed apps concerning ratings and revenue should be a clear indication that users not only develop better apps, but also financially more successful ones. A study by Oliveira et al. (2015) revealed that quality of life is increasing significantly for patients and their caregivers after a patient uses an own medical innovation. This indicates that caregivers are affected similarly by a medical condition as patients. However, they may have better access to solution knowledge in order to enhance health outcomes. In my qualitative sample, several tech-savvy

114

9

Preliminary Conclusions

people developed an app for their relatives who suffered from a chronic condition. In these cases, where no caregiver with a certain set of solution knowledge is available, companies or public entities such as universities may step in and could help to match needs and solutions, for instance by using an online platform. Further, sophisticated user toolkits for smartphone app development may help to catalyze the development of new software ideas (von Hippel and Katz 2002; von Hippel 2001). In this study, I analyzed only medical smartphone apps. Other areas where tangible medical devices are in place require more effort, owing to complex product development and regulatory processes. Again, medical device manufacturers could amplify patients and caregivers’ endeavors by providing access to certain knowledge bases or co-working spaces (Svensson and Hartmann 2018), or by inviting them to jointly develop an innovation further, as is often done today with healthcare professionals (Chatterji et al. 2008).

9.3

Limitations and Further Research

While the study results have great originality and validity, there are limitations: The findings of the interview series are limited to 16 individuals and need quantitative validation. This would allow one to match personal motivation for the app development, background information, disease-related data, and quality of life measures with the analytical app data of a developer’s apps. Further, only the 1,000 apps with the highest rank in three categories for each country at two points in time (mid-2014 and mid-2018) could be considered in this study, owing to technical limitations. I would assume that the number of patient-developed apps is even higher; however, these apps are not downloaded as often as very popular commercial apps and therefore don’t appear in the top 1,000 rankings. Ventola (Ventola 2014) found that healthcare professionals tend to own iPhones rather than Android-based smartphones. The analytical app analysis should be confirmed using data from Google Play and should compare it to this dataset, which is derived from the Apple App Store. The data of some variables had to be transformed for the mediated regression analysis owing to the nonnormality of the real-world data. This impacts on the interpretation of the regression coefficients. Thus, alternative ways to calculate the regression analysis without data transformation would be desirable for better interpretation of the data. Although five-star ratings of digital products and services is recognized as a reliable measure (Chen 2017; Rozenkrants et al. 2017; Stoyanov et al. 2015),

9.3 Limitations and Further Research

115

it would be interesting to find out more why apps developed by patients and caregivers have higher quality. A study with users of user-developed vs. nonuser-developed apps could reveal interesting insights that go beyond the findings of this thesis. Finally, the App Store is a very convenient platform for individuals to share their (health-related) software with others. The platform offers a set of advantages for developers compared to individuals who wish to commercialize a tangible medical device. Thus, the transferability of the data on tangible medical devices should be evaluated. Study 2, which focuses on patients and caregivers who developed and commercialized a medical device for their own and their peers’ needs, will address these questions.

Part III User Entrepreneurs for Social Innovation – The Case of Patients and Caregivers as Developers of Tangible Medical Devices

10

Introduction: User Entrepreneurs for Social Innovation

Rare diseases are a major burden to society and to each individual and their peers who live with such a disease. In 1990, a five-year-old boy suffered from a spontaneous retinal detachment and lost sight in his one eye. His father noticed that the state-of-the-art techniques ophthalmologists had at the time to examine the residual eye were insufficient to properly assess the retina’s condition. In his already established firm, this father and his colleagues began to work on a technology to do ultra-widefield retina scans so as to be able to detect spontaneous detachments. After a first patent was granted, the father sought to sell the concept to established manufacturers; all declined cooperation. Aware of his product’s benefits, he continued developing it further and, 10 years after the incident, brought to market a machine for ultra-widefield retinal scanning. Some years later, his self-developed machine helped to recognize a retinal detachment in his son’s second eye early enough; this time, the doctors were able to save his sight. In 2015, the company was acquired by a Japanese camera and optics company for around US$400 million.1

Healthcare professionals have long been recognized as a valuable source of innovation for the development of medical devices. Both companies and scholars have found substantial evidence that involving them can lead to successful new product developments (Lettl et al. 2006). Particularly valuable are healthcare professionals who have already developed medical devices (Lüthje and Herstatt 2004) and procedures (Hinsch et al. 2014) for their own needs, or who discover new, off-label uses for drugs (von Hippel et al. 2016). 1 Source:

https://www.optos.com/en-gb/about, accessed November 22, 2018.

Electronic supplementary material The online version of this chapter (https://doi.org/10.1007/978-3-658-32041-6_10) contains supplementary material, which is available to authorized users. © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_10

119

120

10

Introduction: User Entrepreneurs for Social Innovation

While the contributions to innovation of healthcare professionals and medical device manufacturers as providers of medical devices and services have been intensively described, the roles of patients and caregivers in innovation in the healthcare sector has seen little attention in academia and industry. Current research indicates that patients’ role in healthcare is transforming from that of a passive consumer of healthcare to a knowledgeable and critical recipient of healthcare products and services (DeMonaco et al. 2019; Oliveira et al. 2015; Pols 2014). This development is levered by two major trends: First, the prevalence of digital health services such as the health-related information resources that are available online, patient-centered online communities, and mobile health services that give patients access to information and control over their own data (Amann et al. 2016; Dwivedi et al. 2016). Second, the dramatic increase in chronic diseases that require patients to often think about their disease and to eventually implement behavioral changes to daily activities (Goodman et al. 2013). Significant unmet medical needs, better access to medical information, and the more active roles of patients and their caregivers often results in innovation in the healthcare sector that holds great potential for scholars and practitioners. I investigate innovative users who develop medical devices for their own unmet medical needs and who subsequently become user entrepreneurs. Users have been proven to be a major source of innovation. While manufacturers expect to benefit from selling a product or a service, users expect to benefit from using their innovations (von Hippel 2009). A sound body of literature on user innovation has emerged over the past decades, and it indicates that users are the source of many important innovations in diverse industries such as healthcare, sports, banking, scientific instruments, and the humanitarian sector (Goeldner et al. 2017; Oliveira and von Hippel 2011; Baldwin et al. 2006; Lüthje et al. 2005; Riggs and von Hippel 1994). Most of the studies have focused on the incorporation of user-developed ideas into incumbents’ research and development activities (Franke et al. 2006; Lüthje et al. 2005). This focus on the firm’s perspective has neglected the endeavors of individuals to commercialize their self-developed inventions themselves (Hienerth et al. 2014). Drawing on this gap, Shah and Tripsas’ (2007) research on user entrepreneurship has extended user innovation theory by appending the commercialization process of user-developed ideas by the innovators themselves. User entrepreneurship has been studied in only a few industries, such as juvenile products (Shah and Tripsas 2007) and sports (Baldwin et al. 2006; Franke and Shah 2003). According to Shah and Tripsas (2007), enjoyment during the development and commercialization process as well as low opportunity costs have been identified as the main drivers of user entrepreneurship. Further, they stated that they would not expect

10

Introduction: User Entrepreneurs for Social Innovation

121

innovative physicians to develop medical devices on their own, but rather assume that such innovations will be developed by established firms or startups, with the healthcare professional barely involved. Several studies have confirmed this finding, showing how product concepts developed by healthcare professionals contribute to corporate innovation in the medical device industry (Chatterji and Fabrizio 2012; Chatterji et al. 2008; Lettl et al. 2006). There is very little evidence that healthcare professionals can become successful entrepreneurs (Smith and Shah 2013). Based on these ambivalent findings, I would not expect to identify patients and caregivers who develop and commercialize medical devices for their own needs; yet, first empirical evidence indicates that they innovate for their own or their relatives’ needs, thereby contributing to innovation in the healthcare sector (Goeldner and Herstatt 2016; Oliveira et al. 2015), but there is no scholarly evidence of the commercialization activities of patients and their caregivers concerning their self-developed innovations. I this section, I seek to address two gaps: First, I aim to extend the user entrepreneurship literature by showing the prevalence of innovative patients and caregivers—developers of medical devices according to their own or their caregivers’ needs. Second, I want to provide data on how these users recognize and subsequently exploit entrepreneurial opportunities in a social innovation associated with their health burden. In total, I conducted 14 case studies on user entrepreneurs who developed and commercialized tangible medical devices. This contributes to the two—emerging—patient-driven innovation and social innovation research streams. In study 2, I make two major contributions: First, I shed some light on the under-researched phenomenon of patients and caregivers as user entrepreneurs. Having need knowledge about their unmet medical needs, patients and caregivers have developed medical devices on their own and have acquired solution knowledge, particularly technical knowledge, legal knowledge, and regulatory knowledge if needed during the stages of the development process. To meet their medical needs, patients and caregivers have even developed high-risk medical devices that require significant efforts to gain approval by regulatory agencies. Thus, I propose that the greater the need for a solution, the greater the effort individuals are willing to take. The second contribution relates to the emerging social innovation research stream and its connection to user innovation. The users in the sample did not maximize their profits, but rather sought to market their devices at reasonable prices so as to offer many other patients access to these devices (Battilana et al. 2012), in order to increase quality of life. In this extreme case, patients and caregivers maximize their utility as non-pecuniary benefits of increasing their own and

122

10

Introduction: User Entrepreneurs for Social Innovation

others’ quality of life—a common indicator for social innovation scholars (Pol and Ville 2009)—that substitutes somewhat for pecuniary remuneration. This benefit outweighs the barriers of high opportunity costs and the little turbulences in the market for medical devices—as initially proposed by Shah and Tripsas (2007). Study 2 is organized as follows: In chapter 11, I discuss the study’s theoretical framework and derive a set of research questions. In chapter 12, I sketch the methods and the dataset that were used. In chapter 13, the empirical findings are presented; in chapter 14, I discuss these findings; in chapter 15, I propose preliminary implications for theory and practice, discuss the limitations and outline avenues for further research.

Theoretical Background and Research Questions

11

Contents 11.1 11.2 11.3 11.4

11.1

Opportunity Recognition and Exploitation in Entrepreneurship . . . . . . . . . . . . . Opportunity Recognition and Exploitation by Patients and Their Caregivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Social Innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

123 124 125 126

Opportunity Recognition and Exploitation in Entrepreneurship

Entrepreneurship scholars seek to better understand how and by whom opportunities for future goods and services are discovered, created, and exploited (Shane and Venkataraman 2000; Venkataraman 1997). Entrepreneurial activities can be seen as a two-step process: opportunity recognition followed by opportunity exploitation (McMullen and Shepherd 2006). Eckhardt and Shane (2003, p. 336) defined entrepreneurial opportunities as “situations in which new goods, services, raw materials, markets, and organizing methods can be introduced through the formation of new means, ends, or means-ends relationships.” Shane (2000) was one of the first to argue that opportunities are recognized rather than searched for. Prior knowledge is crucial to be able to recognize opportunities, especially less obvious opportunities (D. A. Gregoire et al. 2009). For instance, prior employment or a family member suffering from a chronic disease can provide unique insights. Thus, some individuals are in unique positions to recognize entrepreneurial opportunities (Shah and Tripsas 2007). Similarly, Shane and Venkataraman

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_11

123

124

11

Theoretical Background and Research Questions

(2000) emphasized the need for good information and the cognitive properties to value such information. Because several opportunities may emerge at a similar point in time the selection process (including assimilation, organization, categorization, and prioritization of information) must be highly selective (Dutta and Crossan 2005) and is positively influenced by an entrepreneur’s social ties (Ellis 2011). George et al. (2016) summarized six critical factors that strongly impact on opportunity recognition: prior knowledge, social capital, cognition, environmental conditions, entrepreneurial alertness, and systematic search. Opportunity exploitation is determined by the nature of an opportunity as well as an entrepreneur’s characteristics, perceptions, and capabilities (Kohlbacher et al. 2015; Shane and Venkataraman 2000). It is hard to balance opportunity recognition and exploitation—the literature suggests having a short exploration time for innovations with low novelty, while more novel ideas should be carefully exploited as soon as knowledge about the innovation is also available to others (Choi et al. 2008). Being able to exploit an opportunity depends not only on users’ abilities and motivations—De Jong (2013) found that positive attitudes, subjective norms, and perceived control must be instantaneously present for a successful exploitation to be able to emerge. Choi and Shepherd (2004) found that entrepreneurs are more likely to exploit an opportunity if they have more rather than less knowledge about customer demands. In the case of user entrepreneurs, I assume that such users have high need knowledge and thus exhibit one of the key requirements for the successful exploitation of their inventions.

11.2

Opportunity Recognition and Exploitation by Patients and Their Caregivers

In chapter 2, I have elaborated on the interlinkage between user innovation, user entrepreneurship, and user innovation in healthcare. Yet, the connections between those subjects and the opportunity recognition and exploitation of patients and their caregivers will be summarized in this section. Innovative patients and their caregivers often have extended knowledge of their disease (Elberse et al. 2011) as well as technical knowledge, mostly through education (Goeldner and Herstatt 2016). The results of the qualitative analysis in study 1 confirm this finding. Some individuals identify opportunities because they are able to perceive connections between seemingly unrelated patterns (Baron 2006). In healthcare, barriers such as regulations and high capital requirements for clinical testing hamper opportunity exploitation more than in other industries (Chatterji et al. 2008). Patients and their caregivers, particularly those with chronic

11.3 Social Innovation

125

conditions and who have strong constraints in daily life or face dead-end situations (Habicht et al. 2013), often face challenging tasks in their daily caring routines. This prior knowledge may help patients and caregivers to identify more and other entrepreneurial opportunities than professional medical device developers or healthcare professionals (D. a. Gregoire et al. 2009). Owing to constraints such as regulatory approval, fairly high capital requirements, and challenges associated with diffusion, it is often hard to exploit medical device ideas in the healthcare sector (Braun and Herstatt 2008). Thus, studies of patients and caregivers as user innovators found only evidence of peer-to-peer diffusion to other patients or sometimes to healthcare professionals (Oliveira et al. 2015). In this study, I seek to further analyze diffusion endeavors of self-developed medical devices by patients and caregivers.

11.3

Social Innovation

Social innovation is a nascent field in innovation research that has emerged laterally to other research streams on innovation: The scientific discourse on social innovation has mainly focused on definitions (Edwards-Schachter and Wallace 2017; van der Have and Rubalcaba 2016; Cajaiba-Santana 2014; Pol and Ville 2009) rather than integrating social innovation to other fields of innovation. Many innovations are labeled social, for instance, innovation in the public, humanitarian, educational, or healthcare sectors or in the maker movement (Edwards-Schachter and Wallace 2017). As Mulgan (2006) noted, the underlying rationale is the importance of the social dimension of innovation compared to business innovation, which seeks to maximize profit. Scholars agree that the two fields are not disconnected but overlap, with a shared basis: In their review of social innovation and business innovation, Pol and Ville (2009) discuss four conceptions of social innovation. Their analysis of social innovation and institutional change, social purposes, the public good, and needs not addressed by the market revealed that these four have in common the improvement of the quality or the quantity of life (Pol and Ville 2009). As noted earlier, the concept of quality of life is closely related to the subjective wellbeing and the happiness of individuals (Edwards-Schachter and Wallace 2017) and is being used across multiple disciplines such as psychology, medicine, economics, and sociology (Costanza et al. 2006). Thus, it is a well-suited parameter to analyze the wellbeing of individuals across disciplines. In his review of social innovation, Cajaiba-Santana (2014) identified two viewpoints on social innovation: an individualistic perspective, which draws on individuals and their characteristics to develop social innovation, and a structural

126

11

Theoretical Background and Research Questions

perspective, which focuses on social structures, organizations, and the barriers these organizations face. The conjunction between both perspectives opens interesting opportunities for research into social innovation and will be further elaborated on in this study. A bibliometric analysis by van der Have and Rubalcaba (2016) revealed that social innovation is rooted in four distinct intellectual scholarly communities, namely community psychology, creativity research, social and societal challenges, and local development. Conversely, this indicates that social innovation is not originally rooted in innovation research (van der Have and Rubalcaba 2016) The social entrepreneurship field (Short et al. 2009) is another avenue for scholars to investigate how social innovations are developed and which barriers and enablers these innovators encounter along the way (Lettice and Parekh 2010). Bacq and Janssen, who follow an individualistic perspective, summarize the social entrepreneur as a visionary individual “who is able to identify and exploit opportunities, to leverage the resources necessary to the achievement of his/her social mission and to find innovative solutions to social problems of his/her community that are not adequately met by the local system” (Bacq and Janssen 2011, p. 382). In their review, Peredo and McLean (2006) found that social entrepreneurs seek to create social value, have the capacity to envision an opportunity, employ innovation, accept above-average risk, and have scarce resources to pursue a social venture (Peredo and McLean 2006). In net, social innovation is an emergent research stream that looks at innovation not from the perspective of cutting-edge technology, but about a novel way of solving social problems in an interconnected world (van der Have and Rubalcaba 2016).

11.4

Research Questions

To date, there has been very little research at the intersection of user innovation and social innovation (Goeldner et al. 2017) and on how user entrepreneurs contribute to social innovation. In this study, I want to investigate cases of patients and caregivers as user entrepreneurs for social innovation. To evaluate the opportunity recognition and exploitation processes of patients and caregivers, two research questions that address the individual perspective (RQ1) and the structural perspective (RQ2) on social innovation according to the framework of Cajaiba-Santana (2014) were developed:

11.4 Research Questions

127

• What are the reasons for user innovators to further develop their medical device ideas, recognize a business opportunity, and become user entrepreneurs? • How do innovative patients and caregivers exploit their business opportunities, despite the constraints of the healthcare system? By answering these questions, I seek to analyze user entrepreneurship in a highly regulated industry that poses significant barriers to users. Thus, this is a study on an extreme cases of user entrepreneurship that seeks to draw conclusions that are valid for social innovations beyond the healthcare sector (Eisenhardt 1989).

12

Methodology

In this section, the methodology used for data acquisition and analysis is outlined, as well as a description of the procedures that were applied to ensure internal validity, external validity, and reliability (Gibbert et al. 2008). Inductive theorizing based on qualitative data is particularly appropriate for new and complex empirical contexts with little previous work (Bansal et al. 2018). This applies to user entrepreneurs in social innovation and particularly for the cases of patients and caregivers as innovators of medical devices. Case studies are used as a well-suited research strategy to build theory for the observations made during the research (Yin 2013). I opted for a multiple-case study design to gain a better understanding of user entrepreneurs in social innovation (Chandra and Leenders 2012; Hienerth and Lettl 2011; Lettl et al. 2008). A case study design is appropriate to the study’s explorative character and to develop reliable and generalizable propositions (Eisenhardt 1989). Multiple cases are commonly regarded as more robust than single-case studies, since cross-case comparisons foster validity and reduce the findings’ context-dependency (Goffin et al. 2019; Gehman et al. 2018; Lettl et al. 2006; McDermott and O’Connor 2002). The motivations and triggers as well as the development and commercialization processes of medical devices developed by patients and caregivers are investigated within this study. To ensure the findings’ validity, each case was explored using three data sources (Gibbert et al. 2008): semi-structured interviews with the inventor, secondary data such as articles or websites, and patent data. This is in line with other scholars who Electronic supplementary material The online version of this chapter (https://doi.org/10.1007/978-3-658-32041-6_12) contains supplementary material, which is available to authorized users.

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_12

129

130

12

Methodology

selected similar approaches (Schweisfurth and Herstatt 2016; Kalogerakis et al. 2010; Howells 2006). The case sampling focused on diseases where user entrepreneurs and social innovation were observed (Eisenhardt 1989). There is no sampling frame for patients and caregivers who develop medical devices, and the screening of user innovators and particularly innovative patients and caregivers is costly owing to their low frequency in the population (de Jong et al. 2018). Thus, I collected a sample that included cases from several countries and various target diseases in order to increase the findings’ external validity (Yin 2013). The research into patients and caregivers as innovators has shown that only very few patients and caregivers commercially diffuse their inventions (DeMonaco et al. 2019; Oliveira et al. 2015). According to the study of Oliveira and colleagues (2015) on patient innovation in rare diseases, most such innovations are shown to other patients or shared in a social network, but very few innovators further commercialize their innovations. One reason is that only 10% of the 182 innovations was tangible products, while the remaining 90% was descriptions of an activity relating to the treatment or changes in strategies or behaviors relating to a disease (Oliveira et al. 2015). It is known from previous studies that patients and caregivers have developed medical smartphone applications according to their needs (Goeldner and Herstatt 2016; Bullinger et al. 2012). Yet, the commercialization process of a medical app differs greatly from a tangible device, because production, marketing, and regulatory approval are much easier for entirely digital products and services (Goeldner and Herstatt 2016). Thus, I focused on tangible medical devices, and excluded all developers of software and medical apps without a corresponding medical device. Thus, several data sources had to be used in order to identify a meaningful set of innovators. Since I seek to develop theories that are generalizable to other patients, caregivers, and—ultimately—to social innovation, I purposely selected a variety of user innovators. Specifically, I used mixed purposeful sampling, as described by Onwuegbuzie and Leech (2007), including criterion sampling, extreme case sampling, and snowball sampling. I reached out to a diverse interviewees, who varied concerning their roles in the innovation process (patient or caregiver), age, gender, medical device class, and disease. Seventeen patients and caregivers were approached of which 14 participated—an 82% return rate. Of those 14 innovators, three were identified in newspapers, three via Internet searches, five via personal recommendation of patients and caregivers I spoke to, and three through the Internet platform patient-innovation.com1 . The first set of three interviews was 1 Source:

www.patient-innovation.com

12

Methodology

131

held between December 2014 and January 2015, and the remaining 11 in March and April 2016. Since several diseases and medical devices of different complexities were analyzed, the data has high variance. Thus, new cases were added stepwise, data saturation was reached after 14 cases (Morse 1995), resulting in an optimal sample size for cross-case analysis in phenomenological qualitative research (Onwuegbuzie and Leech 2007; Eisenhardt 1989). Three interviews were conducted face-to-face, and the remainder via telephone. The first data source was interview data from 14 semi-structured interviews with developers of tangible medical devices from Austria, Germany, Israel, Sweden, Switzerland, the UK, and the U.S. who had successfully launched their medical devices in the market. The interview guideline is primarily based on established constructs obtained from previous studies: First, I asked how the opportunity for the innovation was recognized (Franke et al. 2006; McMullen and Shepherd 2006; Lüthje et al. 2005; Shane 2000). Second, I examined the exploitation of the innovation (von Hippel et al. 2016; Habicht et al. 2013; Bogers et al. 2010). Also, a set of questions on the disease addressed by each device and every inventor’s educational background was asked. The second data source was archival data obtained from websites, newspapers, product information provided by the interviewees, as well as other online resources such as TED Talks videos. This data initially helped to confirm whether the inventor is a patient or a caregiver, and in many cases gave a sound understanding of the product functionalities. I used this data to confirm and validate the findings obtained from the interviews, since this data is considered to be more objective. The third dataset was patent data on the interviewees’ patents. The patents or patent applications were downloaded from publicly available online sources and were used to complement the first two data sources. The sample included nine patients and five caregivers; four were female developers, while 10 were male.2 At a ratio of about 30%, the number of female developers is similar to other studies of user innovations (Magnusson et al. 2016; Tietze et al. 2015). The innovator age ranged between 17 and 58, with an average of 41 years. The interviews lasted between 20 and 74 minutes (average: about 42 minutes). All interviews were transcribed and coded using the software MAXQDA 11. For data reduction, I followed an inductive approach with three sequential steps: I started with open coding, categorized the themes in a second 2 Two

couples developed an innovation for their children. However, in both cases, the fathers were interviewed, although both innovations were co-developed by the father and the mother.

132

12

Methodology

step and then linked them in a third step (Gioia et al. 2013; Miles et al. 1994). I sought to ensure the results’ rigor: Investigator triangulation was achieved, since multiple researchers did some interviews together, and parts of the analysis were done independently and cross-checked (Gioia et al. 2013; Gibbert et al. 2008). Although the interview data was the main data source for further analysis, archival data, websites, and patent data was used to triangulate the data, increasing validity (Goffin et al. 2019; Shah and Corley 2006). This was more objective and had less potential for retrospective sensemaking bias (Eisenhardt and Graebner 2007).

Findings

13

Contents 13.1 13.2 13.3

13.4

13.5

13.1

Descriptive Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Unmet Medical Needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Opportunity Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3.1 Ideation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3.2 Prototype Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.3.3 Intellectual Property . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Opportunity Exploitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4.1 Product Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4.2 Regulatory Approval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4.3 Production and Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Market Launch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

133 136 137 137 137 139 140 140 141 142 144

Descriptive Findings

The findings indicate that patients and caregivers are a potential source of innovation for medical devices. The innovations reflected the variety of medical devices in the market: nine innovations targeted chronic conditions, while the remaining five diseases were fairly acute but also had significant negative impacts on the patients’ quality of life (see Table 13.1).

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_13

133

134

13

Findings

Table 13.1 A qualitative study on user entrepreneurs for social innovation—characteristics of interviewees ID/role

Country of origin

Disease

Medical device class

Patient 1

Germany

Back pain

Not classified

Patient 2

Israel

Rehabilitation after stroke

I

Patient 3

Sweden

Diabetes

I

Patient 4

U.S.

Paraplegia

Not classified

Patient 5

Switzerland

Paraplegia

Not classified

Patient 6

UK

Aortic dilatation

III

Patient 7

U.S.

Post-surgical wounds

I

Patient 8

U.S.

Post-surgical wounds

I

Patient 9

U.S.

Diabetes

IIb

Caregiver 1

Germany

Cystic fibrosis

IIa

Caregiver 2

Israel

Neuromuscular disorders

I

Caregiver 3

U.S.

Alzheimer’s

Not classified

Caregiver 4

UK

Retinal detachment

IIa

Caregiver 5

Germany

Febrile seizures

IIa

Medical devices1 are classified according to the EU-Directive 93/42/EEC (2007)2 into four classes, according to the risks of their use on humans—classes I, IIa, IIb, and III. A wheelchair poses a low risk to a user (class I medical device) and is only subject to general controls such as Good Manufacturing Practice (GMP) examination, while a pacemaker, which remains in the body for more than 30 days, bears the highest risk (class III medical device) and requires intense clinical testing so as to assure safety and effectiveness prior to market launch (Kramer et al. 2012). In 10 of 14 cases, regulatory approval was obtained either by 1 “A

medical device means any instrument, apparatus, appliance, software, material or other article, whether used alone or in combination, including the software intended by its manufacturer to be used specifically for diagnostic and/or therapeutic purposes and necessary for its proper application, intended by the manufacturer to be used for human beings for the purpose of: diagnosis, prevention, monitoring, treatment or alleviation of disease, diagnosis, monitoring, treatment, alleviation of or compensation for an injury or handicap, investigation, replacement or modification of the anatomy or of a physiological process, control of conception, and which does not achieve its principal intended action in or on the human body by pharmacological, immunological or metabolic means, but which may be assisted in its function by such means.” (EU-Directive 93/42/EEC 2007). 2 Similar guidelines issued by the Food and Drug Administration (FDA) apply in the U.S.

13.1 Descriptive Findings

135

CE certification (in Europe) or FDA approval (in the U.S.). Four innovators stated that their product required no regulatory approval. Of the 10 approved medical devices, five were class I, two class IIa, one class IIb, and one class III.3 Thus, patients and caregivers developed not only low-risk medical devices, but there is evidence that some also develop medical devices that require a long-lasting and capital-intensive regulatory approval. Examples of medical devices in the sample are: two of the nonclassified medical devices were a wearable sensor that notifies caregivers if an elderly person leaves their bed and a wheelchair supplement that allows users to drive their wheelchair in snow. The timer attached to an insulin pen is a class I medical device that helps diabetics to not forget their last insulin intake. The waterproof T-shirt that allows patients to take a shower after surgery is a class I medical device. The inhalation device for children with pulmonary diseases linked to a computer game is a class IIa medical device. The smart insulin pen that tracks insulin intake and exchanges data with a smartphone is a class IIb medical device. The only class III medical device in this sample was the customized surgical mesh implanted around the patient’s aorta, where it supports both the aorta and heart valve, preventing the enlargement and rupturing of a dilated aorta. The innovators’ educational backgrounds were very diverse: the interviewees had completed studies in engineering (5), business (2), psychology (1), medical journalism (1), social work (1), education (1), and musicology (1). One interviewee had a high school education; another was about to finish high school in the next year. Three of the engineers were working in healthcare; one in a field not related to the invention, and the other two in a field relating to the invention. Since one of the latter was diagnosed with the disease, he found that his employer’s products did not fit his needs, which led him to start his own business. Based on four established constructs (Franke et al. 2006), which were adapted to suit this study (see Appendix 3), I asked the interviewees to estimate their lead-userness concerning their medical devices. On a seven-point Likert scale, the average value was 6.7, indicating a very high lead-userness in the sample patients and caregivers. In most cases (9 of 14), a company was incorporated right before the first patent application was filed. In 10 cases, the user founded their own company and commercialized the product themselves. In two cases, the founder had already 3 The

distribution of medical devices into several risk classes in Europe reveals a similar picture: in 2012, 56% of all medical devices brought to the European market were class I, 27% class IIa, 12% class IIb, and 5% class III. Source: https://de.statista.com/statistik/daten/studie/325915/umfrage/medizinproduktein-europa-verteilung-nach-risikoklassen, accessed July 3, 2018.

136

13

Findings

founded a company (not targeting that specific disease) and was able to do the development in this company. In the two remaining cases, the idea was licensed out to a company; however, in both cases it was hard to find a manufacturer, since major medical device manufacturers in the relevant areas refused to accept a patent-pending prototype that was developed by a patient or caregiver. Two other innovators could not find a partner for development and then decided to advance the device until it was ready for sale within the own small business. In most cases, the innovators aimed to develop the medical device on their own. I just liked building things, and this was a challenge for me. It’s important for me to do it myself. In general, I’m a bit humble, but I thought I could do it the best way. (Patient 2)

13.2

Unmet Medical Needs

In all cases, the innovation process started with a strong unmet medical need that was not served by existing medical devices. All the sample inventors searched for solutions on the market that suited their needs, and all were dissatisfied with the existing devices (if there were any). I kept thinking, women have been going through mastectomy surgery in this country for years. Why is there no product to protect me? This really confused me. (Patient 7)

Since nine out of the 14 patients suffered from a chronic condition, they were even more dissatisfied. I got frustrated that these ninety-three percent of people [who do not use an insulin pump], of which I was one, were being left out in the cold. (Patient 9)

This dissatisfaction was the starting point of their innovative endeavor.

13.3 Opportunity Recognition

13.3

137

Opportunity Recognition

13.3.1 Ideation Several opportunity recognition patterns were observed: Four interviewees stated that there was a special single moment when the idea for the medical device appeared serendipitously (Yaqub 2018) during their daily routine. My aunt, who was my grandfather’s primary caregiver, was very stressed when she took care of him (…) Once, I sat night watch for my grandfather. When I saw him stepping off the bed. The moment his foot touched the floor, I came up with the idea… (Caregiver 3) I happened to be on a train, sitting between two seats, and I was surprised at how good this felt. At home, I got some pieces of wood from my brother and tried to redesign such a seat. (…) This moment in the train was really memorable. (Patient 1)

In the remaining 10 cases, the idea was developed slowly and was re-iterated over a longer period in time-consuming and exhausting daily routines. This took between two months and three years. This is particularly stressful for the parents: You need to persuade your child to inhale twice a day for ten minutes, and you need to sit right beside them to check whether they are really doing it. (Caregiver 1)

Three of the 10 abovementioned innovators stated that they used a structured process to find a solution to their problem. Public health scholars have emphasized the significant knowledge gains of patients and their caregivers (Pols 2014; Joseph-Williams et al. 2014) during sickness periods. This study confirms that the sample inventors gained significant medical knowledge while treating their own or their peers’ disease. When you’re in a hospital for three years in rehabilitation, you see many, many patients, and you learn a lot about physiotherapy. (Patient 2)

13.3.2 Prototype Development Next, prototypes were built and the device was used for the first time. The prototyping was done at home, in a private workshop or, in three cases, in the already

138

13

Findings

existing own company. In three cases, universities were consulted and their product development capabilities were used. In three cases, makerspaces were visited, because additional manufacturing was needed for prototyping. The development work was mostly done in spare time, and it took between six months and more than five years until a prototype with all required functions was available. We did it all in our spare time. For a year, we worked every other weekend and in the evenings. (Patient 3)

Three of the caregivers were able to do the prototyping completely on their own, while the remaining two caregivers and all nine patients needed external help— solution knowledge—to be able to develop the prototype. Experts as well as friends and family with the needed complementary knowledge were consulted in order to obtain this knowledge. I consulted experts during all the development steps. (Patient 5)

The innovator’s personal network was important in all cases and was a prerequisite for the successful prototyping. My neighbor helped me. I paid her a consulting fee and she helped me develop the prototype. (Patient 8)

Others were inventors without a formal engineering education: This is my gift. Some people have a gift for music or for drawing. I have a gift for inventing things. I have invented things before. I’ve written patents in other fields. And I’m good with the engineering. I have no formal engineering education, but I’m good at it. (Patient 2)

After a first prototype was available, in 13 of 14 cases, it was used extensively by the inventor or the person they cared about. In one case, the prototype was no longer needed, because work on the prototype started only after the person in question’s rehabilitation was completed. Soon, the interviewees recognized the possibilities of their idea. Seeing potential early on, nine of the inventors decided during prototype development that they would try to commercialize their idea later. The remaining five inventors developed a prototype for own use only and realized or decided later that their idea would also be valuable to others.

13.3 Opportunity Recognition

139

It makes no sense to me that I’m the only one in the world with a solution to this problem. Thousands of people have the same problem. (Patient 5)

The interviewees had also been asked to produce more prototypes, because friends and other peers affected by the same disease also wanted to use the device. A friend asked me for a prototype, but he had some requests I first had to incorporate. (Patient 1)

These requests were a key reason for the interviewees to enhance their idea beyond own usage. Since they themselves had a need, it was clear to them all that their solution would also be valuable to others with the same ailment. Requests from others fueled their desire to further develop their innovation and to finally diffuse it as a product in the market. It was very stressful to us as parents to convince our son to take his medication. (…) And it’s the same in all the other affected families. (Caregiver 1)

13.3.3 Intellectual Property Financial considerations or a business plan were conducted only on a small scale, if at all. The sample patients and caregivers were convinced of their idea from the outset and spent little effort on calculating potential returns on their investments. However, all the innovators emphasized the importance of early intellectual property (IP) protection during the opportunity recognition process. In seven cases, the patent application was submitted within the first year after they had recognized the opportunity and had begun to work on the device. One night I was at my drawing board and I came up with the idea how to lift the wheelchair’s casters off the ground. Well, I filed for the patent the next day. (Patient 4)

In the remaining cases, it took two years (three cases), three years (two cases), or five years (two cases) to apply for a patent. Of the 14 innovators, 12 hired a patent attorney to help them to file the patent, while two filed on their own. Until mid2018, 12 of 14 patent applications were approved by the authorities, indicating that these inventions were not obvious compared to prior art. The high costs associated with a patent application were a hurdle in the development process, but all the interviewees managed to get funding. About half received initial funds

140

13

Findings

during opportunity recognition from investors or governmental funds, while the other half covered the costs on their own or with the help of family and friends.

13.4

Opportunity Exploitation

Exploiting product ideas in the medical device sector is challenging, since many barriers occur during development. After the decision to further develop beyond own use and the securing of IP, the prototype was often shared with others. However, the initial feedback often disappointed developers. I showed the prototype to investors. I showed it to factories. (…) And no one took it seriously. (…) The only ones who took me seriously were some physical therapists who wanted to use the device. (Caregiver 2)

Established producers of related medical devices were also not necessarily interested in acquiring the IP generated outside their company boundaries. I went to see some representatives of the largest producer of such devices. I was convinced they would be very interested and would buy the patent. But they just said no. (…) It really surprised me. (Caregiver 1)

Two of the inventors finally found a partner for further development and licensing, while two others (who wished to partner with a third party) continued developing the medical device on their own.

13.4.1 Product Development The resources needed to further develop their products came from different sources, depending on the complexity of the manufacturing process and the product’s medical device class. The resources spent varied from several thousand euro to more than 10 million euros. While seven of the developers were able to finance the development completely on their own or only with support from family or friends, the remaining ones received external funds from investors or governmental funds. Yet, this was hard to achieve, since particularly high-risk medical devices developed by nonprofessional developers are not considered to be secure investments.

13.4 Opportunity Exploitation

141

It wasn’t possible to raise money from conventional venture capital because they thought there was no market for this product. Over ten years, I was turned down at least once by every venture capital firm in the UK, Europe, and North America. (Caregiver 4)

This external money was mostly delivered only after the proof-of-concept had been successfully demonstrated. In the opportunity exploitation phase, all interviewees (except the two who licensed the idea to an external manufacturer) decided to quit their job and started working on their medical device full-time. In most cases, staff was hired and more external partners were involved in the development. It was in this phase that the transition from user innovator to user entrepreneur occurred. Different sources of external knowledge were needed in this stage to complement the innovator’s competences: Technical knowledge and regulatory knowledge was needed in 11 of the 14 cases. In nine cases, the innovators presented their prototype to healthcare professionals and further discussed their concept so as to gain additional medical knowledge. That doctor from Harvard Medical School […] sent a really incredible and inspiring e-mail saying, ‘This is a huge problem. We need a solution. Please solve this and do something about it.’ (Patient 3) I showed it to physiotherapists I knew, and they told me on the spot that if there was a machine like this, every physiotherapy department would buy one. (Patient 2)

13.4.2 Regulatory Approval After the product was finalized, regulatory approval had to be sought. For class I medical devices or nonclassified devices, the regulatory burden was acceptable— only a fairly low number of tests and documentation is required. For higher-risk medical devices, the regulatory approval was a considerable barrier to the developers. One decided to license the idea to an established manufacturer because he feared the high costs associated with class IIa approval. Another developer stated that the ongoing discussion with authorities and hospitals was nerve-wracking: It is great to transform people’s lives and I am very happy to know that I saved many people. But I am mentally tired of all the arguing and fighting with the hospitals and the regulatory agencies. (Patient 6)

142

13

Findings

Since the formal structures for the medical device approval are designed for companies, small startups and independent developers struggle with the resources needed for receiving regulatory approval (Chatterji 2009). Ultimately, all five high-risk products (classes IIa, IIb, and III) in the sample achieved official approval. A friend is the CEO of an electronic components manufacturer. I told him about my idea and he said that he could deliver ten-thousand units in the next days. Obviously, this was very helpful for further development. (Caregiver 1) We had very good connections in the U.S. that came from this investor. He said: use this legal firm and use this FDA liaison officer. They will ensure that you will present yourself properly. (Caregiver 4)

This regulatory knowledge is key, since regulatory approval, particularly for class IIa or higher products, is a significant burden to nonprofessional developers of medical devices.

13.4.3 Production and Distribution Initial manufacturing was ramped up mainly in the investors’ home country—only two interviewees stated that they began to manufacture in a country with lower labor costs. Reasons for a production site close by were higher expected quality, better responsiveness to changes, fear of infringements on IP in other countries, and already established manufacturing processes for medical devices. In five cases the production was done within the company, while in nine (including the two cases where the idea was licensed to an incumbent manufacturer from the outset), production was outsourced. I am producing on my own, to be more independent. I need to control what is happening. (Patient 5)

Two of the five self-producers later outsourced their production to a manufacturer. I no longer produce them in my garage. I found a manufacturer to build them in higher numbers with better efficiency than I could. (Patient 4)

13.4 Opportunity Exploitation

143

Before launching the medical device in the market, the innovators struggled to determine a price for the device. Owing to their market research, all the innovators knew that their medical device is unique and not yet available on the market. However, most interviewees emphasized their innovation’s social component. Because their main goal was to diffuse their device to other patients, they sought to keep the price reasonable. We argued a lot and finally agreed on this price. I had to tell them clearly that no one would buy the device if is too expensive. (Patient 5) I thought about the other parents, who would want to get it at a reasonable price. (Caregiver 2)

Others just decided on the price on their own. The price? I just guessed. (Patient 3)

Most of the devices were priced according to the value the product delivers or based on existing devices that were somewhat comparable to their new invention. We came up with this price based on what we thought the product’s value is. (Caregiver 3)

The innovations’ social character were also supported during later stages of the development process: some innovators were supported by suppliers and other stakeholders, because the motivation for developing the device had a social element. All interviewees except one told their stakeholders about their own ailment. My suppliers are very friendly and accommodating. They even offered me reduced prices. (…) Everybody tried to show some goodwill. (Patient 5)

Still, many inventors were frustrated during the later stages of the development process, particularly concerning business planning. So, the business side of it has been very challenging and quite frustrating at times. But personally it’s very rewarding. (Patient 3)

Then, sales channels had to be identified. Again, these differed in relation to the medical device’s associated risk: Two inventors of high-risk products built

144

13

Findings

an own distribution network owing to their products’ high complexity and uniqueness. The three remaining inventors of high-risk products collaborated with an established company for distribution. Further, one of these three inventors offers the product for sale online. Concerning the low-risk products, two are available online only, three via an established medical device distributor and online, and four through an established distributor only. I sell it mainly online. (…) I am really satisfied by literally thousands of responses from customers who say, ‘This is the best piece of wheelchair equipment I ever bought.’ (Patient 4)

Online distribution is particularly important for selling a device abroad, since it is challenging for inventors to identify suitable distributors around the globe. Nonetheless, many sample inventors have managed to build a distribution network with an established partner in many countries. Further, two inventors successfully launched a crowdfunding campaign to receive funds for the initial production and to directly distribute to first customers for the initial product batch.

13.5

Market Launch

The timeframe from the start of the ideation process to market launch varied between one and 19 years, and took on average 5.4 years, of which 3.7 accounted for opportunity recognition and 1.7 for opportunity exploitation. Although there is no actual data available in the literature, it is estimated that medical device development within the medical device industry is substantially faster (Fargen et al. 2013). I assume that particularly opportunity recognition is carried out considerably more rapidly in the medical device industry. The final and most significant barrier for the sample innovators appeared only after market launch: the health insurance reimbursement process. In many of the inventors’ countries of origin, healthcare services are mostly delivered to patients for free or at very little cost. Thus, this is only true for products that are reimbursed by health insurances. Since all the sample medical devices came to market without reimbursement, this was an unexpected and therefore a significant barrier. Most diabetes patients in Europe (…) get everything for free. They are not used to paying. Even if they are willing to pay, there aren’t many channels where you can pay for these things. Most pharmacies don’t have diabetes products. None in Sweden do or they are very limited. Because people get it all for free. (…) The mechanism of paying for stuff doesn’t exist. (Patient 3)

13.5 Market Launch

145

This results in significant efforts to convince patients to buy their devices without reimbursement. Nonetheless, all the interviewees stated that they are managing to sell a sufficient number of devices to have a viable and sustainable business. Most interviewees confirmed that the positive feedback is a major reason why they are very satisfied with their new business and want to continue to further develop the device or even to develop new medical devices for the same target group. Of the 14 interviewees, 13 used their product or had their relative using it. This improved their own quality of life significantly. It definitely improved my quality of life. I mean, I have used it every day for five years now. So, I have probably used it fifteen thousand times. (Patient 3) Well, my life is different. My life is very different […] I use it all the time, I love it. (Patient 8)

Only one interviewee had no need to use the device, since his rehabilitation process had been completed by the time he began to develop it. All the interviewees stated that the final result was worth all the efforts invested during the recognition and exploitation of their opportunity.

Discussion

14

I have sought to advance the understanding of the opportunity recognition and exploitation of user entrepreneurs who developed a medical device for their own need or for a person they cared for. In all observed cases, opportunity recognition and exploitation was stimulated by a trigger (Morris et al. 2000), the unmet medical need, and concluded by market launch (see Figure 14.1).

Figure 14.1 Process overview of opportunity recognition and exploitation

All the sample innovators did extensive market screening before starting their innovative endeavor. Since no suitable medical device was available, they began to solve their own or their relatives’ medical problem on their own. Thus, the sample innovations are unique or even superior to alternatives on the market, if there were any at all. This confirms Oliveira et al. (2015), who showed that patients are able to develop innovations that are novel with respect to prior art. The sample patients and caregivers had to rely on external knowledge to further develop their prototype. Owing to their own experience with a disease, their © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_14

147

148

14

Discussion

medical knowledge is typically highly developed. A medical doctor was involved in only one case as a co-developer with the patient. This confirms that the sample innovators already had high medical knowledge after having long dealt with an unmet medical need (von Hippel 1994). Yet, they lacked solution knowledge (Schweisfurth 2017)—sometimes technical knowledge, in almost all cases legal knowledge and regulatory knowledge—to further develop the device into a marketable product. The case of a gas and pipe engineer who developed a support structure for his aorta illustrates how previously available technical knowledge was mainly transferred from an existing knowledge domain into the medical device area (Haefliger et al. 2010). The same is true for a pressure sensor needed for an inhalation device. The inventor, a professional parachutist, took a pressure sensor from a reserve parachute and modified it to suit the needs of the inhalation device he designed for his son, who has cystic fibrosis. The sample patients and caregivers limited their development activities to not only fairly simple medical devices (class I or nonclassified), but also contributed to complex, riskier, and invasive medical devices (classes II and III). A recent study of makerspaces in Swedish hospitals revealed a similar picture: healthcare professionals developed 81% of lower-risk medical devices and 19% of higherrisk medical devices (Svensson and Hartmann 2018). Scholars of user innovation have long emphasized the own benefit a selfdeveloped innovation delivers to the user (von Hippel 2005; Lüthje and Herstatt 2004). After the own problem is solved, users generally have no incentive to further develop their prototype. Research into user entrepreneurship has shed some light on the commercialization process of user innovations: As outlined by Shah and Tripsas (2007), users receive feedback on their innovation by interacting with the public and with communities. In this study, all the innovators knew that many other patients have the same condition. Interactions with the public and communities had already taken place before the innovation was developed. Chronic, noncommunicable diseases are reaching epidemic proportions and are a big burden to societies worldwide (Daar et al. 2007). In 9 of 14 cases, the sample innovators addressed such a chronic disease, presumably because there is no prospect of a cure, which increased the need for a solution. Confirming extant research (Oliveira et al. 2015), 13 of 14 sample innovators stated that their quality of life was increased after they or the person they cared for had used the self-developed medical device. This is striking, because people with chronic diseases tend to have a lower quality of life than the general population (Wikman et al. 2011; Rothrock et al. 2010). Scholars from both public health (Steptoe et al. 2015) and social innovation (Pol and Ville 2009) have emphasized the importance of improving quality of life as a key outcome of innovation. To summarize this

14

Discussion

149

section on the individualist perspective, an increase in quality of life is the main motivation to further develop the medical device idea, to recognize the business opportunity based on the strong unmet need, and to become user entrepreneurs in order to solve their own problems and those of others with the same disease. Regarding users’ motivations to exploit their self-developed medical devices, the results of study 2 contradict Shah and Tripsas (2007), who did not expect medical devices to be brought to the market by innovators owing to the high opportunity costs and the turbulences in the market for medical devices. The 14 case studies convey initial evidence that, in some cases, particularly if there is a high (unmet medical) need, there is a higher likelihood that such barriers can also be overcome, even if opportunity costs are high or if the market is relatively stable. Most likely, Shah and Tripsas (2007) mainly considered healthcare professionals as users in the healthcare system who often face patients with unmet medical needs. For them, leaving their medical office and starting a venture carries high risk. Although there are some anecdotal examples of healthcare professionals who start a venture with their self-developed medical device (Smith and Shah 2013), most develop their innovations with incumbents (Chatterji et al. 2008; Lettl et al. 2006) and continue their regular work. Yet, patients and caregivers often do the development on their own, since established medical device manufacturers don’t accept their concepts for integration into their firms’ development activities. The expected benefits for patients and caregivers are more multifaceted than for companies, which mainly seek to maximize profit. Patients and caregivers seek to help themselves and, if they successfully solve their problem, also others with the same condition (Habicht et al. 2013). Since patients and caregivers often offer their medical device for a below-average price in the market, the nonpecuniary benefits of successfully helping others seem to partly substitute for pecuniary remuneration (Shah and Tripsas 2007; Podolny and Scott Morton 2002). This is a well-known pattern and has been addressed by social innovation scholars (Murray et al. 2010; Mulgan 2006). Nonetheless, from a structural perspective, high barriers such as patent application, regulatory approval, market access, and reimbursement by health insurances remain challenging to manage for nonprofessional developers of medical devices.

Preliminary Conclusions

15

Contents 15.1 15.2 15.3

Theoretical Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 Managerial Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 Limitations and Further Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154

In this section, theoretical and managerial implications of study 2 will be derived. Further, some limitations and ways to overcome these limitations are mentioned. In Part IV, I will integrate the findings from study 1 and study 2 and elaborate on the overall contribution of this thesis.

15.1

Theoretical Implications

This dissertation contributes to theory in several ways. First, I have shed light on a very under-researched phenomenon: user entrepreneurs for social innovation, particularly cases of patients and caregivers who develop a medical device. The innovative behaviors of patients and caregivers are rooted in their medical needs, which are unmet by existing medical devices. Likewise, von Hippel et al. (von Hippel et al. 2016) identified a similar market failure in the case of medical doctors and off-label drug discoveries. I have emphasized the importance of patients and caregivers who, in most cases besides healthcare professionals, are the de facto users of a medical device. With need knowledge of an unmet medical need, they have developed the device on their own and have acquired solution knowledge, particularly technical knowledge, legal knowledge, and regulatory knowledge if needed during the stages of the development process. To meet

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_15

151

152

15

Preliminary Conclusions

their medical needs, patients and caregivers even develop high-risk medical devices that require significant efforts for approval by regulatory agencies. Thus, it can be proposed that the greater the need for a solution, the greater the effort individuals are willing to take. This extends the scope of user innovation research to user entrepreneurs’ opportunity recognition and exploitation process. The second contribution relates to the emerging research field of social innovation (van der Have and Rubalcaba 2016; Cajaiba-Santana 2014) and the poorly understood connection between user innovation and social innovation (Kruse et al. 2019). Building on an unmet medical need, users such as patients and caregivers go beyond their own problem space and address the needs of many others. Although it is well known that users share their ideas with their peers in communities (Hienerth and Lettl 2011; Henkel 2006; Lakhani and von Hippel 2003), developing and bringing an approved medical device to market is a big burden to individuals. Yet, the sample users did not stop their innovative endeavor after they had completed the prototype for own usage, but continued the development over a significant period, because they were aware of the market demand for their device and the market failures of established medical device manufacturers. Financial considerations did not play a decisive role during opportunity exploitation. I argue that, in this extreme case of user entrepreneurship, patients and caregivers maximize their utility, since nonpecuniary benefits of increasing their own and others’ quality of life partly substitute for pecuniary remuneration (Battilana et al. 2012; Pol and Ville 2009). This benefit outweighs the barriers of high opportunity costs and the turbulences in the market for medical devices, as initially proposed by Shah and Tripsas (2007). Following the argumentation of Pol and Ville (2009) that social innovations are for the public good, and concerning needs not addressed by the market, the results of this study can confirm that the sample user entrepreneurs did develop social innovations: the innovations in my sample were priced below industry average in order to increase diffusion of the innovation (de Jong et al. 2018), were developed over an above industry timeframe under high uncertainty and ultimately led to an increase in quality of life of the sample innovators. Although there is a sound body of research into social entrepreneurship (Lettice and Parekh 2010; Short et al. 2009), the links between user entrepreneurs and social entrepreneurs must be further explored. My study is a first step to combining these two research streams. This opens opportunities for further research for scholars from both fields.

15.2 Managerial Implications

15.2

153

Managerial Implications

This study has shown that patients and caregivers develop medical devices according to their own needs. While healthcare professionals often work with incumbents to further develop their concepts (Chatterji et al. 2008), there have been very few interactions between medical device manufacturers and patients during the different stages of the product development process (Shah and Robinson 2007). Although there is evidence that opening a firm’s boundaries is valuable in the healthcare sector (Bullinger et al. 2012; Melese et al. 2009), I found that established medical device manufacturers have not yet developed mechanisms to integrate patient-developed concepts into their R&D processes. Manufacturers are generally skeptical towards concepts not developed within the own organization, particularly if development was done by a nonprofessional developer, such as a patient or a caregiver (not-invented-here syndrome) (Katz and Allen 1982). Yet, the integration of patients, caregivers, and healthcare professionals into the R&D process would allow medical device manufacturers to profit from usage-related resources that companies typically do not own (Schweisfurth and Herstatt 2016). This may be particularly valuable for new idea generation, but also for concept generation and testing. The financial impacts of such user-developed innovation are hard to assess. There is initial evidence from several countries that the overall user spending on innovative user-developed endeavors is close to or even exceeds firms’ R&D spending in a country (von Hippel 2017; von Hippel et al. 2011). Svensson and Hartmann (2018) found that innovations developed by healthcare professionals in a hospital-based makerspace had an about 14 times higher economic impact than the costs of establishing and maintaining the makerspace. Although I did not have access to the sample companies’ financial statements, the interviewees stated that their innovations generated sufficient returns for the inventor and for the firm’s operating costs. While these findings focus on the healthcare sector, they can be transferred to other sectors in which individuals develop solutions for their own unmet needs, such as the public sector or the humanitarian sector. Particularly the latter offers a range of opportunities to meaningfully combine user innovations and social innovations (Goeldner et al. 2017).

154

15.3

15

Preliminary Conclusions

Limitations and Further Research

Despite the findings’ originality and validity, this study has limitations, which may serve as cues for further research. First, the interview findings are limited to 14 individuals and need quantitative validation. A large-scale survey among founders in the medical device industry would be very valuable to quantify the number of user innovators, their motivations, and the barriers they face. These could be mapped regarding the opportunity recognition (individual perspective) and opportunity exploitation (structural perspective) stages of social innovation. Second, only successful innovators who managed to launch their medical device onto the market were analyzed in this study. A separate study with inventors who could not diffuse their medical device ideas would give important clues to better understand the results and to learn more about the significance of the barriers during the opportunity recognition and exploitation process. Despite these limitations, study 2 of this dissertation has contributed to the user innovation and social innovation literatures and has opened many avenues for further research at this vibrant intersection.

Part IV Integration of Findings, Implications, and Conclusion

Summary of Findings

16

In section 1.2, I developed three research questions for this dissertation: • How do patients and caregivers contribute to innovation in the healthcare sector? • What are the characteristics of patients and caregivers as innovators and how do they differ from other innovators in the healthcare sector? • What are the reasons for patients and caregivers diffusing their innovations to others and thus to become user entrepreneurs? In the course of this thesis, I have analyzed the phenomenon of patients and caregivers as innovators in the healthcare sector from different perspectives and found answers to these questions, which I will now summarize. I found evidence that patients and caregivers are a key source of innovation in the healthcare sector. There is sound evidence that patients and caregivers are developing medical devices according to their own and their relatives’ needs. Study 1 revealed that they have developed a substantial number of medical apps that were available in the Apple App Store in mid-2018. Of 1,265 medical apps in the sample, 134 were developed by patients and caregivers. A mediated regression analysis that considered 841,167 user ratings revealed that apps developed by patients and caregivers had significantly better ratings than apps developed by companies or healthcare professionals. It is remarkable that patients and caregivers develop apps with not only better ratings, but also financially successful medical apps. It seems that the unmet medical needs a patient or a caregiver personally encounters in daily life is a key success factor for developing a medical app. Although I only considered medical apps that are equivalent to a medical © The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_16

157

158

16

Summary of Findings

device, whether and how patients and caregivers also develop tangible medical devices remain open questions. In study 2, I conducted 14 case studies on developers of tangible medical devices. Together with the qualitative section of study 1 (in section 7.2), this shed light on the characteristics of patients and caregivers as innovators in healthcare. Medical device companies generally rely on their internal resources during the innovation process (i.e. technical knowledge, legal knowledge, regulatory knowledge, knowledge on production and distribution, and others). Although it has become increasingly popular to include external sources into the innovation process, there is evidence that, to date, mainly healthcare professionals have been included in several stages (Chatterji and Fabrizio 2014; Lettl et al. 2009; Chatterji et al. 2008). Compared to other innovators in healthcare, patients and caregivers draw on very different resources during their innovation processes. Owing to their own experience with a disease—irrespective they are a patient or caring for a loved one—their medical knowledge is highly developed, but distinct from healthcare professionals’ knowledge. This high medical knowledge in their domain and the unmet medical needs—in short, need knowledge—was the starting point for the development. Thus, the sample patients and caregivers had to rely on external knowledge, such as technical knowledge and, in almost all cases, legal and regulatory knowledge—in short, solution knowledge—to further develop their prototypes into marketable products. All 14 cases shared that intellectual property protection was achieved early, which turned out to be a clear advantage for further development. In both studies, I observed some patients and caregivers who sought to license their idea to an established medical device manufacturer. Only few were (after several attempts) successful, while the remainder further developed their products and managed to bring them to the market on their own without help from an incumbent. This indicates that some established medical device manufacturers do not have appropriate processes to incorporate outside knowledge into their innovation process. Yet, it also demonstrates the persistence of these individuals, since they were adamant that their products should also be available to others. The regulatory approval, which applied even for more invasive and thus complex class IIa/b or class III medical devices, and the complexity of the product development involved in such a regulatory approval, was a handicap yet not a deal-breaker for the overall development process of the sample patients and caregivers. The long development timeframe (on average, 5.4 years) undermines their patience and gives a glimpse of the barriers and pitfalls that occur during the process. Finally, I found that, in both qualitative studies (the qualitative part of study 1 and in study 2), owing to the own experience with the unmet medical needs,

16

Summary of Findings

159

there is a strong motivation for patients and caregivers to diffuse their innovation. Although some medical devices in both studies were financially very successful, financial considerations did not seem to be a decisive factor during the early development; rather, it was the pursuit to maximize their utility by improving their own and others’ quality of life instead of maximizing their pecuniary remuneration. Thus, I conclude that this maximization of quality of life indicates that the sample patients and caregivers developed social innovations for the public good, and concern needs currently not addressed by established medical device manufacturers. Since established manufacturers were in some cases unwilling to (co-)produce a medical device, some sample patients and caregivers had to become entrepreneurs in order to be able to diffuse their products. Other sample entrepreneurs decided early on that they wanted to bring their medical device to the market on their own and thus founded a company. The patients and caregivers who developed a medical app did not face such high barriers, since diffusing and marketing an app are fairly simple and don’t require large investments. Overall, I found strong evidence that patients and caregivers are developing medical devices, yet this happens predominantly under the general public’s radar. Thus, I derived a set of implications to better integrate these findings into theory, practice, and health policy.

Implications and Conclusion

17

Contents 17.1 17.2 17.3 17.4

17.1

Theoretical Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Managerial Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Implications for Health Policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

161 164 165 166

Theoretical Implications

The observation of the phenomenon that patients and caregivers develop medical devices according to their own unmet medical needs was the starting point for this thesis. In two separate studies, I shed light on this phenomenon from different perspectives. User innovation theory was the primary theoretical lens I adopted; further, I applied user entrepreneurship theory, social innovation theory and, to some extent, free innovation theory. In this section, I will discuss the interrelationships between the abovementioned—hitherto disconnected—research streams (see Figure 17.1). I have contributed to user innovation theory in several ways: I showed that user-developed products can compete with company-developed ones in the healthcare market. In study 1, I found that 11% of medical apps were developed by patients and caregivers and 29% by healthcare professionals. Thus, 40% of all medical apps in the sample were developed out of a need a person encountered during receiving or providing health-related services. This is in line with research from several other empirical fields, such as banking (van der Boor et al.

© The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Göldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6_17

161

162

17

Implications and Conclusion

2014; Oliveira and von Hippel 2011), scientific instruments (Riggs and von Hippel 1994), and sport (Hienerth et al. 2014) that were in a similar range (Bradonjic et al. 2019). Interestingly, most decision-makers (and most likely also the public) have under-estimated the percentage of user-developed products in a marketplace (Bradonjic et al. 2019). Since study 2’s results did not allow me to draw inferences about the share of the total population, a quantitative investigation with startups in the medical device sector could shed more light on the phenomenon of patients and caregivers as developers of tangible medical devices. Further, study 1 contributed to user innovation theory by providing a largescale, real-world dataset on the market performance of user-developed products that is not based on experimental data (Nishikawa et al. 2013; Poetz and Schreier 2012). The mediated regression analysis, which is based on 837 apps and 841,167 ratings on a five-star scale, is the first to analyze real-world data on the adoption of user-developed products in a marketplace without co-development by intermediaries. The analysis revealed that user-developed apps were rated significantly better than non-user-developed ones and that revenue is highest for patient-developed apps. This emphasizes the importance of usage-related knowledge, which companies typically do not have (Schweisfurth and Herstatt 2016) for the diffusion and the commercial success of medical apps. In study 1, I found that a large fraction of user-developers offers medical apps for free. Although company-developed apps was the largest fraction of free apps (I discussed possible reasons for this in section 8.2), about 50% of apps developed by patients and caregivers were released for free. Since free apps were downloaded more often (in my study, about 17 to 29 times more often), these patients and caregivers seem to be incentivized more by diffusion than by commercial success. In study 2, most sample patients and caregivers emphasized that they currently offered their tangible medical devices for a comparatively low price in order to increase diffusion. Owing to their market research, the patients and caregivers in both studies knew that their medical device was unique and not yet available on the market. Thus, the social component and the desire to increase others’ quality of life partly substituted for pecuniary remuneration. Figure 17.1 depicts how the benefits from own use are transformed into benefits from others using, and thus how user innovation and social innovation were interrelated. Yet, this diffusion process was much more challenging for the developers of a tangible medical device than for the developers of a medical app. While the diffusion of user-developed innovations is generally often cumbersome and without direct benefits for an innovator (de Jong et al. 2018), the diffusion of digital products and services is significantly faster and cheaper. About 50% of patients and caregivers who developed a medical app offered their app for free. Further

Monetary or processrelated benefits

163

+ User innovation

Social innovation

Journeys of patients and caregivers as user entrepreneurs

Others‘ benefit from use

Benefits from own use

17.1 Theoretical Implications

Manufacturer innovation

Paid development and/or no free revealing

Free user innovation

+ Free social innovation

Free innovation

Unpaid development and free revealing

Figure 17.1 Interrelationships between user innovation, free innovation, and social innovation theory

quantitative investigations of the developers of such free user innovations and particularly their motivations and their changes in quality of life would help us to better understand the interrelationships between the—as yet unrelated—research streams. The own unmet medical needs were the starting points for all sample patients and caregivers in both of my studies. They all recognized that their innovations were also valuable to others and began to produce and diffuse their medical devices. Most were motivated by helping fellow patients, but we should not underestimate the desire to achieve profits. While some of the sample patients and caregivers remained social innovators, others began to maximize their profits and to become regular entrepreneurs (see Figure 17.1). Nonetheless, most retained a social component and offered their medical device for a comparatively low price. Further studies should explore this intersection between user entrepreneurs and social entrepreneurs, since this opens several avenues for further research for scholars from both fields.

164

17.2

17

Implications and Conclusion

Managerial Implications

In the medical device industry, there has been a call to integrate diverse stakeholders into the innovation process (Symmank et al. 2015; Bullinger et al. 2012; Melese et al. 2009). Yet, to date, there has predominantly been evidence that it is beneficial to integrate innovative healthcare professionals (Chatterji and Fabrizio 2014; Hinsch et al. 2014; Lettl et al. 2006) or other hospital employees (Thune and Mina 2016) into medical device manufacturers’ R&D activities. I have shown that patients and caregivers developed medical devices on their own and even commercialized them later. Both studies’ results indicated that there was very little interaction between established medical device manufacturers and the sample patients and caregivers, although both sides would have benefited from such exchanges. Particularly established medical device manufacturers have not yet developed mechanisms to integrate patient-developed concepts into their innovation process. This is surprising, since it is well-known that integrating users helps to reduce information asymmetries and increases innovation process efficiency (Henkel and von Hippel 2004). One reason may be that medical device manufacturers perceive healthcare professionals and particularly physicians as their most important users. This may be true in some but not all cases. Another reason for skepticism toward patients and caregivers may be that they are nonprofessional developers. Healthcare professionals often have substantial reputations within their field of expertise and are seen as experts. Patients who have faced a disease for many years don’t have such a reputation, although they may have gained substantial knowledge about a disease and its treatment. Overall, this pattern is linked to the so-called not-invented-here syndrome (Katz and Allen 1982), in which companies are reluctant to accept concepts developed outside the own company’s boundaries. The integration of all three user-related stakeholders (i.e. patients, caregivers, and healthcare professionals) into the innovation process would allow medical device manufacturers to benefit from usage-related resources, which companies typically don’t have (Schweisfurth and Herstatt 2016). This may be particularly valuable for new idea generation, concept generation, and testing. Further, I also encourage medical device manufacturers to—if possible—hire people with the target disease or who know people with it, and to embed them into the organization (Schweisfurth and Raasch 2014). An example is the extremely successful1 Vienna-based digital health startup mySugr, which was founded by 1 The

startup mySugr was acquired by the pharmaceutical company Roche in 2017. Source: https://www.roche.com/media/releases/med-cor-2017-06-30.htm, accessed February 20,

17.3 Implications for Health Policy

165

two diabetics in 2012. According to the company’s website2 , a large fraction of employees live with diabetes. My dissertation has shown that such first-hand experience is extremely valuable and fosters innovation.

17.3

Implications for Health Policy

I have shown that users strongly contribute to the development of medical devices in the field of medical smartphone apps and to tangible medical devices. Considering this, one may question whether and how the increasing prevalence of users in medical device development increases social welfare. Previous research has shed some light on the positive influence of user-directed innovation and the benefits of manufacturers collaborating with users in the development of new products and services that eventually increase social welfare (Gambardella et al. 2017). Given the strong discrepancy between the public support that manufacturers and healthcare professionals now receive, compared to user innovators (Henkel and von Hippel 2004), health policymakers should identify ways to support patients and caregivers who aim to find solutions for their unmet medical needs and thus to improve the quality of their own and others’ lives. Supporting users and particularly patients and caregivers to develop medical devices for their own needs would encourage thousands of people to tackle their unmet medical needs by developing innovations that otherwise would not have been developed or that would not have been further developed owing to a lack of solution knowledge or a lack of resources. Since the user-developed apps in study 1 accounted for almost 1.5 million downloads in Germany, the UK, and the U.S., they impacted on a substantial set of people who aim to improve their medical outputs via a digital tool. An obvious policy recommendation is support for tools and places that foster local innovation. One of the rare examples for such an intervention was recently noted by Svensson and Hartmann (2018): in their study on six newly opened makerspaces in Sweden, they found that these innovations would not have been developed without the makerspace and that the potential returns from the innovations were more than 10 times larger than the initial investments. Thus, such fairly simple and locally applied interventions may already strongly impact on users 2019. Although the financial terms of the deal were not disclosed, a sales amount between e70 million and e100 million was circulated. Source: https://techcrunch.com/2017/07/ 07/diabetes-platform-mysugr-exits-to-roche-for-as-much-as-100m, accessed February 20, 2019. 2 Source: https://mysugr.com/about-us, accessed February 20, 2019.

166

17

Implications and Conclusion

and thus may be increasing social welfare in a community and beyond. Digital counterparts of such makerspaces are hackathons, i.e. a meetup of a serendipitous set of people from diverse backgrounds who work together on an intense, shortterm project and who compete with other teams (Angelidis et al. 2016). Such health-related hackathons are a relatively new phenomenon that has recently been proven to provide meaningful results (Olson et al. 2017). A systematic integration of patients and caregivers into such events would allow them to swiftly integrate missing solution knowledge and to bring medical devices into the market faster. The organizers of such events could also provide guidance along well-known pain points, such as the regulatory approval process, the identification of suitable financial support, or contact with healthcare professionals and health insurances. Policymakers should identify opportunities to integrate such activities into their portfolios so as to better support patient-directed innovation.

17.4

Conclusion

I was actually diagnosed really late, my sister was diagnosed at 11, I was diagnosed at 35. And I was already into the diabetes industry at that point. I started to work for a diabetes company [], but I was not diabetic at that point, and worked there for 5 years. Then I started at another company, and right about that time after I got that job, I got diagnosed, which is crazy right? So here I am, the director of the advanced technologies section in an insulin pump company diagnosed with type 1 diabetes. So I was trying to figure out what changes we need to bring to our product so more people can use it. [] And it turns out that 93% of people who use insulin [with diabetes type 1 and diabetes type 2] don’t use an insulin pump, but pens and syringes. So here I am, I am newly diagnosed type 1, hearing that 93% of people won’t use this product I am developing and by the way I don’t use it, right? But my company did not care, they said: We make insulin pumps here because we can sell insulin pumps for 5,000 bucks. I asked how do we bring the benefits of insulin pumps to pen and syringe users? So the original idea was not to start smart pen company necessarily, but it was to work with smart pens and you know, honestly I couldn’t find anybody who is doing it the way I thought it needed to be done…3

A key challenge for medical device manufacturers is how to integrate outside knowledge into the organization and particularly into the innovation process for new products and services (Symmank et al. 2015; Melese et al. 2009). The abovementioned example, which I encountered during my research, underlines the 3 Source:

personal interview with patient 9 (study 2) on April 20, 2016.

17.4 Conclusion

167

importance of need knowledge for successful innovation. Although the developer’s sister was a diabetic and he was working for a company in this field, his diagnosis with diabetes type 1 and thus the accompanying daily routines were the trigger for his idea and subsequently for the founding of his company. I found significant evidence that patients and caregivers are a key yet unknown source of innovation in the healthcare sector. The sample patients and caregivers developed both medical smartphone apps and tangible medical devices that are used by thousands of other patients across the globe. They did not hesitate to develop even complex, highly regulated medical devices and demonstrated dedication over long periods. In the two studies in this thesis, I highlighted the importance of patients’ and caregivers’ need knowledge owing to their daily struggles with a disease. The sample patients and caregivers were highly knowledgeable about their individual medical subjects. Thus, their medical knowledge is distinct from that of healthcare professionals. Medical device manufacturers should consider including the two complementary need knowledge sets into their innovation process so as to better incorporate the needs of both intermediaries such as healthcare professionals and end-users such as patients. I also presented evidence that patients and caregivers have a strong motivation to diffuse their innovation. In study 1, I found that medical apps developed by patients and caregivers had significantly better user ratings, indicating that these apps are strongly appreciated by other patients. This motivates these developers to further develop and diffuse their medical apps. In study 2, the sample patients and caregivers sought to improve the quality of their and many other patients’ lives and thus were willing to distribute their devices for a comparatively low price. Since they addressed unmet medical needs that were currently not addressed by established medical device manufacturers, their social innovation positively impacted on social welfare. Since digitalization and the increasing prevalence of chronic diseases are advancing, I assume that patients and caregivers as developers of medical devices will also increase in the future. This thesis is a first step to better understanding and valuing this key yet unknown and under-researched phenomenon.

References

Abbott, L.M. and Smith, S.D., 2018. Smartphone apps for skin cancer diagnosis: Implications for patients and practitioners. Australasian Journal of Dermatology, 59(3), pp. 168–170. Adams, P., Fontana, R. and Malerba, F., 2013. The magnitude of innovation by demand in a sectoral system: The role of industrial users in semiconductors. Research Policy, 42(1), pp. 1–14. Agu, E., Pedersen, P., Strong, D., Tulu, B., He, Q., Wang, L. and Li, Y., 2013. The smartphone as a medical device: Assessing enablers, benefits and challenges. 2013 IEEE International Conference on Sensing, Communications and Networking (SECON), pp. 76–80. Amann, J., Zanini, C. and Rubinelli, S., 2016. What online user innovation communities can teach us about capturing the experiences of patients living with chronic health conditions. A scoping review. PLoS ONE, 11(6), pp. 1–26. Angelidis, P., Berman, L., De La, M., Casas-Perez, L., Celi, L.A., Dafoulas, G.E., Dagan, A., et al., 2016. The hackathon model to spur innovation around global mHealth HHS Public Access. Journal of Medical Engineering & Technology, 40(7–8), pp. 392–399. Armstrong, D., Gosling, A., Weinman, J. and Marteau, T., 1997. The place of inter-rater reliability in qualitative research: an empirical study. Sociology, 31(3), pp. 597–606. Bacq, S. and Janssen, F., 2011. The multiple faces of social entrepreneurship: A review of definitional issues based on geographical and thematic criteria. Entrepreneurship and Regional Development, 23(5–6), pp. 373–403. Baden-Fuller, C. and Haefliger, S., 2013. Business Models and Technological Innovation. Long Range Planning, 46(6), pp. 419–426. Bakker, D., Kazantzis, N., Rickwood, D. and Rickard, N., 2018. A randomized controlled trial of three smartphone apps for enhancing public mental health. Behaviour Research and Therapy, 109(February), pp. 75–83. Baldwin, C., Hienerth, C. and von Hippel, E., 2006. How user innovations become commercial products: A theoretical investigation and case study. Research Policy, 35(9), pp. 1291–1313.

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2021 M. Gldner, Patients and Caregivers as Developers of Medical Devices, Forschungs-/Entwicklungs-/Innovations-Management, https://doi.org/10.1007/978-3-658-32041-6

169

170

References

Baldwin, C. and von Hippel, E., 2011. Modeling a paradigm shift: From producer innovation to user and open collaborative innovation. Organization Science, 22(6), pp. 1399–1417. Balka, K., Raasch, C. and Herstatt, C., 2010. How Open is Open Source? – Software and Beyond. Creativity and Innovation Management, 19(3), pp. 248–256. Bansal, P. (Tima), Smith, W.K. and Vaara, E., 2018. New Ways of Seeing through Qualitative Research. Academy of Management Journal, 61(4), pp. 1189–1195. Baron, R.A., 2006. Opportunity Recognition as Pattern Recognition. Academy of Management Perspectives, 20, pp. 104–120. Basole, R.C. and Karla, J., 2012. Value Transformation in the Mobile Service Ecosystem: A Study of App Store Emergence and Growth. Service Science, 4(1), pp. 24–41. Battilana, J., Lee, M., Walker, J. and Dorsey, C., 2012. In search of the hybrid ideal. Stanford Social Innovation Review, 10(3 (Summer)), pp. 51–55. Block, J.H., Henkel, J., Schweisfurth, T.G. and Stiegler, A., 2016. Commercializing user innovations by vertical diversification: The user–manufacturer innovator. Research Policy, 45(1), pp. 244–259. Bodkin, C. and Miaoulis, G., 2007. eHealth information quality and ethics issues: an exploratory study of consumer perceptions. International Journal of Pharmaceutical and Healthcare Marketing, 1(1), pp. 27–42. Bogers, M., Afuah, A. and Bastian, B., 2010. Users as Innovators: A Review, Critique, and Future Research Directions. Journal of Management, 36(4), pp. 857–875. Bogers, M. and West, J., 2012. Managing Distributed Innovation: Strategic Utilization of Open and User Innovation. Creativity and Innovation Management, 21(1), pp. 61–75. van der Boor, P., Oliveira, P. and Veloso, F., 2014. Users as innovators in developing countries: The global sources of innovation and diffusion in mobile banking services. Research Policy, 43(9), pp. 1594–1607. Box, G.E.P. and Cox, D.R., 1964. An Analysis of Transformations. Journal of the Royal Statistical Society. Series B (Methodological), 26(2), pp. 221–252. Bradonjic, P., Franke, N. and Lüthje, C., 2019. Decision-makers’ underestimation of user innovation. Research Policy, 48(6), pp. 1354–1361. Braun, V. and Herstatt, C., 2008. The Freedom-Fighters: How Incumbent Corporations Are Attempting To Control User-Innovation. International Journal of Innovation Management, 12(03), pp. 543–572. Brem, A., Bilgram, V. and Marchuk, A., 2017. How crowdfunding platforms change the nature of user innovation—from problem solving to entrepreneurship. Technological Forecasting and Social Change, (October), pp. 1–13. Brynjolfsson, E., Hu, Y. (Jeffrey) and Simester, D., 2011. Goodbye Pareto Principle, Hello Long Tail: The Effect of Search Costs on the Concentration of Product Sales. Management Science, 57(8), pp. 1373–1386. Budych, K., Helms, T. and Schultz, C., 2012. How do patients with rare diseases experience the medical encounter? Exploring role behavior and its impact on patient–physician interaction. Health Policy, 105. Bullinger, A.C., Rass, M., Adamczyk, S., Moeslein, K.M. and Sohn, S., 2012. Open innovation in health care: Analysis of an open health platform. Health Policy, 105(2–3), pp. 165–175.

References

171

Cajaiba-Santana, G., 2014. Social innovation: Moving the field forward. A conceptual framework. Technological Forecasting and Social Change, 82, pp. 42–51. Camacho, N., Landsman, V. and Stremersch, S., 2010. The connected patient. In S. Wuyts et al., eds. The connected customer: The Changing Nature of Consumer and Business Markets. New York: Taylor and Francis Group, LLC, pp. 107–135. Camerini, L., Schulz, P.J. and Nakamoto, K., 2012. Differential effects of health knowledge and health empowerment over patients’ self-management and health outcomes: A crosssectional evaluation. Patient Education and Counseling, 89(2), pp. 337–344. Chandra, Y. and Leenders, M.A.A.M., 2012. User innovation and entrepreneurship in the virtual world: A study of Second Life residents. Technovation, 32(7–8), pp. 464–476. Chatterji, A.K., Fabrizio, K.R., Mitchell, W. and Schulman, K.A., 2008. Physician-industry cooperation in the medical device industry. Health Affairs, 27(6), pp. 1532–43. Chatterji, A.K., 2009. Spawned with a silver spoon? Entrepreneurial performance and innovation in the medical device industry. Strategic Management Journal, 30(2), pp. 185–206. Chatterji, A.K. and Fabrizio, K.R., 2012. How Do Product Users Influence Corporate Invention? Organization Science, 23(4), pp. 971–987. Chatterji, A.K. and Fabrizio, K.R., 2014. Using users: When does external knowledge enhance corporate product innovation? Strategic Management Journal, 35(14), pp. 1427–1445. Chen, C.W., 2017. Five-star or thumbs-up? The influence of rating system types on users’ perceptions of information quality, cognitive effort, enjoyment and continuance intention. Internet Research, 27(3), pp. 478–494. Choi, Y.R., Lévesque, M. and Shepherd, D.A., 2008. When should entrepreneurs expedite or delay opportunity exploitation? Journal of Business Venturing, 23(3), pp. 333–355. Choi, Y.R. and Shepherd, D.A., 2004. Entrepreneurs’ decisions to exploit opportunities. Journal of Management, 30(3), pp. 377–395. Cohen, J., 1968. Weighted kappa: nominal scale agreement with provision for scaled disagreement or partial credit. Psychological Bulletin, 70(4), pp. 213–220. Cohen, W.M. and Levinthal, D.A., 1990. Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35(1), pp. 128–152. Connor, K., Brady, R.R.W., de Beaux, A. and Tulloh, B., 2013. Contemporary hernia smartphone applications (apps). Hernia, 18(4), pp. 557–561. Cooper, N., Hazeldine, S. and Quaggiotto, G., 2017. Two Paths to Supporting Grassroots Innovation. Stanford Social Innovation Review. Cortez, N.G., Cohen, I.G. and Kesselheim, A.S., 2014. FDA Regulation of Mobile Health Technologies. New England Journal of Medicine, 371(4), pp. 372–379. Costanza, R., Fisher, B., Ali, S., Beer, C., Bond, L., Boumans, R., Danigelis, N.L., et al., 2006. Quality of life: An approach integrating opportunities, human needs, and subjective well-being. Ecological Economics, 61(2–3), pp. 267–276. Daar, A.S., Singer, P.A., Persad, L., Pramming, S.K., Matthews, R., Beaglehole, R., Borysiewicz, L.K., et al., 2007. Grand challenges in chronic non-communicable diseases. Nature, 450(7169), pp. 494–496. Dala-Ali, B.M., Lloyd, M.A. and Al-Abed, Y., 2011. The uses of the iPhone for surgeons. The Surgeon, 9(1), pp. 44–8.

172

References

DeMonaco, H., Oliveira, P., Torrance, A., von Hippel, C. and von Hippel, E., 2019. When patients become innovators. MIT Sloan Management Review, 60(3), pp. 81–88. DeMonaco, H., Ali, A. and von Hippel, E., 2006. The major role of clinicians in the discovery of off-label drug therapies. Pharmacotherapy, 26(3), pp. 323–32. Directive Council, 2007. EU-Directive 93/42/EEC concerning medical devices. Official Journal of the EU, (M5), pp. 1–60. Douglas, E.J. and Shepherd, D.A., 2000. Entrepreneurship as a utility maximizing response. Journal of Business Venturing, 15(3), pp. 231–251. Durbin, J. and Waston, G., 1951. Testing for Serial Correlation in Least Square Regression. Biometrika, 38(1/2), pp. 159–177. Dutta, D.K. and Crossan, M.M., 2005. The nature of entrepreneurial Opportunities: Understanding the process using the 4I organizational learning framework. Entrepreneurship: Theory and Practice, 29(4), pp. 425–449. Dwivedi, Y.K., Shareef, M.A., Simintiras, A.C., Lal, B. and Weerakkody, V., 2016. A generalised adoption model for services: A cross-country comparison of mobile health (m-health). Government Information Quarterly, 33(1), pp. 174–187. Eckhardt, J.T. and Shane, S.A., 2003. Opportunities and Entrepreneurship. Journal of Management, 29(3), pp. 333–349. Edwards-Schachter, M. and Wallace, M.L., 2017. ‘Shaken, but not stirred’: Sixty years of defining social innovation. Technological Forecasting and Social Change, 119, pp. 64– 79. Eisenhardt, K.M., 1989. Building Theories from Case Study Research. Academy of Management Review, 14(4), p. 532. Eisenhardt, K.M. and Graebner, M.E., 2007. Theory building from cases: Opportunities and challenges. Academy of Management Journal, 50(1), pp. 25–32. Elberse, J.E., Caron-Flinterman, J.F. and Broerse, J.E.W., 2011. Patient-expert partnerships in research: how to stimulate inclusion of patient perspectives. Health Expectations, 14(3), pp. 225–39. Elenko, E., Underwood, L. and Zohar, D., 2015. Defining digital medicine. Nature Biotechnology, 33(5), pp. 456–461. Ellis, P.D., 2011. Social ties and international entrepreneurship: Opportunities and constraints affecting firm internationalization. Journal of International Business Studies, 42(1), pp. 99–127. Engler, T.H., Winter, P. and Schulz, M., 2015. Understanding online product ratings: A customer satisfaction model. Journal of Retailing and Consumer Services, 27, pp. 113– 120. Estellés-Arolas, E. and González-Ladrón-De-Guevara, F., 2012. Towards an integrated crowdsourcing definition. Journal of Information Science, 38(2), pp. 189–200. Eysenbach, G., 2001. What is e-health? Journal of Medical Internet Research, 3(2). Fargen, K.M., Frei, D., Fiorella, D., McDougall, C.G., Myers, P.M., Hirsch, J.A. and Mocco, J., 2013. The FDA approval process for medical devices: An inherently flawed system or a valuable pathway for innovation? Journal of NeuroInterventional Surgery, 5(4), pp. 269–275. Field, A., 2009. Discovering Statistics using SPSS 3rd ed., Sage Publications. Franke, N. and von Hippel, E., 2003. Satisfying heterogeneous user needs via innovation toolkits: the case of Apache security software. Research Policy, 32(7), pp. 1199–1215.

References

173

Franke, N., von Hippel, E. and Schreier, M., 2006. Finding commercially attractive user innovations: A test of lead-user theory. Journal of Product Innovation Management, 23(4), pp. 301–315. Franke, N., Keinz, P. and Schreier, M., 2008. Complementing Mass Customization Toolkits with User Communities: How Peer Input Improves Customer Self-Design. Journal of Product Innovation Management, 25(6), pp. 546–559. Franke, N., Schirg, F. and Reinsberger, K., 2015. The frequency of end-user innovation: A re-estimation of extant findings. Research Policy, 45(8), pp. 1684–1689. Franke, N. and Shah, S., 2003. How communities support innovative activities: an exploration of assistance and sharing among end-users. Research Policy, 32(1), pp. 157–178. Franko, O.I. and Bhola, S., 2011. iPad apps for orthopedic surgeons. Orthopedics, 34(12), pp. 978–81. Fuchs, C., Prandelli, E., Schreier, M. and Dahl, D.W., 2013. All That is Users Might Not be Gold: How Labeling Products as User Designed Backfires in the Context of Luxury Fashion Brands. Journal of Marketing, 77(5), pp. 75–91. Füller, J., Schroll, R. and von Hippel, E., 2013. User generated brands and their contribution to the diffusion of user innovations. Research Policy, 42(6–7), pp. 1197–1209. Gambardella, A., Raasch, C. and von Hippel, E., 2017. The User Innovation Paradigm : Impacts on Markets and Welfare. Management Science, 63(5), pp. 1450–1468. Garg, R. and Telang, R., 2013. Inferring app demand from publicly available data. MIS Quarterly, 37(4), pp. 1253–1264. Gehman, J., Glaser, V.L., Eisenhardt, K.M., Gioia, D., Langley, A. and Corley, K.G., 2018. Finding Theory–Method Fit: A Comparison of Three Qualitative Approaches to Theory Building. Journal of Management Inquiry, 27(3), pp. 284–300. George, N.M., Parida, V., Lahti, T. and Wincent, J., 2016. A systematic literature review of entrepreneurial opportunity recognition: insights on influencing factors. International Entrepreneurship and Management Journal, 12(2), pp. 309–350. Gibbert, M., Ruigrok, W. and Wicki, B., 2008. What passes as a rigorous case study? Strategic Management Journal, 29(13), pp. 1465–1474. Gioia, D.A., Corley, K.G. and Hamilton, A.L., 2013. Seeking Qualitative Rigor in Inductive Research: Notes on the Gioia Methodology. Organizational Research Methods, 16(1), pp. 15–31. Goeldner, M. and Herstatt, C., 2016. Are Patients and Relatives the Better Innovators? The Case of Medical Smartphone Applications. Technology and Innovation Management, Hamburg University of Technology, Working Paper No. 91. Goeldner, M., Kruse, D.J. and Herstatt, C., 2017. Lead User Method vs. Innovation Contest – An Empirical Comparison of Two Open Innovation Methodologies for Identifying Social Innovation for Flood Resilience in Indonesia. Technology and Innovation Management, Hamburg University of Technology, Working Paper No. 101. Goffin, K., Åhlström, P., Bianchi, M. and Richtnér, A., 2019. State-of-the-art: The quality of case study research in innovation management. Journal of Product Innovation Management, 36(5), pp. 586–615. Goodman, R.A., Posner, S.F., Huang, E.S., Parekh, A.K. and Koh, H.K., 2013. Defining and measuring chronic conditions: imperatives for research, policy, program, and practice. Preventing Chronic Disease, 10(Mcc), p. E66.

174

References

Greenhalgh, T., 2009. Chronic illness: beyond the expert patient. BMJ, 338(7695), pp. 629–631. Gregoire, D. a., Barr, P.S. and Shepherd, D. a., 2009. Cognitive Processes of Opportunity Recognition: The Role of Structural Alignment. Organization Science, 21(2), pp. 413– 431. Gregoire, D.A., Barr, P.S. and Shepherd, D.A., 2009. Cognitive Processes of Opportunity Recognition: The Role of Structural Alignment. Organization Science, 21(2), pp. 413– 431. Habicht, H., Oliveira, P. and Shcherbatiuk, V., 2013. User Innovators: When Patients Set Out to Help Themselves and End Up Helping Many. Die Unternehmung, 66(3), pp. 277–294. Haefliger, S., Jäger, P. and von Krogh, G., 2010. Under the radar: Industry entry by user entrepreneurs. Research Policy, 39(9), pp. 1198–1213. Hair, J.F., Black, W.C., Babin, B.J. and Anderson, R.E., 2014. Multivariate data analysis 7th ed., Pearson. Halbinger, M.A., 2018. The role of makerspaces in supporting consumer innovation and diffusion: An empirical analysis. Research Policy, 47(10), pp. 2028–2036. Hamdi-Kidar, L., Keinz, P., Le Nagard, E. and Vernette, E., 2019. Comparing Lead Users to Emergent-nature Consumers as Sources of Innovation at Early Stages of New Product Development. Journal of Product Innovation Management, 36(5), pp. 616–631. Harhoff, D., Henkel, J. and von Hippel, E., 2003. Profiting from voluntary information spillovers: how users benefit by freely revealing their innovations. Research Policy, 32(10), pp. 1753–1769. Hartzband, P. and Groopman, J., 2010. Untangling the Web—patients, doctors, and the Internet. New England Journal of Medicine, 362(12), pp. 1063–1066. Hartzler, A. and Pratt, W., 2011. Managing the Personal Side of Health: How Patient Expertise Differs from the Expertise of Clinicians. Journal of Medical Internet Research, 13(3). van der Have, R.P. and Rubalcaba, L., 2016. Social innovation research: An emerging area of innovation studies? Research Policy, 45(9), pp. 1923–1935. Hayes, A.F., 2017. Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach 2nd ed., Guilford Publications. Henkel, J., 2006. Selective revealing in open innovation processes: The case of embedded Linux. Research Policy, 35(7), pp. 953–969. Henkel, J. and von Hippel, E., 2004. Welfare implications of user innovation A. N. Link & F. M. Scherer, eds. The Journal of Technology Transfer, 30(1–2), pp. 73–87. Herstatt, C. and von Hippel, E., 1992. From Experience: Developing New Product Concepts Via the Lead User Method: A Case Study in a “Low-Tech” Field. Journal of Product Innovation Management, 9(3), pp. 213–221. Herstatt, C. and Raasch, C., 2007. So halten Sie Konkurrenten auf Distanz. Harvard Business Manager, 11/2017, p. 66. Hienerth, C., von Hippel, E. and Jensen, M.B., 2014. User community vs. producer innovation development efficiency: A first empirical study. Research Policy, 43(1), pp. 190–201.

References

175

Hienerth, C. and Lettl, C., 2011. Exploring how peer communities enable lead user innovations to become standard equipment in the industry: Community pull effects. Journal of Product Innovation Management, 28, pp. 175–195. Hienerth, C. and Lettl, C., 2017. Perspective: Understanding the Nature and Measurement of the Lead User Construct. Journal of Product Innovation Management, 34(1), pp. 3– 12. Hinsch, M.E., Stockstrom, C. and Lüthje, C., 2014. User Innovation in Techniques: A Case Study Analysis in the Field of Medical Devices. Creativity and Innovation Management, 23(4), pp. 484–494. von Hippel, E., 2009. Democratizing innovation: The evolving phenomenon of user innovation. International Journal of Innovation Science, 1(1), pp. 29–40. von Hippel, E., 2005. Democratizing Innovation, Cambridge: MIT Press. von Hippel, E., 1998. Economics of Product Development by Users: The Impact of “Sticky” Local Information. Management Science, 44(5), pp. 629–644. von Hippel, E., 2017. Free Innovation, Cambridge: MIT Press. von Hippel, E., 1986. Lead users: a source of novel product concepts. Management Science, 32(7), pp. 791–806. von Hippel, E., 1994. “‘Sticky Information’” and the Locus of Problem Solving: Implications for Innovation. Management Science, 40(4), pp. 429–439. von Hippel, E., 1976. The dominant role of users in the scientific instrument innovation process. Research Policy, 5(3), pp. 212–239. von Hippel, E., 2001. User toolkits for innovation. Journal of Product Innovation Management, 18(4), pp. 247–257. von Hippel, E., DeMonaco, H. and de Jong, J.P.J., 2016. Market failure in the diffusion of clinician-developed innovations: The case of off-label drug discoveries. Science and Public Policy, 44(August 2016), p.scw042. von Hippel, E. and Katz, R., 2002. Shifting Innovation to Users via Toolkits. Management Science, 48(7), pp. 821–833. von Hippel, E. and von Krogh, G., 2006. Free revealing and the private and privatecollective model. R&D Management, 36(3), pp. 295–306. von Hippel, E., Ogawa, S. and de Jong, J.P.J., 2011. The age of the consumer-innovator. MIT Sloan Management Review, 53(1), pp. 27–35. von Hippel, E. and Tyre, M.J., 1995. How learning by doing is done: problem identification in novel process equipment. Research Policy, 24(1), pp. 1–12. Holman, H. and Lorig, K., 2004. Patient self-management: a key to effectiveness and efficiency in care of chronic disease. Public Health Reports, 119(3), pp. 239–243. Howells, J., 2006. Intermediation and the role of intermediaries in innovation. Research Policy, 35(5), pp. 715–728. Hussain, M., Al-Haiqi, A., Zaidan, A.A., Zaidan, B.B., Kiah, M.L.M., Anuar, N.B. and Abdulnabi, M., 2015. The landscape of research on smartphone medical apps: Coherent taxonomy, motivations, open challenges and recommendations. Computer Methods and Programs in Biomedicine, 122(3), pp. 393–408. Istepanian, R.S.H., Jovanov, E. and Zhang, Y.T., 2004. Introduction to the Special Section on M-Health: Beyond Seamless Mobility and Global Wireless Health-Care Connectivity. IEEE Transactions on Information Technology in Biomedicine, 8(4), pp. 405–414.

176

References

Jacobs, W., Amuta, A.O. and Jeon, K.C., 2017. Health information seeking in the digital age: An analysis of health information seeking behavior among US adults. Cogent Social Sciences, 3(1), pp. 1–11. Jeppesen, L.B. and Frederiksen, L., 2006. Why Do Users Contribute to Firm-Hosted User Communities? The Case of Computer-Controlled Music Instruments. Organization Science, 17(1), pp. 45–63. Jin, C., Su, Y.-S., De Jong, J.P.J. and von Hippel, E., 2018. Household Sector Innovation in China: Impacts of Income and Development. Available at SSRN 3205423. Johnson, R.B. and Onwuegbuzie, A.J., 2004. Mixed Methods Research: A Research Paradigm Whose Time Has Come. Educational Researcher, 33(7), pp. 14–26. de Jong, J.P.J., von Hippel, E., Gault, F., Kuusisto, J. and Raasch, C., 2015. Market Failure in the Diffusion of Consumer-Developed Innovations: Patterns in Finland. Research Policy, 44, pp. 1856–1865. de Jong, J.P.J., 2013. The decision to exploit opportunities for innovation: A study of hightech small-business owners. Entrepreneurship: Theory and Practice, 37(31), pp. 281– 301. de Jong, J.P.J., Gillert, N.L. and Stock, R.M., 2018. First adoption of consumer innovations: Exploring market failure and alleviating factors. Research Policy, 47(2), pp. 487–497. de Jong, J.P.J. and von Hippel, E., 2009. Transfers of user process innovations to process equipment producers: A study of Dutch high-tech firms. Research Policy, 38(7), pp. 1181–1191. Joseph-Williams, N., Elwyn, G. and Edwards, A., 2014. Knowledge is not power for patients: A systematic review and thematic synthesis of patient-reported barriers and facilitators to shared decision making. Patient Education and Counseling, 94(3), pp. 291–309. Kalogerakis, K., Lüthje, C. and Herstatt, C., 2010. Developing innovations based on analogies: Experience from design and engineering consultants. Journal of Product Innovation Management, 27, pp. 418–436. Kamerow, D., 2013. Regulating medical apps: which ones and how much? BMJ, 347(f6009). Kampker, A., Deutskens, C., Maue, A. and Hollah, A., 2016. Elektromobile Logistik. In C. Deckert, ed. CSR und Logistik. Springer Gabler, pp. 293–307. Kanstrup, A.M., Bertelsen, P. and Nøhr, C., 2015. Patient innovation: An analysis of patients’ designs of digital technology support for everyday living with diabetes. Health Information Management Journal, 44(1), pp. 12–20. Katz, R. and Allen, T.J., 1982. Investigating the Not Invented Here (NIH) syndrome: A look at the performance, tenure, and communication patterns of 50 R & D Project Groups. R&D Management, 12(1), pp. 7–20. Kesselheim, A.S., Xu, S. and Avorn, J., 2014. Clinicians’ contributions to the development of coronary artery stents: A qualitative study of transformative device innovation. PLoS ONE, 9(2). Khechine, H., Pascot, D. and Prémont, P., 2008. Use of health-related information from the Internet by English-speaking patients. Health Informatics Journal, 14(1), pp. 17–28.

References

177

Kohlbacher, F., Herstatt, C. and Levsen, N., 2015. Golden opportunities for silver innovation: How demographic changes give rise to entrepreneurial opportunities to meet the needs of older people. Technovation, 39–40, pp. 73–82. Kostkova, P., 2015. Grand Challenges in Digital Health. Frontiers in Public Health, 3(May), pp. 1–5. Kramer, D.B., Xu, S. and Kesselheim, A.S., 2012. How does medical device regulation perform in the United States and the European union? A systematic review. PLoS Medicine, 9(7). Kranz, M., Möller, A., Hammerla, N., Diewald, S., Plötz, T., Olivier, P. and Roalter, L., 2013. The mobile fitness coach: Towards individualized skill assessment using personalized mobile devices. Pervasive and Mobile Computing, 9(2), pp. 203–215. Kristensson, P., Gustafsson, A. and Archer, T., 2004. Harnessing the Creative Potential among Users. Journal of Product Innovation Management, 21(1), pp. 4–14. von Krogh, G. and von Hippel, E., 2006. The Promise of Research on Open Source Software. Management Science, 52(7), pp. 975–983. von Krogh, G., Rossi-Lamastra, C. and Haefliger, S., 2012. Phenomenon-based research in management and organisation science: When is it rigorous and does it matter? Long Range Planning, 45(4), pp. 277–298. Kruse, D.J., Goeldner, M., Eling, K. and Herstatt, C., 2019. Looking for a Needle in a Haystack: How to Search for Bottom-up Social Innovations that Solve Complex Humanitarian Problems. Journal of Product Innovation Management, 36(6), pp. 671– 694. Kübler, R., Pauwels, K., Yildirim, G. and Fandrich, T., 2018. App Popularity: Where in the World Are Consumers Most Sensitive to Price and User Ratings? Journal of Marketing, 82(5), pp. 20–44. Lakhani, K. and von Hippel, E., 2003. How open source software works: “free” user-touser assistance. Research Policy, 32(6), pp. 923–943. Lakhani, K. and Wolf, R.G., 2005. Why Hackers Do What They Do: Understanding Motivation and Effort in Free/Open Source Software Projects. In J. Feller et al., eds. Perspectives on Free and Open Source Software. Cambridge: MIT Press, pp. 1–27. de Langhe, B., Fernbach, P.M. and Lichtenstein, D.R., 2016. Navigating by the stars: Investigating the actual and perceived validity of online user ratings. Journal of Consumer Research, 42(6), pp. 817–833. Langlois, R.N. and Robertson, P.L., 1992. Networks and innovation in a modular system: Lessons from the microcomputer and stereo component industries. Research Policy, 21(4), pp. 297–313. Lee, G. and Raghu, T.S., 2014. Determinants of Mobile Apps’ Success: Evidence from the App Store Market. Journal of Management Information Systems, 31(2), pp. 133–170. Lettice, F. and Parekh, M., 2010. The social innovation process: themes, challenges and implications for practice. Int. J. Technology Management, 51(1), pp. 139–158. Lettl, C., Herstatt, C. and Gemuenden, H.G., 2006. Users’ contributions to radical innovation: evidence from four cases in the field of medical equipment technology. R&D Management, 36(3), pp. 251–272.

178

References

Lettl, C., Hienerth, C. and Gemuenden, H.G., 2008. Exploring How Lead Users Develop Radical Innovation: Opportunity Recognition and Exploitation in the Field of Medical Equipment Technology. IEEE Transactions on Engineering Management, 55(2), pp. 219–233. Lettl, C., Rost, K. and von Wartburg, I., 2009. Why are some independent inventors ‘heroes’ and others ‘hobbyists’? The moderating role of technological diversity and specialization. Research Policy, 38(2), pp. 243–254. Lilien, G.L., Morrison, P.D., Searls, K., Sonnack, M. and Hippel, E. Von, 2002. Performance Assessment of the Lead User Idea-Generation Process for New Product Development. Management Science, 48(8), pp. 1042–1059. Liu, C.Z., Au, Y.A. and Choi, H.S., 2014. Effects of Freemium Strategy in the Mobile App Market: An Empirical Study of Google Play. Journal of Management Information Systems, 31(3), pp. 326–354. Lumley, T., Diehr, P., Emerson, S. and Chen, L., 2002. The Importance of the Normality Assumption in Large Public Health Data Sets. Annual Review of Public Health, 23, pp. 151–169. Lüthje, C. and Herstatt, C., 2004. The Lead User method: an outline of empirical findings and issues for future research. R&D Management, 34(5), pp. 553–568. Lüthje, C., Herstatt, C. and von Hippel, E., 2005. User-innovators and “local” information: The case of mountain biking. Research Policy, 34(6), pp. 951–965. Magnusson, P.R., Wästlund, E. and Netz, J., 2016. Exploring Users’ Appropriateness as a Proxy for Experts When Screening New Product/Service Ideas. Journal of Product Innovation Management, 33(1), pp. 4–18. Mascie-Taylor, C.G.N. and Karim, E., 2003. The burden of chronic disease. Science, 302(5652), pp. 1921–1922. McDermott, C.M. and O’Connor, G.C., 2002. Managing radical innovation: An overview of emergent strategy issues. Journal of Product Innovation Management, 19(6), pp. 424–438. McMullen, J.S. and Shepherd, D. a, 2006. Action and the Role Entrepreneurial in the Theory of the of Uncertainty Entrepreneur. Academy of Management Review, 31(1), pp. 132–152. Meister, S., Becker, S., Leppert, F. and Drop, L., 2017. Digital Health, Mobile Health und Co. – Wertschöpfung durch Digitalisierung und Datenverarbeitung. In M. A. Pfannstiel, P. Da-Cruz, & H. Mehlich, eds. Digitale Transformation von Dienstleistungen im Gesundheitswesen. Wiesbaden: Springer Gabler, pp. 185–212. Melese, T., Lin, S.M., Chang, J.L. and Cohen, N.H., 2009. Open innovation networks between academia and industry: an imperative for breakthrough therapies. Nature Medicine, 15(5), pp. 502–507. Menard, S., 2002. Applied Logistic Regression Analysis, Sage. Miles, M., Huberman, A. and Saldaña, J., 1994. Qualitative data analysis: A methods sourcebook 3rd. ed., Sage. Miric, M., Boudreau, K.J. and Jeppesen, L.B., 2019. Protecting their digital assets: The use of formal & informal appropriability strategies by App developers. Research Policy, 48(8).

References

179

Morris, M.H., Zahra, S.A. and Schindehutte, M., 2000. Understanding factors that trigger entrepreneurial behavior in established companies. Entrepreneurship and economic growth in the American economy, 12, pp. 133–159. Morrison, P.D., Roberts, J.H. and von Hippel, E., 2000. Determinants of user innovation and innovation sharing in a local market. Management Science, 46(12), pp. 1513–1527. Morse, J.M., 1995. The Significance of Saturation. Qualitative Health Research, 5(2), pp. 147–149. Mudambi, S.M. and Schuff, D., 2010. What Makes a Helpful Online Review? A Study of Customer Reviews on Amazon.com. MIS Quarterly, 34(1), pp. 185–200. Mukherjee, A. and McGinnis, J., 2007. E-healthcare: an analysis of key themes in research. International Journal of Pharmaceutical and Healthcare Marketing, 1(4), pp. 349–363. Mulgan, G., 2006. The Process of Social Innovation. Innovations: Technology, Governance, Globalization, 1(2), pp. 145–162. Murray, R., Caulier-Grice, J. and Mulgan, G., 2010. The open book of social innovations. Social innovator series: ways to design, develop and grow social innovations, The Young Foundation. Narayan, V.A., Mohwinckel, M., Pisano, G., Yang, M. and Manji, H.K., 2013. Beyond magic bullets: True innovation in health care. Nature Reviews Drug Discovery, 12(2), pp. 85–86. Nishikawa, H., Schreier, M., Fuchs, C. and Ogawa, S., 2017. The Value of Marketing Crowdsourced New Products as Such: Evidence from Two Randomized Field Experiments. Journal of Marketing Research, 54(4), pp. 525–539. Nishikawa, H., Schreier, M. and Ogawa, S., 2013. User-generated versus designergenerated products: A performance assessment at Muji. International Journal of Research in Marketing, 30(2), pp. 160–167. O’Neill, S. and Brady, R.R.W., 2012. Colorectal smartphone apps: opportunities and risks. Colorectal Disease, 14(9), pp. e530-4. Ogawa, S., 1998. Does sticky information affect the locus of innovation? Evidence from the Japanese convenience-store industry. Research Policy, 26(7–8), pp. 777–790. Ogawa, S. and Piller, F., 2006. Reducing the risks of new product development. MIT Sloan Management Review, 47(2), pp. 65–71. Oliveira, P., Zejnilovic, L., Canhão, H. and von Hippel, E., 2015. Innovation by Patients with Rare Diseases and Chronic Needs. Orphanet Journal of Rare Diseases, 10(41). Oliveira, P. and von Hippel, E., 2011. Users as service innovators: The case of banking services. Research Policy, 40(6), pp. 806–818. Olivier, J. and Norberg, M.M., 2017. Positively skewed data: revisiting the box-cox power transformation. International Journal of Psychological Research, 3(1), p. 68. Olson, K.R., Walsh, M., Garg, P., Steel, A., Mehta, S., Data, S., Petersen, R., et al., 2017. Health hackathons: theatre or substance? A survey assessment of outcomes from healthcare-focused hackathons in three countries. BMJ Innovations, 3(1), pp. 37–44. Onwuegbuzie, A.J. and Leech, N.L., 2007. A call for qualitative power analyses. Quality and Quantity, 41(1), pp. 105–121. Peredo, A.M. and McLean, M., 2006. Social entrepreneurship: A critical review of the concept. Journal of World Business, 41(1), pp. 56–65.

180

References

Podolny, J.M. and Scott Morton, F.M., 2002. Love or Money? The Effects of Owner Motivation in the California Wine Industry. Journal of Industrial Economics, 50(4), pp. 431–456. Podsakoff, P.M., MacKenzie, S.B., Lee, J.Y. and Podsakoff, N.P., 2003. Common Method Biases in Behavioral Research: A Critical Review of the Literature and Recommended Remedies. Journal of Applied Psychology, 88(5), pp. 879–903. Poetz, M.K. and Schreier, M., 2012. The value of crowdsourcing: Can users really compete with professionals in generating new product ideas? Journal of Product Innovation Management, 29(2), pp. 245–256. Pol, E. and Ville, S., 2009. Social innovation: Buzz word or enduring term? The Journal of Socio-Economics, 38(6), pp. 878–885. Pols, J., 2014. Knowing Patients: Turning Patient Knowledge into Science. Science, Technology & Human Values, 39(1), pp. 73–97. Pramann, O., Imhoff, M. and Albrecht, U. V, 2014. Zukunftstrend Medical Apps—Vom App-Store direkt in die medizinische Anwendung. Bundesgesundheitsblatt, 57(12), pp. 1402–1410. Raasch, C. and von Hippel, E., 2013. Innovation Process Benefits: The Journey as Reward. MIT Sloan Management Review. Remuzzi, G. and Garattini, S., 2008. Rare diseases: what’s next? The Lancet, 371(9629), pp. 1978–1979. Research2Guidance, 2015. mHealth App Developer Economics 2015 (November), pp. 1– 36. Research2Guidance, 2017. mHealth App Economics 2017. Current Status and Future Trends in Mobile Health. Riggs, W. and von Hippel, E., 1994. Incentives to innovate and the sources of innovation: the case of scientific instruments. Research Policy, 23(4), pp. 459–469. Roma, P. and Ragaglia, D., 2016. Revenue models, in-app purchase, and the app performance: Evidence from Apple’s App Store and Google Play. Electronic Commerce Research and Applications, 17, pp. 173–190. Rothrock, N.E., Hays, R.D., Spritzer, K., Yount, S.E., Riley, W. and Cella, D., 2010. Relative to the general US population, chronic diseases are associated with poorer health-related quality of life as measured by the Patient-Reported Outcomes Measurement Information System (PROMIS). Journal of Clinical Epidemiology, 63(11), pp. 1195–1204. Rozenkrants, B., Wheeler, S.C. and Shiv, B., 2017. Self-expression cues in product rating distributions: When people prefer polarizing products. Journal of Consumer Research, 44(4), pp. 759–777. Rutt, C. and Müller-Eschenbach, G., 2013. Alle gegen Einen. Unser Kampf gegen den Krebs, Murmann-Verlag GmbH. Sakia, R.M., 1992. The Box-Cox transformation technique: a review. The Statistican, 41, pp. 169–178. Salvador, F., De Holan, P.M. and Piller, F.T., 2009. Cracking the Code of Mass Customization. MIT Sloan Management Review, (50315). Sánchez-González, G., González-Álvarez, N. and Nieto, M., 2009. Sticky information and heterogeneous needs as determining factors of R&D cooperation with customers. Research Policy, 38(10), pp. 1590–1603.

References

181

Schieppati, A., Henter, J.I., Daina, E. and Aperia, A., 2008. Why rare diseases are an important medical and social issue. The Lancet, 371(9629), pp. 2039–2041. Schreier, M. and Prügl, R., 2008. Extending Lead-User Theory: Antecedents and Consequences of Consumers’ Lead Userness. Journal of Product Innovation Management, 25, pp. 331–346. Schweisfurth, T.G., 2017. Comparing internal and external lead users as sources of innovation. Research Policy, 46(1), pp. 238–248. Schweisfurth, T.G. and Herstatt, C., 2016. How Internal Users Contribute to Corporate Product Innovation: The Case of Embedded Users. R&D Management, 46(S1), pp. 107–126. Schweisfurth, T.G. and Raasch, C., 2018. Absorptive capacity for need knowledge: Antecedents and effects for employee innovativeness. Research Policy, 47(January), pp. 687–699. Schweisfurth, T.G. and Raasch, C., 2014. Embedded lead users—The benefits of employing users for corporate innovation. Research Policy, 44(1), pp. 168–180. Shah, S.G.S. and Robinson, I., 2007. Benefits of and barriers to involving users in medical device technology development and evaluation. International Journal of Technology Assessment in Health Care, 23(1), pp. 131–137. Shah, S.K. and Corley, K.G., 2006. Building Better Theory by Bridging the Quantitative– Qualitative Divide. Journal of Management Studies, 43(8), pp. 1821–1835. Shah, S.K. and Tripsas, M., 2007. The accidental entrepreneur: the emergent and collective process of user entrepreneurship. Strategic Entrepreneurship Journal, 1(1–2), pp. 123– 140. Shah, S.K. and Tripsas, M., 2016. When Do User Innovators Start Firms? A Theory of User Entrepreneurship. In D. Harhoff & K. Lakhani, eds. Revolutionizing Innovation: Users, Communities, and Open Innovation. Cambridge: MIT Press, pp. 285–308. Shah, S.K., Winston Smith, S. and Reedy, E.J., 2012. Who are User Entrepreneurs? Findings on Innovation, Founder Characteristics, and Firm Characteristics. Kauffman Foundation Research Paper Series. Shane, S., 2000. Prior Knowledge and the Discovery of Entrepreneurial Opportunities. Organization Science, 11(4), pp. 448–469. Shane, S. and Venkataraman, S., 2000. The Promise of Entrepreneurship as a Field of Research. Academy of Management Review, 25(I), pp. 217–226. Shaw, B., 1985. The Role of the Interaction between the User and the Manufacturer in Medical Equipment Innovation. R&D Management, 15(4), pp. 283–292. Short, J.C., Moss, T.W. and Lumpkin, G.T., 2009. Research in social entrepreneurship: past contributions and future opportunities. Strategic Entrepreneurship Journal, 3(2), pp. 161–194. Shuren, J., Patel, B. and Gottlieb, S., 2018. FDA regulation of mobile medical apps. Journal of the American Medical Association, 320(4), pp. 337–338. Smith, S. and Shah, S., 2013. Do innovative users generate more useful insights? An analysis of corporate venture capital investments in the medical device industry. Strategic Entrepreneurship Journal, 7, pp. 151–167. Smith, S.W. and Sfekas, A., 2013. How much do physician-entrepreneurs contribute to new medical devices? Medical Care, 51(5), pp. 461–467.

182

References

Steptoe, A., Deaton, A. and Stone, A.A., 2015. Subjective wellbeing, health, and ageing. The Lancet, 385(9968), pp. 640–648. Stevens, D.J., Jackson, J. a, Howes, N. and Morgan, J., 2014. Obesity surgery smartphone apps: a review. Obesity Surgery, 24(1), pp. 32–6. Stock, R.M., Oliveira, P. and Von Hippe, E., 2015. Impacts of hedonic and utilitarian user motives on the innovativeness of user-developed solutions. Journal of Product Innovation Management, 32(3), pp. 389–403. Stoyanov, S.R., Hides, L., Kavanagh, D.J., Zelenko, O., Tjondronegoro, D. and Mani, M., 2015. Mobile App Rating Scale: A New Tool for Assessing the Quality of Health Mobile Apps. JMIR mHealth and uHealth, 3(1), p. e27. Svensson, P.O. and Hartmann, R.K., 2018. Policies to promote user innovation: Makerspaces and clinician innovation in Swedish hospitals. Research Policy, 47(1), pp. 277–288. Symmank, C., Krause, J. and Gurtner, S., 2015. Stakeholder Variety in Healthcare and Their Integration in the Medical Device Development Process. In S. Gurtner & K. Soyez, eds. Challenges and Opportunities in Health Care Management. Cham: Springer International Publishing, pp. 247–258. Terry, M., 2010. Medical Apps for Smartphones. Telemedicine and e-Health, 16(1), pp. 17– 22. Thune, T. and Mina, A., 2016. Hospitals as innovators in the health-care system: A literature review and research agenda. Research Policy, 45(8), pp. 1545–1557. Tietze, F., Pieper, T. and Herstatt, C., 2015. To own or not to own: How ownership impacts user innovation–An empirical study. Technovation, 38, pp. 50–63. Tirole, J. and Lerner, J., 2002. Some Simple Economics of Open Source. The Journal of Industrial Economics, L(2), pp. 197–234. Utterback, J.M. and Abernathy, W.J., 1975. A dynamic model of process and product development. Omega, 3(6), pp. 639–656. Venkataraman, S., 1997. The distinctive domain of entrepreneurship research. In J. Katz & R. Brockhaus, eds. Advances in entrepreneurship, firm emergence and growth. JAI Press, pp. 119–138. Ventola, C., 2014. Mobile devices and apps for health care professionals: uses and benefits. Pharmacy and Therapeutics, 39(5), pp. 356–364. Visser, B., Korevaar, D. and Nolan, T., 2013. Mobile medical apps: dangers and potential solutions. Journal of Telemedicine and Telecare, 19, pp. 229–230. Wenzel, M., Henne, N. and Zöllner, Y., 2014. Beyond the pill: The move towards valueadded services in the pharmaceutical industry. Journal of Medical Marketing, 14(2–3), pp. 91–98. West, J. and Mace, M., 2010. Browsing as the killer app: Explaining the rapid success of Apple’s iPhone. Telecommunications Policy, 34(5–6), pp. 270–286. Wikman, A., Wardle, J. and Steptoe, A., 2011. Quality of Life and Affective Well-Being in Middle-Aged and Older People with Chronic Medical Illnesses: A Cross-Sectional Population Based Study J. S. Ross, ed. PLoS ONE, 6(4), p. e 18952. Wilson, J., 1999. Acknowledging the expertise of patients and their organisations. BMJ, 319(7212), pp. 771–4. Wittich, C.M., Burkle, C.M. and Lanier, W.L., 2012. Ten common questions (and their answers) about off-label drug use. Mayo Clinic Proceedings, 87(10), pp. 982–990.

References

183

Xu, S. and Kesselheim, A.S., 2014. Medical Innovation Then and Now: Perspectives of Innovators Responsible for Transformative Drugs. Journal of Law, Medicine and Ethics, 42(4), pp. 564–575. Yaqub, O., 2018. Serendipity: Towards a taxonomy and a theory. Research Policy, 47(1), pp. 169–179. Yin, P.-L., Davis, J.P. and Chhabra, Y., 2014. Entrepreneurial Innovation: Killer Apps in the Iphone Ecosystem. American Economic Review: Papers & Proceedings, 104(5), pp. 255–259. Yin, R.K., 2013. Case Study Research: Design and Methods 5th ed. Zipkin, P., 2001. The Limits of Mass Customization. MIT Sloan Management Review, 42(3), pp. 81–87.