Hospital Costs in Massachusetts: An Econometric Study [Reprint 2014 ed.] 9780674499270, 9780674499263


222 43 6MB

English Pages 252 [256] Year 1968

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
FOREWORD
AUTHORS' PREFACE
CONTENTS
TABLES
FIGURES
1. INTRODUCTION
2. DATA USED IN THE STUDY
3. COST MODELS USED IN THE STUDY AND PRELIMINARY ANALYSES OF THE 1958–59 DATA
4. EMPIRICAL RESULTS FOR 1958–59: I
5. EMPIRICAL RESULTS FOR 1958–59: II
6. EMPIRICAL RESULTS FOR 1962–63
7. CONCLUSIONS
KEY TO SYMBOLS AND ABBREVIATIONS. APPENDIXES
KEY TO SYMBOLS AND ABBREVIATIONS
APPENDIX 2.1. Reporting Schedules from the Hospital Statement for Reimbursement of the Bureau of Hospital Costs and Finances of the Commonwealth of Massachusetts
APPENDIX 2.2. Criteria for Classifying Hospitals in the Massachusetts Hospital Cost Study
APPENDIX 2.3. Inpatient Costs for the 72 Massachusetts Community Hospitals 145
APPENDIX 4.1. An Analysis of Departmental Costs
BIBLIOGRAPHY. NOTES. INDEX
BIBLIOGRAPHY
NOTES
INDEX
Recommend Papers

Hospital Costs in Massachusetts: An Econometric Study [Reprint 2014 ed.]
 9780674499270, 9780674499263

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

HOSPITAL COSTS IN MASSACHUSETTS

WERTHEIM PUBLICATIONS IN INDUSTRIAL RELATIONS Established in 1923 by the family of the late Jacob Wertheim " f o r the support of original research in the field of industrial cooperation . . . " James J. Healy E. R. Livernash George C. H o m a n s

Derek C. Bok J o h n T. D u n l o p , Chairman

HOSPITAL COSTS IN MASSACHUSETTS An Econometric Study MARY LEE INGBAR A N D LESTER D. TAYLOR

H a r v a r d University Press / C a m b r i d g e , Massachusetts / 1968

© Copyright 1968 by the President and Fellows of H a r v a r d College All Rights Reserved Distributed in G r e a t Britain by Oxford University Press, L o n d o n Library of Congress Catalog Card N u m b e r 68-14258 Printed in the United States of America

FOREWORD BY JOHN T. DUNLOP

This volume symbolizes the new relationship between medical care problems and economics that has arrived in the 1960's. Indeed, a field of medical economics has at last emerged, built upon an appreciation of the distinctive features of medical care institutions and the insightful application of the tools of economic theory and the analysis of mountains of quantitative data through statistical and econometric procedures, including the use of computers. The present intellectual challenge and excitement in this new field, as always, owes a debt to a few pioneers who labored in arid times. The new procedures arid techniques of quantitative analysis and data handling have opened new possibilities of economic diagnosis. But it has no d o u b t been the surfacing to public policy discussions of urgent issues of medical care that has generated intellectual interest, made available public and private research funds for substantial projects on a new scale, and that has attracted economists to work seriously and legitimately on the medical care sector. A variety of factors no doubt contributed to the timing of the recent insistent public and political concern with medical care issues: a decade of explosive expansion in federal support of medical research, rising incomes and educational levels with an appreciation of the potentials of modern science, and the rapid increase in medical and hospital costs. The publication of this volume in the Wertheim Series calls attention to the significant, and p e r h a p s neglected, role of collective bargaining and voluntary health and welfare plans, which rapidly expanded in the 1950's, in creating new centers of power, informed opinion, and concern with medical care in labor unions and management, their joint trustees and consultants. Collective bargaining helped to focus and generate expenditures in the medical care sector, developed widespread new interest outside of medical personnel in medical care costs a n d medical fee schedules, contributed new pressures from the organization of hospital employees, and created new types of specialists interested in the administration of medical care. The field of medical economics will develop most fruitfully with continuing interchange between economists and medically trained personnel and a d m i n i s t r a t o r s — i n d e e d , eventually leading to the generation of several new breeds of specialists. The medical professions and adminisV

vi/FOREWORD trators, beset by financial pressures and stringency, appear to be hospitable to economic analysis, or at least interested, but they need to recognize that a preoccupation with resource allocation and efficiency over the long run may have revolutionary implications for the present medical care system. Economists in turn should be warned that the applications of economic analysis at the industry level are not directly transferable to the hospital industry without considerable caution. It is often useful for pedagogical purposes, particularly with medical personnel, to stress the similarities with industry generally, but there are also some distinctive features to medical care. The measurement of the quantity and quality of output of various medical institutions and the requirements for certain professional standards for medical procedures and the qualification of personnel apart from least costs raise some of the most difficult questions for economists. The nonprofit character of voluntary hospitals does not constitute an obstacle to economic analysis, but the absence of more definite arrangements for capital flows is seriously limiting. The present volume, Hospital Costs in Massachusetts, is to be approached as a contribution to the evolving field of medical economics. There are many factors which influence the variation in costs among hospitals in any one year: size of hospital, type of hospital, level of occupancy, types of services performed, the case-mix, and so forth. The problem for research is to disentangle and measure the separate influence of each significant factor. This study utilized data for 1958, 1959, 1962, and 1963. " T h e typical Massachusetts general community hospital that emerges from this study is one whose costs can be effectively described by the volume and composition of its services, together with its size, level of occupancy, and internal staff structure. U p to 98 percent of the variation in total hospital cost among the 72 hospitals in the sample, and up to 72 percent when costs are expressed in terms of available bed days, can be explained." This significant study is a summary of the result of a painstaking and meticulous analysis of hospital costs among general community hospitals in Massachusetts. The techniques developed for the handling of these data are available in a separate volume, MEROPS: Medical Economics Regression Option Programming System, which no doubt will be useful to other researchers in this field and to administrators. The detailed data— to department levels in hospitals—will also prove of great interest to planners, administrators, and scholars alike. The authors of this volume set high standards for the new field of medical economics.

AUTHORS' PREFACE When this study was initially conceived in 1960, hospital costs were in a period of rapid rise, the implications of which were only beginning to be widely recognized. At present, expenditures are increasing even faster and there is little likelihood that this trend will abate. Population growth, higher incomes, health insurance, and especially Medicare will continue to fuel d e m a n d while the h u m a n c o m p o n e n t in hospital services makes it difficult, if not impossible, to counteract the effect of larger salaries and fewer working hours through improvements in productivity or technological innovations. F u r t h e r m o r e , medical science will doubtless m a k e its contribution to higher medical costs by continuing to enlarge the potential of treatment, much of which will be very expensive. The m a j o r hope is that changes in the structure and organization of the hospital industry will lead to more efficient ways of providing medical care, thus offsetting, or at least moderating, the pressure to increase costs. This will not be easy, as history attests. Moreover, if any systematic search for these ameliorative changes is to be undertaken, a knowledge of the cost structure of the hospital industry more detailed than that currently available is a prerequisite. It is to this problem that the present study is addressed, in the hope of understanding the behavior of hospital costs so as to improve methods for their description, comparison, prediction, and control. We h o p e that our results will be of interest to at least four groups. For the hospital administrator, the investigation and its methods both point to specific factors which influence costs and raise questions, as yet unanswered, which may assist in controlling costs and evaluating performance. These are summarized in C h a p t e r 7. The presentation also includes a demonstration of the precision gained by employing regression methods to forecast the costs of individual hospitals. F o r the hospital planner, whether he be associated with government, hospital, or community, the findings may provide suggestions for effective expansion of facilities. For the student of hospital costs, there is new evidence on the shape of the average cost curve and several new avenues of research are suggested. In addition, several statistical problems peculiar to the data at h a n d are discussed, including the choice of units in which to express costs, levels of aggregation to be used, elimination of r e d u n d a n t information, and what to do a b o u t missing data. F o r our fellow economists, our results demonstrate that traditional methods of analysis are applicable, with some improvisation, to the service industries and that, vii

Vili

/ AUTHORS' PREFACE despite the problems, accounting data are suitable for econometric analysis. Finally, for all groups, the study has developed a number of research procedures and computer methods for applying econometric techniques to the exceedingly diverse and non-uniform data that were initially reported. Although developed from information concerning hospitals, the techniques herein utilized have been generalized and are described in the companion volume to this work, MEROPS: Medical Economics Regression Option Programming System. This volume, in contrast, presents the substantive findings. The first three chapters provide the setting for this study, including the historical background and questions to be investigated, a description of the data to be analyzed and of the hospitals to be included in the sample, and a discussion of the models adopted and of the statistical techniques employed to estimate their parameters. The empirical results are reported in the next three chapters and then summarized in Chapter 7, which also examines their implications for public policy with special emphasis on the findings with respect to hospital size and occupancy rate. The research strategy has been to estimate a cost structure from data reported by a group of Massachusetts hospitals for 1958 and 1959, and then to test this structure with data for the years 1962 and 1963. The discussion proceeds by steps from costs at the aggregate level to those most important at the departmental stage. Also examined are the key questions of the existence of economies (or diseconomies) of scale with respect to hospital size and the effect of occupancy on cost. This basic analysis is then broadened to include the investigation of hospital charges and revenues and then, in order to test the homogeneity of the hospital cost structure in Massachusetts, to encompass several other groups of short-term hospitals. Finally, the data for 1962 and 1963 are used to test the cost relationships estimated from the 1958 and 1959 data. The tests take two forms: first, the 1958-59 equations are used with the 1962 and 1963 values of the independent variables to forecast costs in 1962 and 1963; and, second, several equations are re-estimated with the 1962-63 data, and the resulting estimates of the coefficients compared to those obtained with the 1958-59 data. The exercise is also extended to several other hospital groups and the resulting pattern is compared to that obtained in the hospitals of the primary study. The work of this project has been carried out under the academic sponsorship of the G r a d u a t e School of Public Administration of Harvard University (now the John Fitzgerald Kennedy School of Government). The authors are especially indebted to John T. Dunlop, who was instrumental in mobilizing support for the initiation and continuance of this study and throughout provided invaluable constructive criticisms.

AUTHORS' PREFACE/¡χ Financially, the investigation has been almost completely supported by Public Health Service Research G r a n t N o . H M 00190, "Economics and Administration of Medical Care Programs," from the Division of Hospital and Medical Facilities of the U.S. Public Health Service, with occasional year-end assistance from the Department of Economics through the provision of computer time financed by National Science Foundation G r a n t GP-683 and GP-2723. Moreover, the project could not have been undertaken without the assistance and cooperation of the hospital organizations of the C o m m o n wealth of Massachusetts. In particular, we gratefully acknowledge the assistance of Theodore W. Fabisak, Director of the Massachusetts Bureau of Hospital Costs and Finances, and of Edmund H. Stone, Audit Manager, in making available the data utilized in this investigation. Other statistical information and equally important assistance in evaluation were obtained from the Massachusetts Hospital Association and its Council on Administrative Practice, and from Massachusetts Hospital Service, Inc., now officially Massachusetts Blue Cross, Inc. A large number of individuals from these organizations supplied guidance and provided unusual expertise in hospital matters. A m o n g this group, the Directors of Blue Cross, Richard C. Brockway and Henry D. Jones, and particularly the Director of Research, Mary E. Koen, have been especially generous with their time. Similarly, the Massachusetts Hospital Association has provided a wide variety of specialized knowledge concerning the operation of hospitals and their accounting systems; in addition to our indebtedness to its Presidents, Directors, Council members and staff", we are especially grateful to Anthony P. Reis, William S. Brines, Dr. Dean A. Clark, and Lawrence E. Martin for their aid in these respects. Additional data have been obtained with the assistance of Dr. A. Daniel Rubenstein, Director of the Bureau of Hospital Facilities of the Massachusetts Department of Public Health. Finally, several other nursing and hospital groups have also provided information. For their role in the development of this volume, special thanks are due the staff of this project. The tabular presentations and much of the research assistance do credit to Mary Louis Fisher, who has been the mainstay of the full-time staff in its last three years. Special mention should also be made of the work of Jean Castleman Chase, who previously held that role. Among the students who have assisted during summers or while at school, particularly helpful contributions have been made by David Solosko, and on the more technical side, Erwin Morton, Jr., Lee S. Hyde, Sheila L. Grinell, and Donald S. Licking. A number of others also have worked for this project for various periods of time, including Mary A. Fortune, Mary Jane Fiske, Emily S. Steinberg, Nancy M . Wiesner, Carola M . Dibbell, and Amy Beth Bernstein. Joan M . Bloom and Roberta B.

X/AUTHORS' PREFACE H o c h b e r g have been particularly helpful with the b i b l i o g r a p h y . M i r i a m G . B r a v e r m a n has played a vital role in p r o o f r e a d i n g the final m a n u s c r i p t . In a d d i t i o n , as explained in the MEROPS m a n u a l , the project is indebted to its m a n y p r o g r a m m e r s a n d c o m p u t e r experts. In this g r o u p are A r a m G r a y s o n , R i c h a r d A . LaBrie, R o b e r t R . G l a u b e r , Albert E. B e a t o n , R i t a B. M a h o n e y , J o h n Van Bemmel, a n d R o b e r t J o s e p h B u r n s a n d his m a c h i n e r o o m staff. A grateful w o r d is also d u e G e r a l d D . R o s e n t h a l , Elliot J. Berg, G l o r i a S. Gerrig a n d o u r other colleagues at C u l l o d e n H o u s e w h o s e f o r e b e a r ance, s u p p o r t , a n d g o o d h u m o r m a d e o u r task seem easier. T h e critical assistance a n d c o n t i n u e d s u p p o r t of R u t h P. M a c k ( M r s . I n g b a r ' s m o t h e r ) a n d C a r o l A . T a y l o r ( M r . T a y l o r ' s wife) are a c k n o w l e d g e d with g r a t i t u d e . This volume, in a sense, w o u l d not have been possible w i t h o u t t h e coo p e r a t i o n a n d u n d e r s t a n d i n g (if n o t always the e n t h u s i a s m ) of t h e I n g b a r offspring, D a v i d , Eric, a n d J o n a t h a n . T o the patience of their f a t h e r , D r . Sidney H . I n g b a r , w h o s e role has varied f r o m medical, editorial, a n d scientific c o n s u l t a n t to a r b i t r a t o r a n d t a s k m a s t e r , this b o o k owes its existence. J a n u a r y 1968

M.L.I. L.D.T.

CONTENTS CHAPTER 1/INTRODUCTION

1

A p p a r e n t Causes of Rising Hospital C o s t s P u r p o s e of Present Study

5 8

CHAPTER 2/DATA USED IN THE STUDY

10

Time Period Selection of Hospitals C o m m u n i t y Hospitals Direct D e p a r t m e n t a l Expenses Versus Inpatient Costs: T h e Influence of Accounting Procedures

12 12 16 22

CHAPTER 3 / C O S T MODELS USED IN THE STUDY AND PRELIMINARY ANALYSES OF THE 1958-59 DATA

26

Statistical M e t h o d s T h e regression model T h e factor model Cost Models and their Underlying R a t i o n a l e Choice of I n d e p e n d e n t Variables Levels of Aggregation G e n e r a l m o v e m e n t of d e p a r t m e n t a l costs A factor analysis of d e p a r t m e n t a l expenses Pooling the D a t a for 1958 and 1959 T h e Missing D a t a Problem

27 27 29 30 32 41 · 41 44 45 46

CHAPTER 4/EMPIRICAL RESULTS FOR 1958-59: I

48

Total Hospital Cost Pooling the 1958 and 1959 d a t a Regression results Effects of size a n d utilization Primary, Secondary, and I n d e p e n d e n t Service Costs P r i m a r y services Secondary services I n d e p e n d e n t services D e p a r t m e n t a l Costs C o m p a r a t i v e E x p l a n a t o r y Powers of Models with Different Deflators

48 48 51 56 60 60 62 65 67 67

CHAPTER 5/EMPIRICAL RESULTS FOR 1958-59: II

74

Hospital C h a r g e s Hospital Revenues Costs of the R o u t i n e and Special Service D e p a r t m e n t s P r o b l e m s with Missing Categories and Missing Observations A d v a n t a g e s of Disaggregation Extension to O t h e r G r o u p s of Hospitals Residuals f r o m T w o A B D E q u a t i o n s for Total Hospital Cost

74 78 78 80 83 84 91

xi

xii/CONTENTS CHAPTER 6/EMPIRICAL RESULTS FOR 1962-63

95

Patient Day Model Applied to 1962-63 D a t a 1962-63 Costs, Charges, and Revenues per A B D Predicted f r o m 1958-59 E q u a t i o n s 1962-63 Predictions for O t h e r Hospital G r o u p s Using 1958-59 E q u a t i o n s for the 72 C o m m u n i t y Hospitals

95 100 103

CHAPTER 7 / C O N C L U S I O N S

105

Survey of the Results C o m p a r i s o n s with O t h e r Studies Implications for Policy M a c r o implications Micro implications Afterthoughts

105 109 114 114 117 119

KEY TO SYMBOLS A N D ABBREVIATIONS

125

APPENDIX 2.1/REPORTING SCHEDULES FROM THE HOSPITAL STATEMENT FOR REIMBURSEMENT OF THE BUREAU OF HOSPITAL COSTS A N D FINANCES OF THE C O M M O N W E A L T H OF MASSACHUSETTS 129 APPENDIX 2.2/CRITERIA FOR CLASSIFYING HOSPITALS IN THE MASSACHUSETTS HOSPITAL COST STUDY 142 APPENDIX 2.3/INPATIENT COSTS FOR THE 72 MASSACHUSETTS C O M M U N I T Y HOSPITALS

145

APPENDIX 4 . 1 / A N ANALYSIS OF DEPARTMENTAL COSTS

148

A B D and P D Models Nursing service A d m i n i s t r a t i o n a n d general Laboratory Radiology Operating r o o m Medical a n d surgical supplies Pharmacy Hotel services D e p a r t m e n t a l O u t p u t Models

148 148 150 150 153 153 153 157 157 161

BIBLIOGRAPHY

167

NOTES

203

INDEX

231

TABLES 2.1

M a s s a c h u s e t t s hospitals in 1959: A c o m p a r i s o n by hospital g r o u p o f data from t h e A m e r i c a n H o s p i t a l A s s o c i a t i o n ( A H A ) and f r o m the M a s s a c h u s e t t s Hospital C o s t Study ( M H C S ) .

13

2.2

M a s s a c h u s e t t s hospitals in 1963: A c o m p a r i s o n by hospital g r o u p o f data from the A m e r i c a n H o s p i t a l A s s o c i a t i o n ( A H A ) a n d from the M a s s a c h u s e t t s Hospital C o s t Study ( M H C S ) .

14

2.3

C h a n g e s in aggregate direct d e p a r t m e n t a l expenses: 72 M a s s a c h u s e t t s C o m m u n i t y H o s p i t a l s in 1958 and in 1959.

17

2.4

D i r e c t d e p a r t m e n t a l expense per hospital reporting service: 72 M a s s a chusetts C o m m u n i t y H o s p i t a l s in 1958, in 1 9 5 9 , a n d in 1 9 5 8 - 5 9 .

19

2.5

E x p e n s e per a v a i l a b l e bed day, per patient day, and per discharge: 72 M a s s a c h u s e t t s C o m m u n i t y H o s p i t a l s in 1 9 5 8 - 5 9 .

21

2.6

R e l a t i o n s h i p s between c o r r e s p o n d i n g general ledger expenses, direct d e p a r t m e n t a l expenses, and inpatient costs: 72 M a s s a c h u s e t t s C o m munity Hospitals in 1 9 5 8 , in 1959, and in 1 9 5 8 - 5 9 .

24

2.7

C h a n g e s in aggregate inpatient d e p a r t m e n t a l costs: 72 M a s s a c h u s e t t s C o m m u n i t y H o s p i t a l s in 1958 a n d in 1959.

146

2.8

Inpatient cost per hospital reporting service: 72 M a s s a c h u s e t t s C o m munity H o s p i t a l s in 1958, in 1959, a n d in 1 9 5 8 - 5 9 .

147

3.1

R e l a t i o n s h i p s a m o n g m a j o r e x p l a n a t o r y factors and various measures o f service: 7 2 M a s s a c h u s e t t s C o m m u n i t y H o s p i t a l s in 1 9 5 8 - 5 9 . R e l a t i o n s h i p s o f direct d e p a r t m e n t a l expenses to hospital service expense and the size-volume f a c t o r : 7 2 M a s s a c h u s e t t s C o m m u n i t y H o s pitals in 1958, 1959, a n d in 1 9 5 8 - 5 9 .

42

3.3

First order regression e q u a t i o n s o f direct d e p a r t m e n t a l expenses on measures o f the size-volume factor: 72 M a s s a c h u s e t t s C o m m u n i t y H o s p i t a l s in 1 9 5 8 - 5 9 .

43

4.1

T e s t on pooling 1958 a n d 1959 d a t a : H o s p i t a l service expense: Patient day model for 72 M a s s a c h u s e t t s C o m m u n i t y Hospitals in 1958 a n d 1959.

50

4.2

Regression e q u a t i o n s for hospital service expense: A v a i l a b l e bed day m o d e l for 72 M a s s a c h u s e t t s C o m m u n i t y H o s p i t a l s in 1 9 5 8 - 5 9 .

52

4.3

Inefficient hospital size: T w o sets o f thresholds for hospital

3.2

4.4 4.5

34

service

expense: Patient day model for 72 M a s s a c h u s e t t s C o m m u n i t y H o s pitals in 1 9 5 8 - 5 9 (averaged). Savings from e c o n o m i e s o f scale in hospital service expense for 7 2 M a s s a c h u s e t t s C o m m u n i t y Hospitals in 1 9 5 8 - 5 9 .

58 59

R e g r e s s i o n e q u a t i o n s for p r i m a r y service expense: A v a i l a b l e bed day model for 7 2 M a s s a c h u s e t t s C o m m u n i t y Hospitals in 1 9 5 8 - 5 9 .

61

4.6

Regression e q u a t i o n s for s e c o n d a r y service expense: A v a i l a b l e

day model for 7 2 M a s s a c h u s e t t s C o m m u n i t y H o s p i t a l s in 1 9 5 8 - 5 9 . Regression e q u a t i o n s for independent service expense: A v a i l a b l e bed

64

4.7

day model for 7 2 M a s s a c h u s e t t s C o m m u n i t y H o s p i t a l s in 1 9 5 8 - 5 9 . C o m p a r a t i v e e x p l a n a t o r y power o f models with different deflators:

66

4.8

xiii

bed

xiv/TABLES

4.9

4.10 4.11

4.12 4.13 4.14 4.15

4.16 4.17 4.18 4.19

4.20

5.1 5.2 5.3 5.4 6.1

6.2

6.3

Hospital services, p r i m a r y services, secondary services, a n d independent services for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. C o m p a r a t i v e explanatory power of models with different deflators: D e p a r t m e n t a l services for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression equations for nursing service expense: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression e q u a t i o n s for administration a n d general expense: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression e q u a t i o n s for l a b o r a t o r y expense: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression e q u a t i o n s f o r radiology expense: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression e q u a t i o n s for operating r o o m expense: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression e q u a t i o n s for medical a n d surgical supply expense: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression e q u a t i o n s for p h a r m a c y expense: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression equations for medical supply expense: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression equations for hotel service expense: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Average direct d e p a r t m e n t a l expense for hospitals reporting o u t p u t i n f o r m a t i o n : 72 Massachusetts C o m m u n i t y Hospitals in 1958 a n d in 1959. Regression e q u a t i o n s for expense per unit of o u t p u t : L a b o r a t o r y , radiology, operating r o o m , delivery r o o m : Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression e q u a t i o n s for hospital service charges: Available b e d - d a y model for 72 Massachusetts C o m m u n i t y hospitals in 1958-59. Regression equations for hospital service revenues: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression e q u a t i o n s for routine service expense: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Regression e q u a t i o n s for special service expense: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals in 1958-59. Hospital service expense: Patient day model for 67 M a s s a c h u s e t t s C o m m u n i t y Hospitals in 1962, in 1963, and in 1962-63 c o m p a r e d to 1958-59. Tests of hospital service expense for economies of scale and utilization: Patient day model for 67 Massachusetts C o m m u n i t y Hospitals in 1962-63 averaged. Predictions of 1962-63 hospital service expense f r o m 1958-59 equations: Available bed day model for 72 Massachusetts C o m m u n i t y Hospitals.

69

70 149

151 152 154 155

156 158 159 160

162

164 75 77 79 81

96

99

101

FIGURES 5.1 Differences between actual and predicted expense per A B D by hospital group: Voluntary acute hospitals in Massachusetts in 1958: Hospital, primary, secondary, and independent services.

86

5.2 Differences between actual and predicted expense per ABD by hospital group: Voluntary acute hospitals in Massachusetts in 1959: Hospital, primary, secondary, and independent services.

87

5.3 Differences between actual and predicted expense per ABD by hospital group: Voluntary acute hospitals in Massachusetts in 1959: Nursing service, administration and general, laboratory, radiology, and operating room. 5.4 Differences between actual and predicted expense per ABD by hospital group: Voluntary acute hospitals in Massachusetts in 1959: Hotel service, dietary + housekeeping + laundry and linen, medical supplies, medical and surgical supplies, and pharmacy.

88

89

5.5 A comparison of two methods of predicting hospital service expense per ABD: I: 72 Massachusetts community hospitals in 1959.

92

5.6 A comparison of two methods of predicting hospital service expense per ABD: II: 72 Massachusetts community hospitals in 1959.

93

XV

1/INTRODUCTION

On Friday, December 30, 1966, the following announcement appeared in the New York Times: HOSPITALS' RATES TO G O U P S U N D A Y This referred to hospitals in New York City. Examples of the anticipated increases in charges for semiprivate rooms were: Hospital New York Hospital Lenox Hill St. Vincent's Montefiore Some increases had already taken effect in the fall: Jewish Memorial St. Clare's Joint Disease Unity

Old rate $36.00 $51.00 $45.00 $51.00

New rate $45.00 $55.00 $54.00 $56.00

$40.00 $39.00 $38.00 $74.00

$44.00 $42.00 $47.00 $85.00

These increased room rates reflected an insistent underlying trend in hospital expenditures that had raised the hospitalization component of the Consumer Price Index (CPI) from 57.8 to 168.00 or 191 percent between 1950 and 1966. This is to be compared with an increase of 74 percent for the entire medical care component and 35 percent for the CPI itself. Not only is the upward trend in hospital costs strong, but it appears to be accelerating. Between December 1965 and December 1966, for example, the cost of hospital services increased faster than any component of comparable weight in the CPI—16.5 percent as compared with 6.6 percent for medical care and 3.3 percent for the CPI. 1 The differential rate of advances, moreover, is being maintained; at least, it was during the first half of 1967, when there was an advance of 9.3 percent in the daily service charge of hospitals, as compared with a rise of 3.3 percent in the price of medical care and an increase of 1.1 percent in the cost of all items in the CPI. In the quotation from the New York Times, increased labor costs were cited as the reason for the increased room rates. Such advances in wages and salaries, however, were not isolated phenomena of New York City 1

2/HOSPITAL COSTS IN MASSACHUSETTS hospitals, but are indicative of a general movement that has been occurring across the country for a considerable period. 2 Between 1950 and 1965, for example, the average expenditure per patient day by nonfederal, short-term hospitals in the U. S. increased 185 percent to $44.50 per patient day in 1965, an average increase of about 4.2 percent per year; the rise was even more dramatic between 1960 and 1965, averaging more than 6.5 percent per year, 3 exclusive of the 8.3 percent j u m p to $48.15 in 1966. Moreover, the phenomenon of rising hospital costs is nothing new. Odin W. Anderson, writing in the December 1963 issue of Hospitals, quoted a statement made 60 years earlier by George P. Ludlam, then superintendent of New York Hospital: It is, I think, an acknowledged fact that the per-diem cost of patients per capita is constantly increasing. Also, I think it will be admitted that this increase is not wholly due to advances in the market cost of supplies. It is due in large measure to the advance and development of medical and surgical science which has revolutionized old methods and introduced such as are unquestionably more costly ... constant familiarity with these methods engenders a spirit of extravagance which permeates the whole establishment and which it is exceedingly difficult to check or control,4 Just as sharp increases in hospital costs have a familiar ring, so too do the problems from which they stem. Indeed, the issues have remained so constant that in 1960, without warning readers, the American Hospital Association could reprint in Hospitals an article by Donald M. Morrill that had originally appeared in the August 1929 issue of The Modern Hospital. Morrill noted: It is time that we all gave serious thought to a critical analysis of the economic fundamentals of our hospital work, the structure of which, in spite of markedly changed demands, we have changed but little with the exception of some increase in rates. Frankly, we do not know our capital investments and capital costs, we have developed no rational and universally applicable principles of cost accounting and rate making and we have almost no unit of direct financial comparision, one hospital with another, because of different methods of arriving at comparative units such as cost per day, cost per visit, room cost and ward cost.5 Although the issues surrounding hospital costs remain largely unchanged, major shifts in their isolation and control have occurred as prepayment schemes have increasingly reflected consumer pressures for improved hospital care and the best of scientific medicine. The change in attitude became generally apparent in 1958 when Francis R. Smith, commissioner of insurance of the state of Pennsylvania, prefaced his specific directives to Blue Cross and its participating hospitals: / do not believe that everything has been done to bring about the most efficient and economical management of our hospitals. In fact, I believe very little has been done. I do not believe that everything has been done by hospi-

INTRODUCTION/3 tal administrators, by the Blue Cross organizations, and by the medical profession to eliminate unnecessary admissions and to reduce protracted hospital stays. In fact, I believe, with few exceptions, very little has been done. I do believe, however, unless action is taken immediately in both the above regards, the whole scheme of prepaid medical care through the Blue Cross system will be irreparably injured at the expense of millions of citizens of Pennsylvania resulting in severe personal and financial hardship and suffering.6 This decision heralded the end of the laissez-faire era for hospital costs. It gave a new turn to the concern about the high cost of hospital care. Attention had formerly been centered on how the consumer was to meet the hospital bill, which often was large, unexpected, and overwhelming in terms of income. 7 In the early 1930's, this interest was expressed in the monumental reports of the Committee on the Costs of Medical Care. 8 Similar concerns were the subject of the National Health Survey of 1935-36, the National Health Assembly (1948), and the national Commissions on Hospital Care (1947), on the Health Needs of the Nation (1953), and on Financing Hospital Care (1954-55). 9 By the late 1950's, the interest of the federal government in the field of medical economics was firmly established. Needed statistical data, including those relating to utilization and morbidity, were being collected by the United States Center for Health Statistics and the National Health Survey, 10 by the branch of Health Economics of the Public Health Service, and by the Social Security Administration, as well as by other agencies of the federal government, such as the U. S. Department of Commerce. In addition to the government-sponsored studies of the 1940's and 1950's, a number of nongovernmental g r o u p s " and authors have been investigating the effect on the utilization of hospital facilities of such factors as age, sex, family income, family size, and other demographic, social, and economic characteristics of the population. 1 2 Other studies examined the influence of differences in payment mechanisms on physician and consumer behavior, with particular emphasis on increases in elective surgery with fee-for-service payments 1 3 and the alleged lengthening of hospital stays with prepayment. 1 4 Yet another group took account of diagnostic categories, 15 some on an exceedingly detailed basis, 16 while others considered specialized aspects involving a particular community, 1 7 a single geographic area, 18 a specific prepayment program, 1 9 or a particular type of facility. 20 This work culminated in 1962 and 1963 with a series of conferences 21 and the publication of several papers that analyzed statistically the various factors influencing the utilization of hospital care, thus permitting the demand for hospital care to be forecast under varying assumptions concerning population characteristics. 22 Commissioner Smith's decision contrasts sharply with the implications

4/HOSPITAL COSTS IN MASSACHUSETTS of these studies. He emphasized curbing the "unnecessary" flow of funds and services, rather than seeking more facilities and more adequate sources of finance. Prepayment had permitted consumers to avoid unplanned expenditures arising from sudden illness, while public welfare and medical assistance programs 2 3 had assumed responsibility for those who were either too poor or ineligible for the voluntary programs. As a result, in Pennsylvania as in other areas, the growing burden to state and municipal budgets 24 of medical and hospital care made a challenge like Commissioner Smith's inevitable. True, a few issues had previously been raised: 25 in the early fifties, a number of writers had become aware of the problem of hospital costs, 26 including some who felt that expenditures were already exceeding the bounds of economic returns. 2 7 There had also been debates on the advantages of comprehensive coverage and the disadvantages of fee-for-service, especially as these factors influenced the "unnecessary" utilization of hospitals. 28 Nevertheless, it was the entry of the prepaid, consuming public, either through insurance commissioners, prepayment organizations, or citizens' groups, that ultimately stimulated the concern for the internal structure of hospital costs and the reasons for their rise. One of the earliest, and still one of the best, research projects following Commissioner Smith's proclamation was undertaken by the Citizens Hospital Study Committee of Northeast Ohio (1958-61). In addition to conventional measures of utilization, studies were made of such factors as geographic origin of patients by census tract and the effect of physician's hospital appointments and office locations. Even more important, however, this work extended the earlier investigations by New York groups 2 9 into a full-fledged examination of the structure of hospital costs, including concern with efficiency, wage levels, variations among hospitals, and the effects of changes in occupancy. The pattern of investigation initiated by Commissioner Smith and adopted by the Citizens Committee of Northeast Ohio spread rapidly. Studies were undertaken by academic groups for state insurance commissioners, such as those of Trussell and van Dyke (1960, 1962) at Columbia University. Other universities were requested by Blue Cross organizations to begin inquiries, such as those of McNerney (1962) and his colleagues at the University of Michigan. Several governors established committees chaired by such distinguished figures as Egeberg in California (December 1960), Bowles in Michigan (1962), Gruehn in Maryland (1964), and Folsom in New York (1965). The American Medical Association (1964) established its own Commission on the Cost of Medical Care. In Pennsylvania, meanwhile, the investigations became virtually ongoing, the Hospital Council of Western Pennsylvania and the Gover-

INTRODUCTION/5 nor's Hospital Study Commission both requesting the Pennsylvania Economy League to prepare reports (1959, 1962, 1965, 1966). In New Jersey (1960) and Michigan (1961), the Hospital Service Plan of the state engaged in the inquiries directly. Several of the studies had very interesting consequences. In a few states, they stimulated individual hospitals to establish utilization committees to review both admissions and long-stay cases,30 and inspired hospital and community groups to adopt regionalplanning mechanisms—particularly if federal funds were assisting. In other states, such actions followed the enactment of legislation designed to mitigate the rise in hospital costs. Indeed, in New York the program for "moderating, monitoring, and meeting the cost of hospital care" proposed by the Governor's Committee on Hospital Costs (1965) had the distinction of being implemented before it was published, since several important revisions in the public health, social welfare, and insurance laws were enacted by the New York State legislature simply on the basis of the recommendations of the preliminary summary-report. Finally, partly as a result of the interest in better data that was encouraged by these statewide investigations, and partly as a result of the advances in computer technology (which facilitated the application of more sophisticated statistical techniques), various studies31 of the economics of the hospital industry were undertaken (these are discussed in the concluding chapter). APPARENT CAUSES OF RISING HOSPITAL COSTS In one sense, the cause of rising costs is clear—the price of everything that a hospital uses in providing care has been rising.32 Of the $9.1 billion spent by nonfederal, short-term general and other special hospitals in the United States in 1965, $5.6 billion went to payroll, or $27.44 of the $44.48 spent per patient day. 33 About two thirds of a hospital's expenditures are payments for labor, and increases in wages and salaries of hospital personnel, unskilled as well as skilled, have not only kept pace with the general economy-wide increases in earnings but have begun to rise from their previous charitable levels, both in terms of rates and hours. 34 While salaries were climbing, more employees were needed because declines in hours per employee were not being offset by increases in productivity. 35 In addition, more personnel hours were required because care was becoming more complex, requiring greater variety and higher levels of skills.36 Advances in medical technology are a further reason for rising costs. Quite apart from changing diagnoses and the spread in chronic diseases resulting from increased longevity,37 both what is amenable to treatment

6/HOSPITAL COSTS IN MASSACHUSETTS and the personnel, space, and equipment required to treat a given condition have undergone dramatic changes. Open heart surgery, kidney transplants, and cobalt bombs, to cite the obvious examples, have become accepted notions. 38 Equally important, though less apparent, is the change in routine medical procedures. 39 Cunningham (1961) estimated that the number of square feet required per hospital bed in order to house increases in personnel and equipment had probably doubled since the early forties. 40 Some of this increase, of course, may be related to the growth in the research and educational functions of the hospitals, as well as to patient activities. But whatever the reasons, there has been vast expansion in the hospital plant, especially in the postwar periods. In monetary terms, the assets of nonfederal, short-term hospitals were valued at $17.8 billion in 1966, or at 480 percent of their 1948 level, whereas the number of beds at 768,000 in 1966 was only at 162 percent of 1948.41 Finally, of course, expenditures have increased because hospitals have been providing more patient days of care in addition to more service per day. Between 1948 and 1966 for example, the average daily census of the nonfederal, short-term hospitals rose 163 percent, from 361,000 to 588,000. During the same period, the annual number of admissions nearly doubled, increasing from 15,072 thousand to 26,897 t h o u s a n d — this higher rate reflecting the decline in the average length of stay from 8.7 days in 1948 to 7.6 days in 1958, a level that rose to 7.9 by 1966. Some of this increase in use results from the growth in population, while some stems from the larger proportion in the high-user over age 65 category. Anderson (1962), for example, reports that between his 1953 and 1958 surveys, the proportion of the general population admitted to hospitals remained steady at 12 percent, but the number of hospital days per 100 person-years increased from 87 to 94.42 Some of this increase reflected a slight rise in the average length of stay, from 7.4 to 7.7 days, but a large part was due to the fact that more older people used the institutions. In 1958, 14 percent of the admissions and 23 percent of the patient days were associated with individuals 65 years of age and over as compared with 10 and 15 percent, respectively, in 1953. Moreover, the effects of these population and demographic changes were reinforced by new attitudes of both patients and physicians favoring the institutionalization of medicine. This is evident in the trend to locate physicians' offices at or near the hospitals, 43 in the growing use of the ambulatory services of a hospital for nonurgent care, 44 and in the number and tone of the articles in the public press. Surging hospital costs are not a thing of the past. The costs of equipment, construction, and especially labor will continue to increase. The probable applicability of new minimum-wage laws, including both exten-

INTRODUCTION/7 sive over-time requirements and higher minimum hourly rates, as well as the need to provide competitive wages at the higher educational levels, would guarantee such increases, even without the elevation and diversification in skill levels demanded by advancing medical technology. Thus, unless the industry can effect organizational changes and technical innovations that begin to offset the rises in labor costs, an important, indeed essential, consumer product may be on the verge of destroying its present market structure. 4 5 Rather than speculating fruitlessly about ways to hold down the prices of labor and other required resources, therefore, it would appear that the more useful task for cost analysts in the hospital industry would be to isolate the structural changes—whether organizational or technical— that would lead to increased productivity. Although there are obvious difficulties in increasing productivity in the labor-oriented service industries, recent advances suggest that productivity may be subject to marked improvements via electronic patient-monitoring systems, 46 automated record-keeping and dispensing systems, 47 even computerized diagnostic and screening aids 48 and packaged hospitals. 49 Such proposals have moved a long way from the usual list of cost-saving innovations of the 1950's. 50 Thus, it would appear that the important question is not whether the prices of manpower and equipment are going to go up, but whether increases in supply and improvements in productivity can hold the advances in hospital costs within reasonable bounds. This question has not been ignored in the past. Indeed, the discussion and quotations at the beginning of this chapter merely scratch the surface of an immense literature that deals with the effects of the structure of the hospital industry on its costs and, more generally, the reasons for the increases in costs. N o r is the notion of rationalizing the system new; it is evident in the early discussions of regionalization, 51 and British literature particularly has long been concerned with concepts of alternative costs, maximization of cost-benefit ratios and alternative means to selected ends. 52 A detailed listing of much of this is included in the bibliography. 53 However, in viewing this work, it is important to realize that the decade of the sixties has introduced both new potentials and new problems. The potentials are those offered by electronic data-processing which will eventually permit extensive manipulation of vast quantities of information: it has already enabled existing data to be statistically interrelated so that utilization and cost can be more accurately predicted. The problems are those posed by a society with the affluence to meet its immediate needs without regard to their true, long-run price. At present, the old adage, "penny wise and dollar foolish," would appear to be all too applicable to the hospital field. There are excellent data

8/HOSPITAL COSTS IN MASSACHUSETTS on the relative cost of using one type of disposable supply item rather than another, 54 but very little information on how hospital costs per unit of care vary with hospital size or with the occupancy rate. In fact, the consensus of a recent meeting 55 in Chicago was disagreement on the optimum size of a hospital, if indeed such were really to exist. Even less is known about the determinants of hospital costs at the departmental level and about the interaction of such departmental costs one with another. Yet these and similar questions are of crucial importance to public policy-makers and hospital administrators. If, for example, large hospitals are more efficient in providing a given complex of service than small hospitals (that is, if there are economies of scale), future expansion of hospital facilities should take place in large hospitals, and, if the cost savings are particularly large, there may even be a case for consolidating existing facilities. Similarly, it may be that there are economies of scale in particular departments, in which case it would be more efficient if several hospitals were to share those facilities. Noneconomic factors may weigh against a community's taking advantage of such economies of scale in expanding its facilities, but the decision should be made knowing that the economies could exist. PURPOSE OF PRESENT STUDY This study was designed to extend our understanding of the structural determinants of hospital costs. It applies econometric methods to a large body of data on the operation of Massachusetts hospitals in an effort to estimate the cost structure of the hospital industry in the Commonwealth. The data are available in sufficient detail to permit analyses of the variation in hospital costs at the departmental level as well as at more aggregate levels. This richness of data, together with the range of statistical techniques used in analyzing them, are the characteristics that most sharply distinguish this investigation from other studies of hospital cost during recent years. A secondary, though not necessarily less pressing, objective was to develop a methodology for efficiently organizing, coding, analyzing, and interpreting cost information for the hospital industry. The procedures used in the organization and codification of the data, together with the descriptions of the computer programs developed in the course of the study, are reported in M ERO PS: Medical Economics Regression Option Programming System (Ingbar, 1966). The present volume focuses on analysis and interpretation of the information. The principal questions investigated are: What causes costs to vary among hospitals? Do they vary simply because activity levels differ from hospital to hospital or is there something more fundamental involved?

INTRODUCTION/9 How does hospital size enter the picture? Are larger hospitals really more efficient than smaller hospitals in terms of lower costs per patient day or per unit of service? W h a t is the influence of the occupancy rate on hospital costs? What is the effect, if any, of the composition of services, that is, the product-mix? Finally, are costs at the departmental level determined by the same factors as costs at the aggregate level, and are there any economies of scale at the departmental level? Answers to questions of this sort bear on two types of problems. First, wise public policy concerning the most effective provision of hospital care requires sound information on the cost of providing specified kinds of care in institutions of various types and sizes. Second, hospital administrators need more data concerning their internal cost structures and how they compare to norms for the industry as a whole. This study aims to throw at least some light on both groups of problems and to point the direction in which brighter illumination is to be found.

2/DATA USED IN THE STUDY

Massachusetts data offer several advantages: (1) They were the most carefully assembled, uniformly defined, carefully audited, detailed data available on a statewide basis at the time this study was begun. (2) It is reasonable to assume that Massachusetts hospitals are representative of the nation's hospitals as a whole.' (3) They were close at hand. The Bureau of Hospital Costs and Finances (BHC) of the C o m m o n wealth of Massachusetts provided the greatest share. This bureau was established on J a n u a r y 1, 1954 within the Commission on Administration and Finance at the recommendation of a study g r o u p 2 charged by the Massachusetts legislature with developing a more equitable and uniform reimbursement system for hospitals. Prior to its establishment, each bureau or department of the C o m m o n w e a l t h had made its own arrangements to purchase medical care with public funds. This often resulted in rates being paid by the state that frequently differed—even within the same institution for identical types of hospital service. With the C o m monwealth spending upwards of $20 million in 1952 for the direct purchase of hospital care for welfare and public assistance patients from a b o u t 190 different institutions, in addition to supervising the purchase of another $30 million for industrial accident cases and Blue Cross subscribers, the stakes in economy and greater uniformity were large. 3 T h e B H C was to specify, collect, and, most important, audit information relevant to determining the costs for which a hospital should be reimbursed by the m a j o r contractual purchasers of hospital care—Blue Cross, W o r k m e n ' s C o m p e n s a t i o n , and the public assistance and welfare programs. The concept of reimburseable cost and the system of accounting used in determining these costs had been developed and tested over a long period of time. 4 Each hospital in the C o m m o n w e a l t h is required to file annually with the B H C the Hospital Statement for Reimbursement, commonly referred to as the H C F 300 Report. 5 In filing this report, a facsimile of which is presented in Appendix 2.1, the hospital is conceived as divided into a series of departments, for each of which expense data are to be reported. Thirty-three departmental allocations are requested in all, although each hospital usually obtains the approval of the B H C to add or delete a few items appropriate to its circumstances. These departments are grouped into three m a j o r categories. T h e first includes departments providing hotel, administrative and plant-upkeep 10

DATA USED IN THE services; the second includes those providing patient-oriented professional care, while the third encompasses the so-called special services, such as operating rooms, and the radiology and laboratory departments. The first two contain the departments that are termed routine or general by the BHC because their expenses must be recovered through the room and board rate, whenever it is applicable. The expenses of the special services, on the other hand, are recovered through the ancillary Blue Cross per diem rate or from charges per unit of service for which all patients are billed. In addition to expense data, charge and revenue data are also reported, as are data on the volume of services rendered by the hospital as a whole and by the special service departments. Detailed discussion of the problems and principles implicit in these reports for developing a certified reimburseable cost can be found in the Final Report of the Special Commission to Investigate and Study the Laws Relating to Hospitals and Medical Costs (Massachusetts Senate N o . 958, 1964) and in manuscripts of Fleischman (1959), Lowry (1963), and Anschuetz (1965). Once the H C F 300 Report is filed with the BHC, the data are audited. Beginning with the individual ledger entries, the auditors verify in sample months the accuracy of the statistics for each service and for every type of expenditure presented; they trace the accuracy of the departmental allocations and inpatient apportionment procedures, and, if necessary, they teach the hospitals the correct methods to use. These plus other auditing processes put the quality of Massachusetts hospital cost data in a category by itself in relation to hospital cost data collected on a regular basis in the United States. 6 In addition to the voluminous information available from the H C F 300 Report, supplementary unaudited facts and assistance in interpreting the BHC data were obtained from the other major hospital organizations in the Commonwealth. A m o n g these were the Massachusetts Hospital Association (MHA), 7 Massachusetts Hospital Service, Inc. (Blue Cross), 8 and the Bureau of Hospital Facilities of the Massachusetts Department of Public Health (the Hill-Burton Agency for the state). 9 In particular, the Annual Hospital Statistical Report to the Massachusetts Department of Public Health provides general information for every licensed hospital on the specific categories of service offered, types of surgery performed, and number of patient deaths and autopsies. Other statistical series, such as the number of interns and residents, were obtained from the published annual surveys undertaken by the American Medical Association. D a t a on hospitals are also compiled annually by the American Hospital Association (AHA), and the results published in the August " G u i d e Issue" of Hospitals: Journal of the American Hospital Association. The Association kindly provided these figures on punch cards along with other unpub-

STUDY/Π

12/HOSPITAL COSTS IN MASSACHUSETTS lished data concerning personnel, their professions, length of training and departmental location. Still other statistical information was obtained from special studies, such as the one of contractual physicians by Roemer (1962), while other more descriptive material is to be found in the Anderson-Sheatsley-Health Information Foundation study of hospital utilization in Massachusetts. 1 0 Detailed information on the individual statistical series and their sources is available on request. TIME PERIOD At the time this study was initiated, the year ending September 1959 was the most recent for which audited data were available from the BHC, which meant that five years of data, 1955-59, were potentially accessible. A preliminary graphic analysis of departmental costs from about 40 hospitals for these five years indicated that large unpatterned changes in departmental costs were frequently the result of shifts in accounting practices as the hospitals, under the tutelage of the B H C , evolved toward uniformity in reporting procedures. This analysis also failed to reveal a few departments that could be held responsible for major changes in expenditures over time. Since, even without the problems of shifting accounting practices, five years of data are scanty for effective time-series analysis, it was decided to use a cross-section sample of hospitals; 1958 and 1959 were selected as the years to be studied since they were relatively free of the accounting peculiarities that had plagued the earlier period. Two years of data were selected in order to minimize any extraordinary circumstances that might be unique to a single year. (The method followed in combining the data for these two years is described in Chapters 3 and 4.) Later, the limited data for 1962 and 1963 were used to test the stability over time of the relationships observed in 1958-59. SELECTION OF HOSPITALS As previously indicated, the H C F 300 Report must be filed by any hospital in the state wishing to receive payment for medical care from any program employing the contractual per diem method of reimbursement. Essentially, this means that the report is filed by every general hospital in the state providing acute or short-term care (an average length of patient stay of under 30 days). For these hospitals, filing the H C F 300 Report has become prerequisite to the receipt of at least one half of their income; indeed, some hospitals receive upwards of 80 percent of their income in this way from the contractual third parties. The total hospital population of Massachusetts as reported to the American Hospital Association ( A H A ) is described by major category of ownership and control and by type of care in Tables 2.1 and 2.2 for 1959

r«. ττ η ο ΙΟ ο* ο οο

Co a co j oco nr i ito f )·ο « η V o>—r-.o*o>&>coco-«CNCNCOco • -O ^rsO^^O'COCO^CNCNOCO CN CN CN > —

CO CN ^ f í o * W co ·— «—

o Χ

έ ΐ

I

ι C °N ^ ° to Ό

I

-ε ζ .ΐ Ο »

loo-^rmcNCNf^cocN^ocNcocorv ^

LO 00 00

< — ·—


— I r-» co

I

CN CN

, η K

l •—

. σ»

ι Γ» I->

.«ι I

N

. in κ

oo co CN O C ι CN ι Lm πO •I o CO CO CN I I I

E ri ο

O' o «o O 00 CN o. o co

c o o m o o

u-j > — o o o « o (> η (Ν on cn

in V Tt co" co" co" h".* Ν." CN* ρ—"CN o

U

σ

,ΐ:

Ε Ε ο "
— T J - o c o i o t o o c o « — O O O ^ U O C N n N C O ^ ^ T f O O C N t o>o O o s - n e o ' » Ν Ό Ο Ο ί η τ Τ η ο io >o o«· oocNooo·—no -o ^ m •«î cn o CN o r». r-. r-. cn o CN" i—" CO" o«·" »o* > — co lo* oo r-TcN o ' o o ' ^ o ^ e o c o c o c N r ^ . r ^ . o P · ^ ^ ΙΛ ΙΛ ΙΛ m o O t f Ό · 1 J2 c o. •o ΐ < u oΧ c Χ Χ •- C » > s Q. β, tO~Ό . 5 ϊ < ί Ε £ k. o Ε O o 0 — « >. C >· >» Χ Φ JO •β ¡ ¡ τ ο φ o ^ •A


-—- c o u « u u υ υ — ϋ β .r Ol _ o. α. σ σ χ >s-o TJ Φ U 3c 3c c c ? ? ν υ — - ν — r-c -C ~βΖ S 2 ο t Η * ü u Ζ Ζ Ε Ε u S t 5 U ο ο E E-g Ο- ο 5Ο β « « ϋ u v a -- Ε Ε o O 3 3 Μ £ 2 «η μ vi ο t a. o o « 0-C ^ 0 o c ο. ο 0 « 0 ο 1 1 » U U H h ü U Ο Ο S S 3 3 ΟΧ s

2

î i i 0 , 1 —

Ρ i i

· -c

" s u

8 J o cû

E

g o Ï -o

— * S • μ -i o ϊ u ° -ο Ο Ό C ·• ·• O

HI « « ü u ° ï •S s O Φ Ϊ ® 1 -c -Ξ J .c 0 < c X

o « o

£ o

ϊ u

0 · « s . ï "S * o ·» ¡S II Ü

υ

< 1

-s
— — ά α. .



"Ο ο ® σι α . C Ε

" Ο Μ ' - ι η Ό Χ Ν Ο ^ ΐ Λ ΐ η Ο Ό η ^ ^ η ΐ Λ ' - Λ Ν Ο Ό Ό

•Ο

CN ι - Ο* CO* Ο * CO* Ο* Ο CNCNCN"— F—

Ν

p i c ^

" σ *U M M o o O υ υ -O g σ o Φ

α

o.

t S e 5 ·£

E • i i i φ



2 η

» - S i

ε . MI/) ¡ « U u O X C Ϊ Σ g Ü

o υ ^ Ü ï» o u

H - o h c

DATA USED IN THE S T U D Y / 1 5 and 1963 respectively. Of the 209 hospitals reported to be in the state in these time periods, 140 in 1959 and 143 in 1963 were classified as shortterm general and other special hospitals. A l t h o u g h accounting for nearly 67 percent of the n u m b e r of hospitals and 91 percent of the admissions in 1959, these institutions accounted for only 32 percent of the 65,216 beds in the state and 29 percent of the patient days. These percentages, moreover, have changed little with time—each being exactly one percentage point higher in 1963. The n u m b e r of hospitals a n d the admission percentages remained unchanged f r o m the 1963 level in 1965, but the bed and patient day percentages attributable to the nonfederal, short : term institutions rose to 35 and 32 percent respectively. Of the 140 short-term general a n d other special hospitals in Massachusetts reporting to the American Hospital Association in 1959, 117 have been included in this Massachusetts Hospital Cost Study ( M H C S ) . These 117 hospitals, as reported to the A H A , accounted for 95 percent of the beds, 94 percent of the patient days and 98 percent of the expenditures of nearly $211 million by the short-term institutions in Massachusetts in 1959. These percentages, as shown in Table 2.1, would be altered slightly (to 92 percent, 92 percent and 101 percent respectively) h a d the data reported to the B H C replaced the unaudited A H A data listed by the hospitals themselves. Hospitals were usually omitted from the final M H C S group because they filed the H C F 300 R e p o r t for less than 12 m o n t h s in at least one of the four study years, 1958, 1959, 1962 and 1963. However, several hospitals were omitted for other r e a s o n s . " The other details of the classification scheme are described in Appendix 2.2. Included in the 117 hospitals whose data were analyzed are 104 voluntary, 3 proprietary, and 10 municipal institutions. Of the voluntary hospitals, which include two city teaching establishments, 72 are termed community hospitals; they become the " p i l o t " M H C S sample from which most conclusions are drawn. Each of them was an accredited short-term, voluntary (nonprofit) general hospital that lacked academic, municipal, and religious affiliations. As a group, they accounted for about one half of the beds, one half of the patient days, and slightly under one half of the nearly $200 million spent by the 104 voluntary hospitals in 1959. As indicated in Table 2.2, by 1963 the importance of these 72 hospitals in the voluntary group had increased somewhat. The remaining 45 were excluded f r o m the pilot sample because for various reasons their cost structures might differ from that of the community hospitals. F o r regression analysis to yield valid results, only institutions with a homogeneous cost structure can be included. Thus, the six voluntary teaching hospitals formally certified as affiliated with medical schools had to be omitted because of the inclusion of one of the two

16/HOSPITAL COSTS IN MASSACHUSETTS largest hospitals in the state, the Massachusetts General, whose presence would have dominated statistical results. Similarly, the other teaching hospitals had to be left out because of the presence of the Boston City Hospital 1 2 —the largest hospital in the state. In addition, there was considerable ground for questioning the homogeneity of the cost structure between any teaching hospital and the community group because of the peculiar accounting problems associated with the treatment of research, teaching and training expenses (all of which are allegedly removed from the patient costs calculated for the BHC). In a parallel manner, the cost structures of church-sponsored hospitals are likely to be unduly influenced by the accounting conventions used in imputing wages to religious orders. Maternity hospitals were segregated because of their speciality. Unaccredited and proprietary hospitals were excluded for obvious reasons, 13 but the former might have been included with the community group had their number been greater. Finally, the municipal hospitals had to be treated as a separate entity because their data are probably unduly sensitive to the procedures used to segregate their budgets from those of other public authorities. Restricting the pilot sample to include only community hospitals does not preclude other types from the analysis, however. Once regression equations have been estimated from the data of the 72 community hospitals, they can be used to predict the costs of the excluded hospitals. If the predictions are within the margin of error provided by the hospitals in the sample, and if there is no tendency for the predictions to be either systematically high or low, there will be prima facie evidence that other types have the same cost structure. These tests are reported in Chapter 6. COMMUNITY HOSPITALS While it was believed that the cost structures of the 72 community hospitals would be homogeneous and devoid of the influence of undue size or unusual accounting problems, they were sufficiently varied to be representative of the diversity in size, utilization and services that this investigation proposed to relate to differences in cost. In describing the magnitude and range in variation of the data in the present section, a system of classification is used as an expository device that is derived and discussed in Chapter 3. This groups traditional hospital departments as defined by the American Hospital Association, under three m a j o r headings which are termed primary, secondary, and independent services. The departments in the primary services have costs that move closely with one another but relatively independently of the departments in the other two categories; the same is also true for the departments in the secondary services. The remaining departments, whose costs move neither with one another nor

ν — α. *

?

«

.if co >

ral

•fi co

1 Ο ι ® ι o !

C N W¡

— » V W > co Γ * * — · rx «ο ri — W cö fN n n - i - ^ d i N O p d o

— < N — O O O

J;

— η ci « m s' Ο η ο »

© Γ Ο (Ν 'h«. r^ r«. » -û ' o oo io

O S (h Ι Ο Ό Ν

s μ η co ν η ο > >- 1 Λ 1 η Μ η Μ

SOONO-NS-N'-COTJÍN

(Ν < Ν — · — —

co co — r — f r - O O OO O o o

) 00 — O* O Ν { > Ν Ifl « O C N O O -O " * T co

INMe-O'-O-OOCNNlflfflO' O Ν Ν - Ν η o > 0> 0 > Ν O - ^ (Νοιοο-ιηττίΛνοττη rv Γ Ν — « m C N ^ C O I M — · (Ν C N — « 1

ιη ν ο ο ' co c m co o Ό fN < 0 Ν O ·

oo m — — oo c m © «o co cn co co co η Ό s η (Ν ο Λ Ο 1 η Π Γ Ο Γ) (Ν

fs. ο- Ο · Ο · m (Ν U 1 ΙΛ Ο Ο Ν Ο ' " iC oo m * τ"

_ _ C D - Ό Λ co (ν m o" n Η

Ο >ΰ rΝ (Ν o« m ο· s rí Ν Ν

I — 9 Μη

(η ^ Ι Ο ( Μ 1 Λ »

Ο » •t > ο ο

C N (Ν — ·

C O n ^ ri

β) ΙΟ — Os _ Ω _

« υ " Ο β ο

ile

IV, -o 3 C " O 0 4» s L. 3 ζ

:¡ £ -ο
0) 4 > : û α : < 2 < Σ 2
— ο> CN

io co CO O co lo co lo h» io o co o oo io io -o cn cncor-cs-—(ν — ·- —

βο «Ο — Ilo ο ο co οι - S tf ι - S

CN 0 Γ "O* ^f* LO* LO* CN

-s — — — OOCOCOCO» o

> ; > »

« « > o ·u o 2 ° «o
f. co ό ι— ο

dl— ΓΗ Ο I— o 4 « θ> · • CO LO r - IO p-·. ·— > S « (Ν Ν «Ο ^ Π Ο Ο· ΙΟ Ο Γ-- Ν O ι - O* «Λ co io ·— » o

ίο ·ο r ^ r Z ^ í ο Λ in η' ^ rC N ^ Í © « ^ f CN Ο 00 Ο ΙΟ Ν Π Ο Ο* S p i CN -ο (Ν m ^ ττ ín O CN η ' î O» Ν Ifl O» Οι S TI í s ^ ró τ-' η 00 OΟ Ό Ο Ν Ό* 0> (η ΓΟ «—

J> ίο í ¡n œ í o o η" ν c ó c N - ^ i o i n c N IO CN ττ Ο Ο "Ι CN Ο CN P v o o r v ^ - F ·— ·—" "

Κ oí « ^ Ό

Κ § ·ο " Ν S Γν W> CN

° o· o «o V ο ·— r--ν TJ (Ν

Ο Ο· Ο· Ο io CO o ·ο «o co cn r·«. Ό ^ ΙΟ Γ»» r— O CN O i— O Ό ° Ο» Γ·* © »Ο CN

I

I

I

I

I

I

η

η

η

35 C N O ® w S

σι ι

-S--S

φ — g) « J n £ Ε ω U 3 U "O C "O

>· i o- o> · o> » o ν — o — .τ σ ° O S,) S o *r s o c o c — "Λ o "O 4 > -û "D o c o o >·£ 0£ < _l OL ± < .

C C C O o o C O C Oc ÛÛ σ υ 3 "Ό 0) Ol C

I

υ υ ο. Û.

3 3 3 3 3 3 Ζ Ζ Ζ Ζ Ζ Ζ ι

o — o >o Ό o o o o o oo·^ O -O M N o o o o o o o o — ·—" o o o o co co co ^ ο > ο > η οο ο> m ι ^ (Ν η η (Ν (Ν \ ν

Î s

Λ Ό uì

(Ν·—

i \

ρ > . C O Γ > . Γ-» ^ Ό

Γ Γ ι (Ν Ν Ν Ν

r-- r- r» r-. co co co * ' ι' Γ Γ Γ Γ

Γ Γ Γ

*



CMfN^m Γ Γ Γ

Γ

— · C N C N C S

00 U < 00

Φ Ο. Ι Λ 3 C

υ •β
. >. 0 σ 0 s < < 2 -Ό Ό TJ -Ό Ό I I Φφ Φ α > α. o. .0J¡J1 .0 α S ν Φφ Φ I I ^ .0 _ο ο α 0 «Ξ σ ο to m .2 < < Ο0 σ > > > ο 0 σ " Ο -σ -ο φ φ ο ι/Ι 3 3 C C C o o o o Ζ>

• —

>£ α Ε -ο Ü ο

— c Μ - O ο -

ΟΟ

(J eö •eu < < = > ¡ U L O "3 » E u u oí «5 <
n m N o . r ( s 0 > < Ν > 0 < Ν < Ν Ό » - 0 ^

-

(Ν CN CN 3 3υ «

Ζ -5 5 O

¡ñ >A •Λ Χ o CN «o CN co

Ε

» Ji -o

ε E

"

I f ì l f l ΙΛ V η m m IO **>

ο.

J

M o

Ό Ν Ν

Ρ··» Ό © ττ ο ο β ο ^ ο ο< rv.

«

5" J*

·— « ο ο> (V (Ν JO CN — m κ . ί>

>o Ν » f*» c*> »o Ο· Γ·«· Ο·

Γν

fs-r-f·»

Ό U"> ÍN ·— ο Ό œ ^ ν Ό ο co OJ Ν m

ο

5 α

, ρ , „ „ , Μ Λ Π Μ Π 4 Ό Ό Ό Ό

, CÎ Ό

» » »

_ϊ -D ο I-

o Χ

-ο -ο I) C s

Λ O Ό O Ό ·~ Ό Ό ^ Χ 00 Ο· ΙΟ ^ ΙΟ Ό ι Ό 00 Γν Μ Π Ν Π Ν — — Ν (Ν (Ν

Ο

η

^

Ό Ο· Ο

Γν

62/HOSPITAL COSTS IN MASSACHUSETTS factors for primary services were also important in explaining aggregate cost. The major exception is the nursing education factor, which did not show up at the aggregate level. It is evident from equation (1) in Table 4.5 that ambulatory and surgical activities are clearly the most important factors since together they yield an R 2 of .58. The medical education and physician service factor shows up less strongly with the primary services than with aggregate cost, and the same is true for the private and ward service factors. The importance of nursing education in explaining the primary service costs is obvious. Nursing services account for over 40 percent of expenditures by the primary services, so anything that affects nursing costs will naturally affect the costs of the primary services. The major factor is the presence of a school of nursing in the hospital. Student nurses substitute in many routine nursing tasks, with costs being reduced accordingly. This shows up in the negative coefficients for variables designating the presence of a school. Equation (6) of Table 4.5, for example, indicates that each additional student nurse per bed decreases the primary service cost per ABD by about $ 1.40. Further discussion can be found in Ingbar, Whitney, and Taylor (1966). As was the case with aggregate costs, the occupancy rate is not important in the ABD model for the primary services, but does show up strongly in the PD model (see equation [2,] of Table 4.8). This indicates that the primary service cost, as is true for the aggregate, is not responsive to the level of occupancy. The results of this section indicate that the costs of the primary services are largely determined by their own activities rather than by such macro factors as over-all occupancy. The activities of the radiology, laboratory, and operating room departments are reflected in the ambulatory, radiology, laboratory, and surgical factors, and the nursing education factor reflects the internal structure of the nursing department. The negative coefficient for ward services indicates that it is more economical, all other things being equal, for a patient being treated by the primary services to be in a ward, and the positive coefficient for physicians' salaries indicates that hospital care is more expensive (in terms of hospital costs) when the service of physicians is included. All this is reasonable, of course, for it attests to the basic autonomy of the primary services. They form the backbone of the medical care provided by the hospital, and because of this they affect the costs of other departments. But the causality is largely one way, since there is little, if any, feedback from the other departments into the primary services. Secondary services. The secondary services comprise seven departments which, as a group, accounted for 34.6 percent of total hospital cost in

EMPIRICAL RESULTS FOR 1958-59: 1/63 1959. (See Table 2.3 for a listing.) The departments included in this category do not provide medical care themselves but service those that do. Therefore, the costs of the secondary services should be quite closely related to the activities of the departments comprising the primary services. The equations estimated for the secondary services are tabulated in Table 4.6. T w o things in particular should be observed. T h e first is that the R2 of .50 is about 20 points below that for aggregate cost and the cost of the primary services. This probably shows that factors unique to individual hospitals, but which do not appear as explanatory variables, are more important with the secondary than with the primary services. Variations in age of plant, square feet of floor space, plant layout, and other factors not measured in this study are likely to be important. Also, new technology in hotel areas may be less likely to spread quickly and uniformly (if at all) to other hospitals than comparable innovations in primary services. A relatively new hospital can probably prepare food, for example, more efficiently than an old hospital, but for a similar patient load the operating rooms of both hospitals are likely to be equally efficient, at least with respect to the design of the suites and the nature of their equipment. Since the ages of the 72 hospitals vary greatly, each hospital is likely to have its own history of technological innovations in secondary services, which (since it goes unmeasured) will be reflected in a lower R2. Similarly, the lower R2 may also show that some hospitals purchase one or several of the secondary services from outside vendors. For example, the data do not indicate whether a hospital operates its own laundry or uses contract services, and the same is true of f o o d and drugs. T h e second important fact in Table 4.6 is the highly significant coefficient for the occupancy rate. This suggests that a larger portion of the expenditures of the secondary services responds to occupancy than in the primary services. T h e coefficient of the occupancy rate is still negative (and significant) in the P D model (see equation [3s] of Table 4.8), however, indicating that higher occupancy reduces the cost per P D for the secondary services as well as for the primary services. T h e medical education and physician service, nursing education, ambulatory, radiology, surgical, and size-volume factors are the other important elements for the secondary services (equation [28] of Table 4.6), and of these the first three are the most important statistically. Most of these are related in one way or another to the primary services, which is consistent with the notion that the costs of the secondary services derive to a great extent from the activities of the primary services. A f t e r the occupancy rate, there is some instability in the order of importance of the factors (as measured by their t values), depending upon what variables are used to measure the various factors. For example,

i I í

CL

> i

χ

χ

χ

χ

χ

χ

χ

χ

χ

«^O^-ChO'O'O^'^OO

5 > «ι · ζ

. o· m

κ s o r» "Ο Ό CO ~ o ci Ν η ·— co en

S

(Ν -

β Ό — Ν —

M «

o> o> o ·

s

c o n o Ν ^ 00 -o — CD — r o r r c-> t π o

O

- «Ν m ο ( N r t r i r i r t M f O f O M

r-» m τ? ο · (Ν ^ ^

? S

s Ο

ϋ

θ η ο

τ ^ Ό —

r » o c·* o o m o κι ^f t o o O -*í

00 οο o Ν ·*τ o

o o 00 ΙΛ (Ν O —•

3

·-

ΊΕ

Υ O

Μ (Ν >Ό

O» S ^ Ό

Ν Φ ( Ν CO Φ Ν O O O O*

EN ·— O·—'

C O

O M

O 3

0 0

? . 2

S

«

Y

O

^ N O N O Η ^

Ό CO OO Ν

^

(Ν Λ

Η

Β - o

Η IO UN O OO FO O O

U->

ÍN — RV R-S Ό CN CI OÍ

Ν « Τ &> ^ Ο· OJ I -

O

ut

O

O

Ό

» O« Η O

RV >O OO O

O

Ι— — LOCO ΓΝ Τ O

( Ν CN

CN

—' CN

( N O L T N O A L O I O F N ^ O Ό Ό Ό Χ X X X Χ Ό X X Χ Χ Χ O «O ^Í· O* ΙΟ Χ RV CO E ( N » » O > O O O O ( N O I N

S

C O ·-

ΙΛ

tñ ΙΛ

O

Α>

.1:

Σ>

Sr

•a

F) O· «1 O

^

4)

J !

«— «— 00 O R » R-> RV

'

•Ζ.

-

«Ι Ο (Ν -O

•O OO H»

-O -O FT.EE _ O

O S'

»

«

9

D

J

Ε E

CO CN >— ·— —



O



O

O

O

Η

I N I O IO

Π

ITI ! > O

O


Ο O

«

CD Ο

^ »— ο

CO CO 0 0 co c o O

""

co

a u , , •ο ·ο ·ο to

I) · J

< οο χ χσ> coro c c n o ^

œ ^ s

ο ίο Ό Μ η M o> ΙΛ Ν

Ο ^ »T Ν

m O (Ν CO ji/-) Ν η

^ô ^ Iη

8

•Ο CN m β r»·— β «

Μ Cl rη

^ _o -

cû Ν I m Ό o r--ou"> (NCN —

( N ' - « ' - o ( N i n i n ( s O ' í O ' t n o ^ t N ' f -O Ν

tN π

ΙΟ - o cÑ -O ( ,

» t N o i n N n u i n w n t - o c o o u - i ^ r - k o o o o ( N ^ f N O n i h l f l l f l l f l ^

b

«

E 2

' • ( N ' - ^ O C N - - ·

CNCNCNOCNCN



Ζ

» Ν Ό ΐ Λ ί Ν Ό Ό



fO


m

4 ·

4





• 4 4

4 •

4

4 **-





4

4

1





4

i : σ> ΐΓ> £ co σ ¡3 -a ιο £σι 3

4

· 4 4

4

1 § ;

i

s I a> oc Φ £ φ α» 2 £ o

§

4

I





·

4 •

• 4



4

Ό β) o a

- o ai υ ¿ Ό



4

4



4 4

• •

• o IO +

ι

1 o IN +

1

1 O +



4 4

4

4

I

4

4



4 1

4





σ> aj C ΙΛ „ ® o H

·

·

4

4 4

·

ϊ

4

4

I



• •



4

4

4





ι







® SB'S s? S •o -o § a> O^ ω

4





d) (Λ » H

• 4

S s 4

4

·

4



_ § if cL

2 3 O ü Ό Π

·





4

4 4



·

4

•O ς Ζ Ϊ ? Η fe . ïa S N 0J V T3 C K 0) σ o, _ ¿z o s r Η o o 0 evi







8.S S J2 5 Β Q. V) in α> o 3 χ:

4

4



ε I υ cvj

f

I



• 4 4

I

(soa (lun |onpo (uaojad so s|0npisay

1

1

1 o

1

ϊ

c

jsoo jjiin |Dn|OD juaojad SD S|Dnp|say

94/HOSPITAL COSTS IN MASSACHUSETTS ïn passing, it can be noted that Figure 5.6 confirms the finding that unexplained differences in costs are not related to hospital size. This completes the analysis of the data for 1958 and 1959. The present chapter has extended the cost analysis of Chapter 4 to hospital charges and revenues and to several groups of hospitals not included in the original sample. The results, especially with respect to charges, are encouraging. Charges follow costs very closely, and can be quite well predicted by the activities of the primary services and by the occupancy rate. A n d since most charges eventually become revenue, revenue is explained quite well also. The present chapter also extended the cost analysis of Chapter 4 to two groups of departments defined by the Massachusetts Bureau of Hospital Costs. Nothing particularly new was uncovered, except possibly that the R 2 for the special services, a B H C group consisting mostly of departments in the primary and independent services, is among the highest for any model estimated in this study. The results of section 5 show that in terms of predicting aggregate costs there is something to be gained by disaggregation. The results of section 6 indicate that there is probably more homogeneity in the cost structure of Massachusetts hospitals than might appear at first glance, and that in any further study a larger sample of hospitals could profitably be used. Finally, the results of section 7 indicate that, in most hospitals, costs are within a range of 10 percent of what is to be expected after their differences in size, occupancy, and services have been taken into account. This suggests that regression analysis may be a valuable administrative tool for evaluating the reasonableness of costs in individual hospitals and their departments.

6 / E M P I R I C A L RESULTS FOR 1962-63

A key element in econometric analysis is the testing of a model with data that were not used in its estimation. The present study is f o r t u n a t e in being able to perform such tests with 1962-63 data on several equations estimated in C h a p t e r 4. These tests take three forms. First, the data are analyzed with the P D model, as were the 1958-59 data, for the advisability of pooling the two years of information, for the existence of economies of scale, and for the importance of occupancy. Next, by using actual 1962-63 values for the independent variables, equations with parameters estimated from the 1958-59 data are used to predict 1962-63 costs per A B D for the community hospitals. Finally, using the 1958-59 equations f r o m the 72 community hospitals, 1962-63 costs are predicted in analogous fashion for the other groups of short-term hospitals. T h e m a j o r reason for undertaking these tests is to see how stable the cost structure is over time. If the 1958-59 equations predict 1962-63 within 15 to 20 points, say, of the R2,s of 1958-59, and if the hospitals that were high (low) in 1958-59 remain also high (low) in 1962-63, this would be strong evidence that the cost structure was stable between the two periods. A n d this evidence could be c o r r o b o r a t e d if the coefficients estimated with 1962-63 data were approximately the same as with the 1958-59 data or reflected known changes in prices or product-mix. Since observations are usually at a premium in econometric studies, it would be most useful if the 1962-63 data were analyzed in the same detail as that of 1958-59. But the law of diminishing returns even applies to statistical investigations, and so the 1962-63 data have only been used to test prior findings; nevertheless, some new results have been uncovered. PATIENT DAY MODEL APPLIED TO 1962-63 DATA T h e patient day model was applied to 1962-63 data for 67 of the 72 Massachusetts community hospitals. Except for the five missing hospitals, which were omitted because some of their information appeared suspect,' the results are directly c o m p a r a b l e with 1958-59. T h e relevant equations are tabulared in Table 6.1, and these equations, except for the absence of the d u m m y variable for medical and surgical physician expense, are comparable to the equations in Table 4.1. F o r ease of comparison, the equation with averaged 1958-59 data is given in column 4 of Table 6.1.

95

9 6 / H O S P I T A L C O S T S IN T a b l e 6.1.

MASSACHUSETTS

H o s p i t a l s e r v i c e e x p e n s e : P a t i e n t d a y model for 67 M a s s a c h u s e t t s

C o m m u n i t y H o s p i t a l s in 1962, in 1963, a n d in 1 9 6 2 - 6 3 c o m p a r e d to 1 9 5 8 - 5 9 . Coefficient0 Variable 1962 63 " , averaged

1 9 5 8 5 " ' averaged

1962

1963

Constant

28.12 (4.74)

36.45 (6.09)

30.83 (5.42)

29.86 (3.48)

s

55

4.94 (1.71)

7.67 (2.24)

6.57 (1.96)

10.19 (2.35)

s

77

106.50 (46.60)

136.65 (54.28)

122.38 (50.76)

151.03 (41.90)

s

88

13.34 (2.92)

13.94 (3.44)

13.79 (3.13)

13.94 (2.33)

s

90

19.66 (7.06)

13.20 (8.54)

16.77 (7.66)

11.52 (5.94)

x

38

3.54 (3.19)

4.90 (3.66)

3.87 (3.37)

3.93 (1.93)

x

38

-0.86 (0.97)

-1.34 (1.08)

-1.02 (1.00)

-1.31 (0.61)

x

66

-4.31 (6.29)

-14.39 (7.88)

-7.35 (7.20)

-15.33 (4.59)

R2 Se

.424

.413

4.08

4.94

.428 4.41

Sum of squared residual s for 1962 and 1963: Sum of squared averaged resi d u a l s for 1962 and 1963: Percentage of residual variance due to hospital effects:

.568 2.87 2420.91 1130.88

2(1130.88)

aStandard errors are in p a r e n t h e s e s . For definitions of s y m b o l s , s u b s c r i p t s , and superscripts, see K e y to Symbols and Abbreviations.

The most obvious result is that the R2 with the 1962-63 data is about 14 points lower. This important change might be explained by either or both of two possibilities: (1) with no other change in structure, the variance of the error term could have increased between 1958-59 and 196263, that is, the population R 2 could have decreased between the two periods; or (2) the underlying structure other than the error variance could have changed; for example, the coefficients could have shifted, or,

EMPIRICAL RESULTS FOR

1962-63/97

alternatively, the 1958-59 model could have overlooked some explanatory factors that had become important by 1962-63. As to the possibility that the structural coefficients changed between 1958-59 and 1962-63, the equations in columns 3 and 4 of Table 6.1 provide relevant comparisons. The differences, though fairly substantial in percentage terms for salaried physician expense (s 5 5 ) and occupancy ( x ^ ) , are not sufficiently large to indicate a change in structure—at least with respect to coefficients. The coefficients on outpatient weighted films ( j 8 8 ) and number of beds ( x 3 8 ) are virtually the same, while in the light of the instability of the coefficient for weighted operations (s 7 7 ) noted between 1958 and 1959, the difference between the two coefficients of i 7 7 is not alarming. The implications of the smaller coefficient for the quadratic term for hospital size (X3 8 ) will be discussed in the latter part of this section. Perhaps more revealing than the differences between the coefficients for the years averaged are those between the coefficients for 1962 and 1963 taken separately (columns 1 and 2 of Table 6.1). The coefficients for 1963 are generally closer to those for 1958-59 averaged than are the coefficients for 1962-63 averaged. This is particularly true for salaried physician expense, private patient days (Í90), and occupancy rate. These differences suggest that either something happened between 1958-59 and 1962-63 to make the cost structure more unstable in the short run—that is, on a year-to-year basis—or else there is something peculiar about the 1962 data. Inasmuch as the 1962-63 data were not analyzed as extensively as the 1958-59 data, the latter is a real possibility. One possible explanation for the lower R2 for 1962-63 involves the role of the laboratory, radiology and surgical factors. It was argued in Chapter 4 that they represent the spectrum o f medical care provided, both quantitative and qualitative. In 1958-59, furthermore, the results suggest that number of laboratory tests and x-ray films were probably good indices of the complexity of this bundle of services. Inclusion of these factors, therefore, not only allowed for their own direct impact on costs, but also adjusted for variations in the "quality" of the product among hospitals. Because of the rapid spread of such procedures by 1962, however, these factors probably became poorer measures of differences in complexity and types of medical care. In other words, hospital costs varied among hospitals in 1962-63 just as much as in 1958-59 because of dissimilarity in type, complexity, and intensity of medical care provided, but laboratory, radiology, and surgical activities had become poorer measures of these differences in 1962-63. If this is correct, then it can be argued that, in order to explain the 1962-63 costs, the number of explanatory factors must either be increased, or measures must be used that

98/HOSPITAL COSTS IN MASSACHUSETTS are more sensitive to qualitative differences. For example, perhaps measures of radioisotope activity should supplement simple counts of radiology films. (See Silver, 1965.) Factors unique to the individual hospitals were estimated to account for about 93 percent of the residual variance in the 1958-59 PD model (see Chapter 4). This estimate was obtained by averaging the residuals for the two years, squaring the averages, multiplying by two, adding, and then dividing by the sum of the squared residuals for both years (see equation [2] of Chapter 4). Using the first two equations of Table 6.1, the comparable figure for 1962-63 is 93.4 percent (see the bottom of Table 6.1). This amazing stability in the share of individual hospital factors in the residual variance further corroborates the absence of cyclical factors in the behavior of hospital costs. The size and utilization variables appear to be less significant with the 1962-63 data. The coefficient on the occupancy rate, in particular, is only about half as large with the 1962-63 data as with that of 1958-59, and it is just barely greater than its standard error. Given the apparent instability of the coefficient between 1962 and 1963, however, and the fact that its value in 1963 is close to that for 1958-59, it seems unlikely that a sharp structural break occurred between the two periods. Of particular interest is the difference between 1958-59 and 1962-63 when the occupancy rate is designated by dummy variables. A s in Chapter 4, the values chosen are 65 percent or less and 85 percent or more; and the equation is tabulated in Table 6.2. The difference lies in the substantially larger (and highly significant) coefficient for the hospitals with occupancy 65 percent or less—an important result—and the positive (but insignificant) coefficient for the hospitals with occupancy 85 percent or more. If the data are to be believed, this suggests that the curve relating costs per P D and occupancy has a trough somewhere between an occupancy of 65 percent and 85 percent. There was only a hint of this with the 1958-59 data. Moreover, it is particularly interesting that the R2 with the two discontinuous variables for occupancy is higher than when occupancy in its continuous form is used (see column 3 of Table 6.1 and equation [1] of Table 6.2). This means that there may be some significant nonlinearities in the relationship between costs per P D and the occupancy rate. However, it could also reflect some peculiarities in the 1962 data. With regard to hospital size, the positive coefficient on size and the negative coefficient on the square of size means that the average cost curve is still an inverted U. However, the least efficient size of hospital in terms of costs per P D (after the other factors affecting costs are taken into account) apparently increased between 1958-59 and 1962-63 from about

E M P I R I C A L RESULTS FOR

Table 6.2. T e s t s of hospital s e r v i c e expense for economies of s c a l e and utilization: Patient day model for 67 M a s s a c h u s e t t s Community H o s p i t a l s in 1 9 6 2 - 6 3 averaged. Variable Constant

s

S

s

s

x

55

77

88

90

38

Coefficient0 Equation 1

Equation 2

22.56

38.79

(3.21)

(5.42)

6.12

5.57

(1.82)

(1.75)

150.59

138.99

(47.70)

(48.31)

11.75

13.10

(2.96)

(2.99)

19.53

16.31

(7.13)

(7.24)

5.59

E x c l uded

(3.14) x

38

-1.45

Excluded

(0.94) x

66

-12.56

Excluded

(6.84) -3.56

6.63

dl

(1.25)

(1.91) d2

-0.10

0.94 (1.42)

R2

(1.80) .486

.510 4.08

Se Equation 1

4.18

Equation 2

1 if < 6 5 % o c c u p a n c y 0 otherwise

1 i f < 100 b e d s 1

1 if > 8 5 % o c c u p a n c y 0 otherwise

0 otherwise _

2

l i f > 250 beds 0 otherwise

a

S t a n d a r d e r r o r s are i n p a r e n t h e s e s . F o r d e f i n i t i o n s of s y m b o l s , s u b s c r i p t s , a n d s e e K e y to S y m b o l s a n d A b b r e v i a t i o n s .

superscripts,

1962-63/99

100/HOSPITAL COSTS IN MASSACHUSETTS 150 beds to around 190. This figure is obtained by solving (1)

3.87^38 -

1.02*?g

from the equation in column 3 of Table 6.1 for its maximum. Although the reasons for this increase have not been probed, it may merely reflect the gradual expansion of hospital size with time rather than an outright shift in the optimum size. The coefficients on the size terms, as has already been noted, are less significant in 1962-63 than they were in 1958-59. Assuming that this is not caused entirely by the exclusion of the five hospitals from the 196263 sample, it could mean that the average cost curve had flattened by 1962, or that the shape of the curve had changed in such a way that, although it is still nonlinear, a quadratic equation is no longer a good approximation. Some evidence that this may be true is given in equation (2) of Table 6.2. Here the size terms have been replaced by two dummy variables, the first for those hospitals with 100 beds or fewer, and the second for those with 250 beds or more. The important evidence is the significant negative coefficient on the dummy variable for hospitals with 100 beds or fewer. (The insignificance of the dummy for hospitals with 250 beds or more could indicate that the average cost curve is fairly flat in that area, or it could simply be the result of the paucity of hospitals that large in the sample.) The negative coefficients on the two dummy variables confirm the fact that the average cost curve has a peak somewhere between 100 and 250 beds. And the higher R2 with the dummy variables means that they approximate the curve better than does a quadratic function (see column 3 of Table 6.1 and equation [2] in Table 6.2). Although the least efficient size for a hospital appears to have increased between 1958-59 and 1962-63, the results with 1962-63 data do not upset the findings with 1958-59 data that there are diseconomies of scale with respect to hospital size up to a point, and economies of scale thereafter. In particular, both sets of data indicate: (1) that hospitals with 100 beds or less have lower costs per PD, holding other factors constant, than hospitals with between 100 and 200 beds, and (2) that there are economies of scale beyond 200 beds. As in 1958-59, the actual dollar savings involved are small. A hospital in 1962-63 would have had $.80 lower costs per PD if there were 100 beds instead of 190 beds, or costs would have been lowered by $1.25 per PD if there were 300 beds instead of 190, assuming all other factors were held constant. 1962-62 COSTS, CHARGES, AND REVENUES PER ABD PREDICTED FROM 1958-59 EQUATIONS In this section, equations fitted to 1958-59 data are used to predict 1962-63 costs, charges, and revenues. This is a clear test of the model

EMPIRICAL RESULTS FOR

1962-63/101

since no 1962-63 data were used in estimating the equations. If the models were correctly specified—that is, if the equations contained the correct variables and had the right functional form—and if there were no major changes in cost structure between 1958-59 and 1962-63, the equations would predict 1962-63 as well as they predicted costs in 1958-59. It would be a major act of faith, however, to expect this, for no one really believes that models are perfectly specified or that underlying structures are totally homogeneous among hospitals or completely stable over time. The best that can be hoped for is that the equations capture the essential features in the determination of hospital costs and that they are general enough in terms of functional form and explanatory variables to allow for some technological innovation. The 1962-63 data will be a good test of whether this has been accomplished. To predict the dependent variable in 1962-63, one equation for each of the categories analyzed in Chapter 4 has been selected. Only the ABD model has been tested (see Table 6.3), because the results will easily extend to the other models. The first item in Table 6.3 defines the department, while the first and second columns give the equation used. Column 3 reveals the R2 for 1958-59, and column 4 gives the R2 for the predictions. 2 Finally, column 5 presents the Theil U. The Theil U is a measure of goodness of fit, and is more appropriate to employ in forecasting than

T a b l e 6.3.

P r e d i c t i o n s of 1962-63 h o s p i t a l s e r v i c e expense from 1958-59

equations: A v a i l a b l e bed day model for 72 Massachusetts Community H o s p i t a l s . S e r v i c e or d e p a r t m e n t 0

Hospital

services

Table

Equation

1958-59

number

number

R2

4.2

12

5.1

26

5.2

1962-63 2

Theil

.72

.54

.10

.80

.72

.11

26

.80

.72

.11

R

Charges ^ • b Kevenues Primary services

4.5

17

.71

.50

.12

Secondary

4.6

28

.51

.30

.34

4.7

12

.67

.66

.12

4.10

21

.57

.30

.34 .34

services

Independent Nursing

services

services

A d m i n i s t r a t i o n and general

4.11

7

.46

.30

Laboratory

4.12

35

.75

.39

.17

Radiology

4.13

24

.77

.73

.09 .30

O p e r a t i n g room M e d i c a l and s u r g i c a l Pharmacy Medical Hotel

supplies

services

supplies

4.14

16

.53

.26

4.15

14

.55

.26

.30

4.16

8

.38

.35

.12

4.17

24

.67

.46

.10

4.18

14

.41

.26

.30

° A l I a r e e x p e n s e s per A B D u n l e s s o t h e r w i s e n o t e d . b

Per

ABD.

U

102/HOSPITAL COSTS IN MASSACHUSETTS the/? 2 . It measures whether the relationship between the actual and predicted values is a 45° line through the origin. In the usual situation, it ranges between 0 and 1 (0 denoting perfect forecasts, and 1 denoting no relationship at all). 3 In view of all the things that could be wrong with the original 1958-59 equations, in addition to the structural changes that might have occurred between 1958-59 and 1962-63, they are remarkably accurate in predicting 1962-63 when judged by the levels of the Ä 2 's. This confirms what was found at the aggregate level in the previous section—namely, that the underlying structure of costs remained essentially stable between the two periods. The only apparent systematic phenomenon in the residuals for total cost is that the equation tends to underpredict. For only 15 of the 144 observations are the predictions too high. The R2 does not reflect this tendency, but the Theil U does and is higher because of it. This underprediction could be due to an autonomous trend in hospital costs, caused perhaps by higher employment costs or by more complex services not reflected in any of the independent variables. If this is so, it is in keeping with the positive coefficient on the dummy variable separating 1958 and 1959. The results for hospital service expense carry over to the three subaggregates, which means that there probably were no sharp structural breaks within the components that offset one another. The R2,s for the 1962-63 predictions for primary and secondary services are all about 20 points below 1958-59 levels, while the R 2 for the independent services is nearly equal to the 1958-59 value. As with the total, the tendency to underpredict is also apparent in the equations for primary and secondary services—particularly in the nursing, radiology, and laboratory departments and the hotel services. The equation for the independent services, on the other hand, has a slight tendency to overpredict. The lack of sharp structural breaks is further corroborated by the predictions for the individual departments. The prediction from the radiology equation compares favorably to those of 1958-59, as does the equation for pharmacy costs. The equations for medical and surgical supplies, the operating room, and the hotel services are relatively the poorest—the equations for the latter two, in particular, tending both to underpredict and to have a low R2. When judged by the R2's, the equations for charges and revenues also yield adequate 1962-63 forecasts. As with total cost, however, both equations tend to underpredict. For example, the equation for charges underpredicts 142 of the 144 observations—the average prediction being about $6.00 too low. This compares to an average underprediction of costs of about $4.40. That the charge and revenue equations systemat-

EMPIRICAL RESULTS FOR

1962-63/103

ically underpredict 1962-63 is not surprising, since, given the fact that charges (and therefore revenues) are established to reflect costs, the (apparent) a u t o n o m o u s trend in costs would have to show u p in charges also. 1962-63 PREDICTIONS FOR OTHER HOSPITAL G R O U P S U S I N G 1958-59 EQUATIONS FOR THE 72 C O M M U N I T Y HOSPITALS As with the 1958-59 data, the 1962-63 predictions have been extended to several other groups of short-term hospitals in Massachusetts not included in the pilot sample. Unlike the 1958-59 data, however, only hospital service expense per A B D is predicted for 1962-63. T h e results, for the most part, corroborate what was f o u n d with the 1958-59 data; the equation underpredicts the costs per A B D for the teaching hospitals, overpredicts for the church hospitals, and neither systematically over- or underpredicts for the municipal hospitals. T h e m a j o r shifts appear to be in the other teaching hospitals and the proprietary hospitals where forecasts for 1962-63 are systematically too high. This contrasts to 1958-59 where they were generally too low for the other teaching hospitals and were nearly correct on the average for the proprietary hospitals. In view of the small number in these two groups, the switches may reflect changes unique to particular institutions. T h e results indicate that the underlying cost structure was reasonably stable between 1958-59 and 1962-63. There is some evidence that it may have become a little less stable year to year by 1962, but there is no evidence that cyclical factors were any more i m p o r t a n t . It also appears that the variance of the error terms in the regression models increased between 1958-59 and 1962-63. This could be because laboratory, radiology, and surgical factors became less sensitive as indicators of the spectrum of medical care. There also appears to be a trend in hospital costs that is not reflected in either the o u t p u t or internal staff-structure factors. Evidence is the systematic underprediction of costs in 1962-63 using the 1958-59 equation. This is particularly true for the d e p a r t m e n t s providing the primary and secondary services, especially laboratory, nursing, and radiology departments and hotel services. With regard to economies of scale, the shape of the curve relating hospital costs per P D to hospital size, after taking into account other important explanatory factors, was apparently still an inverted U in 196263, but the least efficient hospital size appears to have increased to a b o u t 190 beds from about 150 in 1958-59. The shape of the curve seems to have altered somewhat, being less well approximated by a quadratic function, but still unquestionably nonlinear. As in 1958-59, hospitals with fewer than 100 beds or more than 250 definitely appear to have lower

104/HOSPITAL COSTS IN MASSACHUSETTS costs per PD than those with beds between the two limits, but the dollar magnitudes involved were still small. Finally, in relation to utilization, there is no question that higher occupancy continued to lead to lower costs per PD in 1962-63, and there is also some evidence that the relationship was more nonlinear in 1962-63 than in 1958-59. Occupancy rates of less than 65 percent appear to be particularly inefficient, and there is some indication, in contrast to 1958— 59, that occupancy rates of more than 85 percent are less efficient than between 65 and 85 percent.

7/CONCLUSIONS

The purpose of this chapter is to integrate the findings of the last four chapters and c o m p a r e them with the findings of other investigations. Also, policy implications are discussed, mindful of the temptation that often exists to overinterpret one's results. Finally, the chapter ends with a few afterthoughts that relate to the more general considerations that arose in the course of this project. SURVEY OF THE RESULTS The typical Massachusetts general c o m m u n i t y hospital that emerges from this study is one whose costs can be effectively described by the volume and composition of its services, together with its size, level of occupancy, and internal staff structure. U p to 98 percent of the variation in total hospital cost a m o n g the 72 hospitals in the sample, and u p to 72 percent when costs are expressed in terms of available bed days, can be explained. These percentages are high for cross-section data. The most important explanatory factor is simply the volume of medical care (as measured by the n u m b e r of patient days, n u m b e r of beds, or any other measure of over-all capacity and service); it alone will explain up to 94 percent of the variation in total cost. This means, in effect, simply that larger hospitals systematically serve more patients and cost m o r e to run than smaller institutions. It was primarily to avoid this tautological a r g u m e n t — t h a t hospital costs and o u t p u t will be positively r e l a t e d — t h a t costs were deflated by some measure of size. This deflation also centers attention on a relationship that is of key interest to economists, hospital administrators, and policy-makers alike—namely, the average cost curve. T h e shape of this curve has within the last few years become one of the most controversial issues in hospital cost literature, and, as is true with most controversies, it seems unlikely that any single study can administer last rites. But the present investigation does offer some new evidence. In terms of unit hospital costs, whether they be calculated per A B D or per P D , the most important explanatory factors are those relating to the activities of the departments that directly provide patients with professional services, including those rendering outpatient care. T h e internal staff structure is also important, especially the presence of residents, interns, salaried physicians, and nursing students. 105

106/HOSPITAL COSTS IN MASSACHUSETTS Staff structure may be measured in various alternative ways. Within the frame of reference of this study, for example, the salaries of interns, residents, and staff physicians add to costs over and above the sum of their actual earnings. Of course, from a broader consumer point of view, some of these additions may offset payments that would otherwise be made directly by patients to private physicians. The presence of nursing students, on the other hand, decreases hospital costs, doubtless because they are able to perform many routine tasks (and no doubt some not so routine) that otherwise would be undertaken by staff nurses. Indeed, for the level of nursing salaries prevailing in 1958 and 1959, it was found that the lower costs of service associated with students more than offset the costs of providing the education itself (see Ingbar, Whitney, and Taylor, 1966, and Hale, 1964). To the extent, however, that nursing students are present together with medical students and salaried staff, their presence also tends to raise over-all hospital costs beyond the direct costs of any relevant salaries. Probably this is because a more structured medical staff, even as indicated by the formal language of the hospital's bylaws, 1 is associated with attracting more difficult cases and a higher probable level of qualitative performance; at least this possibility has been suggested by Bower and Roemer (1963), Müller (in Trussell and van Dyke, 1960), and Letourneau (1961). In addition, the literature documents the fact that differences in professional services have also played a key role in explaining the rise in hospital costs over time. 2 The activities of the radiology, operating room, and laboratory departments are of particular interest, since related explanatory variables are positively associated with hospital costs. This can be attributed partly to the cost of service. However, the parameters may also incorporate a general hospital quality variable which the data cannot measure explicitly, but which explains why these activities determine more than just the direct cost of the departments providing the specific service. The number of weighted operations per ABD, for example, obviously reflects the type of illness treated, and the same appears to be true, though perhaps with more emphasis on severity of illness, for x-ray films and laboratory tests. Even more important, however, it seems plausible that radiological activity, in addition to measuring the volume of service and severity of illness, reflects the quality and mores of medical care. At least this appears to have been true in 1958-59. By 1962-63, however, these services must have become less effective measures of the qualitative aspects, the evidence for this being the lower R2. This finding is not surprising. By the later period, such crude measures as films would be less indicative, and newer, more avant garde services would be required (perhaps measures of isotope treatments or steroid chemistries) in order to capture qualitative differences in care.

CONCLUSIONS/107 The ambulatory factor is most significant in explaining total hospital cost per ABD and is also highly instrumental in revealing differences in costs at the departmental level. There are obvious reasons for this importance, the most immediate being that it alone among the explanatory factors measures the activity of the outpatient department and emergency service, the cost of which has been included in determining the unit costs of hospital care. Furthermore, if emergency services are provided, they must be staffed 24 hours a day and equipped with complex apparatus that remains idle much of the time. Such services, as well as those involved in the more routine aspects of an outpatient department, may be doubly costly, for they may affect the structure of the inpatient services as well. The laboratory and radiology departments, in particular, may have to be staffed and equipped very differently both to fulfill inpatient requirements and to serve the needs of large numbers of ambulatory patients, often on a round-the-clock basis. Finally, a hospital may provide its physicians with capital equipment, including offices and even staff, for use in the private practice of medicine, although the physicians do not reimburse the hospitals for the use of such outpatient facilities from their earnings. Hospital size (as measured by the number of beds 3 ) also helps explain the variation in hospital costs. The 1958-59 data revealed the relationship between hospital costs per PD and the number of beds, after other important factors had been taken into account, to be an inverted U, with the peak at about 150 beds 4 and economies of scale thereafter. The 1962-63 data corroborated this inverted U, but indicated that the least efficient hospital size, everything else being constant, was somewhat larger, having increased to approximately 190 beds. The 1962-63 data also suggested that the average cost curve had flattened in its upper range, particularly around 200 beds. This inverted U is, of course, a surprising result, for economists ordinarily think in terms of a U shaped average cost curve. Equally, it defies the conventional wisdom of the hospital field which believes larger hospitals are more economical. Notwithstanding these traditions and despite the haziness persisting in its profile, however, findings from this study clearly indicate that—whatever its exact shape—the average cost curve does not fall monotonically, at least in the 30 to 330 bed range of hospital size, for which results are relevant. The final factor is the occupancy rate. In contrast to the findings of authors such as Tibbitts (1962), and the Pennsylvania Economy League (1962), but in accord with those of Kurtz and Saathoff (1962), this rate was found to be essentially independent of hospital size. With respect to its relationship to costs, however, results for both 1958-59 and 1962-63 indicate that when the occupancy rate increases, costs per PD are definitely lowered, with a hint of a discontinuity around 65 percent occu-

108/HOSPITAL COSTS IN MASSACHUSETTS pancy. On the other hand, occupancy proves useless in explaining costs per ABD; that is, after other factors have been taken into account, there is no difference in costs per ABD between high- and low-occupancy hospitals. In combination, the PD and ABD results imply that most hospital costs are unresponsive to the actual volume of medical care provided, and therefore should be classified as largely fixed. These findings with respect to occupancy, it might be added, were not altered by employing as a measure of the rate of utilization a more sophisticated index of occupancy pressure that takes account of differences in the likelihood of a single bed being empty as hospital size increases (see Blumberg, 1961, and Rosenthal, 1964. Thus they suggest that hospital expenditures may be set according to an expected average daily census plus a healthy margin for periods of peak loads. To the extent that expenditures are determined by these peak loads, the findings suggest that their height does not vary greatly among hospitals. Of course, as actual occupancy rate deviatès from that expected (in terms of an anticipated average daily census plus the peakload allowance), costs per PD will vary. Increased occupancy means that there are more patients over whom to spread the fixed expenditures, and costs per PD will accordingly decrease, while costs per ABD remain the same. Similarly, decreased occupancy will lead to increased costs per PD but unchanged costs per ABD. Turning now to the internal behavior of costs for the hospital, the factor analysis of Chapter 3 indicated that there are two groups of departments whose costs moved together quite closely. They were called the primary and secondary services, respectively. The remaining departments were put into a third group called the independent services. The costs for these three categories and for several individual departments were then analyzed in the same way as were aggregate costs. The major result obtained at a less aggregate level is that costs for the hospital as a whole can be predicted slightly better by aggregating the predictions from the micro equations than by using the prediction from the aggregate equation itself. At the departmental level, there is some evidence of economies of scale (at least in terms of units of departmental output) for those that have clearly defined outputs. Laboratory and radiology are particular cases in point. Surprisingly enough, the hotel services and the pharmacy are the most difficult departments for which costs can be explained, with consequent results for the secondary services as a group. With regard to hospital charges, it was found that the same factors that explain costs also explain charges. From this, it can be concluded that, in setting charges, hospitals are consistent in taking account of the composition of medical care. Moreover, since most of these charges are eventually recovered, the income of the hospitals can also be explained by the

CONCLUSIONS/109 same factors and to the same extent as charges. It should be emphasized, however, that this does not mean that the determinants of the demand for medical care have been explained. The foregoing is the broad picture of the cost structure for Massachusetts community hospitals in 1958 and 1959. This structure was then tested for other types of hospitals and for a later time period. The tests in which the 72 community hospital equations were used to predict the costs of several other groups of hospitals suggest that, except for the major teaching hospitals, cost structures are similar despite differences in ownership and control. The church hospitals, undoubtedly because of their lower personnel costs, had somewhat lower than predicted costs per ABD, whereas the teaching hospitals, as would be expected, had substantially higher than predicted costs. The 1962-63 data indicate that the cost structure is reasonably stable over time. There is some evidence that the long-run average cost curve was shifting to the right and flattening in the upper range, but it was still nonlinear with a peak at about 190 beds in 1962-63. Also, there is some evidence that radiology, laboratory, and surgical activities were poorer explanatory factors, probably because they were less reliable indicators of the quality of medical care in 1962-63 than in 1958-59—undoubtedly reflecting the general elevation in sophistication of medical practice. 5 COMPARISONS WITH OTHER STUDIES The present study is not the first to report a statistical analysis of hospital costs, although few others have had available an equivalent detail of audited data. Paul Feldstein (1961) in a pioneering application of regression analysis to hospital data was among the first to point to the existence of economies of scale with respect to the size of a hospital, although, earlier, more descriptive studies such as those by Schäfer (1949), Pennell et al. (1954), and the prototype series by Block that appeared in Modern Hospital (1953, 1956-1960) had suggested this possibility, as indeed had some of the work on areawide planning undertaken by the Division of Hospital and Medical Facilities of the U.S. Public Health Service. 6 Feldstein's results, based on cross-section data for 60 hospitals ranging in size from 48 to 452 beds, were obtained in a regression of total cost on a constant term and patient days. The constant, being positive (and statistically significant), implied declining average costs per PD when patient days increased, and it was from this that Feldstein concluded that the long-run average cost curve was declining and that therefore economies of scale were present. The particular basis of his conclusion, it should be noted, is at odds with the data from this study; the equivalent relationship for the 72 Massachusetts hospitals in 1958-59 has a negative constant term. 7 Carr and Feldstein (1966) extended Feldstein's earlier analysis to a

110/HOSPITAL COSTS IN MASSACHUSETTS cross-section sample of over 3000 hospitals, using data for 1963. Besides using a greatly increased size of sample, Carr and Feldstein added several refinements to Feldstein's earlier model. First, they used a size-adjusted measure of bed capacity in order to take into account the lower rates of occupancy required by smaller hospitals. Next, they included the number of patient days (their measure of size) as a square, as well as linearly, in order to allow for nonlinearity in the relationship. Finally, and this is the most important extension, they took into account in two different ways the number of services offered by the hospital. First, they included as explanatory variables both a simple count of the number of services offered and this number multiplied by patient days, the latter to allow for any interaction between the number of services and size. Their alternative was to group the hospitals in their sample into five service-capability groups and then to analyze each separately. In both cases, however, the presence of a service was determined by an unaudited self-report, that is, the hospital's statement listing the service in its annual report to the American Hospital Association entitled Survey of Hospitals Accepted for Listing. The results of Carr and Feldstein suggest a long-run average cost curve that is quite different from the one derived from this study. Economists ordinarily think in terms of [/-shaped average cost curves, not the inverted U this study reveals, and this traditional U is the shape their results imply. Based on the equation for the entire 3147 hospitals, Carr and Feldstein find that hospital costs per PD decline until the average daily census reaches about 185 patients and increase thereafter. The Carr-Feldstein results imply, therefore, that diseconomies of scale set in somewhere after 200 beds. The results of this study, on the other hand, suggest the opposite, hence the obvious question: Can these seemingly contradictory results be reconciled? The results for the five service-capability groups defined by Carr and Feldstein show that the relationship between hospital costs per PD and the average daily census decreases for each service-capability group, and it is only for the largest group that the curve eventually turns up. This group contained 20 through 28 facilities as defined by the American Hospital Association. This turning-up appears to occur around an average daily census of 350 patients. Presumably, the curves for the four other groups would turn up also, but the samples lacked hospitals of sufficient size to test this hypothesis. These results suggest, therefore, that the long-run cost curve does not turn up until something beyond 300 beds. This is not inconsistent with the results of the present study, for, given that the sample of 72 community hospitals contained only a few hospitals with 300 or more beds, nothing can be said about the shape of the cost curve

CONCLUSIONS/111 for hospitals with 350 or more. The curve could again turn up in that range. Alternatively, the Carr-Feldstein results and those of this study could be made compatible if in the mid-size range of our peak and the CarrFeldstein trough, hospitals offered a small amount of a great variety of services. Thus, the cost consequences of a given number of services would be minimized by the Carr-Feldstein method, yet their presence would tend to elevate costs inefficiently by our definitions. That this may be the case is tentatively suggested by some of the data presented by Tibbitts (1962) and Kurtz and Saathoff (1962). In addition, the last-mentioned study of Nebraska hospitals discusses the "situational role performance," that is, the diversity and interchangeability of functions undertaken by a single individual, which may well explain the ability of the small institution to operate as economically, if not always as effectively, as the large hospital. Ralph Berry (1965) tackled the economies of scale—number-of-services problem in the same spirit as Carr and Feldstein. Berry went further, however, in that he grouped the more than 5000 short-term hospitals in the United States not only by number of services but also by their equivalence. In the final analyses, Berry defined 40 service groups, including nearly 800 hospitals. For each group, he regressed an average cost, that is costs per PD, on a constant and the number of patient days. Thirty-six of the 40 regression coefficients were negative—26 of the 36 coefficients were greater than their standard errors—which is pretty strong evidence for the existence of economies of scale. The only inconsistency between Berry's results and those of this analysis with regard to the long-run average cost curve is that Berry has economies of scale over the entire range of hospital size, while this study has diseconomies of scale in the lower part of the range. A possible reason for this inconsistency involves the fact that hospitals provide more services as they get larger. 8 The upward sloping portion of the average cost curve of this study may reflect Berry's finding that a small hospital with a given number of services is likely to be more costly per unit of care than a large hospital with the same number. After a certain size, the economies of scale for a given number of services will more than offset the extra expense of adding more, and the average cost curve will decline. This seems to be a plausible explanation and could certainly account for an inverted U cost curve. 9 Expressed in another way, this would mean that hospitals of around 150 to 200 beds tend to provide greater variety of service than they should for maximum efficiency. Another extensive study into the nature of hospital costs was that undertaken by Fitzpatrick, Gottlieb, and Wirick (1964) of the Bureau of Hospital Administration at the University of Michigan. This investiga-

112/HOSPITAL COSTS IN MASSACHUSETTS tion involved a combined time-series and cross-section analysis of 30 hospitals in western New York state for the years 1948-61. The basic findings were that variations in hospital costs are best explained by technology (as measured by the number of employees per bed), the scope of services (as measured by the ratio of ancillary expenses to total patient-care expenses), and the occupancy rate; the effect of the occupancy rate was negative, while the effects of technology and scope were both positive. The study found hospital size to have little independent influence. Although the authors attempted to allow for the interaction of hospital size, technology and scope of services, the multicolinearity among the variables tends to obscure the relationships. Also, the use of number of employees per bed as the measure of technology is rather dubious, given the fact that labor costs account for about two thirds of total hospital expenditure. The Michigan study results, therefore, are somewhat inconclusive. The findings can also be reconciled with those of several other authors. Most interesting, perhaps, is the parallel with the studies of the British National Health Service conducted by Martin Feldstein (1963, 1965, 1966). Whereas Feldstein's analysis of British hospital data related costs to the proportion of cases in each of nine main diagnostic categories, this analysis relates costs to several measures of hospital services. In his 1966 study, Feldstein found that after correcting for differences in these case mixes and the level of case flows among hospitals, the cost per case declined over the observed range of hospital size until about 900 beds and then turned upward. With no adjustment in the level of case flows, Feldstein found that the cost per case curve was constant over most of its range. When this adjustment was made, however, economies of scale (in terms of cost per case) appeared because large hospitals had fewer cases per bed than small ones. Since there is no systematic difference in the occupancy rate between the large and small hospitals, this implies that the average length of stay per case is longer in the large hospitals. Feldstein's argument, therefore, is that the cost per case could be reduced if the large hospitals would shorten the average length of stay and admit more cases. The validity of his argument depends on whether this can be done without jeopardizing the quality of care. Magid and Quadland (1966) made a study of cost variation among the 35 short-term general hospitals in Connecticut, but conclusions are suspect because explanatory variables are either too closely related to one another or to the variable being explained. Since wages and salaries comprise about two thirds of hospital costs, Magid and Quadland, by using the average number of employees per bed as an explanatory factor, come close, in the same way as in the Michigan study, to explaining something by itself. In spite of these defects, however, it is

CONCLUSIONS/113 interesting to note that Magid and Quadland find emergency room visits10 to be an important explanatory variable and also that economies of scale, although present, are relatively unimportant. Both results are in keeping with those of the present study. They are also in accord with the conclusions reached without applying regression analysis by the Maryland Commission to Study Hospital Costs (1964). This commission noted the large variation in the number of employees in hospitals of about the same size but commented that "none of the figures suggest which size of hospitals is the most economical after giving due weight to the differences in the services rendered."11 This matter is further discussed by Weiss (1966) and MacCoun (1961).12 Where do these investigations leave the issue of economies of scale? The present study points to the existence of diseconomies of scale for hospitals with less than 150 to 200 beds and economies of scale thereafter up to about 330 beds, and this can be reconciled with what appears at first glance to be contrary evidence in other analyses. In dollars and cents, however, the effect of economies of scale over this range is insignificant. Beyond 330 beds, the present study can offer no evidence, but the results of Berry, Carr and (Paul) Feldstein, and (Martin) Feldstein indicate that the economies of scale continue to exist. Obviously, the situation demands further investigation. This need is particularly apparent because other inquiries suggest that economies of scale should be present. For example, Lembcke, Hermansen, and Poland (1959) found a decrease in the net square feet per bed for departments of a general hospital as size increased to 175 beds and over. Coughlin and Isard (1963), Coughlin, Isard, and Schneider (1964), and Coughlin (1964) found that although economies of scale were not present among the three Philadelphia hospitals, all the administrators and designers felt that increasing the volume of various specific services should result in such economies. As part of the Duke Endowment studies, Ernst and Ernst (1961) found economies of scale in departmental operations of the Charlotte Rehabilitation Hospital. Similarly, the Hospital Review and Planning Council of Southern New York (1966) found that economies of scale should exist in obstetrical facilities and maternity programs. Finally, the simple fact that there is an overhead component to hospital costs should imply that economies of scale, even on a long-term basis, should be present. This dichotomy between theory and fact deserves critical examination, especially in terms of the effects upon costs per unit of care of the changes in the sizes of the organizational units that are actually rendering these services. Assessing these effects will clearly require more specific data on size and service rates of the organizational units involved,13 including

114/HOSPITAL COSTS IN MASSACHUSETTS information on the qualitative components of care. In addition, if size is to be defined in terms of those elements of patient care that may exceed the bounds of particular departments, single institutions, or geographic areas, new types of detailed information will be needed. Thus, in appraising hospital costs it should be possible to use measures of size that take account of the existence of group-purchasing agreements, joint teaching programs, common laundries, shared laboratories, consultants, and other sorts of common activity or shared equipment, many programs for which are already in existence. 14 Finally, as methods for evaluating medical practice are improved and as indices of quality are refined, 15 more direct measures of these aspects of medical care can be introduced as possible explanatory factors for differences in costs. Although not attempted here, no doubt many of the measures generated by the Professional Activity Study 16 could be usefully related to cost and revenue information. Indeed, it was found that a measure such as autopsy rates might be helpful in explaining differences in laboratory expense, despite the many questions concerning the meaning of such rates, 17 and the accuracy of diagnostic procedures. 18 The issue is not simply whether to supply the consumer with what Coggeshall (1965) has termed "department store" medicine, "where everything he wants is available and the emphasis is on service." 19 Everyone agrees that "rising expectations" 2 0 must be served. The question is what are the advantages and disadvantages of providing service through a chain, formal or informal, as against using a single-building complex or shopping center system. If the chain store pattern is to be followed, then the question arises as to how much of what type of service should be available in each unit. To deal with this, information on travel times, 21 consumer preferences and demographic factors 22 would be needed to supplement evaluations of the qualitative aspects and secondary benefits of providing service in optimally sized units, however defined. IMPLICATIONS FOR POLICY The policy implications of the findings conveniently fall under two headings: (1) those that bear directly on hospital costs and deal with the direction for the future consolidation or expansion of hospital facilities; (2) those that relate to the individual hospital administrator and the decisions that lie within his province. The first of these can be placed under a "macro" heading, and the second under a "micro" heading. Macro implications. As previously noted, because differences in costs are insignificant in dollars for the 30 to 330 bed range of the community hospitals in this study, economies of scale appear irrelevant in deciding whether a community that "needs" 300 short-term hospital beds should

CONCLUSIONS/115 obtain these in one hospital or two. It does not follow, however, that if this community should decide to build two hospitals, they should be identical and independent. The results suggest that considerable savings might result from the sharing of certain facilities; for example, a single laboratory serving both hospitals could make better use of expensive automated equipment. Similarly, amalgamation could result in the elimination of the stand-by costs of some ambulatory facilities. Such savings could be used to improve the services of key emergency units to handle all degrees of severity of injuries and illnesses on a round-the-clock basis, as recommended by the National Research Council. 23 In addition, such coordination could be used to conserve the skills of highly trained personnel, whose scarcity has been increasingly recognized. 24 The economies that would result from the sharing of facilities flow from two sources. The first involves the finding that hospital costs are largely fixed for the accounting year. Variations in costs per patient day occur in response to changes in the occupancy rate, but only because the actual rate deviates from that anticipated in staffing and equipping the service for the period ahead. For some facilities, laboratories, obstetrical units, 25 nursing, and emergency services 26 being particularly good cases in point, the equipment required and personnel needed bear little relationship to the expected utilization rate. 27 Indeed, for much of the time these units are expected to be idle, hence the emphasis on the importance of "stand-by" costs. 28 If the expenditures for maintaining emergency services were the same for 25 anticipated emergency cases a day as for 10, a pooling of facilities would obviously eliminate considerable overhead expense. 29 The results of this study also suggest that there are economies of scale within some departments, and this is the second source of savings that would result from hospitals sharing certain facilities. The laboratory is again a particular case in point. Pooled use of at least certain equipment of a laboratory would lower unit costs, both by eliminating duplicated stand-by or overhead expenditures and by increasing the size of the laboratory. 30 In this era of increasingly complex biochemical assays, the issue is qualitative as well as quantitative. 31 Auto-analyzers require large volumes of specimens to take full advantage of their capacities, 32 but they also are better managed when a biochemist supervises the technicians. Unfortunately, just where economies of scale may actually exist, and their exact extent, remains largely unexplored; 33 precise information is required on the interaction between the cost of a specific service and the capacity of the organizational unit providing the service. Moreover, the problem is compounded by the fact that these organizational units may be neither hospital nor departmental in scope; for example, each ward may be staffed independently to provide nursing service, and laboratories may be es-

116/HOSPITAL COSTS IN MASSACHUSETTS tablished as separate entities to service specialized departments within a single hospital. The issue of appropriate bed size is clearly relevant to the use of federal funds from the Hill-Burton Program. The fact that for practical purposes unit costs are independent of the size of the particular hospital over a 30 to 330 bed range suggests that from the community's point of view the small hospital is as economical as the large hospital for the treatment of those cases in which the more limited services and facilities of the small institution are appropriate to the medical conditions and needs of the patient. Conversely, the large hospital may be no more costly to the community for those cases where its specialized skills and facilities are required. Hence, to the extent that total expenditure is left unaffected, the issue of appropriate over-all size ceases to be purely economic and revolves instead around consumer preferences and the medical requirements for appropriate care. The medical requirements may depend upon the relationships between the volume of case loads and the ability to use highly skilled specialists effectively, by moving either patients to the skill or the services to the patient. When the medical aspects of the choice are moot, the preferences of the consumer become relevant since the convenience and more personalized care of smaller institutions may offset the atmosphere of scientific excellence tending to surround the larger hospital. Although the findings of this study suggest that the distribution of hospital beds among existing institutions does not influence community cost, total expenditure on hospital care is clearly related to the number of beds in the community. Thus, if unnecessary or inappropriate use results in more beds being built than are needed, a community's expenditure would obviously be higher than otherwise. Current policies, as evidenced in programs of areawide planning and in the use of Hill-Burton funds, assume that the costs of such "unnecessary" beds are high and that these should be eliminated. 34 The Citizens Hospital Study Committee of Northeast Ohio 3 5 estimated the cost of maintaining a bed in readiness for service even when unoccupied as $6700 in 1957. Normal use of the bed added $2400 to this figure. This compares with construction costs of from $10,000 to $25,000 per bed for new facilities—the lower figure relating to the addition of beds (as in a new wing) without complementary facilities.36 The present study indicates that once allowance is made for differences in services offered, the cost of an available bed is the same whether or not it is occupied, and it also suggests that a hospital's expenditures are largely set according to an expected average annual daily census plus a margin for peak periods, rather than being tied directly to the number of beds. Particularly if this latter is true, then the elimination of unnecessary beds in itself would not lead to a material reduction in costs, especially as operating costs are high relative to construction costs. 37

CONCLUSIONS/117 Furthermore, if the unnecessary beds exist but are not in place, staffed, and otherwise ready for immediate patient use (these are not included in available bed days in this study), expenditures for their maintenance may be small. Hence the savings resulting from their removal might be negligible.38 Clearly, the real cost of the "unnecessary" bed requires more precise definition. In contrast, the fact that hospital costs appear to be fixed for the accounting year and unresponsive to changes in demand conditions suggests that large contractual buyers of hospital services, such as Blue Cross, can ignore the existence of over- or under-utilization of existing hospital facilities, except when it involves shifts in the number of hospital beds in the community or changes in the buyer's share of the hospital bill. This appears to be in sharp contradiction to the present emphasis upon the importance of eliminating "unnecessary" days of hospital care, particularly those associated with the last days of a patient's stay (in which little special service is rendered). In the absence of changes in number of beds or greater sensitivity of costs to actual use rates, such a reduction in patient days would only lead to a compensating rise in cost per PD, and that would leave total hospital expenditure unchanged. 39 Micro implications. The results of this study are also relevant to the decisions that the individual hospital administrator makes for his own hospital. From this point of view, one particularly important result is that hospital costs are unresponsive to the occupancy rate. N o doubt this finding itself comes as no surprise to most hospital administrators, 40 but it pinpoints one of the most promising areas for reducing costs. If ways could be found to tie some expenses that are now set according to the number of patient days expected to the number of patient days actually treated—that is, if more of the fixed costs could be transformed into variable costs—total costs would probably decline, since the expected number of patient days must include a healthy margin of error, given peak loads, seasonal shifts, weekend troughs, and other peculiarities in the demand for hospital care. The real advantages of areawide planning, pooled facilities, and group ρ μ Γ ΰ Ι ^ β ^ may well lie in their ability to transform what would have been a fixed cost for the hospital into one that is at least variable at the micro level, thereby lowering the over-all magnitude of fixed costs that the community as a whole must support. To participate effectively in such programs, however, hospitals may find it impossible to act alone; it may, for example, be necessary to institute "special courtesy" staff privileges for physicians using such shared facilities. 41 This investigation has also demonstrated the feasibility of using econometric techniques in combination with high-speed computers to analyze

118/HOSPITAL COSTS IN MASSACHUSETTS the data of the hospital industry. As Heneman (1962) stated, the lack of comparability in staffing requirements, productivity, and costs means that "Administrators Waste Time Chasing Details." The individual hospital administrator could use equations such as were presented in earlier chapters to check where his actual or budgeted costs lie, relative to those predicted for a "typical" hospital in similar circumstances. If his costs lie below those predicted by the regression equation, he can congratulate himself for being a good manager; if his costs lie above, he can, by examining the data departmentally, proceed to find out why. Such "target" costing, to adopt a term from Montacute, 42 could provide the standards that "historical" costing and costs per patient day 43 have failed to supply. In order to undertake such analyses, however, the hospital administrator requires current access to large amounts of data on costs and services, both from his own and other hospitals. The benefits that would accrue from centralized collection and storage of these data are obvious. Even in Massachusetts, which appears to have legislation that would permit a state agency to undertake such responsibilities, little progress has been made. Quite apart from improvements in the type of data collected (for example, those concerning laboratory tests) 44 , the Massachusetts Bureau of Hospital Costs and Finances has lacked data-processing facilities to act as a true central depository. Furthermore, problems with respect to rights of disclosure (to the hospitals themselves as well as to outside groups) and frictions created by linking reimbursement payments to cost reports have virtually precluded the use of the data collected by the BHC for planning and administrative purposes. Moreover, it appears that underlying attitudes are such that the only solution for Massachusetts may be to create a superdepository for data previously collected by other groups, such as the BHC, which would be responsible for processing the information and then disseminating it back to interested parties in accordance with prearranged priority and disclosure principles. 45 To create such a super-body, however, would require both subsidies and the full authority of the state, and it would necessitate the integration of innumerable data processing facilities that are increasingly being created within smaller units, even to the point where data concerning hospitals collected in one segment of a university are not in a form suitable for use in another. Although increasing numbers of practical hurdles are being created as automatic data processing becomes more widespread and more sophisticated, these should not obscure, at least not yet, the potentials that exist today for asking different questions, assembling new kinds of data, and integrating and analyzing these statistics at a speed unimaginable even a decade ago. 46 It is, of course, important that data be reported within a uniform set of definitions so that only comparable information is related. But such a

CONCLUSIONS/119 concept of uniform accounting should not preclude a hospital reporting a statistical item simply because it is unique. Indeed, for many purposes the most desired information is that concerning the unusual service and the particularly detailed cost. Thus, although it might appear that lack of data and inconsistent reporting are the biggest problems facing the cost-conscious hospital administrator, this should no longer be the case. The capabilities of computers to handle large volumes of exceedingly detailed information and the potentials of missing observation techniques mean that the major stumbling block at present is not the collection of data, but rather the lack of even rudimentary access to the data that are already being collected. If the interested hospital administrator is to make analyses of the type illustrated in this study, this obstacle must be overcome. AFTERTHOUGHTS Some final observations seem appropriate. Besides the structural and managerial questions of the hospital industry, two basic motivational problems became apparent in the course of this investigation. Both relate to the lack of incentives that stimulate efficient use of resources and foster equitable distribution of the cost of care. Such incentives, at least for most other industries, are closely associated with the profit motive, and they serve to reconcile the satisfactions and divergent behavior of employees, producers, and consumers. In striking contrast to this, the contractual reimbursement structure in the hospital industry, both in Massachusetts and in other areas, 47 neither rewards excellence nor discourages inefficiency. By adhering strictly to the principle of merely covering "allowable" expenditures (defined to be representative of those costs incurred in providing direct service to the contractual patient), long-run interests tend to be sacrificed to short-run expedients. Not only must expenditures fit into fully "reimbursable" categories, but provision for obsolescence and depreciation may be such that the use and replacement of capital equipment is discouraged, even when it could substitute for more expensive labor. 48 As a result, hospitals often find themselves in the position that saving money is tantamount to lowering income. To the extent that there are monetary incentives to cut costs, moreover, they often encourage the hospital to work against the interests of the patient, particularly the unfortunate who lacks insurance and must actually pay the charges. 49 For the hospital, the larger the consumption of the so-called special services (for which every patient is always billed on an itemized basis), the greater is the daily financial value of the case. Yet such special charges, which this study has found to be quite predictable on a per diem basis, are most likely to be burdensome to those patients who

120/HOSPITAL COSTS IN MASSACHUSETTS are least able to afford the payments but should most seek the care—for example, those undergoing the preventive practice of diagnostic evaluation and those who are suffering from chronic or serious illness. Finally, for both the insured and the uninsured patient, since room and board charges are on a per diem basis, hospital income is largely determined by the number of days that patients are hospitalized. As Roemer (1959) recognized, the payment system thus creates administrative incentives for maximum occupancy of a given bed complement—an incentive, moreover, that may be bolstered by attending physicians who also receive a per diem payment for daily visits to each of their patients in the hospital. 50 In short, in the absence of demand pressures, or peer evaluation of the appropriateness of treatment procedures and utilization rates, payment systems quietly encourage prolonged stays and special services. They compound such effects by simultaneously failing to provide the subtle aspects of design and discrimination that would also encourage better and more effective medical care in the hospital and the use of equally appropriate but cheaper alternative facilities. 51 One of the problems, of course, is that the third-party reimbursement structures assume that, at least in terms of payment status, all medical care is equally urgent. Much as the American Medical Association (its own sponsorship of numerous specialty boards, notwithstanding) publicly acclaims that all doctors are equally competent in all procedures, 5 2 thirdparty payment structures leave no room for optional extras, particularly with respect to what the British term the amenity bed or the luxury of diagnosing what Meador (1965) has termed "nondisease." Although lack of comprehensiveness in prepayment coverage may invite premature specification of a diagnostic syndrome or too early admission of what should have remained an ambulatory case, 53 such unfortunate consequences of contract exclusions only point to the power of the tool. It would appear that there should be better ways to provide optional extras and amenity services within the structure of prepayment, without going to the extremes of either eliminating the luxuries or. requiring the system as a whole to support "cadillac" medicine for all. 54 The core of such improvements must await the ability to establish firm standards that distinguish medically significant services from those that merely make the stay in hospital personally more tolerable. For clearly peripheral items, it seems only appropriate for hospitals to charge what-the-traffic-willbear, particularly since this might help alleviate the perpetual shortage of funds for educational purposes, as well as for needed research and "free care." The whole question of reimbursement, therefore, is greatly in need of detailed investigation. 55 With Medicare, contractual reimbursement will

CONCLUSIONS/121 be taking on even increased importance and pressures on facilities and personnel will mount. Greater sums and larger proportions of total hospital cost will be distributed in accordance with a few formulas, and, as a result, reimbursement procedures will play a more active role, for better or for worse, in shaping the movement of hospital costs. Alternatives to audited costs as a basis for reimbursement will need to be considered, such as the all-inclusive rate discussed by Rosenkrantz and Bornstein (1963), the controlled charge system described by TeKolste (1963 and 1966) and Sigmond (1963), and the hourly charge system reported by Losh (1962). There are also numerous proposals concerning the use of floors, ceilings, sliding scales, discounts, and negotiated rates, that might be considered. 56 In addition, to the extent that the role of costs in reimbursement is altered, it is possible to consider alternative methods of reporting expenditure information to reviewing authorities. The reporting formats used by the Dominion Bureau of Statistics for hospitals in Canada, for example, force each hospital administrator to explain changes in departmental expenditures, both as they have already occurred and as they are anticipated in budgets for future time periods. 57 The British, particularly since the Plowden Committee report of 1961 which proposed the forwardlook budget, have also used the reporting process as a device to promote cost-consciousness and expenditure control. 58 The contribution of this study to such discussions is indirect. To the extent that our results indicate that it is possible to appraise the reasonableness of costs by relating them to services offered, the formulas used for reimbursement purposes can be freed of this responsibility. Instead of merely providing payment for dollars that have been "allowably" spent, the formulas can be constructed to provide incentives toward better care and lower costs. Just how this can be accomplished is not exactly clear, but several observations are in order. First, factors that influence revenues, and thus the break-even points of individual hospitals, should not be confused with those that influence expenditures on medical care by the patient, by each hospital, and by the community as a whole. Furthermore, for the community-oriented voluntary hospital, income in excess of expenditure becomes the better service of tomorrow, because funds not spent in one accounting period can be used in the next to foster growth, both qualitative and quantitative. Second, high utilization may be encouraged, as by present per diem reimbursement policies, as a means of enlarging revenues and thereby covering invariant expenses, but these expenses might themselves be lowered if costs were more responsive to actual utilization or if labor-saving innovations were adopted. Even if this were accomplished, however, cost-based methods of reimbursement would immediately require adjustments in per diems to correspond with

122 /HOSPITAL COSTS IN MASSACHUSETTS lowered expenditures, in which case hospitals might find themselves in the unfortunate position of needing to repay contractual groups for overpayments already received under old per diem rates. Thus, it would appear that some compromise between an inflexible per diem (without any consideration of actual break-even point volumes) and strictly variable break-even point pricing (without any regard for the unpredictability of service levels) might be desired. A policy of cost or charges, whichever is lower, appears to offer the disadvantages of both systems and the advantages of neither. If, as this study suggests, it is possible to appraise the reasonableness of hospital costs independently, it would appear that payment systems can remain somewhat divorced from short-term costs. The reimbursement structure could then be used to create incentives for hospitals to produce the most effective service at lowest cost.

KEY TO SYMBOLS A N D ABBREVIATIONS APPENDIXES

KEY TO SYMBOLS AND ABBREVIATIONS

A . Key to explanatory factors and variables Symbols used to designate rates (where η = variable number as specified by the χ subscript below) x„ Rate per available bed day sn Rate per patient day tn Rate per discharge u„ (or u'n) Rate per unit of departmental output Statistical items and their factor association and variable number Factor 1: Size-volume Capacity x 3 8 — Beds—adults and children X40—Available bed days—adults and children X42—Available bed days—medical and surgical services Total services X43—Laboratory tests—all patients X44—Radiology weighted films—all patients X45—Radiology weighted films—inpatients χ 48—Weighted operations—all patients X50—Weighted operations and weighted deliveries—all patients Maternity services X52 — Available bed days—maternity Radiology treatment services Medical education and physician services X54 — Medical and surgical physician status X55 — Medical and surgical physician expense rate χ 56 — Laboratory physicians— fu 11-time* χ 57 — Radiology physicians—full-time* \6> —Intern and resident status x 62 — Average number of interns and residents per bed Nursing education services χ 63—Nursing education status

* Item not used in 1962-63 regression equations. 125

126/KEY TO SYMBOLS A N D ABBREVIATIONS x M — Nursing education expense rate x 6 5 — N u m b e r of nursing students per bed Factor 2: Utilization X66—Occupancy rate—adults and children (Ratio of patient days to available bed days) Factor 3: Length of stay χ67 — Average length of stay—adults and children (Ratio of patient days to discharges) X70—Ratio of ambulatory weighted operations to weighted operations plus weighted deliveries* Factor 4: Laboratory activity X72 — Laboratory test rate—inpatients Factor 5: Radiology activity X74 — Radiology weighted film rate—inpatients X75 — Radiology treatment status Factor 6: Surgical activity X77 — Weighted operation rate—inpatients in 1958-59, all patients in 1962-63 X78 — Weighted operation and weighted delivery rate X79 — M a j o r operation rate x8o —Operation p r o p o r t i o n — m a j o r to all types Factor 7: Maternity activity χ g) — Maternity status χ 82 — Occupancy rate—maternity χ 83 — Average length of stay—maternity x 85 — Maternity patient day rate x 8 6—Weighted delivery rate Factor 8: Pediatric activity X97 — Pediatric status* χ 98 — Pediatric patient day rate* X99 — Operation proportion—tonsillectomies and adenoidectomies to all types Factor 9: Ambulatory activity χ 87 — Laboratory test rate—outpatients χ 88 — Radiology weighted film rate—outpatients χ 89 — Radiology weighted film proportion—outpatients to all patients Factor 10: Private services X90—Private patient day rate x 9 , — Private patient days as a proportion of all patient days *Item not used in 1962-63 regression equations.

KEY TO SYMBOLS AND ABBREVIATIONS/127 X92 — Private maternity patient days as a proportion of all maternity days Factor 11 : Ward services X93 — Ward patient day rate X94—Ward patient days as a proportion of all patient days Other items X41 —Year (1958 = 0 and 1959 = 1) (1962 = 0 and 1963 = 1) X96—Percent of autopsies to deaths

B. Key to superscripts and subscripts appearing in tables in Chapters 4 and 5 Superscripts to slope coefficients B: t = 1.50—1.99 C: t = 1.00—1.49 D: t = 1.00 The significance of regression coefficients has been tested by the standard t test. The t value is greater than 2 unless otherwise specified. The significance of the multiple determination coefficient has been tested by the F test. A 1 percent significance level has been used unless otherwise specified. Subscripts to slope coefficients The number (1 to 9) below and to the right of the slope coefficient indicates the importance of that factor as measured by the rank order of the significance of the slope coefficients. Other Symbols •Used in Tables 4.8 and 4.9 'Used in calculating residuals analyzed in Figures 5.1-5.6

C. Key to abbreviations Inpatients (inpats)—adults and children, and newborn infants staying in the hospital after the discharge of their mother. A&C—adults and children (excluding newborn infants staying in the hospital after the discharge of their mother). M&S—medical and surgical patients = adults and children minus maternity patients. This abbreviation is combined with "inpats" or " A & C " to specify the inclusion or exclusion, respectively, of data relating to newborn infants staying in the hospital after the discharge of their mother.

128/KEY TO SYMBOLS AND ABBREVIATIONS ABD—available bed days = average number of beds that are in place, staffed and otherwise ready to use multiplied by number of days in the year. PD —patient days. Ρ —discharges, or admissions in the case of pediatric data. Β —beds.

D. Definitions of output measures 1. Twenty (20) dental films or "one ordinary large size film" are each considered to be the equivalent of one weighted film. 2. Three (3) minor operations, one major operation or three weighted deliveries performed in the operating room are each considered to be the equivalent of one weighted operation. 3. One Caesarean section performed in the delivery room is equivalent to three weighted deliveries; seven circumcisions or one delivery are equivalent to one weighted delivery. 4. Weighted deliveries are divided by three before being added to weighted operations. 5. A status measure indicates the absence or presence of a service with a zero or one, respectively. 6. Number of interns and residents is rounded to the nearest multiple of five. 7. Where there is conflict in terminology, as in the distinction between newborns and boarders, the policy of the American Hospital Association is followed rather than that of the Massachusetts Bureau of Hospital Costs and Finances (BHC). Thus, in this study newborns include the infants staying in the hospital after the discharge of their mother, that is, newborns include the so-called "boarders" that are separately defined by the BHC because they qualify for the same contractual per diem rates as other patients. In this study, the term inpatients is adopted whenever data concerning adults and children include information for these boarders. In contrast, the term adults and children (A&C) is employed whenever data are known to exclude boarders as well as other newborns. This policy also is at odds with that followed by the BHC.

APPENDIX 2.1

Reporting Schedules from the Hospital Statement for Reimbursement of the Bureau of Hospital Costs and Finances of the Commonwealth of Massachusetts Schedule Schedule Schedule Schedule Schedule Schedule Schedule Schedule

II IV V V-A V-B V-C VI VI-A

—General Fund Income Summary —Statistical Data —Distribution of Expense for Apportionment —Non-Patient Expenses —Other Expenses —Recovery of Expense —Distribution of Expense by Services —Allocation of Special Service Department Costs

129

ζ

UJ ** η u m & O o β

r

1

CO

_366 00 00

o o

i

α o

o o

CO « í o o c o © u-l r·» r·. fO fT f )

c

(rt

ü

1

X

o o

o o

¡

t

1 83 30«

384 00 00

O

ζ

83

X

(SPFCIFY):

TOTAL OROSS EARNINGS

TOTAL SPECIAL SERVICES

0THFR

25 00

30]

83 301

83 30J

83 30J

30]

30j

83

30]

407 00 00 83 30

1_386 00 00

336 00 00 83 30J

324

324 03 00

1

CENTRAL S T E R I L E SUPPLY

1324 05 00

OXYGEN T H E R A P Y

3θ]

83 30J

] 379 00 00 83

o

o o

1

0

o o

o o

1

«

o o

o o o o CO CO CO CO

o o

o η

1

o

co fO αο eo

PHYSICAL THERAPY

cr

353 00 00 83

o o

1

o o

o o co -

o

3 UJ

m

EARNINGS

83 30

j401 00 00 83 30

( N O T AS

PR 1 V A T E

PATIENTS).

ETC.

X X X

COMBINED PER D I E M - GROSS

30

30

X X X

ΡΟΗ M HCP JM-M

93 83

1 NFANTS

NEWBORN

X

* Identificación Code

13

83

93 83 30

402 00 00 83 30

403

I N C L U D I N G W O R K DONE FOR O T H E R H O S P I T A L S A ID

INCLUDING C L I N I C S ,

11

403

93 83 30

1 403 12 93

10

403

X X X

t

30

83

93

30

30

X X

"

INCOME

INCOME (PER SCHEDULE l l - A )

TOTAL GENERAL FUND

OTHER

o «s-

NET EARNINGS FROM S E R V I C E S TO PATIENTS

403 07

M

1 403 09 93 83 30

o

TOTAL DEDUCTIONS FROH GROSS E A R N I N G S

83

403 06 93 83

403 05 93

oo

RECEIVABLES

30

m

00

TOTAL FREE S E R V I C E ANO ALLOWANCES

83

O r»

η

P R O V I S I O N FOR U N C O L L E C T I B L E

403 02 93

1 403 03 93 83 30 403 04 93 83 30

«*-

X X X

X X X

χ

1

É l

Κ»

2

133

Schedule IV Source Code = 14 (audited)

-Hospital S t a t i s t i c a l Data For the year ended IN-PATIENT

STATISTICS:

1.

B E OCOMPLEMENT

A T ENOOF P R E V I O U S

2.

B E DCOMPLEMENT

A T E N DO F C U R R E N T

3.

MAXIMUM

t.

TOTAL

B E DDAYS

INFANT

BASSINET, 5.

STATISTICS INFANTS

DAYS

(EXCLUSIVE

DAYS

EXCLUDED

IN T H E NURSERY, BY C L A S S

DURING

INFANT

THOSE

DAYS

T H EMOTHER'S

OF ACCOMMODATION

INCLUDED

BASS I NETS)

60800960004*

BASSINETS)

(TOTAL

DAYS) WHEN

AN I N F A N T

OCCUPIES

A

HOSPITALIZATION.

IN-PATIENTS,

EXCLUSIVE

O F NEWBORN

IN ITEM 8 1 :

TOTAL IN-PATIENT DAYS

PERCENTAGE OF OCCUPANCY

DISCHARGES (INCLUDING DEATHS)

AVERAGE LENGTH OF STAY

S0S00Ü01001

S0S00991009

80800001002

80300001008

MAXIMUM BED DAYS AVAILABLE

IVATE

TOTAL

6.

O F NEWBORN

A R E ONLY

AVERAGE BED COMPLEMENT

SEM I - P R

(EXCLUDING (EXCLUDING

AV A I LA Β LE

IN-PATIENT

NEWBORN

PERIOD PERIOD

6 0 8 0 0 9 6 1 0 0 4 *

MATERNITY

STATISTICS

8 0 8 0 0 9 9 1 0 0 6

( I Ν - Ρ A T I E NT S ,

EXCLUSIVE

MAXIMUM BED DAYS AVAILABLE

AVERAGE BED COMPLEMENT

O F NEWBORN

TOTAL IN-PATIENT DAYS

INFANTS):

PERCENTAGE OF OCCUPANCY

DISCHARGES (INCLUDING DEATHS)

SEMI-PRIVATE

60895961404*

TOTAL

7.

NEWBORN

INFANT

DAYS

7A.

NEWBORN

INFANT

DISCHARGES

AFTER

8.

NEWBORN

INFANT

STATISTICS:

¿0895991406

THEDISCHARGE APPLICABLE

O F MOTHER

(INCLUDED

IN ITEM 5 ) .

TO T H E ABOVE

(INCLUOED

IN ITEM

(EXCLUDING

TOTAL

NEWBORN

INFANT

DAYS

TOTAL

NEWBORN

INFANT

DISCHARGES

80895001402

30895001401

COUNT

80802001912

5)

IN ITEM 7 ) .

AMBULATORY S T A T I S T I C S : 1.

CLINIC

2.

EMERGENCY

PATIENT

3.

PRIVATE

CLINIC

4.

PRIVATE

REFERRED

TOTAL

VISITS

PATIENT

•Identification

VISITS

PATIENT

AMBULATORY

rORM HC F 300*0

80900009105

VI S I T S

PATIENT

VISITS

Code

VISITS

AVERAGE LENGTH OF STAY

o co

< Ζ Ο — ( UJ Ο —

Ό O υ

* Λ

«

«

«

N Ν (Μ

I« ®

Ç3 ^

ο ο ο ι Ο Ο Ο I

ο ο

ο ο

ο ο

ο ο

ο ο

1

NORMALLY

s í S O

o

THE

UJ

FOR

»1

00

49-

* OVERHEAD

·
-

(TI

INCLUDING CLINICS,

OV rH

(

00

211



»

o 0 ps.

m Os

201

o 0

TOTAL

o o

SCHEDULE V-B)

o o

o

o

(PER

o o

EXPENSES

o o

o

00

30 3

O

o

00

00

O fO

o o

OTHER

00

I-H

00

co O

00

O O O

00

30 3l

s O O

( S P E C I A L ) 122

184

177

30 3l 30 3Í

o o

co

OTHER ( S P E C I F Y ) : PURCHASED SERVICES

SUPPLIES

00

02 00

175

o oco

SURGICAL

00 00

00

00

00

00

170

O m

&

30 31

O CO

AMBULANCE

00

η

MEDICAL

00

VO VO 1-1

LABORATORY

00 E>

1

(/) UI
O o

1

162

30 3 30 31

00

00

00

00

00

00

149

153

R-» IR» r-t o O

1

ROOMS

o O

O o

RADIOLOGY:

OF PATIENTS - S P E C I A L

ROOMS O o

1

DELIVERY

OPERATING

PROFESSIONAL CARE

co D-

X X ARE

Schedule V-A S o u r c e Code -

Hospital

22

Non-Patient Expenses For the year ended

NON-OPERATING A COST

OF

FUND

EXPENSE

AND

PATIENT

RAISING

RESEARCH MEDICAL

ITEMS

NOT

AND

GIFT

LUNCH

COUNTER,

STATE

TAXES

SHOPS,

FEDERAL

AND

EXPENSE

INCURRED

ESTATE DUTY

AS

IN

TAXES

TELEVISION

RESULT

ON

EXPENSE IN

OF

SHOPS,

ETC.

EXPENSE

INCOME

OTHER

INCLUDED

A

EXPENSE

COFFEE

PRODUCING

AND

NURSES

EXPENDITURES

OTHER'

$ 195 195 195 195 195 195 195 195 195

EXPENSE

RADIO

MADE

BE

EXPENSE

EDUCATION

PRIVATE

TO

EXPENSE

NEWSPAPERS,

REAL

CONSIDERED

CARE:

NON-OPERATING

EXPENSES AND

ON

INCOME

NON-HOSPITAL

OTHER

"AGENCY·

FUND

EXPENSES

GENERAL

AVAILABLE

FUNDS

FOR

F A C I L I T ES

EXPENSES

19S

DESIGNATED

TO

THIS

CARE,

DONATIONS,

CAPTION

SUCH

40* 40 40 40 40 40 40 40 40

09

00

40

195 15 00 40

(SPECIFY):

UNDER

PATIENT

00 00 00 00 00 00 00 00 00

PURPOSES

$ 195 00 00 40

TOTAL (TO SCHEDULE V) •INCLUDED

02 04 05 06 07 11 10 14 08

AS

SHOULD

BE

DEDICATION

THOSE

EXPENSES

EXPENSES

FOR

THE

WHICH

HAVE

OPENING

OF

NO R E L A T I O N S H I P

OR

A

CHARITABLE

NEW

HOSPITAL,

BEARING

ETC.

Schedule V-B S o u r c e Code «

22

Other Expenses For the year ended

PROVISION

FOR

DEPRECIATION

-

PROVISION

FOR

DEPRECIATION

-

FIXED

FOR

DEPRECIATION

-

MAJOR

PROVISION INTEREST RENTALS REAL LEGAL

BUILDINGS

MOVABLE

EXPENSE OF

ESTATE FEES

HOSPITAL AND AND

LAND

OTHER

AND

PROPERTY

BUILDINGS TAXES

EXPENSES

TOTAL (TO SCHEDULE V) *

Identification

%

EQUIPMENT

Code

C O M M O N W E A L T H OF MASSACHUSETTS HCP 300-60

EQUIPMENT

19fi

0.3

00

40

19«

04

00

40

io« 198 198 198 198

ns 06 07 08 09

nn 00 00 00 00

¿.n 40 40 40 40

$ 198 00 00 40

137

Schedule V-C ource Code = 22

Hospital Recovery of Expense For the year ended

ITEMS

OF

INCOME WHICH ARE

ACTUALLY

RENTAL

OF QUARTERS

RENTAL

INCOME

INCOME

FROM M E A L S SOLO -

INCOME

FRO(l LAUNDRY

INCOME

FROM L I N E N S E R V I C E

TELEPHONE

TO E M P L O Y E E S

OF H O S P I T A L

INCOME -

INCOME FROM

EXPENSE:

$

OTHERS

TO OTHER THAN -

TO OTHER

PATIENTS THAN

TO OTHER THAN

PAY S T A T I O N

INCOME

PATIENTS

PATIENTS

EXCLUDED

SCHOOL

INCOME FROM T R A I N I N G

PROGRAMS

(SPECIFY):

(SPECIFY):

THE ABOVE

ITEMS

THOSE G E N E R A L

02 03 04 ni Ofi

00 00 00 no no

40* 40 40 in 40 1 9 7 0 7 n o AO 197 08 00 ¿0 197 09 00 40

197 12 00 40

(SPECIFY):

i

TOTAL (TO SCHEDULE V) NOTE:

197 197 197 1Q7 197

197 10 00 40 197 11 00 40

RECORDS

GRANTS—1N—A10 OR S U B S I D I E S

OTHER

AND

OF

AREAS

SERVICE -

NURSING

INCOME FROM M E D I C A L

RECOVERY

OF

INCOME

FUND E X P E N S E

MUST BE O F F S E T ACCOUNTS

* Identification Code

COMMONWEALTH OF MASSACHUSETTS

HCF 300-S0

BY P R E L I M I N A R Y

IN WHICH THE

EXPENSE

ADJUSTMENT

AGAINST

ORIGINATED.

197 00 00 40

139

Schedule

Vl-A-I

S o u r c e C o d e = 23 (.auaicec; S o u r c e C o d e = 24 Costs (unaudited)

.Hospi tal A l l o c a t i o n of Special

Service Department

For the vear e n d e d

NUMBER

TOTAL EXPENSE

%

STATISTICS OF

OPERATIONS-

OPERATING ROOMS:

WEIGHTED

IN-PATIENTS: ADULTS

AND

NEWBORN AMBULATORY OTHER

CHILDREN

t _

INFANTS SERVICES

NON-PATIENT

SERVICES

TOTAL (TO SCHEDULE VI)

74900

00

01

NUMBER

DELIVERIES

NEWBORN

i

i

C I R C U M C I S IONS

IN-PATIENTS: ANO

100.00%

WEIGHTEO

DELIVERY ROOMS: ADULTS

21*

OF

*

CHILDREN

INFANTS

TOTAL (TO SCHEDULE VI)

7530C

00

01

NUMBER

25

100.00%

$

OF

WEIGHTED OPERATIONS

ANESTHESIOLOGY:

WEIGHTED

t

DELIVERIES

IN-PATIENTS: ADULTS

AND

NEWBORN AMBULATORY OTHER

J

CHILDREN

INFANTS SERVICES

NON-PATIENT

SERVICES

TOTAL (TO SCHEDULE VI)

75700

00

NUMBER

RADIOLOGY - DIAGNOSIS:

01

29

100.00%

*

OF

FILMS-WEIGHTEO

IN-PATIENTS: ADULTS NEWBORN AMBULATORY OTHER

ANO

t

CHILDREN

INFANTS SERVICES

NON-PATIENT

SERVICES

TOTAL (TO SCHEDULE VI)

* Identification Code

C O M M O N W E A L T H OF M A S S A C H U S E T T S HCF 300-B0

76600

00

01

35

100.00%

$

Schedule Vl-A - 2 Hospital

Source

Code

= 23

(audited)

A l l o c a t i o n o f Special S e r v i c e Department Costs

Source

Cods = 24

(unaudited)

For the year »nderf

TOTAL STATISTICS

EXPEHSE

i

NUMBER OF TREATMENTS

RADIOLOGY-THERAPY: 1N—PATIENTS: A D U L T S AND NEWBORN

CHILDREN

L

INFANTS

AMBULATORY

SERVICES

OTHER N O N - P A T I E N T TOTAL

SERVICES

(TO SCHEDULE V I )

77000

00 01 40*

100.00^

$

NUMBER OF LABORATORY

LABORATORY:

TESTS

IN-PATIENTS: ADULTS AMBULATORY

INFANTS SERVICES

OTHER N O N - P A T I E N T TOTAL

$

AND C H I L D R E N

NEWBORN

SERVICES

(TO SCHEDULE V I )

77500

00 01 45

100.009

$

NUMBER OF BASAL

TESTS

METABOLISM:

IN-PATIENTS: A D U L T S AND NEWBORN AMBULATORY

SERVICES

OTHER N O N - P A T I E N T TOTAL

$

CHILDREN

INFANTS SERVICES

(TO SCHEDULE V I )

77702

00 01

45

100.009

$

NUMBER OF EXAMINATIONS

ELECTROCARD1OLOGY: IN-PATIENTS: ADULTS NEWBORN AMBULATORY OTHER

INFANTS SERVICES

NON-PATIENT TOTAL

$

AND C H I L D R E N

SERVICES

(TO SCHEDULE V I )

*Identification

Code

COMMONWEALTH OF MASSACHUSETTS HCF 300-S0

77703

0001

50

100.009

$

141

Schedule

.Hospital Allocation of Special

S e r v i c e Department C o s t s

VI-A-3

S o u r c e Code = 23 (audited) S o u r c e Code = 24 (unaudited)

For the year «>nded

STATISTICS NUMBER

PHYSICAL THERAPY:

OF

TREATMENTS

IN-PATIENTS: ADULTS

AND

NEWBORN

$

CHILDREN

INFANTS

AMBULATORY OTHER

TOTAL EXPENSE

Í

SERVICES

N O N — Ρ A T 1 FN Τ

SERVICES

TOTAL (TO SCHEDULE VI)

77900 00 0 1 4 0 * PERCENTAGE

OF

100.00%

$

USE

OR

AMBULANCE:

NUMBER

OF

TRIPS

IN-PATIENTS: ADULTS

AND

NEWBORN AMBULATORY OTHFR

*

CHILDREN

INFANTS SERVICES

N O N — P A T 1 F NT

SERVICES

100.00%

TOTAL (TO SCHEDULE VI)

MEDICAL AND SURGICAL SUPPLIES (SPECIAL):

$

ACTUAL COST OR I N C O M E BASIS PER S C H I X - B ( C O L . 6)

IN-PATIENTS: ADULTS

AND

NEWBORN AMBULATORY OTHER

Í

CHILDREN

INFANTS SERVICES

NON-PATIENT

SERVICES

TOTAL (TO SCHEDULE VI)

72200 00 8 1 60

100.00%

$

ACTUAL COST OR I N C O M E BASIS PER S C H IX-A ( C O L . 6)

PHARMACY (SPECIAL): 1N—PAT 1ENTS: ADULTS NEWBORN AMBULATORY OTHER

AND

Í

CHILDREN

INFANTS SERVICES

NON-PATIENT

SERVICES

TOTAL (TO SCHEDULE VI)

* I d e n t i f i c a t i o n Code COMMONWEALTH OF MASSACHUSETTS HCF 300-00

73600 00 8 1 60

100.00%

$

APPENDIX 2.2

Criteria for Classifying Hospitals in the Massachusetts Hospital Cost Study Community hospitals Voluntary (nonprofit) hospitals that are not church related or operated (AHA Control Classification Code "23") providing general services (AHA Service Classification Code "10") to short-term patients (average length of stay under 30 days—that is, A H A Stay Classification Code "S"). In addition, hospitals in this group lack affiliation with a medical school (without AHA Approvals Code "A-5") but are accredited by the Joint Commission on Accreditation of Hospitals (AHA Approvals Code " A - l " or Blue Cross " G r o u p 1" or " G r o u p 11" indicating a hospital approved by the Joint Commission on Accreditation of Hospitals that may or may not be, respectively, approved by the American Medical Association for the training of interns and residents). Teaching hospitals Voluntary (nonprofit) hospitals that are not church related or operated (AHA Control Classification Code "23") providing general services (AHA Service Classification Code "10") to short-term patients (AHA Stay Classification Code "S"). In addition, hospitals in this category are affiliated with a medical school (AHA Approvals Code "A-5" or Blue Cross " G r o u p 1A") as well as accredited by the Joint Commission on Accreditation of Hospitals (AHA Approvals Code " A - l " ) . All institutions also usually offer cancer programs approved by the American College of Surgeons and residencies approved by the American Medical Association (AHA Approvals Codes "A-2" and "A-3," respectively). Normally, internships approved by the American Medical Association are also offered (AHA Approvals Code "A-4"). Church-sponsored hospitals Voluntary (nonprofit) hospitals that are church related or operated (AHA Control Classification Code "21") providing general services (AHA Service Classification Code "10") to short-term patients (AHA Stay Classification Code "S"). In addition, hospitals in this category lack affiliation with a medical school (without A H A Approvals Code "A-5") 142

CRITERIA FOR CLASSIFYING

HOSPITALS/143

but are accredited by the J o i n t C o m m i s s i o n on A c c r e d i t a t i o n of H o s p i t a l s ( A H A A p p r o v a l s C o d e " A - l " or Blue C r o s s " G r o u p 1" o r " G r o u p 11" indicating a hospital a p p r o v e d by t h e J o i n t C o m m i s s i o n on A c c r e d i t a t i o n of H o s p i t a l s t h a t may or m a y not be, respectively, a p p r o v e d by the A m e r i can M e d i c a l Association for the training of interns a n d residents).

Other teaching hospitals City or v o l u n t a r y ( n o n p r o f i t ) hospitals ( A H A C o n t r o l Classification C o d e s " 1 4 , " " 2 1 , " or " 2 3 " ) p r o v i d i n g general o r special services o t h e r t h a n psychiatric, tuberculosis or m a t e r n i t y care ( A H A Service Classification C o d e s " 1 0 , " " 4 5 , " " 4 9 , " " 5 0 , " " 5 9 " ) to s h o r t - t e r m p a t i e n t s ( A H A Stay Classification C o d e " S " ) . In a d d i t i o n , hospitals in this category are affiliated with a medical school ( A H A A p p r o v a l s C o d e " A - 5 " o r Blue Cross " G r o u p 1A") as well as accredited by t h e J o i n t C o m m i s s i o n on A c c r e d i t a t i o n of H o s p i t a l s ( A H A A p p r o v a l s C o d e " A - l " ) . All institutions also offer residencies a p p r o v e d by the A m e r i c a n M e d i c a l Association ( A H A A p p r o v a l s C o d e " A - 3 " ) .

Maternity hospitals V o l u n t a r y ( n o n p r o f i t ) hospitals ( A H A C o n t r o l Classification C o d e s " 2 1 " or " 2 3 " ) p r o v i d i n g m a t e r n i t y care ( A H A Service Classification C o d e " 4 4 " a n d A H A Stay Classification C o d e " S " ) . In a d d i t i o n , hospitals in this category lack affiliation with a medical school ( w i t h o u t A H A A p p r o v a l s C o d e " A - 5 " ) b u t are accredited by the J o i n t C o m m i s s i o n on A c c r e d i t a t i o n of H o s p i t a l s ( A H A A p p r o v a l s C o d e " A - l " ) . T h e y also offer residencies a p p r o v e d by the A m e r i c a n Medical Association ( A H A Approvals Code "A-3").

Unaccredited hospitals V o l u n t a r y ( n o n p r o f i t ) hospitals t h a t are n o t c h u r c h related or o p e r a t e d ( A H A C o n t r o l Classification C o d e " 2 3 " ) p r o v i d i n g general services ( A H A Service Classification C o d e " 1 0 " ) to s h o r t - t e r m p a t i e n t s ( A H A Stay Classification C o d e " S " ) . H o s p i t a l s in this category are n o t accredited by the J o i n t C o m m i s s i o n on A c c r e d i t a t i o n of H o s p i t a l s ( w i t h o u t A H A A p p r o v a l s C o d e " A - l " as well as w i t h o u t " A - 5 " ) , b u t they are classified by Blue C r o s s as " G r o u p 111" indicating t h a t a l t h o u g h they lack accreditation they m a i n t a i n c o n t r a c t u a l relations with M a s s a c h u s e t t s H o s p i t a l Service, Inc. These hospitals, of course, are n o t a p p r o v e d for a d v a n c e d training p r o g r a m s of any type (lack A H A A p p r o v a l s C o d e s "A-2"-"A-7").

144/APPENDIX 2.2 Proprietary hospitals Hospitals that are controlled by an individual, partnership, or c o r p o r a tion for profit ( A H A C o n t r o l Classification Codes " 3 1 , " " 3 2 , " or " 3 3 " ) providing general services ( A H A Service Classification C o d e " 1 0 " ) to short-term patients ( A H A Stay Classification Code " S " ) . In addition, hospitals in this group are accredited by the Joint Commission on Accreditation of Hospitals ( A H A Approvals Code " A - l " ) , but they lack, of course, training p r o g r a m s and affiliation with a medical school (without A H A Approvals Codes " A - 2 " - " A - 7 " ) . City hospitals Nonfederal governmental hospitals that are operated by a city ( A H A Control Classification C o d e " 1 4 " ) providing general services ( A H A Service Classification C o d e " 1 0 " ) to short-term patients ( A H A Stay Classification C o d e " S " ) . In addition, hospitals in this category lack affiliation with a medical school (without A H A Approvals C o d e " A - 5 " ) but are accredited by the Joint Commission on Accreditation of Hospitals ( A H A Approvals C o d e " A - l " ) .

APPENDIX 2.3

Inpatient Costs for the 72 Massachusetts Community Hospitals

145

•° t» ' Uοϊ Μ O l i ) . O υ υ ι- £2 o Ε ΐ SS o 1ϊρ2

«o Ν (Ν Ό CN ^ CN ^ ιη Μ ΙΛ Ν Ό

(Ν Ν Ν O ^ M co fi >-· in ττ (Ν

8 5 υ5 = e y rs. co Ο«. — CO CO ΙΟ co o © m io "« a-Κ · co G κ co «Ο Ο (Ν CO CO ^ >0 η Ο· Ν ΙΛ o χ

- Ifì Ο Ν ι i ^ ^ ^ I I

υ ι • ο» οο ο >ο -Ο Ο* fî ~ΟΟ Ο «Ο

S « Ζ «o u O O

E E o υ

ti

Λ

6? CS Ό ® ΙΟ »Ο Ν o o in «ή ^ oò ^ ·ο o m

^ -Ο ο f> >ο Ρ^ •Ο Ό ·-" Ο Ό ÍM (Ν ^ ^ ^ W OÍ CN fN — « Ο Ο

Τ ® Ν (Ν ι CN co ιλ CN η » ο •η o η — 00 Γν (Ν ^ to* co

η ο ^m s Ό ^ η Ο CN^ Ό (Ν Μ CN (Ν ·

o CO ^ ^f IO o I> O > 0> - Ο· (Ν o CN 0 CN α Ο» ΙΛ ' CM fsT co «ο η ιλ Ό η r-» Ά

« > Ο- Μ Π S Ο^^ΤΟιΟΓ— tΌ (Ν ro οο o CN ο^OCNCNOOrj> ·*τ «o ^ — · m οF-mο >η ^ Ο* ΟΟ CO CO CO (Ν



5 2 g:

co co CO O* ΓΝ. V»

Ο CO >— CO co οCNΟ CO es o— CO io o

ΌΝ O Ο CO CO io ! 0N 0 η η -Ο Νπ ι- βCN C C O

0 M

« M >. o û. ε M O C L X

O

— o 0 £ < :

O í ,

¡

;

ζ

c o

APPENDIX 4.1

An Analysis of Departmental Costs Several of the more important departments are analyzed individually in this appendix. Seven departments and the hotel service category, which together accounted for 84.8 percent of total hospital expenditure in 1959, have been selected: 1. Nursing 2. Administration and general 3. Laboratory 4. Radiology 5. Operating room 6. Medical and surgical supplies 7. Pharmacy 8. Hotel service In addition to the usual ABD and PD models for all these areas, equations in which costs are in units of departmental output have been estimated for the laboratory, radiology, operating room, and delivery room departments. The ABD and PD models are analyzed first. The ABD model tables are in this appendix; the PD equations are tabulated in Table 4.9. ABD AND PD MODELS Nursing service. Since this is the major component of the primary services, it is not surprising that the important factors in explaining the costs are the same as in the primary services.1 (The equations are tabulated in Table 4.10.) The R2 is lower than with the primary services, but at .57 it is still respectable with cross-section data. Nursing education is the most important explanatory factor. Its highly significant negative coefficient indicates that the presence of student nurses leads to reduced nursing costs, probably because they perform many tasks that regular nurses would otherwise be required to undertake. Adding student nurses appears to reduce costs per ABD by about $1.05. The other important explanatory factors are ambulatory activity, surgical activity, radiology activity, ward services, medical education and physician status, and utilization (equation [21]). The first three undoubtedly reflect output variations, while ward services and medical edu148

Ζ £ · „ ϊ »ί £

> Ό CO η

(Ν η σ « ίο fV Ν rs Ο· Ν CN 0> CO — Ο IO CN Λ (Ν Ό C O α ο> Ν Ο Ο Ο Ο Ο ο ο ο ο ο ο ο ο ο ο ο ο ο

Τ υ

η

(Ν Ο ο η t rv 00 h» —

^ ifì Κ Ό ο Ο οο «ο ο rv 0 Ό χ χ χ χ ^ «O O· C N Ο« Ο Λ "Ο Ό ^ IN ^ ο ο- η ^ ο — · ο ο ο ο

-1 Ο Ο Λ Ό Λ « 1Λ -Ο m ΙtΟ « 1Í1 Ό ΙΛο ο m ο ο> α η— t o t t »o r> — · οο ό m io «ν es m o γν > — — » — —— . — — . f> ο ο ο ο ο ο ο ο o

σ>» »

o o o o o o o o o o

o o o o o

o o o o o

— · C N C*") • 1 Ό -O Ν Ν S

•il * « « • PI η * β n u i • f «Ν es ν η η

ANALYSIS OF DEPARTMENTAL C O S T S / 1 5 3 Radiology. Radiology activity, as would be expected, is the critical determinant of the expenditures of the department. (The A B D equations are in Table 4.13, the P D equations are in Table 4.9.) The number of outpatient x-ray films per A B D alone gives an Λ 2 of .62, and outpatient and inpatient films per A B D together give an R 2 of .68 (equations [1] and [3] in Table 4.13). 3 Comparing the size of the coefficients, the cost per A B D of an additional film is about a dollar more for an outpatient than for an inpatient. The other important explanatory factors are the ratio of autopsies to deaths, ward services, medical education and physician services (whether measured by salaries or by the number of radiology physicians), utilization, and hospital size (equation [24]). There appear to be some economies of scale for the radiology department with respect to the size of the hospital, and the positive coefficient for the occupancy rate indicates that, as in the nursing department, radiology costs vary with the patient load. This is not surprising, given the frequently high cost of the film itself. 4 Operating room. The number of inpatient weighted operations per ABD is the most important variable in explaining operating room costs per A B D , since it alone gives an R 2 of .40. (The A B D equations are in Table 4.14, and the P D equations are in Table 4.9.) At the departmental level, weighted operations are clearly better than major operations as a measure of surgical activity because the R 2 drops to .33 with major operations (equation [2] in Table 4.14). This may reflect differences in the accuracy of the measures, because the data for weighted operations are audited, while those for major operations are not. The presence of student nurses reduces operating room costs per P D , but salaried physicians and inpatient x-ray films per A B D raise them. A higher proportion of tonsillectomies and adenoidectomies, as would be expected, reduces operating costs per A B D . Medical and surgical supplies. Hospital size and surgical activity are the most important explanatory variables (the A B D equations are in Table 4.15 and the P D equations are in Table 4.9), since these two factors together give an R 2 of .43 (equation [1] in Table 4.15). The strong positive coefficient for hospital size indicates that there are rather marked diseconomies of scale in providing medical and surgical supplies. This may reflect the greater complexity of care provided in the larger hospitals or their tendency to provide patients with supplies rather than relying on patients procuring their own through proprietary pharmacies. The other important explanatory factors are nursing education, medical education and physician services, utilization, and ambulatory activity. The coefficients indicate that student nurses raise the cost of providing medical and surgical supplies, while salaried physicians lower it. A n d whether this is indicative of relative efficiencies is debatable! The strong positive co-

Ό Ό ΙΟ O O o

»O η -Ο O O o

00 fr Ό O O o

HI O· o> Ul IO IO O O O o o o

»Γ X M r t co M fN ΙΛ η « Ο Ο l i

X O co CO V M o l i

>T fr cn Tf S η o

? ? O» 00 co «o - o «— S S en Γo σ I I

io o oo ΙΛ (S

ο· o « η fN

co co ^t ιλ t E ·- o " υ >. .

ε ε ο

m e o io — r-. o co r«. «OCNOO·— Ν π m m {Ν ·—• (Ν fN

3 ·1 S
- η ^ ο> ^ ο· •·· CO ·*5· ·— CO CN Ν Ό O V

•Ο Ό Χ Χ Τ ΤΓ m ο co — •Ό CO ·— O

*

_

Ό Χ ^ — η 00 —

CN CSI

X X X X ν ό τ co ^ β η

CN Ο CN ^

Ό Ό Χ Χ Ο· (Ν CN ο ο co ^ σ· r CO — CO —

ε Ο» 00 O O O

S i ' s

»O O — O O

O O O O O O O O O O

o

O O O O O O O O O O

O

— — CN CO CO

Ν

I *

ΙΛ 1/1 Ό Ν

«

«*

*

I Ο Ά Μ ·· t

TT -o ec η Ν τ irt d o o o

— co co ο — - Ν Ν eo o s α π Ό β Ν ri 9· m ν (Ν in Ρ; Κ Ν S β Ο; ο ò e ο ο ο ïo

Ό Ό -C Ό Ό Ό Ό χ χ χ κ κ χ κ m ο » » « η ο U1 Ν Ν Ο Ο (> Ο « « ν rv κ > ο < ο ο ο ο ο ο ο ο

S J

Ψ 5.

2

. . I Ό Ο C O U * > Ό Tfvouimiflfn^ ο ο ο ο ο ο ο ο

£·

-ï E -D E

. _ E

ο ο ο ο ο ο ο ο ο ο

OOOOOOOOOC

— — < Ν f») '

« s ξ

«« « «

*« *

*

-

O

4)

Ν O m CH

Ό CO m η 0 π ι •—•

1 -Ο Ο cn ο> Ο o ^ O* h» Ό I O O O - C O N I N ^ e o n - c n N - r - I N Í N C N - f N

c o o> o — — 0 · ΐ Λ ΐ Λ m •— O f O O - - - C O r i ( N N N ' V I O W N ' O l O i n o u ï CO C N C N C N C N C N C N C N C N

m

m

? "Ο "Ό « C O



Oy Ο· (Ν (Ν Ν » ΙΟ ^ o - o œ o o o r s i * v >— ΙΛΙΓΝ"— d

o* ΓΝ. η CN r i

o

d

o

Γ ν Ό ^ - Ό o

o

o

o

f-V oo 00 CN CN i o o» T í CO CN en 00 r o CN CN CN en CN CN CN CN CN

«O CN CO τ m TT m CN CO CN c»> o» 00 00 oo 00 00 00 β o o o o o o o o o O s

Ια β

• o o oo Γ-» 00 00 o • n co co oo en

ί Ο

-

>— CN Ό * Ζ

CN

°

i ^

-O -O >0 Ό

I"···. I

ANALYSIS OF DEPARTMENTAL C O S T S / 1 6 1 DEPARTMENTAL OUTPUT MODELS The costs of the laboratory, radiology, operating r o o m , and delivery room departments have also been analyzed in output units of the respective departments—that is, tests, weighted films, weighted operations, and weighted deliveries. There are two major reasons for doing this. The first is to see whether there are any economies of scale with respect to departmental size, while the second is to see whether differences among hospitals in costs in units of departmental output can be more readily explained than differences in costs per A B D or per P D . Table 4.19 gives the mean departmental expenses for those hospitals reporting departmental output data. Needless to say, there are some inherent difficulties (at least in the data available to this study) in trying to filter out economies of scale with respect to the size of the department. The major difficulty is that department size and hospital size tend to be very collinear. This can be taken care of in principle by including both hospital and department sizes as independent variables, but the high collinearity between the two variables makes this an ambiguous statistical exercise. Ideally, what is needed are data on hospitals of the same size but with departments of different sizes. Another difficulty, particularly with the laboratory, is that (as was noted in Chapter 2) the data on departmental output are less reliable than either those on expenses or on bed or patient days. For example, there is a great deal of variability among laboratories as to what constitutes a test. A blood test in which both red and white corpuscle counts are taken may count as one test in one laboratory but as two in another. This would not be a major problem if such variability were random with respect to laboratory size. This is unlikely to be the case, however, with the result that the findings will contain an undetermined bias. Moreover, the larger hospitals tend to report laboratory activity in terms of units rather than in terms of tests, which creates missing observations for these hospitals. Still another problem arises because the size of administrative units may be unrelated to the volume of services. It may be possible for some small hospitals to take advantage of economies of scale by using the laboratory of a large hospital for tests, while some large hospitals may operate their laboratories not as a single department but as small a u t o n o m o u s parts. If these phenomena occur, any economies of scale in the data may be obscured because of inappropriate measures of true scale of departmental operations. A final difficulty involves the lack of data regarding the number of tests, particularly similar ones, performed on each patient. Multiple testing may have different cost implications. Some procedures may be both applied and repeated as a matter of routine in large hospitals, but not in

1 i

*

^ >- Ό r-v 0>

η Ό O» Ν » o o (Ν (N

' ^ ^ ; 00 oo ' «Λ •—

o co oo o> o· r·» r-» r». CN

•O Γ-. ·— CO W> o CN œ

co ^ o·· CO O- W) i n Ν ΓΗ

o s

S

C o-

O TT TT Ν η η «0 Ό Ό

s · ^ e « t «s »i Û. - o

o CL «

í Q -c

«

«

Ο Ο Ο — co co

σ o c¡C (I £ ε ^

-O "O E

φC ε

Xc I

1

« = = W e Ο

ε ο « 3 — I I Ϊ α c «ι Μ Η ε ο — — σι ϊ υ ° · Î ϊ » Ο φ Φ ι» Ϊ V

s
= .95.

222/NOTES TO PAGES 111-114 8. Muller notes in the Trussell report (1960) that hospitals with the greater expenditures per patient day were usually the larger institutions, located in densely populated areas and providing a greater range of patient services, and numerous teaching programs. Also see Letourneau et al. (1961). 9. This observation comes f r o m Ralph Berry. 10. T h e importance of outpatient d e p a r t m e n t s has also been observed, as has our result with respect to number of house staff per bed, by Blumberg (1964) in studying 30 hospitals for the Jewish Welfare Federation of San Francisco. Also see Letourneau, Ulveling and Heuser (1961), who also report correlations between costs and outpatient visits, and between costs and number of residencies and internships. 11. Maryland Commission to Study Hospital Costs (1964), p. 87. Lack of consistent staffing patterns, even in hospitals of comparable size, may in part explain these results. For an interesting discussion of such staffing patterns, or rather of their absence, in social service departments, see the American Hospital Association and the National Association of Social Workers (1957). Heavier case loads were observed in smaller departments, especially those with fewer than three or m o r e nongraduate case workers. Part of the diversity in staffing patterns may reflect differences in the proportion of time spent by nurses and other professionals in information-handling or in other secretarial or nonprofessional duties. F o r a discussion of this problem, see Howe (1963), Weeks and Griffith (1964), Gross (1964), Ingbar, Whitney, and Taylor (1966), and Jydstrup and Gross (1966). Also, Lembcke (1959) presents evidence concerning staffing patterns in Swedish hospitals which suggests that their employee efficiency is markedly higher than in the United States, perhaps because their physicians have incentives to play a much m o r e active managerial role unlike physicians in the United States who, according to Gottlieb (1958), may know little about administrative affairs and financial management in the hospitals with which they are affiliated. Querido (1963) reaches the " s h a t t e r i n g " conclusion in his study of Dutch hospitals that improved organization alone could save one fifth of the hospital beds in the community (p. 133) and that one half of these defects in organization involved deficiencies of procedures within the hospital as against insufficient capacity of hospital facilities or external organizational weaknesses. 12. Also see dissertation by Ro (1966) and article by Cohen (1967), which were received too late to be included in the above discussion. 13. See Weiss (1966). 14. For a description of these p r o g r a m s in Massachusetts hospitals, see Massachusetts Hospital Association, Minority Report (1964), and statements of the association before the Massachusetts Special Commission to Investigate and Study Laws Relative to Non-Profit Hospital and Medical Service C o r p o r a t i o n s (1963 and 1964), which a r e listed in the bibliography (especially, those by Quigley, F e b r u r a r y 7, 1963). Also, see description by Perry (1966) of t h e operation of the group purchasing plan which spends in excess of $5 million annually on behalf of its 60 m e m b e r s at an average saving of 12 percent. For a description of these p r o g r a m s in other areas, see National Commission on Community Health Services (1966) and Illinois Hospital Association (December 2, 1966).

NOTES TO PAGES 114-115/223 15. For a discussion of these problems and a m o r e complete list of references, see: Lembcke (1952), Veterans Administration (1959), Gogan (1961), Anderson and Altman (1962), Donabedian and Attwood (1963), Harrison (1963), California Department of Public Health, Medical C a r e Studies Unit (1965), Ellwood (1966), Donabedian in Mainland (1966), Cottrell (1966), and Segal (1966). Also, see: Dominion Bureau of Statistics, Hospital Indicators', working papers of the Medical Review Unit of the Empire State Medical, Scientific and Educatonal Foundation (1965); and Watson in U.S. Public Health Service (Abstracts of Contributed Papers, 1965). 16. Commission on Professional and Hospital Activities (1961, 1962). 17. For example, see discussion in Annals of Internal Medicine (October 1959). 18. See Garland (1959). 19. Coggeshall (1965), pp. 24-25. 20. New England Journal of Medicine (January 24, 1963), p. 214, for example. 21. Lubin and his colleagues (1966), for example, state that larger distinctive patient facilities could result in higher occupancies and savings to the community in capital outlays and operational costs; they would also enable the physician to reduce the number of hospitals with which he was associated, thereby permitting him to reduce his travel time. Also, see Drosness, Reed, and Lubin (1965) on the application of computer graphics to patient origin studies, and Lindheim (1967). 22. See the reports by the California Department of Public Health (1960) and the Hospital Utilization Research Project (1964 and Summary Report, 1966). Also see bibliographical references in Detloff et al. (1964), and Stageman and Baney (1962). 23. Modern Hospital (December 1966). 24. See Illinois Hospital Association (December 1958). 25. See Rorem and Roth (1966). 26. See Owen (1966) for a guide to the evaluation of such facilities. 27. Also see Goldman et al. (1963) on combining intensive care and recovery units. 28. Data from the Anderson-Sheatsley-Health Information study of patterns of hospital use in Massachusetts (1963, 1965) indicate that approximately 45 percent of the admissions to Massachusetts hospitals required round-the-clock standby facilities and personnel for handling emergencies. Exclusive of the maternity cases, about 32 percent of the admissions involved same-day emergency situations. Five diagnostic categories, moreover, accounted for 44 percent of these same-day admissions: appendicitis, fractures and dislocations, other injuries, pneumonia, and arteriosclerotic heart disease including coronary disease. Accidents alone accounted for 11 percent of the admissions (Ν = 2046) that were surveyed in this 1960-61 study of discharges f r o m a representative sample of 50 general and special short-stay hospitals in Massachusetts. In terms of the time of admission, 75 percent of the cases studied were admitted between 9:00 A.M. in the morning and 6:00 P.M. in the evening. T h r e e percent were admitted f r o m midnight to 6:00 A.M., and another 5 percent f r o m 9:00 P.M. to midnight. 29. Thompson and his colleagues at Yale have documented these issues both for the obstetrical service (Thompson, 1966, and King, Holloway and Yeomans, 1966), where demand appears to follow a Poisson distribution, and for intensive

2 2 4 / N O T E S TO PAGE

115

care units (Judd, 1963). They have also used simulation techniques to study waiting times in outpatient departments (Fetter and Thompson, 1966)—a time that becomes increasingly important for both doctor and patient as nonemergency use of these departments increases. For further discussion of outpatient services, see Solon, Sheps and Lee (1960), Solon (1966), Weinerman et al. (1965, 1966), and New York Academy of Medicine (January 1965). London and Sigmond (May 1961) in a study of the efficiency of small specialized units reported that, in the absence of buffer or swing beds, one third of the beds were empty on the average day. In another study, Blumberg (1961), using the Poisson distribution to predict the likelihood of a patient being turned away, developed the notion that pressure on an empty bed increases as the hospital becomes smaller. These notions were also being developed by Flagle (1962, see bibliography) and the Operations Research Division at the Johns Hopkins Hospital. Included in their work were studies of intensive care units (Flagle, 1960), outpatient clinics (Gabrielson et al., 1959), and centralized staffing for nursing units in accordance with the daily fluctuations in requirements of patients on each ward (Connor et ai, 1961). As previously mentioned, for an interesting discussion of the influence of volume of service upon the costs of obstetrical facilities and maternity care, see the Guidelines and Recommendations of the Hospital Review and Planning Council of Southern New York (January 1966). Interesting work is also being undertaken in England. For example, see studies of outpatient appointment systems and casualty settings by Bailey (1960) and Nuffield Provincial Hospitals Trust (1960). Additional studies are reported by Day (1961), Davies et al. (1962), Nuffield Provincial Hospitals Trust (1962), Brotherston (1963), Btesh (1965), Blue Cross Association (1966), Wirick (1966) and in the annual inventory of research in progress prepared by the Health Information Foundation. For continuing reports of new work, also see: American Hospital Association, "Annual Administrative Reviews"; British Ministry of Health, Hospital Abstracts', Johns Hopkins University, Research in Public Health Administration; University of Michigan, Abstracts of Hospital Management Studies and Medical Care Review (previously entitled Public Health Economics and Medical Care Abstracts). In addition, work in progress is reported in Abstracts oj Contributed Papers to Medical Care Section of the American Public Health Association: in recent years such abstracts have been published by the U.S. Public Health Service. In addition to those appearing in the regular journals of the hospital field, brief summaries can be found in publications of the numerous hospitalrelated organizations, including, for example, the Bureau of Research and Planning of the California Medical Association (see especially their monthly Socio-Economic Report and their annual Reference Book). 30. For other evidence of such economies, see " H o w to Plan the Laboratory," Modern Hospital (1961), Foster (1966), and United States Congress, Detection and Prevention of Chronic Disease Utilizing Multiphasic Health Screening Techniques (1966) with special reference to the statement by Ralph E. Thiers in these hearings. 31. For example, see Roemer, "Is Surgery Safer in Larger Hospitals?" (1959). 32. See Collen (1963).

NOTES TO PAGES 115-118/225 33. T h e Pennsylvania Economy League (1962) found that costs per visit declined with increases in outpatient care and hospital size. Similarly, Sachs (1956) reported that orderly and x-ray technician requirements remain unchanged in an outpatient clinic session whether one or three doctors were present. Nurse staffing, however, had to be adjusted as the number of doctors changed. T h e Public Health Service noted declines in the cost of equipment per bed as hospital size increased from 50 to 200 beds (Publication No. 822, 1961). For other cost advantages from central coordination, see Toomey (1965). 34. See, for example, London and Sigmond (January 1961). 35. Citizens Hospital Study C o m m i t t e e , Part I (1961), p. 14, and Part V (1960), p. 12. 36. See Long (1964); Letourneau and Ulveling (1959); U.S. Public Health Service, Representative Construction Costs (1962); U.S. Department of Labor, Requirements for Hospital Construction (1962); Souder (1963, 1964); and Polner (1963). 37. See Ryan (1958). 38. See Seale (1960), and Deeble (1965). 39. If this occurs, accurate estimates of the savings involved in eliminating "unnecessary" utilization cannot be obtained by multiplying the prevailing average per diem rate by the number of " s a v e d " days as was initially reported by the University of Michigan in early press releases (May 14, 1961) concerning the M c N e r n e y study (1962). For example, in the Wall Street Journal of May 15, 1961, it was alleged that the state of Michigan could annually save $7,500,000 by eliminating "unnecessary" utilization. This estimate was derived by projecting to the population of the state, the findings of the study that 130,000 days of hospital care for 18 illnesses could have been saved had 7 percent of the surveyed patients not left too soon and had another 10 percent not stayed too long. Similarly, Myers (1961 and 1965) in developing the actuarial cost estimates for various federal legislative proposals estimated hospitalization-benefit costs by multiplying the estimated number of eligible beneficiaries by a factor representing the average annual per capita cost of hospitalization (after taking into account any maximum-duration and deductible provisions). T h e average daily cost of hospitalization is thus again assumed to be independent of the average length in days of compensable hospitalization. Thus, changes in either patient days or costs per day a r e immediately translated into changes in the average annual per capita cost of hospitalization used to calculate the expected cost of new legislative proposals. Data f r o m this study indicate that the assumption of such a one-to-one relationship is erroneous. 40. See Brines (1963). 41. T h a t this may be feasible has been demonstrated by the Detroit Memorial Hospital, Hospitals (November 1, 1960), p. 17. 42. Finance in Developing Hospital Service. 43. Inadequacies in costs per patient day are discussed by Hottum (1960) and by Saathoff and Kurtz (October 1962). 44. Various systems for measuring units of activity are well established, for example, see Dominion Bureau of Statistics, Schedule of Unit Values (1960);

226/NOTES TO PAGES 118-120 Massachusetts Hospital Association, Appendix to Guide for Cost Analysis (1961); Gross (1963); California Department of Public Health, Development of a Hospital Service Index (1965); and the scale of the New England Pathological Society available from, among others, the Massachusetts Bureau of Hospital Costs and Finances. Scales for the laboratory a r e also included as part of more general relative value studies used in the determination of physician fees. See, for example, California Medical Association (1962) and its I960 Relative Value Studies (September 10, 1960); and " G u i d e to F e e s " in New Medical Materia (February and May 1962). Also see American Medical Association, Workship on Relative Value Studies (1961). For an interesting discussion of differences in the relative value scales, see Brewster and Seldowitz (1965). 45. For participating hospitals and limited types of information, the Commission on Professional and Hospital Activities and Hospital Administrative Services both act as processing agencies. For a discussion of the Comparative Reports prepared by the latter, see Hospital Research and Educational Trust (1962) and more recent publications. Approximately 2000 hospitals now take part in the Hospital Administrative Services program which was taken over by the American Hospital Association in May 1962. Hospital Management also includes as a regular monthly feature, " H o w ' s Business?" a summary of statistical data, including average patient charges and operating expenses, both per occupied bed per month and per bed per month. C h a r g e s are graphed against expenses and occupancy rates are presented, but— in contrast to Hospital Administrative Services—none of this information relates to individual institutions. 46. See, for example, bibliographical references in Ingbar, MEROPS (1966), Best (1963), Hospital M a n a g e m e n t Systems Society (1964), Harrington (1964), and Rikli, Allen and Alexander (1966). For a discussion of some of the potential uses of the computer, also see the Proceedings of the San Jose Conference on Health Information Retrieval published by the University of Southern California School of Medicine (1959) and the Advanced Systems Development Division of the International Business Machines Corporation. 47. See Michigan Hospital Service (1961), Clark (1961), Titmuss (1963). 48. See Rorem (1954) and discussion of leasing by Holmgren (1963). 49. For examples of peculiarities in charge patterns for drugs, such as 6 percent of studied hospitals charging 10 cents or m o r e for a single aspirin tablet, see Myers (1961). 50. For a discussion of some aspects of the physician's role, see G r o u p Health Association of America, News (January 1965). 51. See, for example, Odoroff, Baney and Stageman (1960) who reported that the costs of operating nursing homes ranged f r o m $3.38 to $13.85. Also see: American Hospital Association, Guide to Effective Regional Planning (1962) and Guiding Principles for Agreements between General Hospitals and Long-Term Care Facilities (1963); Stageman, Baney, and Brooks (1963); Joint C o m m i t t e e of the American Hospital Association and the Public Health Service (1963); U.S. Public Health Service, Nursing Homes and Related Facilities Fact Book (1963); California Association of Nursing Homes (1964); and Vanston, Abbe, and H a m p t o n

NOTES TO PAGES 120-148/227 (1965). Also see discussion on home care in Ingbar and Lee (1964) and on a chronic disease unit in a general hospital by Littauer et al. (1959). For information on the possibility of using overnight facilities instead of hospitalization for ambulatory patients, see White et al. (1963). For specific data concerning nursing homes in Massachusetts, see Kelleher and Shaughnessy (1964). 52. For a discussion of this point, see Hyde et al. (1954). 53. T h e Anderson-Sheatsley-Health Information Foundation study of hospital use in Massachusetts (1961, 1963, 1965) concludes that there is a gray area involving perhaps 10 to 20 percent of the admissions surveyed in which, depending on the circumstances, a patient might have been treated either in or out of the hospital. At the time of their first visit to a physician, 20 to 30 percent of the patients reported that their doctors felt no urgency concerning the hospitalization —delays of from weeks to months often being acceptable. About one half of the patients, f u r t h e r m o r e , reported that they had delayed seeing a doctor after the recognition of symptoms. 54. See Acton Society Trust (Mirfin, 1962) and Rosenthal (1966). 55. For a detailed study of charge practices and their impact on patients, see Bost (1955). 56. Also see the editorials in The New England Journal (September 24, 1964, pp. 684-5 and February 4, 1965, pp. 261-2), which discuss proposals concerning price level as against historical depreciation, use of per case rather than per diem payments, and reimbursement status of " f r e e - c a r e , " hospitalization for employees, and bad debts. Also see: Rorem (1957); Nelson (1959); Romanoff (1959); Trussell and van Dyke (1960); Hospitals (November 1960); Modern Hospital (July 1961), p. 87; American Life Convention (September 1961); Hahn (1962); Minnesota Legislative Research C o m m i t t e e (1962); National Commission on C o m m u n i t y Health Services (1966), p. 59; as well as the numerous policy statements issued by the American Hospital Association (especially 1960, 1963) and the statements concerning utilization and charges issued by the Blue Cross Association in Blue Cross Reports. 57. See also Canadian Hospital Association, Canadian Manual (\959). 58. M. Feldstein, Medical Care (1963).

Hospital

Accounting

A p p e n d i x 4.1 / A n A n a l y s i s of D e p a r t m e n t a l C o s t s 1. T h e r e is a wide literature on the problems and costs involved in providing efficient nursing service. One of the most comprehensive studies on the effective use of nursing personnel is by Conner et al. (1961) and Conner (1961). Pelletier and Thompson (1960) discuss design efficiency, and Coggins (1965) the effect of selfcare on staffing requirements. Jensen (1961) looks at the staffing problems of the small hospital, while Levine, Siegel and De La Puente (1961) and Levine (1962) analyze the diversity of nurse staffing among general hospitals. Hale (1964) discusses the reasons for the shortage of nurses, as does the G r e a t e r Boston Hospital Council (1961). T h e general problem of personnel systems and methods is discussed

228/NOTES TO PAGES 148-161 by Sister Bernard (1960), while the specific problem of salary scales is analyzed by Wallen (1960) for hospitals in the Boston area. T h e Bureau of Employment Security (1952) of the U.S. Department of Labor has developed j o b descriptions for nurses as well as for other hospital personnel. T h e Bureau of Labor Statistics and W o m e n ' s Bureau (1961) undertook extensive surveys of earnings and supplementary benefits in hospitals in 1956-57, mid-1960, and mid-1963, updated only by the quarterly Occupational Wage Surveys that are undertaken for selected cities. Additional information on salaries of nurses and other hospital personnel is compiled by the Illinois Hospital Association (1962-65, annually), but other information of this type and detail is scarce, last being published for New York by Reed in 1959. Also see the American Hospital Association (1959) and Note 34 of Chap. 1. T h e National League for Nursing (1964) undertook an extensive study of the cost and quality of nursing education, as did Pfefferkorn and Rovetta (1940) in earlier years. Wolfe (1962) investigated the optimal length of training for a female nursing aide. Newhouse (1963) undertook special studies with the d a t a for the Massachusetts community hospitals which confirmed the absence of economies of scale in nursing education. O t h e r nursing studies a r e listed in the bibliography to Ingbar, Whitney, and Taylor (1966). See especially Abdellah and Levine (1954, 1958) and Aydelotte and Tener (1960). 2. Also, there is considerable variation in administrative salaries, see C a r n e r (1966). For discussions of the social and political forces that may influence hospital costs, see Viguers (1961), Seeman, Evans, and Rogers (1960) and Seeman and Evans (1961). For discussions of the record-keeping aspect of administrative responsibilities, see McGibony (1963), who also discusses medical and surgical supply services (1965). 3. These high R2 a r e particularly impressive in view of the fact that no account is taken of the difference among hospitals in their use of contractual physicians. For interesting discussions of the financial relationships between radiologists and hospitals, see Snoke (1960) and American College of Radiology, Guide in Relationships with Institutions (1961). An alternate weighted system for measuring the volume of service is also discussed by the American College of Radiology, see their Radiology Relative Values (1963). 4. For details on film and labor costs by procedure in Greenville, South Carolina, see Ernst and Ernst (1961). 5. For discussions of the problems involved in operating hospital pharmacies, see McCluskey (1959), Blumberg (1962), and Muller and Westheimer (1966). With respect to medication orders, see Barker and McConnell (1962), Drew and Blumberg (1962), and Baruch (1963). 6. For an interesting discussion on methods of appraising housekeeping departments, see the description of the reporting form devised by the Veterans' Administration in Modern Hospital 101:140-143 (August 1963). For a discussion of the personnel turnover problem in hotel departments, see Hyden and G r e e r (1961). 7. For a discussion of how group purchasing by hospitals in Massachusetts could significantly reduce the expenses of the hotel services, see Perry (1966).

NOTES TO PAGES

161-163/229

Holmgren (1963) discusses how leasing can be used to advantage in the hotel services (as well as in other departments). 8. Ideally, of course, size should be measured by the capacity of each department, while the occupancy rate should also be departmentally defined. Except for the occupancy and beds of the maternity service, however, there are obvious problems of definition. Consequently, departmental output and the over-all occupancy rate were used instead.

INDEX "unnecessary" beds, 116-117. See also Economies of scale; Size-volume factor Berry, Ralph E„ Jr., I l l , 113 Blue Cross: financing, 2-5 passim, 10, 12; Massachusetts, 10, 11, 12, 44, 78; reimbursement policies, 117, 119-122. See also Charges Boston City Hospital, 16 Bureau of Hospital Costs and Finances (BHC), see Massachusetts Bureau of Hospital Costs and Finances

Adjusted general ledger expenses, see Direct departmental expenses Administration and general service: expenditures, 17-25 passim, 41-45 passim, 145-147 passim·, regression equat i o n s for 1958-59, 70, 150, 151; residuals for other hospital groups, 88, 90; predictions of 1962-63 from 195859 equations, 101 Administrative implications of findings, 117-119 Admissions, see Discharges Ambulance, see Motor service and ambulance Ambulatory: expenditures, 17-25 passim, 41-45 passim; services (visits, films! tests), 38, 39, 80, 107, 133; charges, 76; policy implications, 115 Ambulatory activity factor, 38, 105, 107, 126; relation to hospital service expense, 53, 54; relation to primary service expense, 62; relation to secondary service expense, 63; relation to independent service expense, 65; relation to hospital charges, 74; relation to departmental expenses, 148-165 passim. See also Ambulatory, services

Capacity, methods of measurement, 3132, 34, 125, 133. See also Size-volume factor Carr, W . J o h n , 109, 110, 111, 113 Charges: room and board, 1, 2, .11, 22, 78, 119-122; regression equations for 1958-59, 74-76, 94, 108, 109; predictions of 1962-63 from 1958-59 equations, 101, 102, 103. See also Ancillary per diem Church-sponsored hospitals, 13-16 passim, 109, 142-143; residuals for 195859, 85-91 passim·, predictions for 196263, 103 Citizens Hospital Study Committee of Northeast Ohio, 4, 116 City hospitals, 13-16 passim, 144; residuals for 1958-59, 85-91 passim·, predictions for 1962-63, 103 Coefficient of multiple determination (R or R), 31, 47, 48, 53, 127; interpretation of equations with alternative deflators, 67-73; and missing observations, 82; and Theil U, 101, 102. See also Regression analysis Coggeshall, Lowell T., 114 Community hospitals, 13-17 passim, 109, 142; expenditures, 13-15 passim, 1625; residuals for 1958-59, 84-90 passim, 91-94; predictions for 1962-63, 100-103

American Hospital Association (AHA), 11-16 passim, 110 American Medical Association, 4, 11, 120 Anaesthesiology: expenditures, 17-25 passim, 41-45 passim, 145-147 passim·, services (weighted o p e r a t i o n s and weighted deliveries), 37, 139 Ancillary per diem, 11, 22, 44. See also Charges Anderson, Odin W „ 2, 6, 12 Available bed days, see Beds; Sizevolume factor Beds, 125, 128, 133; in Massachusetts hospitals, 13, 14, 15, 17, 34-39 passim, 125, 128, 133; as measure of capacity and deflator, 32, 47; relation to departmental expenses, 42, 43, 105, 107; optimum size, 56-58, 98-100, 103-104, 109-117; relation to independent service expense, 67; issue of

Comparative coefficient of multiple determination (R* 2 ), 67 Consumer Price Index, 1 Correlation analysis, see Coefficient of multiple determination; Regression analysis

231

232/INDEX Cost functions: selection of, 30-32, 49; comparison of alternate models, 6773; transformation of fixed into variable costs, 80, 114-118; homogeneity, 84-91; use in predicting costs of individual hospitals, 91-94; change between 1958-59 and 1962-63, 95-104 Cyclical factor, 39, 45-46, 48-51, 95-98, 100-104, 106, 109, 127; relation to hospital service expense, 55, 98; relation to hospital charges, 76 Deflators, 125; comparison of alternate models, 67-73. See also Beds; Capacity; Discharges; Patient days; Sizevolume factor Delivery r o o m : e x p e n d i t u r e s , 17-25 passim, 41-45 passim, 145-147 passim; services ( w e i g h t e d deliveries a n d weighted circumcisions), 35, 37, 38, 80, 126, 128, 133, 139; regression equations for 1958-59, 67, 161-164 passim; charges, 76; policy implications, 115 Departmental costs: definition in H C F 300 Report, 10, 11, 22-25, 134-141; expenditures, 16-22, 145-147; interrelationships, 41-45; regression equations, 67-73, 148-164; advantages of disaggregation, 83, 84, 94; residuals for other hospital groups, 84-91; predictions of 1962-63 from 1958-59 equations, 100-103 Dependent variables: definition, 27-28, 31, 46; aggregation, 41-45, 83, 84, 94 Dietary: expenditures, 17-25 passim, 4145 passim, 145-147 passim; regression equations for 1958-59, 71, 72; residuals for other hospital groups, 89, 90. See also Routine services Direct departmental expenses, 22-25, 134-135. See also Departmental costs Discharges, 34-39 passim, 125, 126, 128, 133; in Massachusetts hospitals, 13-15 passim; in community hospitals, 34-39 passim·, relation to departmental expenses, 42, 43; as deflator in regression equations, 67-73 Economies of scale with respect to partmental size, 8, 30-32 passim, 67, 108, 115, 161-164 Economies of scale with respect to pital size, 8, 30-32 passim, 47,

de47, hos107,

109-117; relation to hospital service expense, 56-59, 98-100, 103-104; relation to independent service expense, 67; relation to departmental expenses, 148-164 passim Electronic data-processing: potential applications, 7, 118-119; methods employed, 8 Error term, see Cost functions; Standard error of estimate Explanatory factors, 27-41, 46-47, 125127; use in predicting costs of individual hospitals, 91-94 F-distribution (F-test), 28, 127 Fabisak, Theodore W., see Massachusetts Bureau of Hospital Costs and Finances Factor analysis: method, 29-30; of independent variables, 32-41; of departmental costs, 41-45 Feldstein, Martin S., 112, 113 Feldstein, Paul J., 109, 110, 111, 113 Fitzpatrick, Thomas B., I l l , 112 Fixed costs, 117-118. See also Cost functions General ledger expenses, see Departmental costs General services, see Routine services Gottlieb, Symond R., I l l , 112 Harman, Harry H „ 29, 33, 40 H C F 300 Report, see Massachusetts Bureau of Hospital Costs and Finances Health Information Foundation, see Odin W. Anderson Heneman, Harlow J., 118 Hospital costs: history, 1-7; statewide investigations, 4, 5; questions, 8, 9, 29, 30, 41, 48, 74, 83, 95-97, 100-101 Hospital groups, 109; as defined in Massachusetts, 13-17, 142-144; residuals for 1958-59, 84-91, 94; predictions for 1962-63, 103-104. See also Church-sponsored hospitals; City hospitals; Community hospitals; Maternity hospitals; Other teaching hospitals; P r o p r i e t a r y hospitals; Teaching hospitals; Unaccredited hospitals

INDEX/233 Hospital personnel, influence on hospital costs, 5-7 passim, 13, 14, 39, 90, 109, 112, 113 Hospital services: expenditures, 17-25 passim, 145-147 passim; regression equations for 1958-59, 48-60, 68-69, 83-84; residuals for other hospital groups, 84-87 passim, 90; residuals for individual community hospitals, 91-94; regression equations for 196263, 95-100; predictions of 1962-63 from 1958-59 equations, 101, 102, 103-104. See also Independent services; Primary services; Routine services; Secondary services; Special services Hospital Statement for Reimbursement ( H C F 300 Report), see Massachusetts Bureau of Hospital Costs and Finances Hotel services: regression equations for 1958-59, 72, 108, 157, 160; residuals for other hospital groups, 89; predictions of 1962-63 from 1958-59 equations, 101-103 passim. See also Routine services Housekeeping: expenditures, 17-25 passim, 41-45 passim, 145-147 passim·, regression equations for 1958-59, 71, 72; residuals for other hospital groups, 89, 90 Incentives, 119-122 Income, see Revenues Independent services, 108; definition, 16-17 passim, 41-45 passim·, expenditures, 17-25 passim, 39, 43, 45, 145-147 passim; regression equations for 1958-59, 65-67, 68, 69, 83, 84, 94; residuals for other hospital groups, 84-87 passim, 90; predictions of 1962-63 from 1958-59 equations, 101, 102 Independent variables, see Explanatory factors I n p a t i e n t costs, 2 2 - 2 5 passim, 138, 145-147 Kurtz, Richard Α., 107, 111 Laboratory: expenditures, 17-25 passim, 41-44 passim, 145-147 passim; serv-

ices (tests, units, films), 34-36 passim, 38, 46, 80, 97, 106, 107, 114, 140; missing observation problem, 46, 80, 82, 83; regression equations for 1958-59, 67, 68, 70, 108, 150, 152, 161-164 passim', units of service, 80, 82; residuals for other hospital groups, 88, 90; predictions of 1962-63 from 1958-59 equations, 101-103 passim; policy implications, 115 Laboratory activity factors, 36, 106, 109, 126, 127; missing observation problem, 46, 80, 82, 83; relation to hospital service expense, 55, 97, 103; relation to primary service expense, 62; relation to hospital charges, 76; relation to routine and special service expense, 80; relation to departmental expenses, 148-164 passim. See also Laboratory, services Laundry and linen: expenditures, 17-25 passim, 41-45 passim, 145-147 passim; regression equations for 1958-59, 68, 71, 72, 73; residuals for other hospital groups, 89, 90 Length of stay factor, 36, 40, 41, 126, 133; relation to hospital service expense, 53; relation to expense per discharge, 68 Linen, see Laundry and linen Ludlam, George P., 2 Magid, Dennis J., 112, 113 Maintenance of personnel, expenditures, 17-25 passim, 41-45 passim, 145-147 passim Maryland, State of, Commission to Study Hospital Costs, 113 Massachusetts Blue Cross, Inc., see Blue Cross Massachusetts Bureau of Hospital Costs and Finances: history, 10-12; Hospital Statement for Reimbursement ( H C F 300 Report), 10-12 passim, 17, 22-25, 41, 44, 129-141; procedures, 32, 78, 118, 128 Massachusetts Bureau of Hospital Facilities, 11 Massachusetts Commission on Administration and Finance, see Massachusetts Bureau of Hospital Costs and Finances

234/INDEX Massachusetts Department of Public Health, see Massachusetts Bureau of Hospital Facilities Massachusetts General Hospital, 15, 16 Massachusetts Hospital Association (MHA), 11 Massachusetts Hospital Cost Study (MHCS), definition of hospital groups, 12-17,142-144 Massachusetts Hospital Service, Inc., see Blue Cross Maternity activity factor, 37, 38, 40, 126, 128; missing category problem, 46, 47; relation to hospital service expense, 55; relation to hospital charges, 76; relation to hospital revenues, 78, 80; relation to departmental expenses, 148-164 passim. See also Delivery room, services Maternity hospitals, 13-16 passim, 143; residuals for 1958-59, 85-91 passim; predictions for 1962-63, 103 Medical and surgical physicians, 105, 106; expenditures, 17-25 passim, 41-45 passim, 145-147 passim·, services 35, 40, 107, 128; missing category problem, 80, 82 Medical and surgical supplies: expenditures, 17-25 passim, 41-45 passim, 145-147 passim·, regression equations for 1958-59, 71, 72, 153, 156, 157; residuals for other hospital groups, 89, 90; predictions of 1962-63 from 1958-59 equations, 101, 102. See also Medical supplies Medical education and physician services factor, 35, 105, 106, 125, 128; relation to hospital service expense, 53, 54, 97; relation to primary service expense, 62; relation to secondary service expense, 63, 65; relation to independent service expense, 65; relation to hospital charges, 74; relation to routine and special service expense, 78, 80; missing category problem, 80, 82; relation to departmental expenses, 148-164 passim. See also Medical and surgical physicians, services Medical education and research, 6, 90. See also Medical and surgical physicians; Medical education and physician services factor

Medical records, expenditures, 17-25 passim, 41-45 passim, 145-147 passim Medical supplies (medical and surgical supplies plus pharmacy): regression equations for 1958-59, 72, 159; residuals for other hospital groups, 89; predictions of 1962-63 from 1958-59 equations, 101 Medical technology, influence on hospital costs, 5, 6 Medicare, 120 Missing category and missing observation, 46-47, 80, 82, 83, 119 Models, see Cost functions; Factor analysis; Regression analysis Montacute, Charles, 118 Morrill, Donald M „ 2 Motor service and ambulance, expenditures, 17-25 passim, 41-45 passim, 145-147 passim New England Pathological Society, 80, 82 Newborn expenditures, 17-25 passim, 41-45 passim, 128. See also Delivery room Nonpatient expenditures, 17-25 passim, 41-45 passim, 136, 145-147 passim Nursing education, 105, 106; expenditures, 17-25 passim, 35, 40-45 passim, 65, 145-147 passim·, missing category problem, 46-47, 80, 82 Nursing education services factor, 35, 105, 106, 125, 126; missing category problem, 46-47, 80, 82; relation to primary service expense, 62; relation to secondary service expense, 63, 65; relation to independent service expense, 65; relation to routine and special service expense, 80; relation to departmental expenses, 148-164 passim. See also Nursing education Nursing service: expenditures, 17-25 passim, 41-44 passim, 60, 145-147 passim; regression equations for 195859, 70, 148-150; residuals for other hospital groups, 88, 90; prediction of 1962-63 from 1958-59 equations, 101-103 passim·, policy implications, 115. See also Routine services Nursing students, see Nursing education; Nursing education services factor; Nursing services

INDEX/235 Obstetrical units, see Delivery room; Maternity activity factor Occupancy rate, 35-36, 107, 108, 112, 115-117, 126, 133; relation to hospital costs, 9, 30-32 passim, 40; relation to hospital service expense, 58-60, 97, 98, 104; relation to primary service expense, 62; relation to secondary service expense, 63; relation to independent service expense, 65, 67; relation to departmental expenses, 67, 148-164 passim·, relation to hospital charges, 76, 94; relation to hospital revenues, 78; relation to routine and special service expense, 78, 80. See also Utilization factor Operating room: expenditures, 17-25 passim, 41-44 passim, 145-147 passim·, services (weighted operations, major operations), 34, 36, 37, 80, 97, 106, 125, 126, 128, 139; regression equations for 1958-59, 67, 68, 70, 153, 155, 161-164 passim·, charges, 76; residuals for other hospital groups, 88, 90; predictions of 1962-63 from 1958-59 equations, 101-103 passim Operation of plant: expenditures, 17-25 passim, 41-45 passim, 145-147 passim; regression equations for 1958-59, 71 Other special expenditures, 17-25 passim, 41-45 passim, 145-147 passim

Patient days, 34-39 passim, 125, 128, 133; in Massachusetts hospitals, 13-15 passim·, in community hospitals, 34-39 passim; relation to departmental expenses, 42, 43, 105; as deflator in regression equations, 47, 67-73; estimated daily charge, 76 Pediatric activity factor, 38, 40, 41, 126; relation to departmental expenses, 148-164 passim Pharmacy: expenditures, 17-25 passim, 41-45 passim, 145-147 passim; regression equations for 1958-59, 67, 68, 71, 108, 157, 158; residuals for other hospital groups, 89, 90; predictions of 1962-63 from 1958-59 equations, 101, 102. See also Medical supplies Physical therapy: expenditures, 17-25 passim, 41-45 passim, 145-147 passim; services (treatments), 34

Physicians, see Medical and surgical physicians; Medical education and physician services factor Pooling data, see Cyclical factor Primary services, 108; definition, 16-17 passim, 41-45 passim; expenditures, 17-25 passim, 43, 145-147 passim; regression equations for 1958-59, 60-62, 68, 69, 83, 84; relation to hospital charges, 76, 94; residuals for other hospital groups, 84, 86, 87, 90; predictions of 1962-63 from 1958-59 equations, 101, 102. See also Administration and general service; Laboratory; Nursing service; Operating room; Radiology Principal component solution, 29, 30, 33 Private services factor, 39, 126, 127; relation to hospital service expense, 55, 97; relation to hospital charges, 76; relation to hospital revenues, 78; relation to routine and special service expense, 78; relation to departmental expenses, 148-164 passim Proprietary hospitals, 13-16 passim, 144; residuals for 1958-59, 85-91; predictions for 1962-63,103 Purchased services, 17, 63

Quadland, Michael C „ 112, 113 Radiology: expenditures, 17-25 passim, 41-44 passim, 145-147 passim; services (films), 34-38 passim, 97, 106, 107, 126, 128, 139-140; regression equations for 1958-59, 67, 70, 108, 153, 154, 161-164 passim; charges, 76; revenues, 78; residuals for other hospital groups, 88, 90; predictions of 1962-63 from 1958-59 equations, 101-103 passim Radiology activity factor, 36-37, 106, 109, 126, 128; relation to hospital service expense, 54, 97, 98, 103; relation to primary service expense, 62; relation to secondary service expense, 63; relation to hospital charges, 74; relation to hospital revenues, 78; relation to routine and special service expense, 78, 80; relation to departmental expenses, 148-164 passim. See also Radiology, services

236/INDEX Recovery of expense, expenditures, 17-25 passim, 41-45 passim, 137 Regression analysis, 27-28, 47. See also Coefficient of multiple determination Reimbursement policies, 10, 119-122. See also Blue Cross Repairs and maintenance, expenditures, 17-25 passim, 41-45 passim, 65, 145-147 passim Residual analysis: other hospital groups, 84-91; use in predicting costs of individual hospitals, 91-94 Revenues, 130-131; regression equations for 1958-59, 77-78, 94, 108, 109; predictions of 1962-63 from 1958-59 equations, 101, 102; and methods of reimbursement, 119-122 Roemer, Milton I., 12, 120 Routine services, 11, 22; regression equations, 78, 79, 80 Saathoff, Donald E., 107, 111 Secondary services, 108; definition, 16-17 passim, 41-45 passim; expenditures, 17-25 passim, 43-45, 145147 passim·, regression equations for 1958-59, 62-65, 68, 69, 83, 84; residuals for other hospital groups, 84-86, 87; relation to charges, 94; predictions of 1962-63 from 1958-59 equations, 101, 102. See also Dietary; Hotel services; Housekeeping; Laundry and linen; Medical supplies; Medical and surgical supplies; Operation of plant; Pharmacy; Upkeep of capital Sheatsley, Paul B., see Odin W. Anderson Size-volume factor, 34-35, 105, 107, 125; relation to departmental expenses, 41-45, 67, 148-164 passim·, relation to hospital service expense, 53, 56-59, 97, 98-100, 103-104; relation to secondary service expense, 63, 65; relation to independent service expense, 67; relation to routine and special service expense, 78. See also Beds; Economies of scale. Smith, Francis R., 2, 3, 4 Social service, e x p e n d i t u r e s , 17-25 passim, 41-45 passim, 145-147 passim

Special services, 11, 22, 23, 44; regression equations, 78, 80, 81, 94 Specific hospital effects, 63; estimates, 50-51,98 Standard error of estimate (S e ), 27-29 passim, 31, 47-51 passim Surgical activity factor, 37, 106, 109, 126, 128, 139; relation to hospital service expense, 53-55 passim, 97, 103; relation to primary service expense, 62; relation to secondary service expense, 63, 65; relation to hospital charges, 74, 76; relation to hospital revenues, 78; relation to routine and special service expense, 78, 80; relation to departmental expenses, 148-164 passim. See also Operating room, services

T-test, 28,47,48, 127 Teaching hospitals, 13-16 passim, 109, 142, 143; residuals for 1958-59, 85-91; predictions for 1962-63, 103. Theil U, 101, 102 Unaccredited hospitals, 13-16 passim, 143; residuals for 1958-59, 85-91; predictions for 1962-63, 103 Uniform accounting, 22, 23, 118-119. See also Massachusetts Bureau of Hospital Costs and Finances Upkeep of capital: expenditures, 17-25 passim, 41-45 passim, 136; regression equations for 1958-59,71 Utilization factor, 35-36, 40, 107, 108, 115-117, 126; studies in United States, 3-6 passim', deflation, 47; relation to hospital service expense, 53, 56, 58-60, 98, 104; relation to primary service expense, 62; relation to secondary service expense, 63; relation to independent service expense, 65, 67; relation to departmental expenses, 67, 148-164 passim; relation to expense per patient day and expense per discharge, 68; relation to hospital charges, 76, 94; relation to hospital revenues, 78; relation to routine and special service expense, 78, 80. See also Occupancy rate

INDEX/237 Ward services factor, 39, 127; relation to hospital service expense, 55; relation to primary service expense, 62; relation to hospital charges, 76; relation to hospital revenues, 78; relation to routine and special service

expense, 78; relation to departmental expenses, 148-164 passim Wirick, G r o v e r C . , I l l , 112

Year, see Cyclical factor

WERTHEIM PUBLICATIONS IN INDUSTRIAL RELATIONS Published by Harvard University Press * J. D. Houser, What the Employer Thinks, 1927 Wertheim Lectures on Industrial Relations, 1929 * William Haber, Industrial Relations in the Building Industry, 1930 * Johnson O'Connor, Psychometrics, 1934 * Paul H. Norgren, The Swedish Collective Bargaining System, 1941 Leo C. Brown, S. J., Union Policies in the Leather Industry, 1947 * Walter Galenson, Labor in Norway, 1949 Dorothea de Schweinitz, Labor and Management in a Common prise, 1949 * Ralph Altman, Availability for Work: A Study in Unemployment

EnterCom-

pensation, 1950 John T. Dunlop and Arthur D. Hill, The Wage Adjustment Board: Wartime Stabilization in the Building and Construction Industry, 1950 * Walter Galenson, The Danish System of Labor Relations: A Study in Industrial Peace, 1952 Lloyd H. Fisher, The Harvest Labor Market in California, 1953 Theodore V. Purcell, S. J., The Worker Speaks His Mind on Company and Union, 1953 Donald J. White, The New England Fishing Industry, 1954 Val R. Lorwin, The French Labor Movement, 1954 Philip Taft, The Structure and Government oj Labor Unions, 1954 George B. Baldwin, Beyond Nationalization: The Labor Problems of British Coal, 1955 * Kenneth F. Walker, Industrial Relations in Australia, 1956 Charles A. Myers, Labor Problems in the Industrialization oj India, 1958 * Herbert J. Spiro, The Politics of German Codetermination, 1958 Mark W. Leiserson, Wages and Economic Control in Norway, ¡945-1957, 1959 J. Pen, The Wage Rate Under Collective Bargaining, 1959 Jack Stieber, The Steel Industry Wage Structure, 1959 • O u t of print.

Theodore V. Purcell, S. J., Blue Collar Man: Patterns of Dual Allegiance in Industry, 1960 Carl Erik Knoellinger, Labor in Finland, 1960 Sumner H. Slichter, Potentials of the American Economy: Selected Essays edited by John T. Dunlop, 1961 C. L. Christenson, Economic Redevelopment in Bituminous Coal: The Special Case of Technological Advance in United States Coal Mines, 1930-1960, 1962 Daniel L. Horowitz, The Italian Labor Movement, 1963 Adolf Sturmthal, Workers Councils: A Study of Workplace Organization on Both Sides of the Iron Curtain, 1964 Vernon H. Jensen, Hiring of Dock Workers and Employment Practices in the Ports of New York, Liverpool, London, Rotterdam, and Marseilles, 1964 John L. Blackman, Jr., Presidential Seizure in Labor Disputes, 1967 Mary Lee Ingbar and Lester D. Taylor, Hospital Costs in Massachusetts: An Econometric Study, 1968 Studies in Labor-Management History Lloyd Ulman, The Rise of the National Trade Union: The

Development

and Significance of its Structure, Governing Institutions, and Economic Policies, 1955 Joseph P. Goldberg, The Maritime Story: A Study in

Labor-Management

Relations, 1957, 1958 Walter Galenson, The CIO Challenge to the AFL: A History of the American Labor Movement, 1935-1941, 1960 Morris A. Horowitz, The New York Hotel Industry: A Labor Relations Study, 1960 Mark Perlman, The Machinists: A New Study in Trade Unionism, 1961 Fred C. Munson, Labor Relations in the Lithographic Industry, 1963 Garth L. Mangum, The Operating Engineers: The Economic History of a Trade Union, 1964 David Brody, The Butcher Workmen: A Study of Unionization, 1964 F. Ray Marshall, Labor in the South, 1967 Philip Taft, Labor Politics American Style: The California Federation of Labor, 1968

Published by M c G r a w - H i l l Book Co., Inc. Robert J. Alexander, Labor Relations

in Argentina,

Brazil,

and

Chile,

Negotiations,

1963

1961 Carl M. Stevens, Strategy

and Collective

Bargaining

John T. Dunlop and Vasilii P. Diatchenko, Labor Productivity,

1964