THE FLORIDA STATE UNIVERSITY

COLLEGE OF SOCIAL SCIENCES AND PUBLIC POLICY

COMPETITION FOR OUTPATIENT SURGERIES: ASC ENTRY, HOSPITAL

SAFETY AND PHYSICIAN-OWNERSHIP

By

NITIN DUA

A Dissertation submitted to the

Department of Economics
in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy

Degree Awarded:
Fall Semester, 2011
 UMI Number: 3502836

All rights reserved

INFORMATION TO ALL USERS

The quality of this reproduction is dependent on the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.

UMI 3502836
Copyright 2012 by ProQuest LLC.
All rights reserved. This edition of the work is protected against
unauthorized copying under Title 17, United States Code.

ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106 - 1346
Nitin Dua defended this dissertation on July 25, 2011.
The members of the supervisory committee were:

Gary M. Fournier
Professor Directing Dissertation

Patricia Born
University Representative

Thomas W. Zuehlke
Committee Member

Tim R. Sass
Committee Member

The Graduate School has verified and approved the above-named committee members,
and certifies that the dissertation has been approved in accordance with the university
requirements.

ii
 First and foremost, this dissertation is dedicated to the two most important women in
my life. My late mother Lata Dua and my wife Neha. Without their unconditional love and
unbelievable trust in my capability, I would have never reached this point. I also dedicate
this dissertation to my father Bharat Dua, who never stopped me from going after my
dreams.

iii
 ACKNOWLEDGMENTS

Firstly, I would like to acknowledge my gratitude to my advisor, Professor Gary M. Fournier.

I sincerely appreciate his guidance and support in helping me write this dissertation. I would
also like to thank my committee members Professor Thomas W. Zuehlke, Tim R. Sass and
Patricia H. Born for their continued encouragement and invaluable guidance and time.
I am grateful to Dr. Farasat Bokhari for providing unconditional help. It is also impor-
tant for me to acknowledge the continued love and support of Dr. Carol Bullock who not
only encouraged me at every point but also provided a home away from home.
Finally, I would like to thank the faculty members of the Economics Department at
Florida State University who offered helpful advice and comments.
This dissertation was partially funded through the generosity of Florida State University,
Pepper Institute on Aging & Public Policy Dissertation Fellowship program.

iv
 TABLE OF CONTENTS

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

1 Introduction 1

2 Hospital Safety, Endogenous Entry and Competition by Ambulatory Surgery

Centers 5
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Developments in competition between hospitals and surgical centers . . . . 7
2.3 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Patient Data and Hospital Safety Measures . . . . . . . . . . . . . . . . . . 14
2.4.1 Patient Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4.2 Hospital Safety Measures . . . . . . . . . . . . . . . . . . . . . . . . 15
2.5 Empirical Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5.1 Patient Choice Model . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5.2 Geographic Market and Model Estimation . . . . . . . . . . . . . . . 20
2.5.3 Entry Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.6.1 Nested Logit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.6.2 Entry Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.8 Tables and Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

3 Practice Splitting and Physician Investment in Ambulatory Surgery Cen-

ters: Is Poor Safety at Hospitals Responsible? 52
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.2 Background & Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.3 Patient Data and Physician Profile . . . . . . . . . . . . . . . . . . . . . . . 61
3.3.1 Sample Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.3.2 Physician Quality, Hospital Safety Measures and Other Key Variables 64
3.3.3 Physician Ownership Data . . . . . . . . . . . . . . . . . . . . . . . 68
3.4 Empirical Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.5.1 Hazard Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

v
 3.5.2 HOPD-ASC Split Analysis . . . . . . . . . . . . . . . . . . . . . . . 80
3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.7 Tables and Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

4 Conclusion 92

Appendices 100

A Daly-Zachary-McFadden Conditions 101

B Entry Model - Specification Robustness 103

C Coefficient Estimates from Log Odds Regression - Specification Robust-

ness 104

D Human Subjects Approval Memorandum 108

Biographical Sketch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

vi
 LIST OF TABLES

2.1 Outpatient facilities in Florida, 1997-2008 . . . . . . . . . . . . . . . . . . . . 36

2.2 Single level CCS procedures and ASC Share (%) among Medicare, Florida . 37

2.3 Patient choice of provider, Florida 1997-2006 . . . . . . . . . . . . . . . . . . 38

2.4 Variables Used in the Nested Logit Estimation . . . . . . . . . . . . . . . . . 39

2.5 Average Summary Statistics - High-Risk and Low-Risk Hospitals, 1997-2008 42

2.6 Nested Logit Results - Nervous System and Musculoskeletal Patients . . . . . 43

2.7 Nested Logit Results - Eye Surgery Patients . . . . . . . . . . . . . . . . . . 44

2.8 Nested Logit Results - Digestive System Patients . . . . . . . . . . . . . . . . 45

2.9 Average Marginal Effects (% point change) - Nested Logit Results . . . . . . 46

2.10 Simulated Share (%) Change (When Closest Hospital turns High-Risk) . . . 49

2.11 Descriptive Statistics for Entry Model Variables . . . . . . . . . . . . . . . . 50

2.12 Standardized Coefficient Estimates from the Entry Model - Probit Results . 51

3.1 Summary of ASC Adoption Behavior . . . . . . . . . . . . . . . . . . . . . . 88

3.2 Summary of the Main Variables . . . . . . . . . . . . . . . . . . . . . . . . . 89

3.3 Coefficient Estimates from Cox Proportional Hazard Analysis . . . . . . . . . 90

3.4 Coefficient Estimates from the Log Odds Regression . . . . . . . . . . . . . . 91

B.1 Standardized Coefficient Estimates - Entry Model (Specification Check) . . . 103

C.1 Coefficient Estimates from the Log Odds Regression (Clustered Standard Errors)106

C.2 Coefficient Estimates from the Log Odds Regression (Robust Standard Errors) 107

vii
 LIST OF FIGURES

2.1 Changes in Physician Practice Setting, Florida 1997-2008 . . . . . . . . . . . 36

2.2 Unique Zip Code Choice Sets, 2002 (Digestive System) . . . . . . . . . . . . 40

2.3 Excess Nursing and Surgery Complications, Florida Hospitals 1997-2008 . . . 41

2.4 Predicted Choice Probability for the ith Closest Facility . . . . . . . . . . . . 47

2.5 Simulated Marginal Effect - Loss of Market Share for a High-Risk Hospital
(Both, Low Nursing and Surgery Safety) . . . . . . . . . . . . . . . . . . . . . 48

3.1 Specialty Wise Break-up of Physicians, Florida 1997-2008 . . . . . . . . . . . 86

3.2 Excess Nursing and Surgery Complications, Florida Hospitals 1997-2008 . . . 87

viii
 ABSTRACT

Ambulatory surgery centers (ASCs) have emerged as a preferred setting, in lieu of traditional
hospital outpatient departments (HOPDs), for many types of surgical care. Critics claim
that the profit incentive, cream skimming and conflict of interest due to physician ownership
are the primary features of these centers. In this dissertation, I examine factors affecting
patient demand for ASCs, importance of the profit incentive and the role that poor hospital
safety has played in the emergence of these centers. Using detailed ownership data on ASCs,
I further examine the differential impact that variation in ownership structure has on a
physician’s decision to split practice between a HOPD and an ASC. In the second chapter,
I carry out a two part analysis where in the first part a nested logit model of patient choice
is estimated with a focus on hospital safety and other facility-choice determinants, such as
distance, that drive the entry of new ASC facilities. In the second part, results from the
choice model are used to construct an expected profitability measure and the importance
of area level hospital safety is examined. I find that profitability is an important variable
but at the same time low nursing and surgery safety at incumbent hospitals is an important
reason for ASC entry. The third chapter builds on the previous analysis by investigating
the motivations of individual physicians behind the ASC growth. Physician quality as
well as ownership interest is brought in the focus and the results suggest that high quality
physicians are more likely to send patients to ASCs over HOPDs. Further, a beneficial
aspect of physician ownership comes to light, as it is found that the owners have a bigger
incentive to respond to safety problems at the hospital. These results provide policy makers
with information on favorable aspects of ASCs and should be taken into account in any well
informed debate in the area of outpatient regulation. Future work in this area can benefit
from access to even better and detailed measures on quality of physicians as well as ASCs.

ix
 CHAPTER 1

INTRODUCTION

Outpatient care has come to account for a huge share of the health care delivery in the
United States. Outpatient procedures formed 80% of all surgeries carried out in 2003
(Mechanic et al. (2005)) and almost half of these procedures were carried out in ambulatory
surgery centers (ASCs). These centers exist in a multitude of health care areas, tend to
provide focused outpatient care for selected services and compete with the hospital based
outpatient departments (HOPDs) for patient share as well as quality physicians. Meanwhile,
hospitals face the challenge of maintaining and improving patient safety. The recent national
health care quality report submitted by AHRQ to the Congress finds that the problems
of adverse events and patient safety in hospitals have not gotten much better in recent
years (AHRQ, 2010). One out of seven inpatient Medicare patients experienced an adverse
event during hospitalization, and particular concerns also remain about the rate of hospital
acquired infections (HAIs).
Along with these developments, new issues have also come forth regarding the ownership
structure of ASCs and their possible contribution to the rising health care costs. Virtually
all ASCs are owned by individual physicians or group(s) of physicians, with many cases
involving joint ventures with national investor-owned hospital and outpatient chains (Lynk
and Longley (2002), Casalino et al. (2003), Mechanic et al. (2005)). The critics of ASCs
prefer to specifically highlight the conflict of interest that arises out of physician invest-
ment but ignore the possibility of other motivations behind the growth of ASCs and the
positive impact that they may have on patient welfare. For instance, presence of ASCs in
the market increases patient choices, reduces scheduling difficulties for physicians and pro-
vides convenience to patients in the form of greater individual attention and shorter travel

1
distances.
The primary objective of this dissertation is to explore how the growth of ASCs intersects
with hospital safety performance and further, evaluate the role that physicians play as key
participants in the outpatient care market. In this context, I also highlight the part that
physician quality and physician ownership of ASCs perform in augmenting patient welfare.
I rely on rich patient level as well as physician level data sets, made available by Florida’s
Agency of Healthcare Administration and the Department of Health and apply various
empirical techniques to answer the complex questions that have come up with the growth
of outpatient market.
My dissertation is divided into two related chapters. Chapter 2 measures the impact
that hospital safety has on the patient choice of surgery setting and as a result on the entry
decision by an ambulatory surgery center. The patient choice itself is a collective term that
represents the choice of surgery setting jointly determined by a patient’s own characteristics,
the physician who is treating the patient and the patient’s insurance company. I estimate a
nested logit patient demand model that separates the patient choice into two parts. First,
the patient along with the referring physician decide between HOPD and ASC as surgery
setting (nests) and then choose from among the different hospitals or ASCs, that exist in
each nest, depending upon the choice of the nest. The model controls for the key variable
of hospital safety. The hospital in the patient choice set can be a high-risk hospital due to
either nursing or surgery complications. The results from this model suggest that a hospital
that is high-risk due to both nursing and surgery complications, can on average loose 4 to
9 percentage points of market share in the early years of 1998 to 2002. ASCs nearby to the
hospital stand to gain anywhere between 50 to 90%, depending on the procedure type, of
this loss in the hospital market share.
Similar to Chernew (2002), I extend the choice model results to an empirical model of
new ASC firm-level entry. The choice model outcomes are used to construct predict volumes
that a potential ASC entrant will serve upon entry in any locality in any year as a function
of patient characteristics located in the area as well as the existing market structure of
the area. These predicted volumes serve as a measure of expected profitability of the new
entrant. The key finding in this model is again with respect to the safety at the existing

2
hospitals in the area of entry. The results from the entry model indicate that on average an
increase in the number of high-risk hospitals in the market service area of a potential ASC
entrant by 5 or more can increase the ASC entry probability by 20-30% over the baseline
probability.

In Chapter 3, I further extend the study by bringing the focus on individual physicians.
As mentioned before, virtually all ASCs have some form of physician ownership either by
the way of joint ventures with national chains or through individual and physician group
ownership. Using the licensure data on ASCs situated in Florida, I identify the ownership
mechanism and then analyze the differential impact of hospital safety for controlling in-
terest1 and non-controlling interest physician owners. This chapter also asks the question,
that is it possible that high quality doctors practicing at low safety hospitals are more likely
to adopt ASC practice?

I divide the empirical analysis into two sections with the first one focused on building a
cox proportional hazard model to analyze the individual physician’s motivation to join an
ASC. I find that physicians with high outpatient volumes are significantly more likely to
adopt ASC practice while, those with a high inpatient volume component are significantly
less likely to do so. High quality physicians identified by a greater number of staff privileges,
are also significantly more likely to adopt ASC practice. Using a measure of physician quality
similar to that applied in Burke et al. (2007), I find that physicians with ‘star’2 status are not
significantly more likely to adopt ASC practice, but once they adopt the practice they are
more likely to treat their patients at ASCs. I also find that the leading motivation behind
adopting ASC practice is poor nursing safety at the hospitals. Physicians with majority of
practice at a hospital that is high-risk due to nursing complications are 41% more likely to
adopt ASC practice. The second part of empirical analysis looks into the practice splitting
decision of the physician with the split taking place between HOPDs and ASCs and finds
that physicians with a controlling ownership stake at an ASC are significantly more likely
to punish the HOPDs for poor nursing and surgery safety performance by migrating their
patients to ASCs.
1
Controlling interest implies ownership interest of greater than 5%.
2
‘Star’ status measure as used in this study is explained in detail in section 3.3.2

3
 The results presented in this dissertation have important implications for policy govern-
ing competition in the arena of healthcare as well as that focused on hospital safety issues.
Among other important features of outpatient healthcare and ASC growth, I empirically
evaluate two key aspects of hospital safety and physician ownership of health facilities, such
as specialty hospitals and ASCs, in two different but related chapters of this dissertation.
Main empirical findings suggest that poor patient safety at hospitals is one of the important
reasons for the growth of ASCs as well as for the increasing dissatisfaction of physicians with
the hospital practice. Hospitals that are high-risk to the patients due to nursing and surgery
complications are likely to lose significant outpatient share to ASCs. Further, I show that
compared to non-owner physicians, owners are more likely to respond to poor safety at the
hospitals by shifting their patients to ASCs from HOPDs. These results together indicate
the nature of competition between hospitals and ASCs and its potential to improve patient
welfare. First, the specialized, convenient and focused ASC environment provides patients
with a good quality alternative to poor safety hospitals. Second, loss of market share is a
significant threat to hospitals and they are likely to respond by improving safety as well as
taking other steps like investing in specialty departments, private rooms and nursing staff
training. Finally, a previously unexplored benefit of physician ownership comes to light as I
show that owner physicians have easy access to and more control at their own ASCs which
helps them react strongly for the benefit of their patients by migrating them away from
high-risk hospital environment to a relatively safer ASC setting.

4
 CHAPTER 2

HOSPITAL SAFETY, ENDOGENOUS ENTRY

AND COMPETITION BY AMBULATORY
SURGERY CENTERS

2.1 Introduction

There is dissatisfaction over the lack of improvement in the quality of care in US hos-
pitals. The recent national health care quality report by AHRQ to the Congress finds the
problems of adverse events and patient safety in hospitals have not gotten much better in
recent years (AHRQ, 2010). The report found that one out of seven inpatient Medicare pa-
tients experienced an adverse event during hospitalization, and particular concerns remain
about the rate of hospital acquired infections (HAIs) which have also worsened compared
to earlier years.
Along with the challenge of maintaining and improving patient safety, over the last few
years, hospitals have also faced increased competition from specialty care providers called
ambulatory surgery centers (ASCs). These providers operate on a small scale, involve physi-
cian ownership and focus on specific procedures that can be performed on a single day basis.
Their presence in the market increases patient choices, reduces scheduling difficulties for
doctors and provides convenience to patients in the form of greater individual attention and
shorter travel distances. Incumbent, acute care hospitals, on the other hand, recognize the
effects of ASC competition in siphoning off outpatient surgical cases and reducing revenue.
The next two sections provide an overview of these developments and the literature.
The focus of this study is on two distinct, but related empirical aspects of the market for
outpatient surgical care. First, the patient demand for surgical procedures is empirically

5
examined in the framework of a multinomial choice model including hospital outpatient
departments and ASCs. The nature of the problem easily lends itself to a nested logit
framework to estimate demand for specific surgeries that can be done both at hospital
outpatient departments and ASCs. Patients and their referring physicians are assumed to
first decide between a hospital setting or an ambulatory surgery center and then choose
among different acute care hospitals within the hospital nest or among different choices in
the ASC nest.
A key question addressed in this study is the direct effect of patient safety measures
on patient choice of surgical setting. Safety is measured with indexes derived from the set
of AHRQ patient safety measures including post-surgical infection and complication rates.
If primary physicians base their patient referrals upon knowledge of the relative safety of
local hospitals, and if the available measures of these dimensions are good, they should have
explanatory power in how patients get allocated across available options. For this study,
surgeries are defined in narrow categories of close substitutes, based on clinical classification
software’s (CCS) multi-level classification, including operations on the digestive system,
the eyes, the nervous system and on the musculoskeletal system. Data comprise a twelve
year panel including a census of surgeries performed in hospitals and ASCs in Florida.
The empirical strategy is explained in section 3.4, and the results are given in the section
following it.
The second part includes an empirical model of new ASC firm-level entry by locality
and year, as a function of its expected profitability and other determinants, including the
quality of care available at local hospitals. This sequence of estimation is similar to Chernew
(2002) where the authors study the entry of hospitals in CABG market based on payer type
returns. However much the safety indexes are informative in the model of patient choice
of health facility, it is hypothesized that an additional effect on ASC entry may also occur.
With 221 entry episodes in the 1997-2008 panel across Florida, the model addresses whether
the safety of care provided by local hospitals has any explanatory power in the decision to
open an ASC. The hypothesis is that high quality surgeons operating in an incumbent
hospital can profitably disassociate from safety problems by establishing a separate ASC to
achieve better control. ASCs are modeled as profit maximizing firms that enter the market

6
once the threshold predicted volume of surgeries is reached, but their ability to profitably
enter depends further on the safety of existing care in the market.
The following section provides a background to the outpatient health care in US. It is
followed by a literature review and a description of the data. Empirical strategy is described
in the next section, followed by an interpretation of the empirical results. The last section
concludes with a discussion on potential implications and further directions of this research.

2.2 Developments in competition between hospitals and

surgical centers

In the recent past there has been a surge in the number of small stand alone units
that provide outpatient health care. These providers exist in a multitude of health care
areas and are known as freestanding ambulatory surgery centers (ASCs). Type of ASCs
vary between those that provide focused outpatient care for selected services and those that
perform multiple services under one roof.1 Services are limited to planned surgical episodes
requiring less than a 24 hour hospital stay. These facilities are different from hospital based
outpatient departments as they are not located on hospital premises and are often not
affiliated with a general acute care hospital. Most of these facilities are owned by individual
physicians or group of physicians but there is also an increasing trend towards consolidation
with national investor-owned hospital and outpatient chains (Lynk and Longley (2002),
Casalino et al. (2003), Mechanic et al. (2005)). Nationwide, the number of freestanding
ASCs rose from 2,314 in 1996 to 3,400 in 2002 (Mechanic et al. (2005)) while, in Florida
itself their numbers increased from 218 in 1997 to 365 in 2008 (Table 2.1).
Technological developments in the fields of minimal invasive surgery and anesthesia,
better architectural designs for operating rooms and supporting facilities, more convenient
suburban locations and an entrepreneurial zeal among the physicians combined with their
desire for a higher degree of control over procedure scheduling and other facility level op-
erations are some of the reasons for ASC growth (Poole (1999), Mechanic et al. (2005)). It
is also argued that these facilities benefit from efficiencies due to a narrow focus on fewer
1
In 1997, half of the ASCs in Florida could be considered specialty facilities exclusively performing
surgeries related to specific body organs or systems. By 2008, the proportion of ASCs that were specialty
care providers grew up to 60%.

7
disease groups/surgeries. However, the research evaluating various aspects of their role in
health care is limited.
The state of Florida has witnessed its own share of growth in competition from ASCs.
Data supplied by the Agency for Health Care Administration (AHCA) in Florida reveals
some of the trends in ASC evolution and in empirical estimation. The agency collects a
census of patient visits for both inpatient and outpatient care as well as infrastructural data
on health care facilities operating in the state. Those data reveal that the number of ASCs
has increased by 67% over the period of 1997 to 2008. Growth in the number of hospital
based outpatient departments, in comparison, has been stagnant.
As shown in figure 2.1 growth of ASCs is also associated with a gradual change in
physician preferences for outpatient surgery at ASCs. This change reflects the fact that
less than half of all outpatient surgeries are now done in a hospital setting (Haugh, 2006)
as compared to mid 80’s when more than 90% outpatient surgeries were performed at a
hospital. The sample selected in the figure includes only those surgeons whose work involves
outpatient procedures done at hospitals and ASCs, and who at any point during 1997-2008
performed surgeries at both settings. The figure shows the dramatic change in the relative
share of ASCs for elective surgeries.2
ASCs in Florida have also gained substantial market share in several specific procedures.
For instance, using single level Clinical Classification of Services (CCS) codes and data
from quarter 2 of selected years, Table 2.2 depicts trends (and ASC share of total Medicare
outpatient market, in parenthesis) for some of the outpatient procedures. These are major
types of procedures that have witnessed substantial increases in share and are also the
ones that affect a large number of patients. Besides the procedures shown in the table,
there are some others that have witnessed major growth in shares of around 15 to 25
percentage points over the period of 6 years. For instance, 70-90% of eye care related
procedures like Glaucoma, Lasik surgery and therapeutic procedures on eyelids, cornea etc
were performed at ASCs in 2006. Similarly, more than 60% of ‘Esophageal dilatation’3 and
2
The sample omits other specialists doing major interventions like heart surgeries strictly on inpatient
basis. In addition, this share was calculated for only those physicians who on average served at least 300
elective surgery patients every year and at least in one year during 1997-2008 were working both at ASCs
and hospitals. There are 2247 such physicians in the state of Florida.
3
This technique is used to stretch or open the blocked portion of the food tube.

8
‘Upper gastrointestinal endoscopy’4 Medicare cases were also performed at these centers (a
15 percentage point increase from 2001).
As pointed out earlier, ASCs have quite a few advantages over acute care hospitals in
the provision of outpatient services. Advocates for the hospitals, however, argue against
the establishment of ASCs, as outpatient services are a profitable area of their operations
and thus help them cover their costs for other socially relevant but unprofitable services
(Casalino et al., 2003). Not surprisingly, hospitals themselves are getting more involved as
participants in the ASC business model, probably due to the efficiency advantages gained
from a dedicated outpatient care environment (Smallet (2008),Mechanic et al. (2005)).
Another noteworthy feature of the changes in the outpatient care market is the effect
on hospital behavior. Data records as well as independent reports indicate that certain
inpatient procedures that earlier were a part of the basket of services primarily provided at
acute care hospitals are now increasingly being performed at ASCs (Russo et al. (2007)).
ASC entry has not only reduced the cost of services for the patient by reducing length of
stay but has also made certain surgeries/procedures much more convenient than previously
and may have lead to a shift in patient preferences of surgery setting. Thus, drawing of
patients away from inpatient and outpatient departments at hospitals is likely to affect the
decision making process, whether led by the physician or the patient, about the setting
of the surgery; that is, whether the surgery should be performed at hospital outpatient
department or at an ASC, after accounting for all important medical aspects of the surgery.
Table 2.3, using the Florida data from second quarter, presents some trends over the
period of 1997-2006 in the number of surgeries/procedures done at inpatient hospitals,
ASCs and hospital outpatient departments in Florida. The table illustrates the shift of
major surgeries from inpatient to outpatient arena. Shown are some important health
care procedures traditionally carried out as both inpatient and outpatient. For instance,
in the procedure of ‘Laminectomy’, the total number of patients over time (incidence)
has not changed much but inpatient hospitals have lost a fourth of their share by 2006,
compared to 1997, while, ASCs and hospitals outpatient both share almost one third of the
Florida market in 2006. We also witness visits more than doubling at hospital outpatient
4
Visual examination of the upper intestinal tract using a fiber optic or video endoscope.

9
for ‘Tracheoscopy’ over the given period. Most of this growth seems to have come at the
expense of hospital inpatient departments given that the total number of visits have not
changed much. For the growth areas like ‘Insertion of Catheter’ and ‘Colonoscopy & Biopsy’
we find that while the number of overall visits have more than doubled in the first case and
almost tripled for the second, the hospital inpatient and outpatient departments have either
lost share (‘Insertion of Catheter’) or gained very little relatively (‘Colonoscopy & Biopsy’).
ASCs on the other hand, have seen almost quadrupling of total visits and an increase in
their share to almost three fourths of the market in both cases.
Growth of ASCs has wide reaching impact on the health care sector. It has increased
patient choice and has provided patients with an element of convenience as the surgery
centers are likely to locate close to the patient’s residence and due to their small size
provide a feeling of individual attention that is difficult for big hospitals to offer. ASCs
have a capability to impact not only the patient welfare but also the financial goals of
policy makers. Promoting competition and supporting technological developments in the
mode of health care delivery are important alternatives that policy makers can pursue to
control the costs and increase efficiency of the system. To do so, policy makers must have
access to good and actionable information about the nature of competition from ASCs and
their effects on hospital market outcomes.

2.3 Literature Review

Empirical literature that addresses the ASC entry impact on patient welfare is scarce
but there is substantial work done in the area of hospital quality as well as entry and
competition.
Plotzke and Courtemanche (2010) and Courtemanche and Plotzke (2010) studied the
effects of ASC entry on hospital surgical output and the relationship between profitability
of a surgery and its likelihood of being carried out in an ASC, respectively. In the first case,
the authors used a linear probability model to predict the probability that the setting of a
surgery will be an ASC while controlling for profit rate and time fixed effects besides other
patient level controls. Their results suggest a higher probability of profitable surgeries, after
controlling for surgery type, to be performed at an ASC. However, the authors also point

10
out that the patient health and procedure complexity may be even more important drivers
of surgery location choice. Further, it is important to note, and the authors observe it
too, that the ASC growth may also have market expansion effects or that lower prices at
ASCs may lead to an increase in quantity demanded. Also, procedures done at ASCs can
inherently be profitable given that the risk profile of the patient at an ASC is likely to be
lower.
Courtemanche and Plotzke (2010), evaluated the ASC entry effect on hospital output.
The authors found a 2-4% decline in a hospitals annual outpatient surgeries but no signif-
icant change in inpatient surgeries after ASC entry. Also, they found that the impact on
hospital output was stronger due to the first ASC entry than the later entries. Definition of
hospital geographic market is a very important element of competition analysis and their
paper used two different approaches of fixed radius (11.5 miles) and variable radius market
definition. Both of these methods have a problem in that they do not account for possible
substitutions between providers that can take place if one evaluates the geographic market
from the patient’s perspective. The radius approach especially does not account for patients
who are residing at the boundary of the defined market and are thus likely to go outside
the fixed radius. Another issue with Plotzke’s approach is the homogeneous treatment of
ASCs i.e. the scope and service attributes are not treated differently in terms of their effect
on hospital volumes.
Recent research by Weber (2010) comes close to our study in terms of some empirical
techniques and the data set used. However, the author is interested in estimating a demand
curve, as a function of travel time, for health care facilities (hospitals and ASCs) providing
outpatient care by using Agency for Health Care Administration data for Florida from
2003-2004 (we use the same data set covering the time period of 1997 to 2008). The author
uses a multinomial logit choice model with random parameters to estimate patient choice
for outpatient surgery and uses an effective method for computing the welfare impact of
ASCs on patients. Since there is a high estimation cost of mixed logit specifications in
the form of increased convergence times for big data sets, the author restricts estimation to
choice models for individual procedures based on CPT codes. This is not a problem when the
researcher’s interest is in estimating specific demand functions, but for studying competitive

11
effects of entry it is important to define markets broadly enough to cover all procedures that
are generally provided under one roof. Therefore, in this dissertation markets are groups of
procedures that are either carried out on the same body parts or on the same body system.
For example, all operations on the eye form one self contained market. Instead of fixed
effects as applied in Weber (2010) actual measures for hospital safety can be employed to
obtain the precise impact of safety on patient choice as well as ASC entry.
The motivation for modeling patient choice directly emerges from the need to recognize
the value that ASCs may provide from the patient point of view. Due to technological
changes, many surgeries, including for example, hernia repair, bariatric surgery and removal
of gall bladder, that were previously handled with an inpatient hospital stay can now be
done in an outpatient setting (Russo et al., 2007). This evolution and the structural benefits
like small size of the facility, likely short distances from home, focused nursing staff and less
administrative delays accompanied with the same day outpatient surgery aspect, overall
provide a convenience factor for the patients that make ASCs very attractive. On the flip
side, however, patients need to consider characteristics like lack of emergency departments
and potential risk of other complications that an ASC may not be equipped to take care of.
For example, surgeries like ’Mastectomy’ that were typically performed on inpatient basis,
faced controversy and legal action in different states when performed in an outpatient setting
(Russo et al., 2007). Agency for Health Care Administration (AHCA) data, from 2001-06,
show a fall in the share of outpatient Mastectomy surgeries performed at ASCs, from 11.4%
in 2001 to 3.5% in 2006 reflecting this risk element and the regulatory crackdown.
The nature of ASC distance to emergency departments was recently explored by Neu-
man et al. (2011). Emergency services at the hospital are highly valued by the patients as
protection against potential health complications that may arise during surgery. Accord-
ingly, authors claim that the distance to emergency departments is an important factor in
evaluating impact of the growth of ASCs and a resulting increase in distances of ASCs from
hospitals can adversely affect the patient welfare. They find that from 2005 to 2007, the av-
erage patient-to-emergency department distance increased by 12.4%, resulting mainly from
an increase in the share of procedures performed at ASCs and an increase in the distance
between ASCs and emergency departments. They also find that the change in distances

12
occurred predominantly within procedures that carried elevated odds of hospital admission
relative to the lowest risk procedures. However, there are some serious issues with their
results that the authors fail to highlight appropriately. First, the 12.4% increase in distance
refers to an average change from 1.49 km to 1.68 km. This is hardly a big enough magni-
tude of change to cause concern and given that their results are based on Florida data, it is
also important to take into account that a flat geographical topology (based on regression
estimates of actual distances on actual time) ensures that it takes on average a minute to
travel a mile (or 1.6 km) in Florida. Therefore, an average change of 0.19 km is roughly an
increase of 7 seconds in travel time. Second, they choose a subjective measure of surgery
risk, the odds of a surgery leading to hospital admission relative to odds after eye surgery.
Given that eye surgeries have a very low incidence of complications, all other procedures
will seem to have a high risk status.
Unlike the case of ASCs, the association between competition and quality of care has
been studied extensively in health literature, in the context of increased competition be-
tween acute care hospitals. Kessler and Geppert (2005) used the competition measure first
suggested by Kessler and McClellan (2000) and evaluated the effect of competition on pa-
tients with different levels of health severity. They found that in an increasingly competitive
market patients with higher levels of illness severity also get more intensive treatment while
patients with lower levels of severity receive less intensive treatment. Their results sup-
port the notion that there are efficiency enhancing effects of competition, but it is unclear
whether similar conclusions can be applied to the competitive effects from ASCs.
Barro et al. (2006), on the other hand, studied hospital competition in the light of
new entry. They examine the impact of entry by cardiac specialty hospitals on the overall
cardiac care quality in the market. They found that markets with specialty hospital entry
have lower spending for cardiac care, but, these hospitals tend to attract healthier patients
and provide a higher level of intensive procedures. Specialty Hospitals as well as ASCs
can have diverse effects on the health care market. If their entry promotes competition and
leads to better allocation of patients among providers, based on patient severity, then policy
makers should facilitate their growth. In contrast, if their entry has social costs in the form
of undue and unnecessary increases in services, a medical arms race, then these negative

13
external effects need to be accounted for in policy making. Barro et al. (2006) study the
effect on costs and outcomes at the hospitals following specialty hospital entry.
The entry model proposed in this dissertation is similar to the two step analysis carried
out by Chernew et al. (2002). Their work aimed to infer payer type returns in bypass
surgery based on hospital entry behavior. Authors used patient flows from a choice model
as an input into their entry model and tried to estimate the likelihood of hospital entry as a
function of the payer type returns. Like them, I also rely on predicted volumes along with
market level safety measures but, instead of predicting ASC entry, I model the probability
of a county witnessing ASC entry in any given year and use county level predicted volumes
at the best zip code location (as explained in section 3.4).
Evidence that poor hospital safety may drive the entry of new ASCs would help substan-
tiate the hypothesis that these facilities are welfare-enhancing. When local hospital care
is subject to poor safety performance, such as excessive post-operative and other hospital-
acquired infections or poor nursing quality, surgeons operating in the incumbent hospitals
can profitably disassociate from these safety problems by establishing a separate ASC to
achieve better control. ASCs are modeled as profit maximizing firms who only choose to
enter the market when the predicted volume of patients is greater than the threshold vol-
ume that gives zero profit. Once a market attains a size adequate to provide the threshold
patient volume and cover fixed costs, entry depends further on the safety level of existing
care in the market. If so, then the likelihood of a positive welfare effect is strengthened.
This study is directed towards answering these questions.

2.4 Patient Data and Hospital Safety Measures

2.4.1 Patient Data

For this study, inpatient discharge data and outpatient visit data involving surgical pro-
cedures and hospital financial data from 1997 to 2008 are collected from Florida Agency for
Health Care Administration (AHCA). Patient characteristics in the data include detailed
clinical and demographic information. Patients’ procedures, coded in Clinical Classification
System (CCS) codes were used to select the set of surgery types affected by ASC compe-
tition. The multi-level CCS classification system is used for the choice model. It groups

14
procedures by body systems or condition categories (AHRQ (2007)) and allows us to ana-
lyze different health care markets as defined by the CCS type. Separate choice models are
estimated for those CCS types that form the majority of ASC services. These are opera-
tions on the nervous system and musculoskeletal system (CCS 1 & 14), operations on the
eye (CCS 3), and operations on the digestive system (CCS 9). Upon observing the actual
combination of CCS types performed at the same facility, CCS 1 & 14 were grouped to
define a self contained market. These three CCS categories account for approximately 90%
of the total outpatient procedures done at ASCs.
Table 2.4 provides the details of variables used in the choice model specification. In-
cluded are the patient’s payer category (Medicare, Medicaid, Commercial Insurance, and
Other), demographics (age, race, gender), and distance measures derived from patient res-
idential zip code centroid. Distances in miles to each hospital as well as between each
hospital and each ASC in Florida are derived using a program that extracts actual dis-
tances from Google maps. Distances between hospitals and ASCs were calculated using
actual facility addresses. Attributes of each acute care hospital include its control type (for
profit, not for profit, or government), teaching status, nursing intensity, capital intensity,
and the total number of beds available. Two hospital safety safety measures based on the
Agency for Healthcare Research and Quality (AHRQ) safety indexes are used. The methods
for constructing these safety indexes are detailed next.

2.4.2 Hospital Safety Measures

Of great interest to the estimation are the facility characteristics relating to safety - the
nursing safety and surgery safety indexes. First, the underlying components of these two
indexes are constructed using AHRQ software. The AHRQ software is basically an algorithm
that uses hospital inpatient discharge data to compute risk-adjusted safety indexes called
the patient safety indexes (PSIs). 5 These indexes track the occurrence of adverse events
and focus on conditions and complications experienced by patients during their hospital
stay. According to the AHRQ (2003), PSIs screen for medical problems that patients face
due to their exposure to the healthcare system and that can be prevented by appropriate
5
Risk adjustment is based on computed hospital fixed effects. First, using logistic model predicted value
of complications is calculated for each discharge and then subtracted from the actual outcome. Then, this
difference is averaged over each hospital to get the risk adjusted rate.

15
changes at the system or provider level. The report also clarifies that adverse events refer to
injuries or medical harm caused due to medical mismanagement and not by the underlying
health condition of the patient. It is also important to point out that the AHRQ safety
indicators are different from AHRQ inpatient quality indicators as those indicators are
more concerned with measuring the quality of care received by inpatients and is therefore,
applicable to studies interested in evaluating hospital performance in the area of lower
mortality and efficient utilization. However, in this paper, primary interest lies in exploring
the safety issues that are unrelated to a patient’s health condition and therefore indicate
the safety culture through out the hospital’s different departments including the outpatient
services. To construct these PSIs, Florida hospital inpatient data files are used. These data
include all patient discharges with a major surgery diagnosis-related group (DRG) for all
the years from 1997 to 2008.
In the second step, following closely the methods detailed in Encinosa and Bernard
(2005) and Bazzoli et al. (2008), year and hospital specific nursing and safety indexes are
constructed. Using the year and hospital specific PSIs, two aggregate measures of patient
safety over all major surgery discharges, by year and by hospital, are constructed. The
first measure, surgery safety index, consists of nine PSIs that are most closely related to the
actual surgery, while largely independent of post-operative nursing care. These include com-
plications in anesthesia, postoperative hemorrhage or hematoma, postoperative hip fracture,
postoperative physiologic and metabolic derangements, postoperative pulmonary embolism
or DVT, postoperative respiratory failure, postoperative sepsis, postoperative wound de-
hiscence and finally, accidental puncture or laceration. The second measure, nursing safety
index, is constructed from three PSI indicators previously recognized as related to the nurs-
ing activity - post operative hip fracture, decubitus ulcer, and selected infections due to
medical care.6 These two measures distinguish nursing and surgery as sources of adverse
events in hospital, and may produce different effects in the model.
After computing the individual PSIs we subtract from each hospital’s rate the overall
average rate, for the same PSI, of hospitals in the Florida data set.7 This gives a measure
6
As explained in Bazzoli et al. (2008), postoperative hip fracture is used in both type of indicators as it
can occur because of either nursing or surgery errors.
7
Only those individual PSIs are used for which the population at risk was at least equal to 30 patients
at the hospital.

16
of excess number of incidents that took place in the hospital. Finally, each of these excess
measures (as mentioned above, 9 for surgery and 3 for nursing) are weighted by the propor-
tion of patients at risk for the indicator. To illustrate for the case of the nursing indexes,
first, the excess measure for each of the three nursing PSIs is weighted by the number of
patients at risk at the hospital. These weighted excess measures are then summed up and
divided by the total number of patients at risk for all 3 nursing related PSIs.

2.5 Empirical Strategy

2.5.1 Patient Choice Model

The empirical analysis is divided into two parts. First, a choice model is used to study
characteristics and drivers of patient choice of location for surgery, i.e. patient choice8
among ASCs and acute care hospitals’ outpatient departments. Using a choice model, the
first order effect of safety is examined i.e. how is the market share of a hospital affected by
lower safety performance. Patient choices from the model are then used to predict volumes
for an hypothetical entrant in the next part. The choice predictions feed into a model that
predicts probability of entry as a function of expected profits (as measured by expected
volumes) and existing market conditions, including surgery and nursing safety rates of each
hospital in the market.
The patient choice decision for a surgery is modeled within a nested logit framework.
The patient decision is considered as a nested choice as it is reasonable to believe that the
patients who choose a hospital over an ASC may be less healthy and more worried about the
potential complications that may arise during a surgery and that can be much more easily
handled at a hospital due to the presence of an emergency department and the availability
of combined medical experience from different fields. On the other hand, certain procedures,
like lasik surgery or endoscopy, involve much less risk and conceivable complications than, for
example, spine surgery and consequently the patient or physician may be predisposed to the
idea of getting the surgery done in a smaller and convenient ASC facility. Nesting hospitals
8
Patient choice here actually implies the choice of location made by the patient and her physician together.
This choice is likely to be guided by the insurance plan restrictions, especially in the case of Non-Medicare
plans. However, keeping with the traditional use of the term in case of choice models and for ease of
description I will call it the patient choice.

17
and ASCs in different branches further helps by allowing different substitution patterns in
both nests. This is an important benefit of the methodology as it not only weakens the
Independence of Irrelevant Alternatives (IIA) assumption9 but also strengthens the market
share predictions by taking into account that new ASC entry will take away business from
both hospitals and ASCs but certainly in different magnitudes. Finally, nesting also allows
to account for information that is specific to hospitals or ASC nests. For instance, for-
profit/ non-profit status of the hospital along with the safety measures for hospital nest and
distance to the closest emergency center in case of ASCs.

Since there are a number of procedures that can only be treated at hospital outpatient
departments, procedures selected for the study are the ones that are amenable to be done in
both hospital and ASC outpatient environments. The patient is assumed to choose between
all the facilities that can provide the surgery within the expanse of a defined geographical
market. This choice problem is then expressed within the class of nested logit random
utility models where an individual maximizes utility over available nests, and conditional
on a nest, maximizes utility over available choices within the nest. Each patient i (where
i=1 to I ) makes a choice among alternative health care facilities j (where j=1 to J ) in
nest Bk (k = hospital or ASC) that can provide the required procedure. Thus, with utility
expression Uij , where j ∈ Bk , patient’s utility function, given knowledge of clinical condition
and facility choices, can be written, using the terminology by Train (2003), in the following
form:

Uij = Wik + Yij + εij , j ∈ Bk (2.1)

Here, Wik represents a set of variables that vary by nests but are constant over alternatives
within nest k. In the model, these are individual specific regressors like, age, gender,
race, payer type etc. (see Table 2.4) that affect patient’s choice of nest. The second set of
variables, Yij , vary across different choices within the nest and include choice characteristics
as well as interaction terms between patient demographics and choice characteristics. In
9
IIA assumption implies that the relative probability of choosing between any two alternatives is indepen-
dent of all other alternatives available. This relative probability stays constant even when one or more of the
alternatives are perfect substitutes. This problem of IIA assumption is also knows as the red-bus-blue-bus
problem in choice literature (Train (2003)).

18
particular, Yij can be summarized as:

Yij = βRj + ηXi Rj + νDistij + δDistij Xi + θDistij Rj (2.2)

The term Xi represents the patient characteristics, for example, demographics and a set of
dummies for payer types. Distij is the mileage from patients i0 s home to the provider j 0 s
location and Rj is facility characteristics including its size (number of beds), the measures
of safety at the hospital and ownership type. Table 2.4 provides details on the facility
characteristics, Rj , which are defined similarly in both the nests but with some differences
in specific regressors by nest. Certain features like teaching status and profit status are
not relevant to the ASC nest, while others (such as specialty status and distance to the
emergency center) are not relevant to the hospital nest.
The nested logit model assumes that an individual can compare utility derived from
each choice and choose the alternative that gives the maximum utility, so choices are based
on net utilities (Jones, 2000). The stochastic error term is given by εij and is assumed to
follow a Generalized Extreme Value (GEV) distribution. The error terms εij are correlated
within a nest and the probability of facility j 0 s selection by patient i in nest k, for a given
procedure, can be written as:

Pij = Pij|Bk PiBk (2.3)

Here, Pij|Bk is the conditional probability that the alternative j is chosen given an alternative
in nest k is chosen, and PiBk is the marginal probability of choosing an alternative in nest
Bk . The product of marginal and conditional probability then gives the joint probability of
a specific choice. This can be further written as:

eWik +λk Iik eYij /λk

Pij = PK W +λ I P Yij /λk
(2.4)
l=1 e j∈Bk e
il l il

where,
X
Iik = ln eYij /λk
j∈Bk

Here, Iik is the inclusive value and λk is the dissimilarity coefficient, for nest k, indicating
the degree of independence among choices in the nest.

19
2.5.2 Geographic Market and Model Estimation

It must be noted that the sampling design is subject to certain considerations. I use
the choice based sampling strategies to select relevant hospitals, ASCs and patients and to
construct unique zip code specific choice sets. Consistent with multinomial choice analysis,
the choice set for a patient in any zip code area should be self-contained and include every
facility available for the procedures sampled. This means, first, that the analysis should
not overlook any other ‘outside’ facilities where evidence reveals that patients in the local
area are able to choose, and sometimes actually choose, for surgical care. These outside
hospitals and ASCs are competing for local patients. In short, diverse zip code level choice
sets are constructed. Varying the hospital choices by areas as small as a zip code allows
for considerable heterogeneity across the total service areas of any given hospital or ASC.
Further, since separate models are estimated by type of procedures (CCS type - 1 & 14, 3
and 9)10 , the choice sets from the same zip code may also be different for different procedure
types.11 For example, figure 2.2 shows choice sets, for digestive system patients, for two
specific zip codes in Florida along with distances to different ASCs (color shaded) and
hospitals (non-shaded). One of the zip codes is located in Tallahassee (32303) and the
other is in Miami (33009). Since Miami has a much higher population density, zip codes in
Miami have more choices and in this particular case more hospital choices than ASCs. The
average distance to any of the facilities in the choice set is 8.6 miles in Miami and 9.3 miles
in Tallahassee.
Kessler and McClellan (2000) and Gowrisankaran and Town (2003) point out the prob-
lem of endogeneity that can arise if competition is defined on the basis of actual patient
choices. In cases where the interest lies in understanding competitive effects, the estimates
can get biased if the market is calculated from actual patient choices because these choice
decisions may become increasingly affected by unobservable quality aspects of the health
care facilities. Kessler and McClellan (2000) also cite other problems associated with mea-
10
CCS 1 & 14 stand for operations on the nervous system and musculoskeletal system , CCS 3 includes
operations on the eye and operations on the digestive system are included under CCS 9.
11
To reduce the sample to a manageable size, choices that are made by less than 2% of the patients from
a zip code are dropped. Zip codes with less than 10 patients are also excluded. In Fournier and Gai (2007),
sampling issues are discussed at length and authors also explored the sensitivity of the model’s predictions
to changes in the sampling design.

20
suring market sizes based on patient choices and introduce the method of using probabilistic
patient choices based on the estimates of multinomial logit models that account for both
patient and facility characteristics and help build competition measures at patient zip code
residence level. This method is also adopted here to identify the likely potential patient
volume of a new entrant from predictions.

The patient choice model is estimated separately for each year from 1997 to 2007 to
account for any changes in patient choice parameters caused due to a changing health care
environment and the resulting payer response. Separate estimation also allows to account for
patient choice set changes over time as the set of facilities continuously expands. To facilitate
the numerical convergence, the parameters are estimated on a random sample of zip code
choice sets, including 40,000 patients12 for each CCS type, i.e. nervous and musculoskeletal
system, eye care and digestive system surgeries, to obtain converged parameters.13 These
estimates are used to get choice predictions over the 11 year period.

Previous literature in health economics has shown in detail the importance of distance
and travel time for patient choice and the preference for the closest hospital. In Florida, the
distance to the closest hospital or closest ASC can vary across different regions. To account
for this effect I include ordinal constants for each choice in both the nests. The order of
choices is coded for each nest and is decided based on the closeness of each facility choice
to the centroid of the patient zip code. In nested logit models, even with variable choice
sets across individuals, the estimation software still constructs a universal choice set with
all the choices in it and therefore, leads to heavy computational costs. There are also other
estimation problems that result when random samples from the data are taken as there may
not be enough observations left for each of the specific choices, resulting in dropping of the
choice from the model altogether. Including ordinal constants allows to limit the universal
choice set as each choice is now identified by its order, as the closest choice to the patient,
the second closest and so on. The model then gives the effect of varying characteristics
of the choices without taking into account the actual identity of the choice. This helps in
12
I also experimented with bigger sample sizes that form 25% and 50% of the overall CCS wise data and
found little or no change in the parameter estimates
13
This results in a total of 33 choice models (11 years x 3 CCS types) collectively accounting for 120,000
patients each estimated year or approximately 10% of CCS type patient population depending on the year.

21
reducing convergence time for the nested logit models and also to account for all possible
choice characteristics in estimation.

Hospital level measures for nursing and surgery safety, used in the nested logit model,
are also relevant to the choice set definition. There is substantial variation in aggregate
nursing and surgery complication rates across hospitals statewide, especially in the case
of excess nursing complications. However, within localized zip code choice sets it is often
the case that the hospital complication rates are clustered close to each other. Therefore,
in the nested logit specification a discretized version of both nursing and surgery safety
rates is used. The discretized rates indicate if a hospital is high-risk due to poor nursing
or surgery safety. The high-risk hospitals, in turn, are defined within a choice set as those
hospitals that have a complication rate one standard deviation14 greater than the average
complication rate of the choice set. This implies that the same hospital may be considered
a high-risk hospital in some of its markets and a low-risk in some others. This conforms
with the notion that patients choose from the facilities that are relevant to their geographic
choice set. By this definition, more than 35% of the hospitals, on average across the years,
can be considered high risk hospitals.

Table 2.5 presents a comparison of the hospitals that have nursing complication rates
and/or surgery complication rates one standard deviation higher than the average (High-
Risk) in their choice set with those that do not (Low-Risk). On average, across the 12 year
period, there does not seem to be substantial differences in covariates, other than safety
itself, between hospitals categorized as high-risk and those categorized as low-risk. Size
measures like number of beds and inpatient volumes are quite similar for both nursing and
surgery cases. Compared to the hospitals that are high-risk due to nursing safety, high-
risk hospitals by surgery safety standards are more likely to be non-profit or government
hospitals. Hospitals that have higher complications in nursing safety are also likely to have
higher surgery complications while, the same is not true for hospitals that are high-risk for
surgery safety.

14
Measures based on more than 1 standard deviation reduce the number of hospitals that can qualify as
a high-risk hospital to a relatively small group.

22
2.5.3 Entry Model

It is assumed that the entry decision depends on the volume that an ASC expects to
serve upon entry and the area-level safety records. I approximate the expected volumes to
be served by computing the volume predictions post-entry, using previous year’s predictions
and after assuming an hypothetical entrant in the market. In this way, the methodology
in this study is similar to Chernew et al. (2002), except that the focus here is on the entry
events by CCS type at the county level. There are a total of 221 entry events that took
place in Florida during 1998 to 2008, out of which 212 were in 34 urban counties (There are
a total of 67 counties in Florida.). This makes sense as ASC entrants are mostly for-profit
facilities that would tend to locate in well populated markets. Accordingly the sample is
restricted to these counties as potential entry locations.
To simplify the dynamic structure, it is assumed that the entry decision for time t is
made at time t − 1 and entrants use the existing market structure to predict the share of
patients they will get. A potential entrant will evaluate the expected volume of patients
to be served by locating in each zip code in a county and then choose the one that yields
the maximum number of patients. Each potential entry zip code in a county is determined
using the existing distribution of health care facilities i.e. hospitals and ASCs in the market.
Using this method helps to account for local zoning laws and also provides an estimate of
potential zip codes where an entrant can expect to attract patients from. For example,
if a hypothetical entrant decides to enter a particular zip code then it can form its own
expectation of the potential servable market size using the geographical distribution of
patients served at existing facilities (in the entry zip code). This will then give rise to a
number of zip codes that will be sending patients to the entrant’s zip code. Choice model
results are then used to estimate the probable market share that an entrant will get from
each of these zip codes.
Next, ASC facilities can be divided according to two major business models - entrants
that operate as multi-specialty facilities and provide almost all types of multi-level CCS
treatments under one roof. These are also more likely to be partly owned by corporate
chains like Symbion Healthcare, HCA etc. The second type are specialty facilities that
focus on one or two related CCS categories, for example, orthopedic specialties, eye care

23
specialists, urology centers etc. Likewise, different types of ASCs (by model and by service
type and locality) are likely to face different prospects in terms of number of competitors,
set up costs and predicted patient volumes in each market; accordingly, the entry decision
is modeled for each type of entrant. These four types of entrants account for the majority of
entry events over the 11 year period - Multi-Specialty, Nervous System & Musculoskeletal,
Eye Surgery, and Digestive System. These types account for 90% of all urban area entries
(193 out of 212). In 17 cases, two entrants serving the same CCS type enter at the same
location. These ties are treated as single-entry events which brings the number of entry
events down to 176.
The estimated parameters from the choice model provide a basis for predicting the
aggregate patient volume that a new entrant would achieve, across every zip code in its
potential service area. Estimated choice probabilities for new ASC entrants, as well as all
other facilities in the market, are estimated from the model, accounting for various patient
and facility level control variables and their interactions. These predicted probabilities
are interpreted as predicted market share. Combined with the size of the overall patient
population, the choice model thus provides a specific measure of the expected patient volume
an entrant would achieve at every possible location. The prediction for the multi-specialty
entrants is computed differently by switching the specialty status variable from 1 to 0, for
the entrant, and then aggregating the volumes from each estimated procedure type to get
the predicted volumes. The advantage of the prediction for the analysis of entry is that
predicted measures are derived from patients’ estimated probabilities of visiting a facility
rather than being based on the actual visits.
In constructing these estimates, the first step is to create weighted zip code level market
shares for each potential new facility-location as a prediction from the patient choice model.
Expected volume by entering in zip code j can be summarized as:
X
Vd
olj = pbz ∗ Nz (2.5)
z∈Ij

Here, pz is the choice probability of a potential entrant in zip code z and Nz is the number
of patients in the zip code.15 Summing up the expected volume over each zip code z that
15
Since the data provide access to the universe of patients treated in Florida hospitals and ASCs, the

24
an entrant will serve by entering zip code j provides the total expected volume. The set
of zip codes z ∈ Ij are referred to as the market service area of the entrant in zip code j.
Maximum volume from the entry zip codes in a particular county is chosen as the predictor
of entry at a county level.
Next, consistent with the measures of nursing and surgery safety used in the nested
logit choice model, variables for safety included in the entry model are also relevant to the
potential market service area of the entrant. Since the hypothesis is that the entry decision,
after accounting for the expected profitability, depends on the area-level safety, two different
area-level measures are used in the model - number of high-risk hospitals and the volume
of patients treated at the outpatient departments of high-risk hospitals. These measures,
like the predicted volumes, are defined by the market service area of the entrant.
The probability of an urban county witnessing a type a (where a = 1 to 4 for 4 different
types of entrants described before) ASC entry during any time period from 1998 to 2008,
conditional on the predicted volume, and county level hospital safety measures is modeled
with a standard probit:

a a
P (Entryc,t ) = Φ(β1 a ∗ V
d olc,t + β2 P SIc,t−1 + β3 time) (2.6)

I compute the probability of entry by different types by using dummies for each type in
the model interacted with the type specific variables. Here, a represents the entrant type,
while, V
d ol is the predicted volume by type. P SI stands for alternative area-level hospital
risk measures from previous year and time represents time dummies.
Our main interest in the entry model is interpreting how hospital safety measures may
drive physician groups to invest in setting up ASCs. Moreover, the effect of safety on entry
probability can be broken into two parts - direct and indirect. The direct effect is modeled
with the help of the entry probit, i.e. entry is a function of the overall safety in the area,
after controlling for the predicted patient volumes. The prevalence of high-risk hospitals in
the locality improves the prospects of a new ASC entrant indirectly by raising the entrant’s
predicted profitability. In addition, an indirect effect occurs because low prevailing safety
in the area may increase the attractiveness of a new non-hospital facility. When safety risks
actual patient numbers are used to create the zip code level population measure. This is better than using
Census numbers coinciding with our data period, as they are rough extrapolations based on the 2000 census.

25
are relatively stronger and recognized at local hospitals, the choice model predicts larger
shares for new entrant ASC facilities, and this in turn raises the entrants predicted patient
volume.

dPentry ∂Pentry ∂Pentry ∂ V

d ol
= + (2.7)
dP SI ∂P SI ∂ V ol
d ∂P SI
According to the above equation, it is expected that the total derivative of entry prob-
ability, with respect to safety measures, will be significant and positive. All three terms
on the right hand side of the equation are expected to be positive. The first term on the
right hand side is the effect of the increase in number of high-risk hospitals in the entrant’s
market service area on the entry decision given the predicted volume (i.e. marginal ef-
fects coefficient on safety measure in the probit model). The second partial is the effect
of predicted volume on entry (i.e. coefficient on predicted volumes in the probit) which
should be positive if volume drives entry; finally, the last partial is the increase in expected
ASC share (volume) due to a hospital being high-risk (its marginal effect in the patient
choice model), that again we expect to find positive. It is possible that any one of these
effects stand-alone is not strong enough but the overall impact, it is expected, should be an
increased probability of ASC entry.

2.6 Results
2.6.1 Nested Logit

In the first stage of the estimation, nested logit choice models are estimated using a
random sample from the entire state for each year. This sample includes observations
for three CCS types with 40000 cases each (approximately 10% of all cases). Separate
estimation is done for nervous system and musculoskeletal procedures (CCS 1), eye care
procedures (CCS 3) and digestive system procedures (CCS 9). Tables 2.6, 2.7, and 2.8
present these results. The models have been estimated using eleven years of Florida data
from 1997-2007 but for presentation purposes results are included from the selected years
of 1998, 2002 and 2006.
The primary coefficients of interest are those associated with the choice specific variables,
including safety rate measures, distance, specialty status of ASCs as well as the nest specific

26
variables like, age, race, severity (measured by number of diagnosis and procedures) are
always significant. Coefficients from the nest choice equation suggest that older people are
more likely to choose ASCs while males are more likely to choose hospitals over ASCs. White
patients also seem to be relatively more inclined towards hospital treatment for nervous
and digestive system surgeries. This result is different for eye care patients, and somewhat
mixed, as it seems that over the years the trend towards getting treatment at ASCs has
gotten more positive. This in itself may be a result of increasing acceptability of ASC
treatment over time among insurance companies. The signs on payer type coefficients also
suggest a similar trend. In period around 1998, Medicare as well as Commercial insurance
patients were less likely to choose ASCs over hospitals for surgery. However, over the next
few years, new ASC entry leading to more options, technological changes and lower prices
at ASCs gave rise to a more favorable trend for ASCs.
Patients with severe health status are also understandably more likely to choose hospi-
tals. The likelihood ratio (LR) test on the dissimilarity coefficients for hospital and ASC
nest shows that the coefficients are significantly different from each other for all the es-
timated models and therefore this test supports the decision to nest hospitals and ASCs
separately. For consistency with stochastic utility maximization, the dissimilarity coeffi-
cients are required to lie within the unit interval (Daly-Zachary-McFadden conditions). In
the early years of 1997 and 1998 these coefficients take a value much greater than 1 in some
of the CCS type estimations. These estimates are reconciled with utility maximization in
an appendix to this dissertation (appendix A).
Since the nested logit model has been estimated with various interactions and results
in coefficient estimates, to get a clearer understanding of the precise impact of important
variables, it is important to look at the marginal effects. Past literature estimating nested
logit models (Brown and Theoharides (2009), Sahn et al. (2003), Puig-Junoy et al. (1998),
Cameron (1985)) has always relied on presenting coefficient estimates and their standard er-
rors along with either elasticities or marginal effects, formulae for which are mathematically
not very different from each other. 16 The literature has relied on the fact that significance
on coefficient estimates also applies to the resulting estimates of marginal effects and elas-
ticities. This study also follows the previous literature in presenting coefficients to show
16
Using the terminology from section 3.4, marginal effect of a change in variable X on choice probability

27
significance and marginal effects for measuring the impact. The marginal effects of the co-
variates, expressed as percentage point change in choice probability, for selected regressors
are presented in Table 2.9, separated by the CCS type as well as the nest. For the hospital
nest distance, nursing safety, and surgery safety17 are examined while, for the ASC nest,
impact of specialty status, distance from the patient, and distance to the emergency center
are presented. These are basically average marginal effects as first an analytical derivative
is computed and then it is averaged over all observations.18 Also, the table presents average
effects across the prediction years of 1997-2007 along with the variation in signs on the coef-
ficients reflected by number of positives and negatives (in 11 regressions for each CCS type).
The average marginal effects suggest that distance always has a negative impact on choice
probabilities (see figure 2.4), but the impact is relatively bigger for musculoskeletal patients
and digestive system patients, a fall in market share of between 0.4 to 0.6 percentage points,
compared to that of eye care where choice probability falls by 0.2 to 0.4 percentage points.
The result makes sense as eye care surgeries are more specialized in nature and patients may
be more likely to travel far for the treatment. Patients are also more sensitive to distance
in case of eye surgeries at ASCs than at hospitals. Distance to the emergency center is not
important in case of eye surgeries, perhaps because these surgeries are less risky than the
other ASC procedures. Marginal effects also show that patients are highly likely to visit a
specialty ASC for eye and digestive system procedures but not so much for the nervous and
musculoskeletal system procedures.
Turning to the effect of the main variables of nursing safety and surgery safety in Ta-
ble 2.9, results suggest that being a high-risk hospital (safety rate one standard deviation
above the choice set average) has a strong negative impact on choice probabilities and there-
fore, hospital market share. The impact is stronger if the hospital is high-risk due to low
Pij , can be written as:
1 − Pij|Bk
 
∂ Ûij
M Eij = + (1 − PiBk ) Pij|Bk Pij (2.8)
λk ∂Xj
Elasticity is not very different and can be expressed as:

1 − Pij|Bk
 
∂ Ûij
Elasticityij = + (1 − PiBk ) Pij|Bk Xj (2.9)
λk ∂Xj

17
As mentioned before safety rates in the Nested logit Model enter as dummy for a high-risk hospital.
18
Since, coefficients from the nested logit are in general all significant at 1% or 5%, it is safe to assume
that the marginal impact is significant too.

28
nursing safety than surgery safety and is also stronger for nervous system and digestive
system procedures. This again seems to be a reflection of the inherent higher relative risk
of procedures that are in general carried out under the nervous system (eg: Arthroscopy,
Laminectomy, Insertion of catheter) and digestive system (eg: Colonoscopy & Biopsy, Her-
nia repair) categories as against those carried out under eye care (eg: Lens & Cataract
procedures, Therapeutic procedures on eye lids). The choice probability for a hospital that
is high-risk for nursing safety can be lower by anywhere between 1 to 2 percentage points
while, for a hospital that is high-risk due to surgery complications, the probability is lower
by 0.5 to 1 percentage point. Even when the impact is of the opposite sign, in a few cases
for the surgery measure, than expected (positive), the average effect for Medicare patients
is still negative, highlighting the fact that freedom of choice available to Medicare patients
allows them to pick the best hospital. In case of eye procedures, however, the surgery
measure takes an unexpected sign around half of the times.
These results further suggest that if a hospital is high-risk by both nursing and surgery
safety standards then the overall impact on patient choice will be consistently negative.
Also, since patients overwhelmingly prefer the closest facility to their residence (see figure
2.4), it is of interest to examine the negative effect on the market share of the closest hospital
to the patient if it is high-risk in both nursing and surgery safety. To analyze this effect, one
can simulate the marginal impact on the closest hospital if it was the only hospital in the
choice set that was high-risk due to both high nursing and surgery related complications.
Looking at the change in choice probability of the closest hospital when it goes from being
a low-risk hospital to a high-risk hospital in terms of safety performance, while holding
all other characteristics of the facility and the patient constant. The marginal impact is
analyzed for choice set sizes that vary by the number of hospitals and ASCs in them.
This impact is presented in Table 2.10 for all three CCS types and on average it ranges
between 3 − 6 percentage point loss in share for the closest hospital, for the musculoskeletal
procedures, 1 − 8 percentage point for eye care procedures and between 2 − 8 percentage
point for digestive system procedures. These results point out the significant threat of
ASCs to hospitals. When the closest hospital becomes a high-risk hospital, its market share
loss can translate into a loss for the entire hospital nest with ASCs gaining a significant

29
proportion. The closest ASC also stands to gain more than the second closest hospital.
This effect seems to be stronger during the period of 1997 to 2004. For instance, in 1998
an approximate loss of 9 percentage point by the closest hospital in the eye surgery market
translated into a 3 percentage point average gain for the closest ASC. Further, roughly 90%
of the share lost by the closest hospital turned into share gain for all ASCs in the patient
choice set (8 percentage point gain for ASC nest). ASCs also stand to gain approximately
90% of what hospitals lost in other years too in case of eye procedures. Similar calculations
suggest that ASCs stand to gain more than 60% of what hospitals lose in case of nervous
system procedures and more than 50% in case of digestive system procedures. Also, only
in the case of digestive system procedures does it appear that the second closest hospital
stands to gain as much as the second closest ASC, when the closest hospital turns high-risk.
The significance of the impact is also clear from figure 2.5 that plots the loss in share for
all three CCS types, for the closest hospital, when it becomes a high-risk hospital (high
nursing and surgery complications) from a low-risk hospital, for selected years.

2.6.2 Entry Model

For the second stage, results from the nested logit estimation are utilized to construct
out of sample choice predictions by the CCS type for a hypothetical entrant with average
characteristics in each of the potential entry zip codes. Potential entry zip codes themselves
were chosen based on the location decisions of existing health care facilities in Florida
including incumbent hospitals and ASCs. These facilities are restricted to eligible areas set
by the local level zoning laws. Using these predictions an estimate of the expected volume
measure is constructed by CCS type, county and year to include as regressors in the entry
model.
The entry event is modeled for four types of entrants - specialty entrants in three of the
CCS types discussed above and a multi-specialty entrant providing all three of the CCS
procedures at one facility. Specialty status of an ASC was controlled for in the nested
logit estimation and allows for simulating an entry event by a specialty or a multi-specialty
facility. To model multi-specialty ASC entry decision, the predicted volumes are summed
up in each of the three CCS type predictions and it yields an overall estimate of volume

30
that the entrants expect to serve. Predicted volume by type is entered in the model by
interacting it with type dummies.
The entry model results are presented in Table 2.12. It includes results from selected
specifications of the Probit model. These specifications differ by the safety measure used.
While models (2) and (3) use nursing related complications to define high-risk hospitals,
models (4) and (5) use surgery complications. For ease of interpretation and to provide
an understanding of the magnitude of the results, variables are entered as standardized
scores (z-scores) and coefficients represent the impact of one standard deviation change in
the dependent variable on the entry probability. Descriptive statistics for the entry model
variables are given in Table 2.11. Time fixed effects were included in other specifications,
but the results did not change qualitatively. Hence, those results are not included here.
The first Probit, model (1), presents baseline results including only the predicted vol-
umes. Predicted volume measures are significant drivers of entry for all four types with a
positive impact on the entry decision that is stronger for entrants in orthopedic (nervous
and musculoskeletal) and digestive system specialties. However, when the standardized
measures are used, as presented here, the predicted volume for the eye surgery entrants is
not significant. In model (2), variable measuring the number of high-risk hospitals (based
on nursing safety) is included and found to have a significant impact on the entry proba-
bility.19 The marginal effect (change in probability) for this variable suggests that a one
standard deviation increase in the number of high-risk hospitals located in the potential
market service area of an entrant leads to a 2.4 percentage point increase in the probability
of entry by an ASC. The alternative measure of safety based on the surgery complications,
shown in model (4), suggests a 3.5 percentage point increase in the entry probability. Pre-
dicted volume measures are also significant with the exception of eye surgery entrants. The
magnitude of the safety surgery measure is stronger than the nursing safety measure and is
significant with 0.01 p-value. Given the mean number of high-risk hospitals in the market
19
Although, market size has been accounted by the predicted volume measures as they are constructed
using zip code populations, it is still useful to check if the number of high-risk hospitals is positively correlated
with the number of all hospitals and therefore, indicating a residual impact of market size on entry. Appendix
B checks for this effect by including another variable in model (2) & (4) that controls for all hospitals in
the market service area of the entrant. As a result the nursing measure loses significance while the surgery
measure is still significant but with a somewhat lower magnitude than before.

31
service area (see Table 2.11), a one standard deviation change implies on average 5 more
hospitals in the area becoming high-risk by surgery safety standards. 20

Models (3) and (5) account for an alternative variable that measures the direct impact
of potential market that an entrant stands to gain and that is being treated at the high-
risk hospitals. The measure is based on the number of patients treated in the outpatient
departments of high-risk hospitals with their actual residence under the service area of a
potential entrant. Estimated marginal effects in both specifications suggest a strong and
significant impact of a change in this volume on the probability of entry. The predicted entry
probability increases by 2.8 percentage point when more patients get treated at hospitals
that are high-risk due to surgery complications. Based on the average number of patients
treated (Table 2.11) , this translates into a 2.8 percentage point increase in entry probability
when an additional 8000 patients get treated at high-risk hospitals (for surgery safety).
Given that on average there is a 11% probability of an ASC entry across the years, the
marginal effect of a change in safety measures on entry probability, ranging between 2 to
3.5 percentage points, is quite substantial and reflects an almost 30% increase over the
baseline probability.
Overall, the results in this study provide strong evidence for the impact of both nursing
and surgery safety at hospitals on ASC entry growth. However, this impact seems to vary
by the perspective of the agents involved i.e. patients and ASC entrants. In the patient
choice model, evidence for the impact of high-risk for nursing safety hospitals is stronger
than the impact of high-risk for surgery safety hospitals. As shown in Table 2.5, hospitals
need not be high-risk in both nursing and surgery safety at the same time and quite likely
they may be compensating for deficiency in one measure, say surgery safety, by doing better
on the other, nursing safety. The marginal effects from the estimation suggest a very strong
impact of being a high-risk nursing safety hospital on patient choice (Table 2.9) that can
go up to as large as 7 percentage point lower likelihood of being chosen (when the patient
is covered by a commercial insurance provider). Further, the simulated marginal effects on
20
Although the predicted volume measures in the entry model are already accounting for the market size,
the impact of the number of high-risk hospitals variable may still be correlated with the impact of overall
market size. Therefore, for robustness of the results, appendix B also includes the overall number of hospitals
in the regression. The impact in case of surgery safety still remains significant, while in the case of nursing
safety measure it is washed out.

32
hospital shares (Table 2.10) suggest a strong threat to hospitals based on their geographical
location and safety performance, especially in the early years of ASC growth. For instance,
if all the other characteristics of the hospital choices faced by a patient were the same, then
upon turning into a high-risk hospital by both nursing and surgery safety standards, in
some areas the closest hospital could lose up to 12 percentage point of it’s market share (in
bigger markets, in year 2002, see figure 2.5).
The entry model provides alternative evidence about the importance of the relationship
between low safety and the potential market for an entrant. ASCs are profit driven ventures
and areas with low safety will be viewed by entrants as attractive potential markets only
when a substantial volume of patients are being treated at the high-risk hospitals. The
results from the entry model seem to suggest that both nursing and surgery safety at the
area level are significant predictors of ASC entry and that the number of high-risk surgery
safety hospitals has a relatively stronger impact.

2.7 Conclusions

For insured patients, characteristics of health care other than charges play a more im-
portant role in the decision of where to have a surgical procedure performed. Attributes
such as distance of the patient from the provider are a major factor affecting the patient’s
selection of a provider. Moreover, with an increase in hospital acquired infections (e.g.,
MRSA) patients and their doctors’ concerns about safety and quality of care in hospitals
play an increasingly important part in the surgery setting decision. The relative role and
importance of such factors is an indicator of patient welfare and needs to be evaluated
thoroughly to gain a better understanding of competitive effects of ASCs.
The empirical findings in this study reveal that hospital quality and safety record is
inherently a factor in the emergence of ASCs. First, poor safety performance makes the
traditional hospital less attractive to patients and their doctors, opening up opportunities
for ambulatory surgery centers to attain market share. This impact is strongest when the
closest hospital to the patient is a high-risk hospital. In such a scenario, all ASCs located
close to the patient stand to gain. Second, the results further suggest that when poor
safety records are recorded for hospitals in the market service area of the entrant, there

33
is a significant positive effect on the rate of new entry of ASCs. This effect is higher in
magnitude and consistent in significance when the hospitals are low performers on surgery
safety. These results capture a previously unexplored, but key parameter of the rivalry that
goes on between these diverse facilities which is the measurable role of safety. Opponents
of ASCs have argued that the entry is driven solely by profit motives. We find that even
though profits play an important role, given that predicted volumes have a significant effect
on entry, it is also important to account for the fact that the entry decision is dependent
on the area level safety and therefore has positive implications for patient welfare due to
increased convenience and likely better quality of specialized ASCs.
The results support the hypothesis that physicians practicing at high-risk hospitals can
disassociate themselves from safety problems by moving their practice to ASCs. Specifi-
cally, when surgery complications at the hospitals reflect poorly on the practicing physi-
cians/surgeons reputation, they may be more inclined to invest in their own ASCs.
A possible limitation to the entry model exists in cases where hospitals react to ASC en-
try by adjusting safety covariates. Although the safety measures used in the entry model are
from the previous year, there could still be endogeneity due to possible negative correlation
of a hospital’s safety measures across the years. This problem is likely if hospitals can adjust
their risk status by making strategic investments that have a direct impact on safety and
that take effect gradually over years. For instance, hospitals can invest in hiring and training
more nursing staff and increase the number of nurses per patient at the hospital to reduce
nursing complications. Similarly, investments can be made in technology and infrastructure
to minimize surgical errors. Barro et al. (2006) studied the effects of competition on costs
and quality of medical care, in the context of specialty hospitals and they assumed that
there are exogenous changes in hospital quality as a result of entry instead of endogenous,
strategic reactions to entry or the threat of entry by incumbent hospitals. To the hospital,
deliberate improvements in the quality or safety of care may enhance reputation, boost
patient share in the market, and mitigate the threat posed by this new competition. Lack
of accounting for hospital response may affect the measurement of safety impact on ASC
entry and in this case may lead to estimates that are biased downward. 21 Future research
21
The bias is likely to be downward because the unobserved hospital response (omitted variable) is likely

34
can address this issue with the help of instrumental variables (IVs) that are correlated with
the safety at the hospital but are not likely to impact the ASC entry decision. Some of the
financial measures for the hospital are likely candidates for IV techniques. Investments by
the hospital in nursing intensity, for example, are going to affect the likelihood of a hospital
being high-risk for nursing complications but are not likely to affect the ASC entry and can
therefore be used to account for endogenous hospital response.

to follow a period of poor safety performance and will therefore be positively correlated with the safety
measures in the entry model but negatively correlated with the entry probability itself.

35
 2.8 Tables and Figures

Table 2.1: Outpatient facilities in Florida, 1997-2008

Outpatient Care 1997 2001 2004 2008

ASCs 218 244 302 365
Hospital Outpatient 199 197 200 208

Figure 2.1: Changes in Physician Practice Setting, Florida 1997-2008

36
Table 2.2: Single level CCS procedures and ASC Share (%) among Medicare, Florida

CCS Procedure 2001 2003 2006

Colonoscopy and Biopsy(76) 21670 30151 37485
(50) (60) (71)
Insertion of catheter/spinal stimulator(5) 9741 12730 16254
(55) (63) (77)
Decompression peripheral nerve(6) 976 1139 1420
(48) (49) (60)
Other OR therapeutic nervous system procedure(9) 454 599 1224
(31) (33) (57)
Plastic procedures on nose(28) 90 113 130
(42) (50) (56)

37
 Table 2.3: Patient choice of provider, Florida 1997-2006

Procedure(CCS code)& Facility type 1997 2000 2003 2006 Growth

Laminectomy, excision intervertebral disc(3)
Inpatient 5129 4619 4352 3903 (-24%)
ASC 228 385 488 904 (296%)
Hospital Outpatient 373 576 825 1049 (181%)
Total 5730 5580 5665 5856 (2%)
Insertion of catheter/spinal stimulator(5)
Inpatient 979 977 1013 679 (-31%)
ASC 10684 22494 31176 39922 (274%)
Hospital Outpatient 12982 19057 16141 10308 (-21%)
Total 24645 42528 48330 50909 (107%)
Tracheoscopy and laryngoscopy with biopsy(35)
Inpatient 628 469 400 328 (-48%)
ASC 207 256 222 244 (18%)
Hospital Outpatient 877 1283 1965 1761 (101%)
Total 1712 2008 2587 2333 (36%)
Colonoscopy and biopsy(76)
Inpatient 3997 4751 5219 5044 (26%)
ASC 19821 41173 71914 103083 (420%)
Hospital Outpatient 27979 46963 50212 41189 (47%)
Total 51797 92887 127345 149316 (188%)
Endoscopy/endoscopic biopsy, urinary tract(100)
Inpatient 654 554 548 452 (-31%)
ASC 2007 3001 3791 4512 (125%)
Hospital Outpatient 2324 2491 2274 2273 (-2%)
Total 4985 6046 6613 7237 (45%)

38
 Table 2.4: Variables Used in the Nested Logit Estimation

Variable Definition

Wik , Xi Male: dummy indicating gender

White: dummy indicating race
Age: patient age at admission
Number of diagnoses: number of other procedures
Number of procedures: number of other diagnoses
Payer- Medicare: patient insured by Medicare or Medicare-HMO
Payer- Medicaid: patient insured by Medicaid or Medicaid-HMO
Payer-Commercial: patient insured by Commercial insurance, HMO or PPO
Payer-Other/Self: patient insured by other State/Local Govt insurance or Self pay

Rj , k = hospital For Profit, Not for Profit, Government: dummy indicating hospital control type

Teaching: dummy for teaching hospital

Beds: total number of beds available at the facility
Nursing intensity: nursing hours divided by patient days

Capital intensity: dollar value of capital asset divided by inpatient days,

(include land, land improvement, buildings, fixed equipment, leasehold
improvement, movable equipment, construction in progress)

Nursing Safety Index (PSIN ursing): dummy for safety relating to nursing care,
(enters as an indicator for a high-risk nursing safety hospital)

Surgery Safety Index (PSIS urgery): dummy for safety relating to surgical care,
(enters as an indicator for a high-risk surgery safety hospital)

Rj , k = ASC Distance to emergency center: distance in miles to the closest emergency center
Beds: total number of beds available at the facility
Specialty: dummy indicating if ASC specializes in the CCS type

Distij Distance: travel distance between patient zip code centroid and facility

Note: Distance is interacted with all patient demographics and facility characteristics other than safety.
Surgery and Nursing rates are interacted only with payer type and control type of hospital.

39
Figure 2.2: Unique Zip Code Choice Sets, 2002 (Digestive System)

40
Figure 2.3: Excess Nursing and Surgery Complications, Florida Hospitals 1997-2008

41
Table 2.5: Average Summary Statistics - High-Risk and Low-Risk Hospitals, 1997-2008

Nursing Complications Surgery Complications

High-Risk1 Low-Risk2 High-Risk Low-Risk
Variables Mean SD Mean SD Mean SD Mean SD
Excess Nursing Complications Rate3 2.78 3.83 0.12 2.73 0.44 3.22 0.42 3.13
Excess Surgery Complications Rate3 0.32 1.49 0.16 1.48 1.30 1.42 0.04 1.22
Beds 330 242 286 213 330 260 283 205
Inpatient Volume 14665 12315 13278 11103 14080 11135 12056 9035
For Profit (%) 0.42 0.49 0.42 0.49 0.35 0.48 0.43 0.50
Non Profit (%) 0.47 0.50 0.48 0.50 0.53 0.50 0.48 0.50
Government (%) 0.11 0.31 0.09 0.29 0.12 0.32 0.09 0.29
Teaching (%) 0.11 0.31 0.06 0.24 0.09 0.28 0.06 0.24
Nursing Intensity4 0.04 0.05 0.05 0.08 0.05 0.06 0.05 0.07
Capital Intensity5 1.39 3.49 1.42 2.47 1.61 4.07 1.39 2.44
Average N6 81 183 70 187
Notes:
1
High-Risk hospital - Hospital with a safety rate greater than one standard deviation above the
average safety rate in the patient choice set.
2
Low-Risk hospital - Hospital with a safety rate less than one standard deviation above the
average safety rate in the patient choice set.
3
Excess rate of complications per 1000 patients.
4
Nursing hours divided by patient days.
5
Dollar value of capital asset divided by inpatient days.
6
Average number of High Risk and Low Risk hospitals across 1997 to 2008.

42
Table 2.6: Nested Logit Results - Nervous System and Musculoskeletal Patients

1998 2002 2006

Explanatory Variable Coeff. S.E. Coeff. S.E. Coeff. S.E.
Bottom Level Equation
Hospital Specific
Nursing Safety 0.955a 0.165 -0.800a 0.161 0.568a 0.114
Nursing Safety interacted with:
a a
Medicare -1.078 0.136 0.045 0.128 -0.371 0.099
Medicaid -0.44b 0.199 0.248 0.162 -0.224c 0.137
Commercial -1.609a 0.139 -0.237c 0.127 -0.597a 0.100
For Profit Hospital -0.225c 0.131 0.946a 0.125 -0.338a 0.082
Not For Profit Hospital -0.518a 0.129 0.516a 0.118 -0.204a 0.078
Surgery Safety 0.160 0.152 0.164 0.131 0.655a 0.103
Surgery Safety interacted with:
Medicare -0.360a 0.127 -0.86a 0.117 -1.004a 0.098
Medicaid -0.200 0.200 -0.248c 0.148 -0.233c 0.125
Commercial -0.322a 0.126 -0.593a 0.113 -0.974a 0.096
For Profit Hospital 0.080 0.121 0.865a 0.103 0.020 0.084
Not For Profit Hospital 0.308a 0.110 0.321a 0.093 0.282a 0.067
Teaching Hospital -0.651a 0.064 -0.366a 0.039 -0.289a 0.037
Nursing Intensity -2.056a 0.741 -0.465a 0.143 -1.455 2.425
Distance * Nursing Intensity -0.038 0.034 0.039a 0.005 0.222 0.147
Capital Intensity -0.093a 0.026 0.019b 0.009 0.022a 0.008
Distance * Capital Intensity 2.25E-04 0.002 -0.002a 4.06E-04 -0.001c 0.001
For Profit Hospital 0.229a 0.052 -0.697a 0.047 -0.225a 0.039
Not For Profit Hospital 0.736a 0.051 -0.147a 0.035 0.024 0.036
ASC specific
Specialty ASC -1.09a 0.054 -0.391a 0.035 -0.021 0.026
Distance to Emergency Center -0.17a 0.008 -0.032a 0.006 0.007 0.005
Distance * ASC Specialty 0.02a 0.002 0.016a 0.002 0.001 0.002
Common
Beds 0.003a 9.45E-05 0.001a 6.27E-05 0.001a 5.92E-05
Distance -0.007 0.006 0.003 0.006 0.015a 0.004
Distance interacted with:
Beds -2.3E-05a 3.87E-06 -5.70E-06b 2.40E-06 -9.00E-06a 2.24E-06
Age -0.001a 8.38E-05 -0.001a 7.26E-05 -0.001a 6.23E-05
Male 0.003 0.003 0.003 0.002 0.01a 0.002
White -0.002 0.003 -0.001 0.003 -0.01a 0.002
Number of procedures -0.003a 0.001 3.83E-05 3.88E-04 1.11E-04 3.25E-04
Number of diagnosis 0.001 0.001 -0.004a 0.001 -0.001a 3.07E-04
Medicare -0.019a 0.005 -0.017a 0.005 -0.020a 0.004
Medicaid -0.018a 0.007 -0.009c 0.006 -0.015a 0.005
a
Commercial -0.010 0.004 -0.011a 0.004 -0.011a 0.003
Top Level Equation (Base - ASC)
Age -0.013a 0.001 -0.008a 0.001 -0.008a 0.001
Male 0.109a 0.028 0.096a 0.023 0.062a 0.022
White 2.208a 0.033 0.243a 0.032 0.100a 0.03
Medicare 2.385a 0.063 -0.072 0.074 0.085 0.069
Medicaid 3.027a 0.109 0.072 0.091 0.289a 0.093
Commercial 2.621a 0.058 -0.523a 0.068 -0.37a 0.064
Number of procedures 0.281a 0.009 0.038a 0.004 0.015a 0.004
Number of diagnosis 0.509a 0.013 0.267a 0.007 -0.021a 0.004
Dissimilarity parameters
Hospital 1.869a 0.054 1.300a 0.040 1.073a 0.035
ASC 1.235a 0.034 1.131a 0.031 1.055a 0.029
p value from LR test for IIA () 0.000 0.000 0.000
Upper limits on the dissimilarity coefficients, for hospital nest, implied by Börsch-Supan conditions are
are 2.31, 1.66 and 1.39 for years 1998, 2002 and 2006, respectively.
Upper limits on the dissimilarity coefficients, for ASC nest, implied by Börsch-Supan conditions are
are 1.53, 1.31 and 1.43 for years 1998, 2002 and 2006, respectively.
a
p-value < 0.01
b
p-value < 0.05
c
p-value < 0.10

43
 Table 2.7: Nested Logit Results - Eye Surgery Patients

1998 2002 2006

Explanatory Variable Coeff. S.E. Coeff. S.E. Coeff. S.E.
Bottom Level Equation
Hospital Specific
Nursing Safety -0.250 0.274 -1.954a 0.309 -0.253 0.177
Nursing Safety interacted with:
Medicare -1.139a 0.207 0.805a 0.221 0.336a 0.130
a
Medicaid -1.122 0.365 0.862a 0.302 0.881a 0.193
Commercial -1.691a 0.235 0.431c 0.237 0.328b 0.140
For Profit Hospital 0.712a 0.225 1.362a 0.245 -0.007 0.153
Not For Profit Hospital 1.058a 0.218 0.581a 0.231 -0.286b 0.145
Surgery Safety -2.794a 0.299 -1.667a 0.277 0.752a 0.125
Surgery Safety interacted with:
Medicare 0.436c 0.233 -0.504a 0.199 -0.030 0.102
Medicaid -0.154 0.384 -0.125 0.289 -0.667a 0.188
Commercial 0.955a 0.242 -0.208 0.213 -0.306a 0.116
For Profit Hospital 1.426a 0.223 1.15a 0.256 -1.01a 0.146
Not For Profit Hospital 2.51a 0.214 2.027a 0.240 -0.59a 0.102
Teaching Hospital -1.979a 0.111 -0.548a 0.068 0.357a 0.047
Nursing Intensity 9.841a 0.992 2.75a 0.202 0.315 2.799
Distance * Nursing Intensity 0.072 0.046 -0.015a 0.006 0.213 0.158
Capital Intensity 0.032 0.023 0.013c 0.007 0.066a 0.010
Distance * Capital Intensity -1.5E-05 0.001 0.001a 2.71E-04 -1.55E-04 0.001
For Profit Hospital -0.554a 0.073 -0.577a 0.067 0.100 0.077
Not For Profit Hospital -0.071 0.065 -0.097c 0.055 0.692a 0.076
ASC specific
Specialty ASC 0.625a 0.025 1.002a 0.031 1.058a 0.032
Distance to Emergency Center 0.027a 0.004 -0.008c 0.005 0.015a 0.004
Distance * ASC Specialty 0.021a 0.001 0.032a 0.002 0.022a 0.001
Common
Beds 0.003a 1.06E-04 0.001a 8.54E-05 8.28E-05 7.17E-05
Distance -0.009c 0.006 0.015b 0.006 0.005 0.005
Distance interacted with:
Beds -3.5E-05a 4.77E-06 8.34E-06b 3.51E-06 9.97E-06a 3.19E-06
Age -4.89E-04a 6.98E-05 -4.06E-04a 7.97E-05 -2.34E-04a 6.65E-05
Male 0.004a 0.001 0.006a 0.002 0.001 0.002
White -0.039a 0.002 0.007b 0.003 0.001 0.002
a a
Number of procedures -0.008 0.001 0.001 4.32E-04 0.003a 3.67E-04
Number of diagnosis -0.003a 0.001 -0.004a 0.001 -0.003a 3.82E-04
Medicare 0.028a 0.003 -0.051a 0.004 -0.037a 0.003
Medicaid 0.013b 0.006 -0.02a 0.006 -0.012b 0.005
a a
Commercial 0.044 0.004 -0.032 0.004 -0.034a 0.003
Top Level Equation (Base - ASC)
a a
Age -0.022 0.001 -0.027 0.001 -0.035a 0.001
Male -3.74E-04 0.032 0.091a 0.032 0.116a 0.033
White 1.846a 0.039 -0.142a 0.045 -0.162a 0.042
Medicare 1.386a 0.075 -0.378a 0.083 -0.266a 0.072
Medicaid 2.169a 0.124 -0.058 0.122 -0.151 0.112
Commercial 1.678a 0.080 -0.634a 0.084 -0.631a 0.073
a a
Number of procedures 0.516 0.015 0.133 0.007 0.089a 0.007
Number of diagnosis 1.008a 0.019 0.478a 0.011 0.064a 0.007
Dissimilarity parameters
Hospital 1.841a 0.059 1.253a 0.045 0.790a 0.027
ASC 1.011a 0.03 1.285a 0.028 1.110a 0.026
p value from LR test for IIA () 0.000 0.000 0.000
Upper limits on the dissimilarity coefficients, for hospital nest, implied by Börsch-Supan conditions are
are 1.55, 1.35 and 1.23 for years 1998, 2002 and 2006, respectively.
Upper limits on the dissimilarity coefficients, for ASC nest, implied by Börsch-Supan conditions are
are 2.81, 3.03 and 3.16 for years 1998, 2002 and 2006, respectively.
a
p-value < 0.01
b
p-value < 0.05
c
p-value < 0.10

44
 Table 2.8: Nested Logit Results - Digestive System Patients

1998 2002 2006

Explanatory Variable Coeff. S.E. Coeff. S.E. Coeff. S.E.
Bottom Level Equation
Hospital Specific
Nursing Safety 0.381a 0.120 -0.83a 0.142 0.047 0.102
Nursing Safety interacted with:
Medicare -0.631a 0.104 0.225c 0.119 -0.004 0.091
a a
Medicaid -0.479 0.149 0.525 0.147 -0.172 0.122
Commercial -1.128a 0.110 -0.035 0.117 -0.196b 0.090
For Profit Hospital 0.226a 0.091 0.337a 0.099 -9.3E-05 0.067
Not For Profit Hospital 0.215a 0.087 0.524a 0.094 -0.268a 0.069
Surgery Safety 0.303a 0.112 0.668a 0.107 0.312a 0.085
Surgery Safety interacted with:
Medicare -0.566a 0.099 -1.338a 0.105 -0.529a 0.082
Medicaid -0.706a 0.155 -0.378a 0.126 -0.328a 0.105
Commercial -0.436a 0.098 -1.187a 0.100 -0.659a 0.081
For Profit Hospital -0.085 0.083 0.555a 0.082 0.101 0.068
Not For Profit Hospital 0.311a 0.075 0.204a 0.075 0.522a 0.058
Teaching Hospital -0.967a 0.056 -0.816a 0.041 -0.049c 0.029
Nursing Intensity -8.791a 0.643 -0.496a 0.113 -32.924a 4.777
Distance * Nursing Intensity 0.042 0.032 0.036a 0.004 1.182a 0.184
Capital Intensity 0.028 0.019 -0.020 0.015 0.051a 0.007
Distance * Capital Intensity 0.005a 0.001 0.005a 0.001 -2.83E-04 3.77E-04
For Profit Hospital -0.082b 0.037 -0.292a 0.031 -0.187a 0.034
Not For Profit Hospital 0.385a 0.036 -0.010 0.027 0.082a 0.031
ASC specific
Specialty ASC 0.777a 0.029 0.567a 0.024 0.773a 0.025
Distance to Emergency Center -0.020a 0.006 -0.049a 0.004 -0.043a 0.004
Distance * ASC Specialty 0.023a 0.002 0.015a 0.001 0.008a 0.001
Common
Beds 0.002a 8.88E-05 0.001a 5.75E-05 0.001a 4.82E-05
Distance -0.013a 0.005 0.012a 0.005 0.032a 0.004
Distance interacted with:
Beds -1.18E-05a 2.93E-06 0a 2.43E-06 -1.12E-05a 2.41E-06
Age -0.001a 6.81E-05 -0.001a 6.95E-05 -0.001a 5.97E-05
Male 0.007a 0.002 0.007a 0.002 0.001 0.001
White -0.034a 0.002 -0.01a 0.002 -0.013a 0.002
a a
Number of procedures -0.006 0.001 -0.001 3.50E-04 3.31E-07 3.03E-04
Number of diagnosis -0.002a 0.001 -0.005a 0.001 -0.004a 3.67E-04
Medicare 0.019a 0.004 -0.034a 0.004 -0.038a 0.003
a
Medicaid -0.002 0.007 -0.021 0.005 -0.023a 0.004
Commercial 0.015a 0.004 -0.024a 0.004 -0.035a 0.002
Top Level Equation (Base - ASC)
Age -0.017a 0.001 -0.021a 0.001 -0.022a 0.001
Male 0.014 0.028 0.033 0.022 0.066a 0.022
a
White 1.621 0.033 0.022 0.030 0.209a 0.027
Medicare 1.445a 0.067 -0.499a 0.093 0.005 0.068
Medicaid 2.515a 0.119 -0.208c 0.119 0.597a 0.094
Commercial 1.738a 0.064 -0.806a 0.090 -0.471a 0.064
Number of procedures 0.604a 0.015 0.095a 0.005 0.035a 0.004
a a
Number of diagnosis 0.745 0.014 0.337 0.008 -0.03a 0.004
Dissimilarity parameters
Hospital 1.202a 0.044 0.944a 0.032 0.768a 0.029
ASC 0.755a 0.023 0.643a 0.019 0.735a 0.020
p value from LR test for IIA () 0.000 0.000 0.000
Upper limits on the dissimilarity coefficients, for hospital nest, implied by Börsch-Supan conditions are
are 2.41, 1.63 and 1.27 for years 1998, 2002 and 2006, respectively.
Upper limits on the dissimilarity coefficients, for ASC nest, implied by Börsch-Supan conditions are
are 1.49, 1.34 and 1.58 for years 1998, 2002 and 2006, respectively.
a
p-value < 0.01
b
p-value < 0.05
c
p-value < 0.10

45
 Table 2.9: Average Marginal Effects (% point change) - Nested Logit Results

Average, 1997-2007 Min Max Positives Negatives

Nervous & Musculoskeletal
Hospital Nest
High-Risk (Surgery) -0.66 -2.81 0.51 3 8
High-Risk (Surgery), Medicare -1.62 -3.65 -0.05 0 11
High-Risk (Nursing) -1.86 -3.66 -0.18 0 11
High-Risk (Nursing), Medicare -1.51 -4.36 0.55 2 9
Distance from the patient -0.39 -0.52 -0.33 0 11
ASC Nest
Specialty -2.44 -6.65 1.40 2 9
Distance from the patient -0.41 -0.48 -0.37 0 11
ASC Distance to Emergency Center -0.30 -1.29 0.33 4 7
Eye Surgery
Hospital Nest
High-Risk (Surgery) -0.55 -2.51 1.24 4 7
High-Risk (Surgery), Medicare -0.56 -3.12 1.15 5 6
High-Risk (Nursing) -0.89 -1.76 0.49 1 10
High-Risk (Nursing), Medicare -0.68 -1.82 0.64 2 9
Distance from the patient -0.13 -0.18 -0.06 0 11
ASC Nest
Specialty 15.97 12.86 18.85 11 0
Distance from the patient -0.38 -0.50 -0.15 0 11
ASC Distance to Emergency Center 0.11 -0.31 0.48 8 3
Digestive System
Hospital Nest
High-Risk (Surgery) -0.65 -1.89 0.54 1 10
High-Risk (Surgery), Medicare -1.18 -3.35 0.83 1 10
High-Risk (Nursing) -1.99 -3.37 -0.51 0 11
High-Risk (Nursing), Medicare -1.29 -3.52 -0.16 0 11
Distance from the patient -0.50 -0.57 -0.41 0 11
ASC Nest
Specialty 9.93 8.16 12.07 11 0
Distance from the patient -0.61 -0.87 -0.38 0 11
ASC Distance to Emergency Center -0.39 -0.84 0.25 1 10
Notes:
The table presents average marginal effects.
The safety related (High-Risk) marginal effects are presented at an all level as well as by Medicare and Commercial payer type.
‘Specialty’ indicates the marginal effect for being a specialty ASC.

46
Figure 2.4: Predicted Choice Probability for the ith Closest Facility

47
Figure 2.5: Simulated Marginal Effect - Loss of Market Share for a High-Risk
Hospital (Both, Low Nursing and Surgery Safety)

48
Table 2.10: Simulated Share (%) Change (When Closest Hospital turns High-Risk)

1998 2002 2006

Loss/Gain (%)
Market Share Change for1 :
Nervous System & Musculoskeletal
Closest Hospital2 -5.51 -3.83 -2.63
Second Closest Hospital 0.65 0.63 0.4
Closest Ambulatory Surgery Center 1.84 0.99 0.54

All Ambulatory Surgery Centers3 3.79 2.35 1.68

Eye Surgery
Closest Hospital -8.72 -5.84 -1.25
Second Closest Hospital 0.05 0.24 0.14
Closest Ambulatory Surgery Center 3.29 2.22 0.35

All Ambulatory Surgery Centers 8.04 5.1 0.93

Digestive System
Closest Hospital -3.49 -8.62 -1.81
Second Closest Hospital 0.79 1.93 0.35
Closest Ambulatory Surgery Center 0.84 1.99 0.4

All Ambulatory Surgery Centers 1.72 4.52 1.07

Notes:
1
Change measured for various facilities when the closest hospital to the patient becomes a
high-risk hospital
2
Closeness is measured from patient zip code residence.
3
Change in the choice probability of the ASC nest.

49
 Table 2.11: Descriptive Statistics for Entry Model Variables

Variable Mean Std. Dev

V
d ol - Multi-Specialty 25.27 13.99
V
d ol - Nervous & Musculoskeletal 8.16 5.50
V
d ol - Eye Surgery 23.99 18.35
V ol - Digestive System
d 12.94 7.44
Volume treated at high-risk hospitals (nursing) 6.86 6.80
Volume treated at high-risk hospitals (surgery) 7.67 7.95
No. of high-risk nursing safety hospitals 10.21 6.49
No. of high-risk surgery safety hospitals 9.00 5.21
Notes:
Predicted volumes and volumes at high-risk hospitals are in 1000’s.
Volume at high-risk hospitals is the outpatient volume served in the potential entrant’s market service area.
Number of high-risk hospitals is measured for the potential market service area of the entrant.

50
 Table 2.12: Standardized Coefficient Estimates from the Entry Model - Probit Results

Model (1) (2) (3) (4) (5)

Baseline Nursing Complications Surgery Complications
V
d ol - Multi-Specialty entrant 0.277a 0.195a 0.171b 0.168b 0.179b
(0.075) (0.081) (0.09) (0.081) (0.081)
V ol - Nervous & Musculoskeletal
0.270a 0.199a 0.180b 0.181b 0.179b
d
entrant
(0.076) (0.081) (0.088) (0.079) (0.081)
V
d ol - Eye Surgery entrant 0.102 0.040 0.003 0.020 0.006
(0.081) (0.084) (0.094) (0.085) (0.087)
V
d ol - Digestive System entrant 0.266a 0.187b 0.170b 0.158b 0.171b
(0.076) (0.082) (0.089) (0.082) (0.082)
Safety Measures (PSI)
No. of high-risk hospitals 0.128a - 0.190a -
(0.048) (0.047)
Volume of patients treated at
- 0.124b - 0.152a
high-risk hospitals
(0.059) (0.049)

d(Pr entry)/d(PSI) 2.4 ppt 2.3 ppt 3.5 ppt 2.8 ppt
Baseline Entry Probability 11.7%
Notes:
Standard errors in parenthesis.
Explanatory variables enter as z-scores in all the specifications.

Number of high-risk hospitals is measured for the potential market service area of the entrant and the effect is estimated
separately for high-risk hospitals due to nursing complications and high-risk due to surgery complications.
Outpatient volume treated at high-risk hospitals is measured for the potential market service of the entrant.
a
p-value < 0.01
b
p-value < 0.05
c
p-value < 0.10

51
 CHAPTER 3

PRACTICE SPLITTING AND PHYSICIAN

INVESTMENT IN AMBULATORY SURGERY
CENTERS: IS POOR SAFETY AT HOSPITALS
RESPONSIBLE?

3.1 Introduction

Recent growth in the number of physician owned medical care facilities, such as, specialty
hospitals and ambulatory surgery centers (ASCs), has led to concerns regarding over utiliza-
tion of health services, conflict of interest due to potential for self-referral (Mitchell (2008),
Hollingsworth et al. (2010), Jon R. Gabel et al. (2008)) and the overall impact on health
care costs (Barro et al. (2006). The health literature has largely focused on the role played
by physicians as entrepreneurs and the possible adverse impact of their ownership decisions
on patient welfare in the form of higher costs, cherry-picking and demand-inducement. Fur-
ther, in the context of ASCs, recent literature has failed to find evidence of a favorable
impact on patient welfare. Results have indicated that ASCs either lead to increased num-
ber of procedures (Courtemanche and Plotzke (2010), Plotzke and Courtemanche (2010))
or provide only a modest welfare benefit (Weber (2010)). However, in asking the question
of ASC impact on patient welfare, the role of low hospital quality in motivating physicians
to move practice to ASCs has been ignored. Similarly, other factors that may induce a
decline in physician’s hospital based activity in favor of ASC based practice have not been
studied either.
This study examines the potential welfare implications of ASC growth, while highlight-
ing the importance of physician incentives as the driver of competition and change. The

52
focus is specifically on the movement of physicians from hospital outpatient departments
(HOPD) to ASC settings and the various motivations that drive such decisions. For in-
stance, it is important to evaluate whether these developments are a response to safety
concerns at acute care hospitals, or if they induce higher quality care in hospitals. In a
similar vein, one of the key issues is the direct effect of the match between physician quality
and hospital quality on the physician’s choice of practice setting. The underlying hypoth-
esis being that relatively high quality physicians may be more likely to lead the change
and therefore, may be at the forefront of the migration from a hospital based practice to
an ambulatory surgery center environment. The result, good doctors fleeing poor quality
hospitals.
To explore this hypothesis, reliable measures of physician and hospital quality are re-
quired. Burke et al. (2007) measure physician quality by defining ‘star’ status for a physician
which is, a physician who has completed residency at a top-ranked hospital. Burke et al.
(2007) find that the diffusion of stents in angioplasty by the physicians without star status
depends positively on the number of stars they interact with at the same hospitals, whereas
no social influence was found in the opposite direction. These results extend to doctor
quality measures that include age of the surgeon, the overall volume of surgeries performed
and whether he or she was an early adopter of technology (Burke et al., 2009). Measures
along these lines may show distinct patterns among physician selecting ASCs as primary
choice of practice location.
In Chapter 2 of this dissertation, initial evidence is found that ASC entry is more
likely if more patients in an area are treated at high-risk hospitals. This lends credit
to the hypothesis that high quality physicians operating in an incumbent hospital can
profitably disassociate from quality or safety problems by establishing a separate ASC to
achieve better control. I imagine a model of profit maximizing physicians who initially are
not affiliated with an ASC and perform all outpatient surgeries at the hospital outpatient
department. Over time, some portion of these physicians will migrate to ASCs. The timing
of their first migration to ASC and their degree of integration may entail different business
considerations; e.g. building a new facility would require a certain threshold volume of
surgeries, ownership interests entail organizational capital, while a contract relation with

53
an existing owner may include incompleteness. In all cases, however, the ability to profitably
move to an ASC setting may be viewed differently for high quality physicians relative to
low quality ones, and further depend on perceived safety of existing care in the market.
Similarly, the kind of ASC that a physician moves to, multi-specialty or single specialty,
may also shed light on physician incentives. For instance, practice at a single specialty
ASC can arguably provide a higher degree of administrative control to the physician and
accordingly physicians seeking a bigger control of their practice may be more inclined to
move to such ASCs.
The medical cases that are included in the study are confined within the specific cat-
egories of services that are amenable to be performed in a freestanding outpatient setting
like ASC. For example, operations on the digestive system, eyes, and the nervous and mus-
culoskeletal system. The empirical analysis is based on a panel data set from 1997-2008
including a census of outpatient visits identifying the physician performing them in HOPDs
as well as ASCs in Florida. The state of Florida provides a physician profile database that
includes biographical data on individual physicians and helps to construct quality measures
on the physician. Agency of Healthcare and Research Quality (AHRQ) software is used to
construct safety measures for the hospitals in the study (described in section 3.3.2). Safety is
measured with covariates most closely associated with surgical care, and includes measures
on post-surgical infection and complication rates due to poor nursing quality. If physicians
utilize knowledge of relative quality of local hospitals when choosing surgery setting, then
the AHRQ measures will have explanatory power in physician choice of surgery setting. Fur-
ther, if high-quality physicians recognize potential reputational damage from practicing at
low-quality hospitals, they will be more likely to move their practice to an ASC where they
can achieve much higher control on quality as well as other important operational decisions.
In this case, physician involvement with ASCs may be a welfare enhancing characteristic of
such facilities and welfare computations must account for it.
The next section provides a background of ASC growth in Florida and physician in-
volvement in it. The section also highlights important literature on ASCs as well as on
the physician-hospital relationship. In the following sections, information on data sources
and the empirical strategy is detailed. The last two sections provide key results and the

54
conclusion.

3.2 Background & Literature

The hospital-physician relation in United States is not defined by an employer and

employee contract, but based on a mutual understanding that serves interests of both the
parties. Hospitals are supposed to provide infrastructure, such as, equipment and nursing
staff, for physicians to treat patients who cannot be treated in a clinical setting, in turn,
physicians serve the hospital by practicing as the medical staff, performing on-call services in
the emergency department, taking positions on the various administrative bodies inside the
hospital, attracting patients (in case of specialist physicians famous in their line of medical
practice) and by playing a role in quality control at the hospital. In such a setting, the
importance of the match between hospital performance and physician expectations of the
work environment cannot be overemphasized. I hypothesize that patient safety at hospitals
interacts with the physician quality in a way that impacts the physician decision to set up
or move practice and therefore, plays an important role in determining the strength of the
hospital-physician relationship. When hospitals fail to maintain adequate safety standards,
physicians may get concerned about their own reputation getting negatively influenced and
therefore likely to move their practice to other hospitals or specialty facilities such as ASCs.
This impact could be stronger for high quality physicians.
In context of ASC growth, besides safety at hospitals, other reasons may also motivate
physicians to move away from hospitals and must be empirically examined. For instance,
physicians performing a high volume of procedures in hospital outpatient departments may
find investing in an ASC a profit earning opportunity, wherein they can benefit from the
fee-for-service payment system as they get to earn facility fee over and above the physician
fee. Such physicians may decide to capitalize on the reputation that they have earned in
the community due to experience and a well-known high volume practice, by setting up
their own ASC or moving practice to an existing one. Casalino et al. (2008) using results
from a longitudinal community tracking survey (CTS) find that there is an increasing trend
in hospital-physician relations wherein physicians end up choosing one of two mutually
exclusive tracks - hospital employment or separation from a hospital. Hospitals hire a

55
large number of specialist physicians and those physicians who are not employed at the
hospital prefer to separate themselves from the hospital by refusing to serve on medical
staff committees or emergency call and by investing in specialty facilities. Their study
also highlights the importance of entrepreneurial behavior in fostering the growth of ASCs,
as physicians who seek to gain more regular work hours and shelter from an increasingly
complex and unstable market prefer to be employed by the hospitals.
Burns and Wholey (1992) apply a behavioral approach to study physician loyalty and
state that physicians have a peculiar relationship with hospitals as they are both members
of the hospital as well as consumer. They perform as medical staff as well as consume
hospital services by utilizing equipment and staff at the hospital in their own practice.
They suggest and test physician convenience, inertia and organizational commitment in the
form of decision making role at the hospital as the important factors affecting physician
loyalty to the hospital. They also hypothesize that access to other alternatives as well
as dissatisfaction from the hospital may also serve to motivate a physician in exiting the
hospital environment or reducing her share of work at the hospital. Carlson and Greeley
(2010) list results from various surveys indicating growing dissatisfaction among physicians
towards their hospital and it’s negative impact on health care quality. A similar hypothesis
applies to this study as easy access to new and convenient (to both patient and physician)
forms of care delivery such as ASCs and poor safety record of hospitals may be a potent
combination in driving physicians to invest in or join ASCs.
In another paper, Burns et al. (2001) study the trend in strategic alignment between
hospitals and physicians. They categorize physician commitment into two types, one that
is driven by instrumental or utilitarian considerations and the other that is driven by ad-
ministrative involvement of the physician within the hospital. Older physicians, physicians
with a long tenure, salaried employees, and those who admit a higher percentage of their
patients at the hospital or exhibit more inpatient activity fall in the first category and are
more likely to be committed than others. Similarly, physicians with decision making roles
within the hospital in the form of administrative committee members or board members
may also be more committed. Their results, understandably, show that the utilitarian rea-
sons are stronger drivers of commitment. I also hypothesize that utilitarian measures are

56
more likely to drive physicians to move practice. Physicians are more likely to leave when
they have more alternatives available to them, when they are dissatisfied with the hospital
environment and/or when they seek to gain more control over their practice. However,
with respect to inpatient volumes, arguably, the inpatient activity by the doctor is also a
measure of physician quality as such cases are generally more complex and therefore higher
volume indicates physician skill. It is therefore an empirical question as to how does high
inpatient volume affect physician decision to move practice.
Literature suggests that a hospital physician’s motive to join an ASC can be thought
of in terms of ’push’ and ’pull’ factors and how these factors interact with the physician’s
own practice. Push factors are those that are at operation within the hospital and include,
besides other variables, hospital quality/safety, hospital profit type, teaching status and
volume of patients treated by the physician at the hospital inpatient and outpatient de-
partments while, pull factors are those operating outside the hospital and include variables
like, availability of alternatives in the physician’s geographic location, profit potential by
earning facility fee as owner of an ASC, and convenience of practicing at a smaller facility
with fewer administrative tasks. These factors further interact with attributes of the physi-
cian’s practice such as reputation and skill measured by the quality of physician’s education
and training, board certification, inpatient practice and staff privileges. These attributes
collectively indicate the overall quality of the physician. Previous literature like, Burke et
al. (2007), Burke et al. (2009) and Gardner and Vishwasrao (2010) have relied on widely
used and accepted US News & World Report ranking system to rate the residency program
attended by the physician and ascribe quality to the physician based on the rank of the
residency hospital. Besides using the same ranking system we also utilize NIH rankings that
are based on the research funding earned by the residency school of the physician. Further,
this study claims that measures like board certification, number of staff privileges, and the
number of publications that a physician has, indicate a high demand for her services and
therefore, are evidence of reputation and quality.
One of the key hypotheses established in this study is that high quality physicians are
more likely to leave hospitals for ASCs. Safety problems at the hospitals are likely to
combine with other incentives and further motivate the physicians to adopt ASC practice.

57
This requires that physicians be aware of the safety record of the hospital, be affected
by it and take it into account when taking important practice related decisions. Barr
et al. (2008) in a qualitative study, looked into physician views on public reporting of
hospital quality. They found that referring physicians primarily rely on hospital volume
and outcome information as well as patient experience and safety measures to decide upon
hospital quality. Also, most physicians in their study claimed to be aware of their own
hospital’s quality reports. In another study, Blendon et al. (2002) point out the results
from the report by the Institute of Medicine. The report stated that each year, more
Americans die as a result of medical errors made in hospitals than as a result of injuries
from automobile accidents. It is such preventable complications (AHRQ measures) and
errors that are used in this paper as a measure of hospital safety. Majority of the physicians
in Blendon et al. (2002) study suggested two solutions to control errors, and laid primary
responsibility on the hospital administration. Physicians believed that hospitals need to
develop systems for preventing medical errors and need to increase the number of nurses.
Both these solutions should be relatively easy to apply at small focused facilities such as
ASCs and hence, the proposition of joining ASCs may be more attractive for physicians
tired of safety problems at hospitals.
The physician move to ASCs, however, is akin to new technology adoption, as the ASC
environment is different from and lies somewhere between the traditional clinic and hospi-
tal environment in terms of size and scope. The technological developments in the fields of
minimal invasive surgery and anesthesia, (Poole (1999), Mechanic et al. (2005)) have made
it possible for physicians to carry out specific procedures like colonoscopy, upper gastroin-
testinal endoscopy, and cataract surgeries in a smaller but focused ASC environment. The
move for physicians to ASCs gradually picked up speed during the last decade. Although,
Medicare started covering procedures at ASCs in 1980, the real growth came during the
90’s. The number of Medicare certified ASCs grew at an accelerating rate from 2,314 in
1996 to 3,400 in 2002 and 4707 in 2006 (Winter (2003), Mechanic et al. (2005), Casalino
et al. (2008)). Nationally, physicians now own 80% of these ASCs either alone or in joint
venture with a hospital or other corporation (Casalino et al., 2008).
High quality physicians (‘Stars’) and community leaders are often the first ones to adopt

58
new changes in the field of medical care. This strengthens the belief that the growth
of ASCs may have also been impacted by such behavior. Burke et al. (2007) studied
the diffusion of medical innovation by looking at the adoption of ’stents’ (for coronary
angioplasty) by the physicians, after FDA approval in 1995. In their case, the adoption
behavior was characterized by asymmetric social influence where ‘Star’ physicians had a
strong influence on ‘Non-Stars’ while, there was no influence in the opposite direction.
‘Star’ physicians, the paper claims, have also a higher ability and opportunity to learn
about new methods from external sources as well as personal experience. Similarly, in the
case of ASCs, heterogeneity among physicians based on their incentives, practice profile,
location and unobserved entrepreneurial zeal is a driving factor.
Burke et al. (2007) applied a hazard model to examine the hazard of stent adoption
by all physicians and even when they did not find much difference in the propensity to
adopt between stars and non-stars, they did find evidence for influence of star physicians.
Hazard models are particularly suited for studying such technology adoption behavior as
they allow one to account censoring of information for units of analysis that do not witness
the relevant event during the recorded data period. Escarce et al. (1995) also applied a
proportional hazard model to the case of diffusion of laparoscopic cholecystectomies. They
studied various attributes of physicians that may make them more susceptible to adopting
the new technology. For instance, younger physicians and those graduated from a US
school are more likely to be technology friendly, fee-for-service physicians and large practice
physicians were considered to be more likely to adopt early and so were physicians practicing
in more competitive markets. Their results supported competitive behavior on the part of
the physicians and financial reasons as primary drivers of adoption.
In contrast to physician-hospital studies, the economic literature on physician-ASC rela-
tions is still in a nascent stage with only a handful of studies delving into physician behavior
in the context of ASC growth. One of the recent studies by David and Neuman (2011),
presents an interesting analysis of the division of practice by physicians (mainly, Gastroen-
terologists) across different settings, primarily between HOPDs and ASCs. Authors use the
same data as in this study, provided by the Agency of Healthcare Administration (AHCA)
of Florida and test their hypothesis that physicians who do not split their work between

59
hospital and ASCs and are exclusively ASC based (non-splitters, approximately 5% of their
physician sample) are more likely to treat patients with higher-risk profile at the ASC, as
compared to those who work at both the locations (splitters). Using the Charlson comor-
bidity index, authors show that the colonoscopy patients treated by splitters at ASCs are
less complex cases than those treated by non-splitters. However, their results are limited to
a small proportion of physicians not splitting and confined to a narrow range of medical risk
as the mentioned non-splitter behavior weakens with increasing medical risk. Further, an
important factor unaccounted in their study is the differences in the type of ASCs. Anal-
ysis of ASC licensure data from Florida (provided by AHCA) shows that almost half of
the ASCs are multi specialty facilities employing various doctor owners specializing across
different medical specialties. Non-splitters working at such ASCs may be exposed to lower
risk than those who are practicing at single specialty ASCs.

The licensure data also shows that approximately 40% of ASCs in Florida are owned
by corporations like HCA, AmSurg, Health South, Novamed, Symbion etc. Among these,
there are also ASCs owned by hospitals such as Baptist System, Morton Plant and Naples
Collier Health System. Physicians who choose to work at an ASC, their choice of the type
of ASC may shed light on their incentives. For instance, hospital owned as well as HCA
owned ASCs are mostly multi-specialty facilities, with approximately 40-60 physician joint
investors practicing across different specialties. AmSurg and Symbion owned ASCs, on the
other hand, are generally single specialty ASCs with focus on selected procedures in the
areas of eye care, orthopedics and the digestive system. The number of physician joint-
owners at such facilities range between 2 to 6. Physicians who choose single specialty ASCs
over multi-specialty ASCs to set up practice, among other things, must desire a higher
degree of control over their practice and may be more likely to leave a hospital with a
poor safety record. Alternatively, physicians who are working at a hospital owned ASC
may actually be only extending their relation with the hospital by joining it’s ASC and not
reacting to quality or safety changes within the hospital. Such differences in incentives are
accounted for in this study by utilizing the licensure data that provides detail on ownership,
along with the physician profile and practice data.

60
 3.3 Patient Data and Physician Profile

The data for this study comes from two different sources - the Agency of Healthcare
Administration (AHCA) and the Department of Health(DOH), Florida. Hospital financial
data and the patient level data were acquired from the AHCA. This study utilizes patient
records from 1997 to 2008, from both inpatient discharge and outpatient visit data collected
by AHCA. Besides the patient level variables such as age, race, gender, procedure and
diagnosis, the data also includes the operating room physician’s Florida medical license
number. These license numbers in turn are linked to the second data source which is the
physician profile information collected by the DOH. The profile data includes information
on physicians, such as, name, address, education, post-graduate training, state level medical
licenses, staff privileges, faculty appointments etc. These three data sets provide a significant
insight into the practice profile of each physician and help to take into account both the
site of practice of the physician for outpatient work and her inpatient practice. Further, it
helps us build a panel data set on physicians through their various practice sites over a long
period of 12 years, quarter by quarter.

3.3.1 Sample Selection

As explained later in the section 3.4, the empirical analysis is divided into two parts.
The first part includes a hazard model while, the second consists of a fixed effect regression
model. The unit of analysis in both the models is a physician quarter i.e. quarter by quarter
practice information on each physician in the sample. The sample itself is selected using
the inpatient and outpatient practice information on the physicians from 1997 to 2008.
During this period Florida also witnessed entry and exit of physicians from the market.
As a result, we do not have a balanced panel. I choose to model only Medical Doctors
and their decision to choose practice location and ignore those who are either Podiatrists
or Osteopathic Medicine Doctors. Physicians who treat less than 360 outpatients over
the entire period from 1997 to 2008 or are present in the data for less than 6 consecutive
quarters (out of 48 total) are dropped from the sample. Also dropped are physicians who
treat less than 60 patients a quarter on average across quarters during which they practice
in Florida. These conditions leave us with a total of 3159 physicians and ensure that only

61
those physicians are included who have had a substantial presence in Florida during the
time period coinciding with the data. Further, to specify the group of physicians that can be
considered at-risk for joining an ASC at the start of the data observations in 1997, physicians
who were already treating a majority of their patients at an ASC are excluded from the
sample. The act and timing of a physician adopting ASC as a practice site is defined as
that quarter in which a physician treats 30 or more patients at an ASC for the first time
and accordingly, any physician who was already treating more than 30 patients in the first
quarter of 1997 is also not included in the at-risk sample. Next, since ASC environment is
suitable only for outpatient services and procedures and major surgeries like Cardiovascular
or Oncologic surgery are fairly complicated and suited only for an inpatient environment,
physicians who specialize in such procedures are also not included in the sample. Similarly,
physicians with specialization in Emergency Medicine are also excluded.
For the hazard model there are some additional restrictions imposed on the sample.
Physicians are considered to be included in the data set from the first quarter when they
treat 10 patients in Florida, inpatient, outpatient or both. Since for hazard analysis lagged
volume information is needed on the physicians, a physician is included in the analysis only
from the 7th quarter after he or she first treats 10 patients in Florida. Any physician who
joins an ASC before the 7th quarter is excluded form the hazard analysis. This rule is
applied differently for physicians who were practicing in Florida at the beginning of 1997
and were licensed to practice in Florida in 1996 or before. Such physicians are included in
the analysis from the 5th quarter onwards (i.e. 1st quarter of 1998). This again ensures
that we include those physicians in the at-risk sample who have practiced for a substantial
period in Florida and who we can observe for a continued period of time before they decide
to change practice pattern. Analogous to the ASC adopting definition, physicians who do
not treat at least 30 patients in any of the observation quarters are considered non-adopters
and their decision to join an ASC is assumed to be censored at the end of 2008. For
physicians who retire or decease before the end of 2008, their decision to join an ASC is
also censored. Further, physicians who do not reside in Florida anymore are excluded from
the analysis. This exclusion decision helps to take into account the impact of physician
distance to hospitals and ASCs on practice setting decisions. Since the analysis depends

62
in individual level behavioral data, there are a few physician cases where the practice
information is somewhat scattered over time with considerably large periods of time that
the physician was missing from the data. This could happen because a physician moved to
another state and returned after a long period, or because she decided to get post graduate
training in a particular specialty or for simple reasons like taking a sabbatical or going on
a long vacation. Observations on such physicians are not included in the analysis. All the
above conditions, lead to a sample with a total number of 1543 physicians out of which 855
join an ASC at some point during 1998 to 2008. Table 3.1 summarizes the physician sample
while accounting for adoption behavior and entry by new physicians who are either recent
graduates or have move to Florida from another state. There are a 1030 physicians who
fall under the at-risk group of physicians at the beginning of year 1998. At an average of
12 new physicians every quarter, a total of 513 new physicians join the at-risk group during
the period. The ASC adoption rate is relatively higher during the early years from 1998
to 2003 after which the exits start stabilizing with fewer physicians than before moving to
ASCs every quarter.
Next, using latitude-longitude information, I compute travel distances from the physi-
cian’s primary address to each of the hospitals and ASCs in Florida using the Euclidean
method, also known as the “as the crow flies” method. Having the distances to all facilities
helps to account for the opportunity factor behind the move to an ASC. Since physicians
are likely to choose practice locations closer to their own address, the analysis controls for
the number of hospitals and ASCs located, in the previous period, within a small radius of
the physician’s primary address. A higher number of ASCs or hospitals in close proximity
indicate availability of different options to the physician and should increase the probability
of a physician moving practice to an ASC or alternatively, choosing primarily a hospital
practice over an ASC.
The physicians included in the study sample specialize in various medical areas, as
shown in figure 3.1. Roughly one third of the physicians are Internal Medicine specialists
who are generally involved with, in this case, procedures in Gastroenterology. Orthopedists
form one tenth of the sample while, the next two major specialties are in the areas of
Ophthalmology and Urology. Raw data also indicates that physicians specializing in Internal

63
Medicine, Ophthalmology, Orthopedics and Urology are more likely to adopt ASC as a
practice location. This is primarily because these major procedure categories account for
a large number of procedures that can be carried out in the outpatient arena and have
benefited the most from minimally invasive technology changes. For instance, procedures
like GI Endoscopy, Cataract surgery, Laparoscopic and Arthroscopic procedures are now
generally performed in an outpatient environment.

3.3.2 Physician Quality, Hospital Safety Measures and Other Key

Variables

There are two categories of quality measures that are of interest in this study. First is
an indicator of physician quality while the second is of hospital quality. The underlying
hypothesis, as stated before, is that high quality physicians should be more inclined to
disassociate themselves from quality problems at the hospitals. Even if that’s not precisely
the case, it is still likely that high quality physicians have an incentive to move to an ASC
and achieve a tighter control on their own practice as well as a direct control on patient
safety and facility environment. Further, physicians irrespective of their own quality, when
working at low quality hospitals are likely to get frustrated and have it in their own interest
to move practice to an ASC to protect self reputation as well as to capitalize on the profit
opportunity that results due to poor performance of the hospitals.
There is no general consensus in the literature on the best method to measure physician
quality. In case of major inpatient procedures and surgeries, patient outcome data can be
used to construct procedure specific measures on physician quality, for instance, use of beta
blockers for cardiac patients or use of DVT prophylaxis for stroke rehabilitation patients
(CMS.gov, 2011). Such measures, however, cannot be aggregated in a useful way to indicate
overall physician quality. A few of the major health insurers, like Aetna and United, publish
basic physician rankings but detailed information is neither easy to come by nor can it be
compiled in a way to cover a complete roster of physicians practicing within a state. Hence,
in this paper, I apply a set of physician specific measures that together indicate various di-
mensions of physician quality. Following Burke et al. (2007), I first create a ‘star’ physician
indicator. A ‘star’ physician is a medical doctor who completed her most recent post gradu-
ate training (residency or fellowship) at a top-30 medical school. The definition of a top-30

64
school is based on both NIH research funding (ResidentPhysician.com (2005)) as well as the
US News & Health Report (2005 & 2010) and is different by physician’s medical specialty.
A physician is recorded as a ‘star’ if her medical school was among the top-30 in any of the
three listed ranking systems. Alternatively, a highly selective ranking is also used to define
a good quality physician and is based on US News & Health Report’s honor roll system.
Under this system 14 medical schools nationwide are considered to be top schools across all
major medical specialties. Although US News & Health Report measures of hospital quality
are widely accepted in the health care industry and are scientifically constructed1 , it is still
impossible to be completely objective about measuring individual physician quality based
on the ranking of the residency hospital. Similar problems will exist with any method used
to measure individual physician quality in a precise manner. So for the robustness of the
analysis, an alternative measure of physician quality is included that relies on top quality
doctor reports published by ‘Castle Connolly Medical Ltd.’(Castle and Connolly (2010)).
The Castle Connolly Top Doctors list is built with the help of a physician research team
and is basically a peer review system based on the surveys of physicians and healthcare
professionals. The review primarily includes physician’s medical education, training, hos-
pital appointments, administrative posts, professional achievements, and malpractice and
disciplinary results. As a result this quality measure is more encompassing in nature and
accounts for different aspects of physician quality.
Besides the ‘star’ physician measure, three more measures are controlled for. First
is an indicator for board certification (Based on American Board of Medical Specialties
(ABMS)) in physician’s primary specialty. Brennan et al. (2004) point out the importance
of certification in the entire quality movement. They focus on the empirical evidence that
shows validity of board certification and the close relationship of certification examination
scores with measures of physician competence and clinical outcomes. They also highlight
the results of a Gallup poll that demonstrated that certification and its maintenance by
the physician is highly valued by general public. The second measure relies on the number
of hospital medical staff privileges held by the physician as a signifier of physician quality.
The underlying hypothesis is that high quality physicians in a community are more likely
1
Methodology for constructing these measures is available on the website of RTI International, the re-
search company that builds these measures.

65
to be in demand at different hospitals and may also find it easier to be accepted for staff
position across different hospital medical boards. Staff privileges are considered to be a
good indicator of physician quality due to the process involved in acquiring staff privileges.
Generally, a physician seeking staff privileges at a hospital needs to make an application to
the board after which a credentialing committee takes into account various qualifications
of the physician before granting her request. For instance, a hospital board is likely to
judge an applicant using board certifications, recommendations, state licensure, residency
training etc. (Marshall et al. (2002)). Finally, I also include a measure of physician’s
publication record constructed using the Web of Science database. Although, this measure
also indicates physician quality, it is more directly related to a physician’s research skills
and physicians who have faculty appointments are more likely to produce a higher number
of publications. As a result, physicians with faculty appointments and a high number of
publications may actually be less likely to move practice and lack the incentive to invest
in an ASC or to practice at one. To account for this behavior, empirical analysis includes
the number of publications by physicians who do not have any faculty appointments as a
measure of quality.
Table 3.2 presents selected descriptives for both adopters and non-adopters of ASC
practice. The physician volumes have been averaged over the entire period of their practice
in Florida and as expected ASC adopters tend to have higher overall volumes as well as
higher outpatient volumes than non-adopters while, the non-adopters treat slightly higher
inpatient volumes on average. Both, those physicians who adopt ASC setting and those
who do not, have a similar number of ASCs close to their primary address but the adopters
have access to a slightly fewer number of hospitals. A smaller proportion of adopters have
faculty appointments and in line with that they also do not seem to publish much. Although,
the absolute number of ‘stars’ no matter which method is relied upon are higher among
the adopters, in proportional terms only the ‘Castle Connolly’ measure indicates a higher
percentage of top quality physicians at ASCs. Interestingly, almost 40% of the adopters
start out at an ASC that is either owned by HCA or by another hospital chain such as
Baptist Health for example. This seems to imply that for many physicians ASC practice
could also be an extension of their relationship with the hospital where they are already

66
practicing.
Next, we need hospital level quality measures that vary with time. I rely on AHRQ
measures on patient safety. According to the AHRQ guide on patient safety indicators
(AHRQ (2003)), patient safety is an alternative measure of hospital quality that is concerned
with measuring the risk-adjusted incidence of potentially preventable complications that
result from patient exposure to health care system. The measure is different from AHRQ
provided inpatient quality measures as those measures are focused on inpatient outcomes
of mortality as well as utilization of procedures at the hospital while, the safety measures
are more general in nature and apply to the overall culture of safety at a hospital. Safety
complications take place due to poor medical management at the hospital and not due to
the underlying medical condition of the patient and therefore, measures focusing on safety
are much more likely to have a direct relation with a physician’s decision to set up practice.
Using the AHRQ patient safety indexes (PSIs), I construct two aggregate level hospital
safety safety measures - the nursing safety and surgery safety indexes. Following closely the
methods detailed in Encinosa and Bernard (2005) and Bazzoli et al. (2008), risk-adjusted Pa-
tient Safety Indexes2 (PSIs) are computed. The Florida hospital inpatient data file includes
all patient discharges with a major surgery diagnosis-related group (DRG). The AHRQ-
provided algorithm uses these data for computing the safety indexes. These indexes track
the occurrence of adverse events and focus on conditions and complications experienced by
patients during their hospital stay. Using the PSIs, I construct, by year and by hospital, two
aggregate measures of patient safety over all major surgery discharges. The first measure,
surgery safety index, consists of nine PSIs that are most closely related to the actual surgery,
while largely independent of post-operative nursing care. These include complications in
anesthesia, postoperative hemorrhage or hematoma, postoperative hip fracture, postopera-
tive physiologic and metabolic derangements, postoperative pulmonary embolism or DVT,
postoperative respiratory failure, postoperative sepsis, postoperative wound dehiscence and
finally, accidental puncture or laceration. The second measure, nursing safety index, is con-
structed from three PSI indicators previously recognized as related to the nursing activity
2
Risk adjustment is based on computed hospital fixed effects. First, using logistic model predicted value
of complications is calculated for each discharge and then subtracted from the actual outcome. Then, this
difference is averaged over each hospital to get the risk adjusted rate.

67
- post operative hip fracture, decubitus ulcer, and selected infections due to medical care.3
These two measures distinguish nursing and surgery as sources of adverse events in hospital,
and may produce different effects in the model.
Further, as in Bazzoli et al. (2008) I construct aggregate nursing and surgery compli-
cation rates. After computing the individual PSIs I subtract from each hospital’s rate the
overall average rate, for the same PSI, of hospitals in the Florida data set.4 This gives us
a measure of excess number of incidents that took place in the hospital. Finally, each of
these excess measures (as mentioned above, 9 for surgery and 3 for nursing) are weighted
by the proportion of patients at risk for the indicator. To illustrate for the case of the
nursing indexes, first, the excess measure for each of the three nursing PSIs is weighted by
the number of patients at risk at the hospital. These weighted excess measures are then
summed up and divided by the total number of patients at risk for all 3 nursing related
PSIs.

3.3.3 Physician Ownership Data

Besides the patient level data and physician profile data, I also obtained ASC licensure
files from AHCA that contain important details on the ASC management, address, profit
status, controlling interest5 owners and their names, and names of corporate owners in
case of joint ventures.6 Licensure files also contain information on the closest hospital that
the ASC has an agreement with to send patients requiring emergency inpatient care. Any
changes in the facility ownership are also accounted for with the help of these files as ASC
owners need to renew the license every two years.
The major difficulty, however, in using information from licensure files is that there are
approximately 400 ASCs in Florida and each has multiple file submissions, over the years,
which are saved as digital documents that need to be perused and required information has
3
As explained in Bazzoli et al. (2008), postoperative hip fracture is used in both type of indicators as it
can occur because of either nursing or surgery errors.
4
Only those individual PSIs are used for which the population at risk was at least equal to 30 patients
at the hospital.
5
AHCA defines controlling interest owners as any owners with greater than or equal to 5% share in the
organization
6
Corporate owners like Health South and HCA generally own ASCs under various holding companies and
to identify the name of the actual corporation chain, the name of the holding company and its head office
address has to be matched with the state annual reports.

68
to be extracted manually. The main variable of interest in the licensure files is physician
owner identity and ownership share. The owner identity is easier to track in case of specialty
ASCs that are smaller in size and are in general owned by fewer physicians. I find that
multi-specialty ASCs are often owned by large groups of physicians or groups of physicians
in joint venture with corporations like HCA, Health South, Surgical Care Affiliates etc.
Some of the multi-specialty ASCs (around 39) are also owned by hospitals. The structure
of corporate owned multi-specialty ASCs is such that the corporation owns more than 50%
interest in the ASC and the rest is shared by on average 40-60 physicians who all also
practice at the ASC. There is some fluidity in physician ownership at such ASCs as it is
observed that some of the owners leave after 2-3 years at the ASC which makes it even more
difficult to keep track of owner identity. However, it is safe to assume that any physician
who is practicing at a major corporation owned ASC, also owns a small ownership interest
in the ASC and is in all likelihood a less than 5% shareholder (i.e. not a controlling interest
owner).7 There are some other corporations too that are quite active in Florida market,
namely AmSurg, Symbion and Novamed. These corporations generally own specialty ASCs
in joint ventures with a few physicians. For instance, AmSurg tends to partner on average
with 5-6 physicians at a time with established practices. In such ventures, I found AmSurg to
be owning 51% interest while the rest is generally divided equally among physician owners.
Novamed on the other hand tends to specializes in eye surgery centers while, Symbion’s
interests span over both specialty and multi-specialty ASCs. In some cases, owner identity
is ascertained using annual reports submitted to the state department of Florida. The
department requires annual reporting by all organizations in Florida and the information
is available to the general public through the state’s website.
Using the licensure files, I am able to track approximately 1800 physician owners who
have a controlling interest. These numbers further come down when we apply other condi-
tions listed previously in section 3.3.1. The physician owner names have to be searched on
the state department’s website to identify ME license numbers which then are linked with
the patient level data for empirical analysis. Based on these files and work history of the
physicians, I know that 41% of the physicians in the sample have a non-controlling interest
7
This was confirmed from the licensure files that are required to list names of controlling interest owners

69
in an ASC that is jointly owned by a national level hospital chain, 11% have a controlling
interest in an ASC that is jointly owned by a corporation that specializes only in ASCs (eg:
AmSurg), 20% have a controlling interest in an ASC that is run by physician group(s) or
by an individual physician. Approximately 29% have either no ownership interest or a non-
controlling interest. These ownership shares are for a larger group of physicians (1097) than
those that enter the hazard analysis. As in the second stage analysis, I am able to account
for those physicians who had already adopted ASCs at the start of the data period. The
next section details the empirical strategy and how the ownership information is utilized.

3.4 Empirical Methodology

Although the opportunity to invest in an ASC is lucrative for entrepreneurial physicians

for various reasons, including a profit share in the facility fee and a higher degree of control
on own practice, there are other important characteristics of the health market and its
players that are overlooked and ignored by the critics of ASCs. The two-pronged relationship
between hospitals and physicians - both collaborative and competitive at the same time is
an important aspect of the growth of ASCs. It is hypothesized in this study that poor
safety at the hospitals plays a role in motivating physicians to search for or invest in other
outlets of health care delivery such as ASCs and further, that high quality physicians are
more likely to get frustrated with poor hospital safety environment and more likely to find
the opportunity of setting up their own independent practice at an ASC. High quality
physicians are also more likely to be motivated by factors such as an increased authority
on administrative aspects of the practice as well as the ease of scheduling that they can
achieve by moving to an ASC.
Decisions regarding setting up practice or dividing it across different facilities, however,
depend on, among other things, the breadth and depth of a physician’s practice. Physicians
who treat patients ranging across a wide spectrum of health status and medical complexity
find it in their own, as well as in the patient interest, to maintain a substantial practice at
hospitals too. Medically intensive cases need to be treated in an environment where access
to the emergency department is easy and instant and therefore, hospital practice is often
an integral part of physician’s practice. Poor safety performance of the hospital where a

70
physician carries out the majority of her work, however, is likely to adversely affect the
hospital-physician relationship and even when the physician cannot completely sever the
ties with the hospital, she may react to safety problems by reducing the share of her work
performed at the poor safety hospital.
To account for this behavior, the empirical strategy is divided into two parts. In the
first part, I model the decision to adopt an ASC as a practice setting with a hazard rate
model. The decision depends on, among other things, lagged information on hospital safety
performance as well as physician’s own quality. Once a physician has joined an ASC or
ASCs, I ask the following question - what determines the split of physician’s work between
a hospital outpatient department and the ASCs? The question is similar to the one raised
by David and Neuman (2011) in their work on splitter and non-splitter physicians. However,
here the interest is in further utilizing the rich information on ASC ownership, broken up
by individual physicians as well as by joint ventures between physicians and national level
chains, to explore it’s impact on the splitting decision by physicians. Therefore, in the
second part, I use the information on the group of physicians who join an ASC during our
data period (1998-2008), from the period after they have joined an ASC and evaluate the
impact of hospital safety as well as physician specific variables, specifically ownership status,
on the splitting decision. The second stage regression allows to further utilize information
on ASCs that the hazard rate model does not as the ASC information is specific to the
period after adoption.
It is imagined that at the start of each quarter, there is a group of physicians who can
be considered at-risk for joining an ASC or equivalently investing in one. Such physicians
should have a substantial outpatient practice and involvement in medical specialties for
which procedures and surgeries can be carried out in the outpatient setting. This means
physicians specializing in Oncology or Cardiac Surgery are excluded from the at-risk group.
The decision to adopt an ASC setting depends on, besides other physician and hospital
characteristics, the safety performance of the physician’s principal hospital. Principal hos-
pital is defined as the hospital where physician performed the largest share of his overall
hospital work in the previous quarter. Every quarter a certain number of physicians adopt
the ASC practice, where the ASC adoption timing, as previously explained, is measured as

71
the quarter where the physician treated 30 or more patients in an ASC environment for the
first time. Thus, the hazard for adoption can be written in the form of a cox proportional
hazard model as:

h(t|xj ) = h0 (t) exp(α0 + αj xj ) (3.1)

The vector xj represents the set of variables included in the hazard analysis (Table 3.2).
We include key variables on physician quality, physician practice profile and hospital safety
performance. Number of hospital staff privileges, board certification, number of publications
and the ‘star’ status are included as physician quality measures. ‘Star’ status is alternatively
measured by, post graduate training from a school ranked among the top 30 in US, training
from an honor-roll school and being listed as a top doctor in Castle Connolly publication.
To capture the practice profile, I use an experience variable (measured as number of years
from the graduation) along with the variables measuring average inpatient and hospital
outpatient volumes of the physician over the previous 6 quarters. Further, to account
for the availability of an opportunity to move practice to an ASC or to another hospital,
two separate measures are included representing the number of hospitals and number of
ASCs situated within six miles of the physician address in the previous quarter. The
physician address stays constant through out the period and represents the physician’s
primary location which is either physician group office or home address. Assuming that the
physician is likely to choose practice location taking traveling convenience into account, the
six miles measure is the most intensive in terms of expecting an impact on the adoption
hazard. 8 All specifications also include primary hospital fixed effects. The primary hospital
is the one where the physician treated majority of her patients in the previous quarter.
Physicians who joined an ASC at the start of the first quarter in 1997 or who joined
an ASC directly after obtaining their Florida medical license have been excluded from
the hazard analysis as there are not enough quarterly observations to explore their practice
decisions. It is, however, likely that these physicians were influenced to make the decision of
joining an ASC for similar reasons as those who joined after a substantial time had passed.
Information on hospital safety is likely to travel within communities through physicians
8
We also try out another measure based on 10 mile radius of the physician and find qualitatively a similar
impact.

72
who practice at the hospital as well as through hospital staff like nurses. If the hospital
safety in certain communities is exceptionally bad, then physicians who are starting out in
such communities, may be predisposed towards ASC practice. I am able to include these
physicians along with those who have been analyzed in the hazard analysis in the second
part of our empirical analysis.
In the second part, those physicians are analyzed who have adopted ASC practice. The
primary interest in this analysis is to evaluate the role of ownership and how it interplays
with the safety at hospitals. Ownership information has always been an elusive piece of
the puzzle when it comes to deriving an understanding of physician-owner reaction to dif-
ferent incentives. Extracting the physician ownership information from more than 1200
ASC licensure files and in more than half of the ASC cases matching it with Florida state
department’s annual reports as well as ASC business websites helped to generate an under-
standing of ownership structure of these facilities. As explained before, there are primarily
four types of ownership structures that exist in the ASC business. There are joint ventures
between big national level hospital chains and multiple physician groups as well as indi-
vidual physicians that form the first type. Corporations involved here are primarily, HCA,
Health South and Surgical Care Affiliates9 . Physicians at such ASCs in general do not have
a controlling interest (greater than 5% ownership share) but almost always have a small
ownership interest (roughly, 0.5-2%). The second type involves joint ventures where there
are national level ASC corporations like AmSurg, Symbion and NovaMed involved. The
ownership pattern here is much more balanced as the corporation tends to own majority
interest (51%) and handles the management side of the ASC, while, physicians (generally
part of the same group that tends to be small in size with 5-10 physicians) also own a
controlling interest. The third type is non-corporate physician ownership either in the form
of solo physician owner, physician group owning the ASC or joint venture between two
or more physician groups. The common factor, however, is that physicians involved, all
own controlling interest and most such ASCs tends to be specialty facilities. Finally, the
fourth type is the residual category where the facility generally tends to be a multi-specialty
owned by various physicians and physician groups but with no single physician enjoying a
9
Health South owned Surgical Care Affiliates until the middle of year 2007 and therefore, we treat both
these as the same corporation for analysis purposes.

73
controlling interest.
I hypothesize that compared to non-owners, owner-physicians are more likely to divide
the outpatient practice such that relatively more of their practice is situated at the ASC
than at hospital outpatient department. This may be necessary so that they can maintain
their hospital staff privileges to treat cases that are medically too complex to be treated at
the ASC or because the patient or her payer requires the procedure to be carried out at the
hospital. Furthermore, I test whether owner physicians are more likely to react negatively
to the safety problems at the hospital and are therefore, more likely to treat more patients
at their ASC as a response to poor safety at the hospitals. Strong reaction from controlling
interest owners is possible because of the compatibility of the profit incentive with a desire
to treat own patients in the best possible environment. As a result, the interaction of
physician profit incentive with poor safety at the hospital, in sum, may actually be beneficial
for general patient welfare. We test this hypothesis with the help of a hospital fixed effects
regression, where the dependent variable is the log odds of physician share of outpatient
work (ph ) done at the HOPD vis-a-vis that done at the ASCs. This dependent variable
captures the split of work for a physician practicing at both HOPD and ASC. The fixed
effects regression can simply be expressed as:
 
ph
log = β0 + β1 N ursingh + β2 Surgeryh + β3 Distanceih + β4 InV oli +
1 − ph
β5 Qualityi + β6 Ownershipi +

β7 Ownershipi ∗ N ursingh + β8 Ownershipi ∗ Surgeryh (3.2)

The main variables, other than the ownership information, in the second part analysis are
similar to the ones used in the proportional hazard analysis. Safety performance of the
hospital h in the area of nursing and surgery is represented by N ursingh and Surgeryh . If
the physicians are adversely affected by the poor safety at the hospitals, then their odds of
treating patients at a poor safety HOPD will be lower than at the ASC. Further, to account
for the convenience factor, a measure of straight line distance (Distance) from the physician
primary address to the HOPD is also included. Physicians who treat a substantial number
of inpatients are more likely to be careful about maintaining their hospital privileges and
are therefore, more likely to have a stronger relationship with the hospital even during the

74
period when the hospital does not perform as well on safety. The variable InV oli controls for
this possibility by accounting for the number of inpatients treated at the hospital during the
last quarter by the physician i. Physician quality is represented by Qualityi and includes
similar variables as the ones in hazard analysis. Finally, the ownership is measured by
dummies for hospital corporation & physician joint venture, ASC corporation and physician
joint venture and controlling interest physician ownership with the base category formed by
non-controlling interest physician ownership type. These dummies are interacted with the
hospital safety measures of nursing and surgery to measure the differential impact of safety
on the ownership type.
For the log-odds analysis, those quarters are dropped where a physician treats less
than 30 patients overall including inpatient and outpatient practice as overall low volume
quarters may involve periods of low activity, vacation or absence due to training. Also,
physician-facility-quarter combinations involving less than 10 patients are dropped. One of
the major complications that come up with measuring the impact of safety on a physician’s
HOPD-ASC split of practice is that a big proportion of physicians tend to work at more
than one HOPD and more than one ASC during the same quarter. Since the primary
interest is in measuring a physician’s response to poor safety at the hospital, we combine
the ASC quarterly volumes done by the physician and compute the log odds of the share
of work at each HOPD that the physician operates at to that of her share of work done at
all ASCs. Thus, for instance, there will be effectively two observations in each quarter for
a physician who serves at two HOPD’s.

3.5 Results

The main results are divided into two sections, with the first one describing results from
the cox proportional hazard analysis and the second detailing the outcome of the fixed effects
analysis on physician splitting behavior between HOPD and ASCs. One of the major issues
that is faced with measuring hospital safety is that the distribution of the safety measures
tends to center closely around the state mean value with a long right tail. This means that
even when the absolute number of complications is high, the physicians may not have a lot
of choice in terms of choosing a practice setting because the hospitals may be similar to each

75
other in safety record. The figure 3.2 highlights this trend and shows that this is more of a
problem for the surgery safety measure and not so much for the nursing safety measures of
the hospital. A likely outcome of such performance across hospitals is that there may be a
safety threshold above which it is easy to spot a poor performer for both physicians as well
as their patients. To account for this likelihood then, I discretize the safety measures and
consider a hospital as high-risk, for nursing or surgery, when the corresponding safety rate
is one standard deviation above that of the average safety rate in the hospital’s cohort.
The hospital cohort is defined using the physician’s primary address. Physicians, like
their patients, are likely to travel small distances to their practice location and therefore,
when choosing a hospital to practice at and send their patients to, they will choose out of
those hospitals that are located close to their own primary address. Similarly, the relative
quality of these hospitals is what matters the most to a physician when accounting for
patient safety issues. The data indicate that a Florida physician on average travels 12
miles, with the median distance of 3 miles, to her principal hospital. Therefore, to be
inclusive I constructed a 15 mile radius around the physician’s primary location and used
the average nursing and surgery safety rate of all hospitals in the radius to construct the
one standard deviation measures (nursing and surgery) of hospital safety. One of the issues
with the method could be the rare event when a physician chooses or adjusts location based
on the safety of the nearby hospitals. This is not a problem in the empirical analysis, as
I find that the physicians choose office address depending upon where the main office of
their group practice is located and when not a part of a group practice they use their home
address as the primary address. Another safeguard built in the analysis is the fact that in
the data sample the physician address stays constant throughout the analysis period.

3.5.1 Hazard Analysis

The coefficient estimates from cox proportional hazard model are presented in Table
3.3. After accounting for the physician location specific safety measures, we are left with a
total of 1336 physicians who form the overall at-risk group across the time period of 1997-
2008.10 Out of these, 737 physicians adopt ASC practice by the end of the period in 2008.
10
A total of 207 physicians are dropped from the analysis as a result of using physician location specific
safety measures. To test if the hazard results were affected by this change in the sample size, we estimated

76
We basically estimate three specifications, differing from each other due to the different
‘star’ status measure included. All the specifications also include principal hospital fixed
effects. The variables are divided into four major groups reflecting - physician practice
profile, physician quality measures including ‘star’ status, opportunity for or ease of moving
practice and hospital safety. Specification (1) uses post graduate training from a top-30
school as the ‘star’ status measure, specification (2) includes the US News ‘honor roll’
schools to define ‘star’ status while, the last specification (3) uses the ‘Castle Connolly’ list
of top doctors to identify ‘stars’. The coefficient estimates are found to be quite similar
across the three regressions with the exception of the coefficients on ‘star’ status measures.
Physician practice profile is primarily summarized by her average hospital inpatient
and outpatient volume in the last six quarters. Physicians with a substantial inpatient
component to their practice are not as likely to join ASCs given that the majority of their
work is based at the hospital and it is far more convenient to carry out the outpatient
work at the same hospital’s outpatient department. A big outpatient component, on the
other hand, pulls the physician towards the ASC setting by providing an incentive to gain
not only more control on the practice but to also benefit from the profit opportunity in
the form of ASC ownership as the physician can then benefit from an already established
reputation in the outpatient arena. The results in all three specifications in the Table 3.3
support this claim. I find that an increase in the average quarterly outpatient volume by 100
patients increases the probability of joining ASC practice by 9%. 11 The impact of lagged
hospital inpatient volume is much stronger and as hypothesized, in the opposite direction.
An average increase of 100 inpatients in a quarter, decreases the probability of joining an
ASC by approximately 31%.
It was expected that the hazard of joining an ASC will be weaker for the physicians with
faculty responsibilities as they are likely to be involved with the conflicting task of teaching
and research besides handling their own medical practice. Also, the hospital environment
the same model specification but with actual safety rates from the last year on two different samples, one
with 1553 physicians and the other with 1336 physicians. The results were found to be qualitatively similar
suggesting no bias was introduced due to the change in sample size as a result of inclusion of discretized
safety measures
11
The table presents coefficient estimates that can be converted to hazard ratios by taking the exponent
of the coefficient term. For instance, the coefficient on lagged outpatient volume measure in specification
(1) exp(0.08815) is equal to the hazard ratio of 1.092 implying a 9% increase in the hazard.

77
is more likely to suit faculty pursuits. However, the results do not suggest a significantly
negative impact of faculty appointment on the probability of joining an ASC. Experienced
physicians, on the other hand, are found to be more drawn towards hospital practice as
a 1 year increase in experience decreases the likelihood of joining an ASC by 3%. The
majority of the variables on physician quality and hospital safety seem to support the main
hypotheses but the ‘star’ status fails to provide a strong evidence in the hazard results.
Results indicate that the physicians who have been granted admitting privileges at various
hospitals are significantly more likely to adopt ASC practice. As mentioned before, this
variable is considered to be an important measure of physician quality as the hospital
boards have their own credentialing mechanism that aims at providing privileges to good
quality doctors. Physicians with a postgraduate degree from a top-30 school are also 15%
more likely to adopt ASC practice, but the coefficient estimate is not found to be significant
(p value of 11%). The number of publications has been entered with a squared term and
captures the convexity of the impact of the number of publications by physicians who do
not have faculty appointments on their probability of adopting ASC practice. Both the
stand alone term as well as the squared term are significant at 10% (p value 7%) and 1%
respectively indicating that the physicians with the number of publications greater than
10 have a significantly higher probability of adopting ASC practice. ‘Star’ status measures
constructed using ‘honor roll’ and ‘Castle Connolly’ list have a positive but insignificant
impact on the hazard of ASC adoption. The dummy variable indicating board certification
status of the physician is also insignificant as more than 90% of the physicians in the at-risk
group are board certified.
On the other key variables of hospital safety, I again find support for the main hy-
potheses. However, the impact is concentrated largely on the nursing measure as the safety
measure is positive but insignificant. The interpretation of the safety measure has to be
understood in terms of the impact of increasing complications at the hospital on physician’s
probability of adopting ASC practice. The hazard results indicate that if a physician’s prin-
cipal hospital is a high-risk hospital due to a high number of nursing complications then
there is a 41% higher probability that the physician will adopt ASC practice. The result is
significant at 5% (p value 2%). This result also makes sense in the context of the anecdo-

78
tal evidence from the health care market. One of the reasons ASCs have become popular
among physicians and patients is the convenience factor. Due to their smaller size, and a
higher nurse to patient ratio than hospitals, it is easier for them to attract business. Fur-
ther, physicians cannot exercise much administrative control on nursing quality and nursing
intensity issues at the hospitals and therefore, when frustrated with the quality, may find
it easier to divide their work over more hospitals or between hospitals and ASCs. In re-
sults not presented here, the hospital safety measures were also interacted with the doctor
quality measures to evaluate the hypothesis that high-quality doctors may be more likely
to leave poor safety hospitals. I did not find evidence for the hypothesis as the interaction
terms turn out to be insignificant. It is difficult to comment on the precise reason behind
lack of evidence for this hypothesis. Other specifications with the inclusion of interactions
between safety measures and physician specialty were also estimated and the results did not
suggest any differences in physician response to safety problems at the hospital by medical
specialty.12

In summary, the results from the proportional hazard model generate new evidence
supporting the claim that high-quality doctors are the first ones to adopt ASC practice
and poor safety at the hospitals is one of the driving factors behind the migration to
ASCs. However, it is also found that physicians tend to move a portion of their outpatient
practice to ASCs and not the complete practice indicating that hospital privileges still
remain an important aspect of physician’s practice. Results suggest that physicians may
get frustrated with the safety issues at the hospital, but their punishment to the hospital
by taking business out of the hospital, may vary with their own characteristics as well as
their ownership status at the ASC. The next section provides further analysis of physician
decision of splitting outpatient work between hospital (HOPD) and ASC, once the physician
has moved part of the practice to an ASC.

12
In these regressions, four major medical specialties were included with other specialties forming the base
group. Three of the four physician specialties were the ones that include procedures covered in Chapter
2 - Internal Medicine (Procedures in CCS 3 - Digestive System), Orthopedics (Procedures in CCS 1&14 -
Nervous & Musculoskeletal System) and Ophthalmology (Procedures in CCS 9 - ). The fourth category
covered all physicians with specialty in General Surgery. The results in these regressions, with respect to
most variables, match the ones presented here and further suggest that physicians with specialty in Internal
Medicine, Orthopedics or Ophthalmology are significantly more likely to adopt ASC practice.

79
3.5.2 HOPD-ASC Split Analysis

The second part of the analysis focuses only on physicians who adopted ASC setting
and moved part of their practice to an ASC or multiple ASCs. There is a small proportion
of physicians who move their practice exclusively to ASCs but they have been ignored in
this analysis. 13 The main objective of the second stage logistic model is to explore the
factors that affect the ratio of physician practice split between HOPDs and ASCs, primarily
the role of ASC ownership. Due to the restrictions mentioned in section 3.3.1, we are not
able to account for a certain number of physicians in the hazard analysis who adopted the
ASC practice because either there were not enough number of quarters on them before they
adopted ASC or because they started out their practice at an ASC. In this analysis, those
physicians are included too which brings up the number of physicians who split their practice
to 1097. The dependent variable in the regression is the log odds of the work done at an
HOPD versus all the ASCs physician practices at. There are 3 specifications estimated,
just as in the case of hazard analysis, varying by the ‘star’ status measure included. The
results are presented in the Table 3.4.
Most of the variables in the second stage regression are similar to the ones used in the
hazard analysis. Since the outpatient volumes will be endogenous in this model, inpatient
volumes are used, to account for the physician practice profile, in the previous quarter at
the same hospital where the outpatient department is located. The inpatient volumes in-
dicate the depth of a physician’s practice and are likely to make the physician relatively
more loyal to the hospital. Physicians for whom inpatient treatments is a major practice
component, maintaining relationship with the hospital should be more important and as
a result their response to the safety performance issues at the hospital may be less elas-
tic. I include the same physician quality measures as before in the hazard analysis and
also include a variable indicating the distance of the physician to the HOPD in question.
The additional set of variables in this regression are the ownership indicators and their
interaction with the hospital safety measures. The ownership variables, as defined before,
include indicators for - physician owners with controlling interest in an ASC with no cor-
porate involvement, physician owners without controlling interest in a joint venture (JV)
13
Since such physicians will have a share of work at ASCs that is equal to 100%, the log odds for them
will be undefined.

80
with a hospital corporation and physician owners with controlling interest and in a JV with
a corporation specializing in ASC business (ASC Corporation). The residual category is
physicians without controlling interest in an ASC that has no corporate involvement either
(non-owners). These ownership measures are interacted with the safety indicators for the
hospital to highlight the differential impact that hospital safety performance has on division
of work by physician owners with and without controlling interest an ASC.
The results in this part provide further insight into the physician-hospital-ASC dynamic.
The physician ‘star’ status measures are found to be much more informative in this analysis.
Physicians who completed postgraduate training at a top 30 school or were listed as a top
doctor in the ‘Castle Connolly’ publication have approximately 5% lower odds of treating
their patients at a HOPD vis-a-vis an ASC. However, physicians graduating from a ‘honor
roll’ school have 14% higher odds of favoring an HOPD over the ASC environment for their
patients. A possible reason for this discrepancy is that the ‘honor roll’ ranking includes
only 14 hospitals from across the nation and may be a little too narrow in its scope when
accounting for physician quality (While 35% of physicians graduated from top 30 school,
16% are listed as top doctors in ‘Castle Connolly’ publication, only 11% graduated from an
‘honor roll’ school). The measure on publications by physicians without faculty privileges,
unlike the hazard model, is not significant in its impact on the physician decision of splitting
practice between HOPD and ASCs.
The most interesting results, however, come from the ownership variables and their
interaction with safety. Physicians who have a controlling interest in an ASC (JV with
ASC corporations like AmSurg and Novamed or without a JV) are significantly less likely
to treat their patients at a HOPD even when there are no safety issues at the HOPD.
The odds are 44% lower in case of only physician control and 52% lower when a ASC
corporation is involved, compared to non-owners. This finding is in line with conventional
wisdom as ownership provides the profit incentive and therefore the physicians prefer to
perform majority of the services at the ASC. However, interestingly, the odds of treating
patients at an HOPD decrease further to 57% and 64% lower when the HOPD in question
is high-risk due to nursing complications. The surgery complications, on the other hand,
turn the odds to 50% lower in case of controlling interest physicians. 14 In case of ASC
14
Results in chapter 2 of this dissertation indicated that safety is more of a concern with respect to ASCs

81
Corporation JVs, the surgery complications do not seem to impact the physician’s choice
of practice setting. In contrast, physicians who do not have a controlling interest and are
operating at an ASC jointly owned by a hospital chain do not seem to get affected by the
safety issues and also have roughly 13% higher odds of treating their patients at the HOPD
when compared to non-owners. This makes sense if one takes into account the fact that for
many physicians, ownership in a joint venture with a corporation like HCA or with a hospital
entity could just be an extension of the hospital practice and therefore, be guided by purely
profit earning incentives or by the administrative decisions that were originally taken at the
hospital. Finally, it is important to note that the results in the log-odds model are based
on physician level grouped data which can introduce the problem of heteroscedasticity in
the model as the observations on the same physician will not be independent of each other.
The fact that the same physician is observed over time can also cause the same problem.
The problem can also exacerbate because first, the same physician may be treating patients
at more than one HOPD during the same quarter and second, different physicians may
be heterogenous in terms of practice volumes and the share of practice will not be able to
account for size differences in physician volumes. Appendix C of this dissertation takes this
feature of the model into account and further confirms the relationship between ownership
and hospital safety with the help of two alternative standard error corrections.
Turning to the other covariates, inpatient volume in the last quarter has the same
qualitative effect as the inpatient volume measure in the hazard analysis. As in this case,
physicians with big inpatient component are also more loyal to the hospital’s outpatient
department and an extra 10 inpatients increase the odds of choosing a HOPD over ASC by
9%. Physicians are also sensitive to the travel distance and an increase in travel distance by
1 mile decreases the odds by 4%. Board certified physicians and those who have a faculty
specializing in digestive system and musculoskeletal system procedures than those in eye care (see Table
2.9). To further examine those results from the perspective of physician level analysis, specifications were
estimated including interactions between physician’s medical specialty, ownership status and hospital safety.
Medical specialties primarily accounted for procedures in the area of digestive system (Internal Medicine),
eye care (Ophthalmology) and musculoskeletal system (Orthopedics) and other surgeries (General Surgery).
In results not included here, evidence supports the findings in chapter 2 and suggests that physicians with
specialization in Internal Medicine or Orthopedics are significantly more likely to treat their patients at
ASCs vis-a-vis an HOPD. Further, physicians with controlling interest ownership at an ASC and medical
specialty in Internal Medicine, Orthopedics or General Surgery are significantly more likely to respond to
safety problems by migrating their patients to an ASC in comparison to those who are non-owners and
specialize in other procedures like Urology or Otolaryngology.

82
appointment are more likely to choose HOPD’s. However, as previously mentioned, around
94% of the physicians are board certified which makes it difficult to interpret the results
on the board certification measure. In case of faculty members, the tilt of the split in
favor of HOPDs makes sense as the faculty responsibilities and incentives are more tightly
aligned with the hospital setting as both teaching as well as research work is better suited
for the hospital environment. Number of staff privileges have a significant and negative
impact indicating that higher staff privileges are likely to bring the odds of practicing at
a HOPD lower. However, unlike the hazard model, it is difficult to interpret the staff
privileges as only a physician quality measure in this case. Physicians who are practicing
at various locations are also likely to divide their work in smaller proportions across each of
the facilities they practice at. Somewhat different from the hazard analysis is the result that
conditional on adopting ASC practice, experienced physicians have a small but significantly
higher preference of ASCs over HOPDs. A 1 year increase in experience implies 1% lower
odds of working at a HOPD.

83
 3.6 Conclusion

The role played by physicians in the growth of ambulatory surgery centers cannot be
overemphasized. However, the part that hospitals and national level hospital and ASC
chains have played in the growth of outpatient healthcare industry also needs to be high-
lighted. The debate till now has centered only around the cost and utilization issues that
have come forth due to the growth of specialty facilities. This research adds another dimen-
sion to the debate by bringing safety issues at the hospitals into focus and how they have
motivated physicians to set up practice at and invest in specialty centers like ASCs. The
results in this study also put a positive light on the physician ownership of such facilities
and how ownership generates incentives that may be well aligned with patient welfare.
The findings in this study indicate that poor nursing safety is one of the leading causes
in motivating physicians to adopt ASC practice. There is a 41% higher probability of a
physician adopting an ASC when the principal hospital she works at has a high number of
medical complications due to nursing errors. The empirical analysis does not yield evidence
supporting the hypothesis that ‘star’ physicians are the first movers in the migration to
ASCs, but it is found that once a ‘star’ physician has joined an ASC, the odds of her
treating patients at the ASC are significantly higher. This evidence, however, exists for
physician quality measures that either themselves are based on a peer review system (Castle
Connolly measure) or rely on graduation from a medical school ranked among the top 30
schools in US. Other physician quality measures relying on number of hospital staff privileges
and publications also indicate that high-quality physicians are more likely to adopt ASC
practice but in case of publications the evidence favors only physicians with substantially
higher number of publications. These results also bring forward the importance of hospital-
physician relationship and the lack of complete substitutability between HOPDs and ASCs.
Physicians who treat a wide variety of patients with different medical risk backgrounds are
reliant on their hospital privileges and need access to inpatient and outpatient facilities as
well as the emergency departments at the hospital. This in many cases may weaken their
response to the safety issues at the hospital.
Further, these results capture the, often elusive, impact of ownership on the physician’s
splitting of practice across different facilities - HOPDs and ASCs. I find that physicians

84
who have controlling interest at an ASC are much better placed practice-wise in terms of
responding to the safety problems at the hospitals. Their profit incentive is well aligned
with the poor performance of hospitals they work at. Both poor surgery and nursing safety
at the hospital significantly increases the likelihood of an owner shifting patients away from
an HOPD to the ASC. Understandably, this impact does not exist either for non-owners or
for those who are non-controlling interest owners at a hospital run ASC. ASC practice for
them is likely to be just an extension of their relationship with the hospital.
One of the concerns in this study with respect to the results on physician practice
splitting behavior is the exclusion of physicians who at some point after joining an ASC
decide to exclusively practice at either ASCs or at hospitals. Around 12% of the physicians
in the data do so and lead to a potential selection problem. Physicians who have rejected the
hospital practice completely may have done so as an extreme response to safety problems
at the hospitals and the results in this study do not account for such behavior. A probit
model that estimates the probability of a physician’s practice at a HOPD may be able to
yield a correction (Heckman’s selection correction model) that can account for this selection
problem. Further, it is also important to note that all the results presented in this paper
account for the unobservable hospital characteristics with the use of hospital level fixed
effects. I also use safety measures that are lagged by a year to study the impact on individual
physician decisions. However, it is possible that hospitals learn about and keep track of
their own safety performance from year to year and take steps to improve. Such steps will
then be time varying measures that cannot be accounted for with the help of fixed effects. In
such a case, the impact of last year safety performance on the physician decision to migrate
patients to an ASC may be weakened. If such a reaction exists or if there is a selection
problem as mentioned before, then the results in this paper present a lower bound on the
size of physician response to safety issues.

85
 3.7 Tables and Figures

Figure 3.1: Specialty Wise Break-up of Physicians, Florida 1997-2008

86
Figure 3.2: Excess Nursing and Surgery Complications, Florida Hospitals 1997-2008

87
 Table 3.1: Summary of ASC Adoption Behavior

Kaplan Cumulative
Physicians Adopters New Meier Adoption
Period Censored
at risk (exits) Entrants Survivor Probabil-
Function ity

1998, q1 1030 27 0 0 0.974 0.026

1998, q2 1003 37 1 0 0.938 0.062
1998, q3 965 24 2 1 0.915 0.085
1998, q4 940 21 5 14 0.894 0.106
1999, q1 928 34 4 41 0.861 0.139
1999, q2 931 42 3 26 0.822 0.178
1999, q3 912 30 3 9 0.795 0.205
1999, q4 888 25 4 3 0.773 0.227
2000, q1 862 33 2 24 0.743 0.257
2000, q2 851 27 1 13 0.720 0.280
2000, q3 836 14 5 8 0.708 0.292
2000, q4 825 23 1 8 0.688 0.312
2001, q1 809 17 1 30 0.674 0.326
2001, q2 821 18 3 15 0.659 0.341
2001, q3 815 15 2 8 0.647 0.353
2001, q4 806 25 2 7 0.627 0.373
2002, q1 786 19 3 25 0.612 0.388
2002, q2 789 23 5 11 0.594 0.406
2002, q3 772 22 6 6 0.577 0.423
2002, q4 750 23 2 10 0.559 0.441
2003, q1 735 35 2 25 0.532 0.468
2003, q2 723 22 5 12 0.516 0.484
2003, q3 708 19 4 7 0.502 0.498
2003, q4 692 18 4 3 0.489 0.511
2004, q1 673 21 2 21 0.474 0.526
2004, q2 671 14 3 9 0.464 0.536
2004, q3 663 9 1 3 0.458 0.542
2004, q4 656 13 4 4 0.449 0.551
2005, q1 643 20 1 18 0.435 0.565
2005, q2 640 24 3 10 0.419 0.581
2005, q3 623 16 3 11 0.408 0.592
2005, q4 615 17 1 8 0.397 0.603
2006, q1 605 3 3 19 0.395 0.605
2006, q2 618 19 2 9 0.382 0.618
2006, q3 606 6 6 12 0.379 0.621
2006, q4 606 18 6 10 0.367 0.633
2007, q1 592 15 3 22 0.358 0.642
2007, q2 596 15 6 15 0.349 0.651
2007, q3 590 8 3 3 0.344 0.656
2007, q4 582 11 1 3 0.338 0.662
2008, q1 573 11 3 14 0.331 0.669
2008, q2 573 10 7 6 0.326 0.674
2008, q3 562 6 3 7 0.322 0.678
2008, q4 560 6 557 3 0.319 0.681

88
 Table 3.2: Summary of the Main Variables

Variables All Physicians ASC Adopters Non Adopters

Statistic Std. Dev Statistic Std. Dev Statistic Std. Dev
Means
Overall Volume1 166 115.34 182 126.86 141 88.18
Outpatient Volume2 137 108.19 154 119.56 111 79.47
Hospital Outpatient Volume3 93 82.25 83 82.09 109 79.89
Hospital Inpatient Volume4 34 32.70 32 30.01 36 36.46
No. of Hospitals within 6mi 4 3.47 4 3.07 5 3.97
No. of ASCs within 6mi 5 3.86 5 3.81 5 3.95
Number of Staff Privileges 3 2.08 4 2.16 3 1.87
Publications 5 22.14 2 11.07 9 30.39
Publications (Non Faculty Physi-
1 4.19 1 4.79 1 3.29
cians)
Proportions
Faculty Appointments (%) 20 15 26
Board Certification (%) 92 94 90
Postgraduate - Top 30 School (%) 33 33 33
Postgraduate - US News ‘Honor
10 10 10
Roll’ School (%)
Castle Connolly Top Doctor (%) 21 23 20
Started at an ASC Owned by a
− 38 −
Hospital Chain5 (%)
Started at a Specialty ASC6 (%) − 34 −

Number of Physicians 1543 855 688

Notes:
1. Total volume served by the physician has been averaged over all the quarters during 1997-2008 when the physician was
practicing in Florida.
2. Total outpatient volume served by the physician has been averaged over all the quarters during 1997-2008 when the physician was
practicing in Florida. This includes pre and post ASC adoption period.
3. Total hospital outpatient volume served by the physician has been averaged over all the quarters during 1997-2008 when the physician
was practicing in Florida. This includes pre and post ASC adoption period.
4. Total inpatient volume served by the physician has been averaged over all the quarters during 1997-2008 when the physician
was practicing in Florida. This includes pre and post ASC adoption period.
5. A physician is considered to be starting out at an ASC owned by a hospital chain when the first ASC that the physician starts practice
at is owned by a hospital or hospital chain like HCA.
6. A physician is considered to be starting out at a specialty ASC when the first ASC that the physician starts practice at
specializes in a major body part or body system. For eg: Eye care centers, gastroenterology or urology centers.

89
Table 3.3: Coefficient Estimates from Cox Proportional Hazard Analysis

Specifications1
(1) (2) (3)
Honor Castle
Variables Top 30
Roll Connolly
Hospital Safety1
High Risk (Nursing) 0.346b 0.346b 0.347b
(0.146) (0.146) (0.146)
High Risk (Surgery) 0.039 0.044 0.043
(0.143) (0.143) (0.143)
Physician Quality
‘Star’ Status 0.143 0.084 0.021
(0.092) (0.139) (0.1)
Staff Privileges 0.079a 0.081a 0.081a
(0.027) (0.027) (0.027)
Publications3 -0.037c -0.034c -0.034c
(0.021) (0.02) (0.02)
Publications Squared 0.001a 0.001a 0.001a
(0.000) (0.000) (0.000)
Board Certification -0.004 -0.008 -0.013
(0.096) (0.095) (0.096)
Physician Practice Profile
Outpatient Volume (Avg. last 6 qtrs) 0.088c 0.094c 0.095c
(0.052) (0.052) (0.052)
Inpatient Volume (Avg. last 6 qtrs) -0.369b -0.386a -0.39a
(0.154) (0.154) (0.154)
Faculty Appointment -0.114 -0.098 -0.094
(0.13) (0.13) (0.13)
Experience -0.03a -0.03a -0.03a
(0.005) (0.005) (0.005)

Hospitals within 6 mi -0.079a -0.080a -0.082a

(0.031) (0.031) (0.031)
ASCs within 6 mi 0.012 0.014 0.015
(0.024) (0.024) (0.024)
Notes:
1. All specifications include principal hospital fixed effects

2. High-Risk Nursing and Surgery variables are binary indicators of high-risk

status of the hospital due to nursing or surgery complications. A high-risk hospital
has a surgery complication rate 1 standard deviation greater than the average
complication rate of its cohort.
3. The variable publications measures the number of articles published by physicians
who do not have faculty appointments.
a
p-value < 0.01
b
p-value < 0.05
c
p-value < 0.10

90
 Table 3.4: Coefficient Estimates from the Log Odds Regression

Dependent Variable : Log Odds of HOPD Practice v/s ASC Specifications1

(1) (2) (3)
Castle
Explanatory Variables Top 30 Honor Roll
Connolly
Physician Ownership & Hospital Safety2
Controlling Interest Owner3 -0.590a -0.599a -0.591a
(0.024) (0.024) (0.024)
Controlling Interest Owner * High-Risk (Nursing) -0.260a -0.233a -0.243a
(0.062) (0.062) (0.062)
Controlling Interest Owner * High-Risk (Surgery) -0.118b -0.123b -0.116b
(0.060) (0.060) (0.060)
Controlling Interest Owner (ASC Corp JV)4 -0.741a -0.740a -0.735a
(0.030) (0.030) (0.030)
Controlling Interest Owner (ASC Corp JV)* High-Risk (Nursing) -0.299a -0.286a -0.300a
(0.076) (0.076) (0.076)
Controlling Interest Owner (ASC Corp JV)* High-Risk (Surgery) -0.028 -0.056 -0.033
(0.074) (0.075) (0.074)
Non Controlling Interest Owner (Hosp Corp JV)5 0.120a 0.119a 0.123a
(0.022) (0.022) (0.022)
Non Controlling Interest Owner (Hosp Corp JV)* High-Risk (Nursing) -0.058 -0.047 -0.052
(0.054) (0.054) (0.054)
Non Controlling Interest Owner (Hosp Corp JV)* High-Risk (Surgery) 0.038 0.031 0.035
(0.053) (0.053) (0.053)
High-Risk (Nursing) 0.171a 0.156a 0.166a
(0.046) (0.046) (0.046)
High-Risk (Surgery) 0.036 0.042 0.036
(0.045) (0.045) (0.045)
Physician Quality
‘Star’ Status -0.054a 0.128a -0.048a
(0.016) (0.024) (0.019)
Publications6 0.001 0.001 0.001
(0.002) (0.002) (0.002)
Board Certification 0.082a 0.084a 0.087a
(0.019) (0.019) (0.019)
Physician Practice Profile
Inpatient Volume Previous Quarter 0.009a 0.009a 0.009a
(0.001) (0.001) (0.001)
Staff Privileges -0.014a -0.013a -0.014a
(0.005) (0.005) (0.005)
Faculty Appointment 0.162a 0.152a 0.160a
(0.023) (0.023) (0.023)
Experience -0.002a -0.002b -0.001c
(0.001) (0.001) (0.001)
Distance to the HOPD -0.038a -0.040a -0.038a
(0.002) (0.002) (0.002)
Number of Physicians = 1097

Notes:
1. All specifications include fixed effects for the HOPD
2. High-Risk Nursing and Surgery variables are binary indicators of high-risk status of the hospital due to nursing
or surgery complications. A high-risk hospital has a surgery complication rate that is 1 standard deviation greater
than the average.
3. Controlling Interest Owners are physicians who have a 5% or greater ownership share in the ASC. Such ASCs are
either individually owned or by a small group of physicians and many of these ASCs are either single specialty facilities
or small size multi-specialty facilities.
4. Controlling Interest Owners in a JV with a ASC Corporation such as AmSurg or Novamed. These physicians also
have a 5% or greater ownership share in the ASC. The corporation tends to own the management with 51% share
and the rest is generally owned by physician group(s).
5. Non-Controlling Interest Owners are in a JV with a Hospital Corporation such as HCA or Health South. These
physicians do not own a controlling interest and generally at such ASCs, the corporation tends to
own the management with greater than 51% share and the rest is divided among 40-60 physicians.
6. The variable publications measures the number of articles published by physicians who do not have faculty
appointments.
a
p-value < 0.01
b
p-value < 0.05
c
p-value < 0.10

91
 CHAPTER 4

CONCLUSION

The growth of ambulatory surgery centers has wide reaching effects on the outpatient health-
care sector. ASCs have the potential to impact not only the patient welfare but also the
financial goals of policy makers. In this dissertation, the somewhat controversial issue of
ASC development and physician involvement in the form of ownership is studied, for the
first time, in an entirely different but important context of hospital safety.
The recent health literature on specialty facilities such as ASCs and specialty hospitals
has largely focused on the resulting changes in utilization and the role of physician own-
ership wherein the ownership has generally been assumed to be per se a negative aspect
of ASCs due to the likely conflict of interest. Prior literature has also failed to find favor-
able evidence of the impact on patient welfare. Results have indicated that ASCs either
lead to increased number of procedures (Courtemanche and Plotzke (2010), Plotzke and
Courtemanche (2010)) or provide a welfare benefit that is not substantial (Weber (2010)).
However, in asking the question of ASC impact on patient welfare, the role of low hospital
safety has been ignored. If ASCs tend to come up in areas with hospitals that have a
poor track record in patient safety then their growth could be a good outcome for patient
welfare. Even when the investment in ASCs is driven by a profit incentive, poor safety at
the hospitals is aligned well with this incentive. Poor performance by incumbent hospitals
generates the window of opportunity in the form of dissatisfied patients who can then be
treated at a smaller, focused and convenient alternative in the form of ASCs. Further,
physicians themselves play a key role in the growth of ASCs and their desire to gain more
control over their own practice and to disassociate from safety problems at the hospital
also plays an important role in ASC development. If high quality physicians tend to treat

92
more of their patients at ASCs then it is another aspect of the ASC growth that indicates
possible increase in patient welfare.
One of the main contributions of this dissertation is to highlight the invaluable role of
hospital safety in promoting competition in the outpatient healthcare market. This has been
achieved with the help of empirical analysis that utilized the detailed patient and physician
level data that are available in the state of Florida. In chapter 2 I use individual patient
level data to estimate a patient demand model. I rely on a nested logit framework that is
extremely well suited to extract information when there are a large number of individual
observations and control for various patient and facility characteristics that affect patient
choice. For instance, attributes such as distance of the patient from the provider and payer
type of the patient are major factors affecting the patient decision to visit the provider. In
the same model, I also account for hospital safety measures based on nursing and surgery
complications. The choice model then provides an understanding of the relative role and
importance of such factors in patient’s choice of healthcare facility. The results indicate
that a ‘high-risk’ hospital, based on occurrences of nursing complications, faces a choice
probability that on average is 1 to 2 percentage points lower than the other hospitals. With
a second measure, based on occurrences of surgery complications, this impact is smaller
but significant and indicates on average 0.5 to 1 percentage point lower probability of being
chosen by the patients. If the closest hospital to the patient turns high-risk on both nursing
and surgery safety then the impact on choice probability is even stronger, the hospital looses
4 to 9 percentage points of it’s market share in the early years of 1998 to 2002. Almost 50
to 90% of the market share lost by the closest hospital is then absorbed by the ASCs in
patient’s choice set.
The application of nested logit choice model also helps to take care of a major problem
that specifically affects the results from the popular alternative of conditional logit modeling.
This problem is the property of Independence of Irrelevant Alternatives (IIA) that underlies
conditional logit models, and leads to the assumption that in one to one comparison of two
related choices, an existing third choice is completely irrelevant. The nested logit model
helps to get around this assumption by allowing for different substitution patterns in both
the nests - HOPDs and ASCs. This is an important benefit of the methodology as it not only

93
weakens the IIA assumption but also strengthens the market share predictions by taking
into account the fact that new ASC entry will take away business from both hospitals and
ASCs but certainly in different magnitudes.
Another contribution that the chapter makes to the study of new firm entry is by
providing an alternative application of the model built by Chernew (2002). Unlike their
application, I use the choice model estimates to construct predicted volumes that a potential
ASC entrant can aim to serve by entering a particular location, in this case a particular
zip code in a county. This method helps to get around the potential endogeneity of actual
facility volumes that may be correlated with unobserved quality of the facility as well as
provides a measure of potential servable market size even in an area where the entry does
not take place. The results from the entry model further support the claim that safety
problems encourage entry by ASCs. Probit model estimates suggest, after accounting for
expected profitability, an increase of 2 to 4 percentage points in the entry probability of an
ASC when there is an increase in the number of high-risk hospitals in the market service
area of the entrant or alternatively, when there is a high volume of patients being treated
at high-risk hospitals.
In sum, the empirical findings from chapter 2 reveal that hospital quality and safety
record is inherently a factor in the emergence of ASCs. High-risk hospitals are not as
attractive to patients and therefore, open up room for new ambulatory surgery centers to
enter the market and attain greater volume shares. This impact is strongest when the
closest hospital to the patient is a high-risk hospital; in such a case not only the closest
ASC to the patient but all ASCs in a patient’s choice set stand to gain. Further, even when
the expected profitability is accounted for, there is a residual impact of safety problems
observed in the form of results that suggest that an increase in the number of high-risk
hospitals in the market service area of the entrant significantly increase the likelihood of
ASC entry.
It is also important to note that hospitals need not be passive participants in the mar-
ket and may react to ASC entry by adjusting the safety covariates. The results here are
safeguarded by using safety measures in the entry model that are lagged by a year. Further,
it is also worth noting that the response to new ASC entry will be strongest by hospitals

94
located closest to the entry point while, the safety measures used in the analysis capture
the performance of hospitals spread over the entire market service area of the ASC that on
average has more than 8 hospitals. However, endogeneity could still be an issue due to the
possible correlation between safety measures over the years. This problem is more likely
when nearby hospitals adjust their risk status by making strategic investments that have
a direct impact on safety and that take effect gradually over years. Future research will
address this issue with a joint model that incorporates the hospital response.
The results from chapter 2 indicate that even after controlling for profitability measures
that themselves take into account the safety at incumbent hospitals, there is still a significant
impact of measures of safety on entry. This is a possible indication of the importance of
other factors that lead to the emergence of ASCs. One of such factors, is the physician
herself. As suggested before, physicians have played a key role in the growth of ASCs. Their
entrepreneurial zeal combined with their desire for a higher degree of control over procedure
scheduling and other facility level operations (Poole (1999), Mechanic et al. (2005)) has
been a driving force. Thus, another important contribution of this dissertation is to provide
evidence on physician role while, for the first time identifying physician ownership in the
ASCs. Also, in chapter 3 the issue of hospital safety is further examined from a physician
owner’s point of view.
Tracking down physician ownership in medical facilities is a challenging task. The
previous literature has often depended upon using the physician’s work load and practice
share at facility to identify ownership links. However, measures that are based on physician
share of work are likely to be endogenous to any model interested in extracting the impact
of ownership on physician practice volumes. Using all (more than 1200) ASC licensure files,
starting from the year of ASC incorporation, along with the annual reports submitted by the
ASC owners to the state of Florida I assessed the ownership interests of various physicians
practicing at the ASCs. Where necessary, ASC websites were used to complete the missing
information. I also track down the ownership interests of various national hospital and ASC
corporate chains in the ASCs located in Florida with the help of the licensure files combined
with the web search on corporate owners. The exercise yields rich information on not only
the controlling and non-controlling interest owners but also on the ownership structure of

95
these centers. For instance, the extent of hospital involvement in the ASC business was
unknown until now. According to the licensure files, 39 out of roughly 400 ASCs in Florida
are owned by hospitals such as Sarasota Memorial, Sacred Heart and Baptist Health. Such
ownership pattern is likely to weaken the impact of safety on ASC emergence as well as on
physician migration to ASCs as a response to safety problems.
In chapter 3 I also examine another issue of physician involvement in ASCs, which is the
quality of the physicians who adopt ASC practice. Measuring physician quality, however,
is a complex issue, as has been established in previous health literature. For robustness
of the results, different measures of physician quality are used to study the propensity of
high quality physicians to practice at ASCs. I use ‘star’ status of the physician along with
the number of staff privileges and the number of publications as measures of quality. ‘Star’
status is itself defined in three different ways. Physicians who complete their post graduate
training (residency or fellowship) at a school ranked among the top 30 medical schools in
US (using the US News & World Report and school ranking according to NIH funding),
or alternatively physicians who are listed as top doctors in the Castle Connolly publication
‘America’s Top Doctor’s’. As an extremely selective1 measure of ‘Star’ status I also use post
graduation from the US News honor-roll schools that includes only 14 hospitals nationwide.
In using ‘Star’ status as a measure I follow work by Burke et al. (2007), Burke et al. (2009)
and Gardner and Vishwasrao (2010) who have had reasonable success with the measure. I
expand on their work by using alternative measures of the quality status.
The main findings from the chapter further provide support to the importance of safety
in emergence of ASCs through a direct impact on physician willingness to practice at a
high-risk hospital. I find that poor nursing safety is one of the leading causes in motivating
physicians to adopt ASC practice. There is a 41% higher probability of a physician adopting
an ASC when the principal hospital of the physician becomes ‘high-risk’ due to nursing
errors. Unlike chapter 2, however, I find more support for and a stronger impact of nursing
safety than surgery safety on physician’s decision. This result may be an indication of
1
The extreme selectiveness of this measure should be obvious from the fact that to be selected a hos-
pital had to have a score that is 3 standard deviations above the mean score in each relevant medi-
cal specialty. Explanation of honor-roll system is provided in detail on the US News & World Report
website - ”http://health.usnews.com/health-news/best-hospitals/articles/2010/07/14/best-hospitals-2010-
11-the-honor-roll”.

96
the possibility that a physician can take steps to ensure a surgically safe procedure for
the patient and therefore be less elastic to those problems at the hospital, while in case
of nursing safety, the error will lie outside of the physician’s control and her response may
be more elastic to such errors at the hospital. The results also indicate that even though
‘star’ physicians may not be more likely to adopt ASC practice, they are significantly more
likely to treat patients at the ASC as against the HOPD. Physicians with a high number
of hospital staff privileges are also more likely to adopt ASC practice. Such physicians are
considered to be high quality too as staff privileges are a likely indicator of physician demand
across different hospital locations. Although typically ASC physicians do not publish much
(1 publication for those without faculty responsibilities), I find that physicians with greater
than 10 publications are significantly more likely to adopt ASC practice.
A major contribution of this chapter to the literature is the exploration of the relation-
ship between physician ownership and physician response to safety problems at the hospital.
As one would expect, physicians with controlling interest are significantly more likely than
non-owners (without controlling interest) to treat their patients at an ASC. For instance, I
find 44-50% lower odds of an owner’s practicing at an HOPD compared to the ASC setting.
These odds fall further down if the HOPD in question is high-risk due to nursing or surgery
complications, implying that physician owners are more responsive to the safety problems
at the hospital. This could happen for multiple reasons. Owners, driven by profit incen-
tive, may find poor safety performance as an opportunity to shift their patients away from
the hospital to their own ASC. At the same time the results are also an indication of the
flexibility that ownership provides to the physicians in terms of responding to the safety
problems at the hospital by scheduling surgeries away at their own ASCs. Non-owners on
the other hand may be stuck with the hospitals where they have staff privileges. Thus, this
aspect of physician ownership may actually be beneficial for the patients.
Another feature of ASC ownership that comes to light in this dissertation is the dif-
ferential impact of ownership when hospital chains are involved. Physicians who work at
ASCs that are owned by the hospital itself are first, not more likely to treat their patients at
the ASC than at HOPD, and second, they are also not responsive to safety problems. The
behavior indicates that ASC practice for such physicians is likely to be just an extension of

97
their relationship with the hospital and such ASCs should accordingly be treated differently.
One of the weaknesses of the physician level analysis is the likelihood that hospitals may
improve with time, especially when bad safety performance goes public or when the gradual
exodus of patients as well as physicians to ASCs reaches a critical point. Since the analysis
relies on individual physicians as the unit of observation and lagged year safety measures
for the hospital are used, endogeneity of the safety measures is not likely. However, the
correlation of hospital response with patient migration over time may still have an effect
on our results. In that case, the impact of last year safety performance on the physician
decision to migrate patients to an ASC may be weakened and the results in may present a
lower bound on the size of physician response to safety issues. In future research, a possible
check for this effect would be through the application of instrumental variables technique.
However, it may be difficult to find measures of hospital characteristics that affect safety at
the hospital but not the physician share of work.
Another issue faced in this dissertation that future research should be better able to
deal with is the lack of availability of measures on ASC quality. An attempt was made to
construct useful measures of quality that tried to take into account information available in
the AHCA data. However, I found that patient level diagnosis information in the data is
not rich enough to measure quality at outpatient facilities. The AHCA data also includes
information on patient discharge status at ASCs which could have been an important mea-
sure on ASC quality as patient discharge to a hospital from an ASC or patient death at an
ASC can be useful measures indicating potential complications of ASC treatment. However,
again the attempts to find a quality measure were unsuccessful due to lack of variation in
the data, as on average across the data years, only 0.06% of the ASC patients were ever
transferred to a hospital and 0.0004% of the patients died at the ASC. As mentioned before,
future studies should be able to deal with this problem as over time with the proliferation of
ASCs across the country the policy interest in them has also increased manyfold. Therefore,
it is likely that detailed data relating to patient level diagnosis as well as ASC infrastructure
that is required to construct quality measures should also be readily available in the near
future. In the meanwhile, one of the ways this dissertation gets around the problem of ASC
quality is by utilizing the measure of ASC distance to the emergency center as a quality

98
indicator. Patients worried about potential medical complications, all things similar, would
prefer an ASC that is located closer to an emergency center than farther away. Results in
chapter 3 support this hypothesis.
In this dissertation, importance has also been placed on the relevance of distance to
all the participants in the market including physicians as well as the patients. The health
economics literature frequently recognizes the role that travel time and distance play in
affecting the patient demand (Gowrisankaran and Town (2003), Kessler and McClellan
(2000), Tay (2003) etc.) for health care providers. I follow the tradition and include various
measures that capture the sensitivity of the patients towards distance to different facilities
and also that of the physicians towards practice locations. While most of the literature
relies on the straight line distance, I include actual travel distances between patients and
their healthcare choices among hospitals and ASCs. The more precise measure of distance
arguably provides us with better predictions on the expected volumes that a potential
ASC entrant can expect to serve. The results suggest that patient choice is significantly
and negatively affected by the increase in distance to the hospital or the ASC. Distance
also plays an important role in case of riskier procedures as in such cases patients prefer
ASCs that are located closer to the emergency centers. Keeping in mind the importance
of distance and a similar desire of the physician for convenience as the patient, the safety
measures constructed for hospitals in chapter 3 are based on the distances between the
physician primary address and the nearby hospitals.
Finally, an important contribution of this dissertation is to provide health policy makers
with a better understanding of and actionable information on ASCs. While, the growth of
ASCs has obviously increased patient choice and has provided patients with a convenience
factor due to proximate location of ASCs, the results in this dissertation suggest that these
centers provide an alternative to poor safety at the hospitals. Urban areas with high-risk
hospitals are more likely to witness ASC entry and due to a possible loss in market share,
hospitals may get more attentive towards the safety environment they provide to their
patients. Thus, increased competition in outpatient care has a potential to improve overall
patient safety in the market. Results also show that high quality physicians are more likely
to treat their outpatients at ASCs than at HOPDs providing further evidence of benefits of

99
treatment at an ASC. High quality physicians in turn may prefer ASCs due in part to the
higher degree of control that is achievable at a smaller, more focused healthcare delivery
system. Lastly, evidence suggests that physician ownership should not be treated as a
negative aspect of ASCs in general and probably more thought needs to be given to this
issue. Results indicate that even though ownership leads to a migration of patients away
from the HOPD, it also provides the physician with an incentive to react to the poor safety
performance of the hospital which in itself is a beneficial outcome for patients.

100
 APPENDIX A

DALY-ZACHARY-MCFADDEN CONDITIONS

According to the Daly-Zachary-McFadden (DZM) conditions in Börsch-Supan (1990), for

consistency with stochastic utility maximization, the dissimilarity coefficients are required
to lie within the unit interval. This condition ensures the non-negativity of the density
function. However, Börsch-Supan (1990) pointed out that the condition is too stringent
and suggested that the nested logit specification should be viewed as an approximation to
the true underlying demand system and that the stochastic utility maximization should not
be expected to hold globally, but only for the data points sensible for specific application
of the choice model.
Herriges and Kling (1996) while correcting the mistake in Börsch-Supan (1990), lay
down the relaxed consistency conditions. These conditions, in turn, are derived from the
differentiation of the joint probability term in the nested logit. Relying on their theorem
1 and its corollary, the necessary restrictions (we call them Börsch-Supan conditions), in
a two level nested logit model, on the dissimilarity coefficients using the terminology from
this paper can be written as:
1
λk ≤ , k = hospital or ASC
1 − PiBk
and,
4
λk ≤
3(1 − PiBk ) + [(1 + 7PiBk )(1 − PiBk )]1/2
Here, PiBk is the marginal probability of choosing nest k and λk is the dissimilarity
coefficient. Out of the two conditions given above, it’s the second condition that is more
restrictive and provides an upper bound that is comparable to the DZM’s bound of 1
for global maximization. Herriges and Kling (1996) in their paper, provide a descriptive

101
analysis of what upper limits these conditions imply for different nest choice probabilities.
For instance, with the nest choice probability of 0.5, the upper bound comes out to be 1.28.
We estimated nested logit models for a twelve year period from 1997 to 2008, for three
CCS type models of Nervous System and Musculoskeletal System, Eye Care and Digestive
System. The dissimilarity coefficients in our estimations from 1997, 1998 and 1999 are
much higher than 1 for the hospital nest, especially in the case of Eye Care models. Using
the average marginal probability of nest choice from our estimation sample, we find that
the Börsch-Supan conditions are easily met in all 36 of our estimations, except for the Eye
Care models in year 1997, 1998 and 1999. The limits implied and the actual dissimilarity
coefficients, for the results not presented here, are available on request from the authors.

102
 APPENDIX B

ENTRY MODEL - SPECIFICATION

ROBUSTNESS

Table B.1: Standardized Coefficient Estimates - Entry Model (Specification Check)

High − RiskN ursing High − RiskSurgery

a
V
d ol - Multi-Specialty 0.195 0.175b 0.168b 0.158b
(0.081) (0.081) (0.081) (0.081)
V
d ol - Nervous & Musculoskeletal 0.199a 0.181b 0.181b 0.170b
(0.081) (0.081) (0.079) (0.081)
V
d ol - Eye Surgery 0.040 0.017 0.020 0.010
(0.084) (0.084) (0.085) (0.085)
V
d ol - Digestive System 0.187b 0.169b 0.158b 0.150c
(0.082) (0.082) (0.082) (0.082)
No. of high-risk hospitals 0.128a -0.018 0.190a 0.143b
(0.048) (0.077) (0.047) (0.074)
No. of all hospitals 0.184a 0.061b
(0.077) (0.074)

d(Pr entry)/d(PSI) 2.4 ppt - 3.5 ppt 2.6 ppt

Baseline Entry Probability 11.7%

a
p-value < 0.01
b
p-value < 0.05
c
p-value < 0.10

103
 APPENDIX C

COEFFICIENT ESTIMATES FROM LOG ODDS

REGRESSION - SPECIFICATION
ROBUSTNESS

Using physician level grouped data in a regression with log-odds as a dependent variable
may introduce heteroscedasticity in the model. First, multiple observations on the same
physician over time and across different HOPD’s (when a physician practices at more than 1
HOPD in a quarter) may invalidate the assumption of independence between observations.
Second, the practice splitting model is based on physician share of practice at different
facility types (HOPD or ASC) and does not take into account the actual patient volumes
treated by the physician. Therefore, heterogeneity in physician practice size can further im-
pact the homogeneity assumption as the disturbance variances may be larger for physicians
with disproportionately higher patient volumes. If heteroscedasticity is present, then the
resulting standard errors may be biased. To account for this possibility, two alternative cor-
rections are applied to the results in Table 3.4. Table C.1 corrects for heteroscedasticity by
clustering at the physician level. Clustering takes into account the fact that the errors are
not likely to be independent within the same physician when there are multiple observations
on the physician. It allows for variances that vary by the physician clusters and also takes
into account the size of the cluster. Table C.2, on the other hand, makes the correction
with the help of robust standard errors, also known as Huber - White standard errors or
the ‘sandwich’ estimator. The ‘sandwich estimator’ does not assume correlations that are
varying by groups and instead weighs the variance by estimated residuals for each obser-
vation. Therefore, clustering at the physician level imposes further structure on the model
by assuming that correlations may exist within the set of observations for each physician.

104
In this case, however, the assumption is likely to be valid given the grouping by physicians
and clustering correction should be preferred over the robust standard error correction.
Since the aforementioned are both standard error corrections, coefficient estimates stay
unchanged while the standard errors increase in their magnitude. The t-tests from both the
tables are very similar to each other and provide further evidence for the robustness of the
results on the interaction of hospital safety with physician ownership of ASCs. Physician
owners are significantly more likely to migrate their patients to ASCs when the HOPD
in question turns high-risk due to nursing complications. In case of the interaction of
controlling interest ownership with surgery safety, the estimate turns insignificant after
the corrections. Another cost of applying the standard error corrections is the loss of
significance on physician level variables of quality, experience and staff privileges. This can
happen because clustering imposes additional structure on the data and most physician
level variables are fixed over time. Therefore, it is possible that there may not be enough
variation in the data with respect to these variables and it may be harder to identify these
estimates with precision in a model that includes HOPD fixed effects as well as standard
error corrections.

105
Table C.1: Coefficient Estimates from the Log Odds Regression (Clustered Stan-
dard Errors)

Dependent Variable : Log Odds of HOPD Practice v/s ASC Specifications1

(1) (2) (3)
Castle
Explanatory Variables Top 30 Honor Roll
Connolly
Physician Ownership & Hospital Safety2
Controlling Interest Owner3 -0.590a -0.599a -0.591a
(0.083) (0.083) (0.083)
Controlling Interest Owner * High-Risk (Nursing) -0.260a -0.233b -0.243b
(0.107) (0.108) (0.107)
Controlling Interest Owner * High-Risk (Surgery) -0.118 -0.123 -0.116
(0.109) (0.109) (0.108)
Controlling Interest Owner (ASC Corp JV)4 -0.741a -0.740a -0.735a
(0.116) (0.116) (0.116)
Controlling Interest Owner (ASC Corp JV)* High-Risk (Nursing) -0.299b -0.286b -0.300b
(0.143) (0.142) (0.142)
Controlling Interest Owner (ASC Corp JV)* High-Risk (Surgery) -0.028 -0.056 -0.033
(0.118) (0.12) (0.119)
Non Controlling Interest Owner (Hosp Corp JV)5 0.120 0.119 0.123
(0.077) (0.077) (0.077)
Non Controlling Interest Owner (Hosp Corp JV)* High-Risk (Nursing) -0.058 -0.047 -0.052
(0.100) (0.100) (0.100)
Non Controlling Interest Owner (Hosp Corp JV)* High-Risk (Surgery) 0.038 0.031 0.035
(0.097) (0.098) (0.097)
High-Risk (Nursing) 0.171b 0.156b 0.166b
(0.077) (0.077) (0.077)
High-Risk (Surgery) 0.036 0.042 0.036
(0.083) (0.084) (0.083)
Physician Quality
‘Star’ Status -0.054 0.128 -0.048
(0.056) (0.079) (0.07)
Publications6 0.001 0.001 0.001
(0.007) (0.007) (0.007)
Board Certification 0.082 0.084 0.087
(0.064) (0.064) (0.064)
Physician Practice Profile
Inpatient Volume Previous Quarter 0.009a 0.009a 0.009a
(0.001) (0.001) (0.001)
Staff Privileges -0.014 -0.013 -0.014
(0.016) (0.016) (0.016)
Faculty Appointment 0.162b 0.152b 0.160b
(0.078) (0.077) (0.077)
Experience -0.002 -0.002 -0.001
(0.003) (0.003) (0.003)
Distance to the HOPD -0.038a -0.040a -0.038a
(0.007) (0.007) (0.007)
Number of Physicians = 1097

Notes:
1. All specifications include fixed effects for the HOPD and the standard errors are clustered at the physician level.
2. High-Risk Nursing and Surgery variables are binary indicators of high-risk status of the hospital due to nursing
or surgery complications. A high-risk hospital has a surgery complication rate that is 1 standard deviation greater
than the average.
3. Controlling Interest Owners are physicians who have a 5% or greater ownership share in the ASC. Such ASCs are
either individually owned or by a small group of physicians and many of these ASCs are either single specialty facilities
or small size multi-specialty facilities.
4. Controlling Interest Owners in a JV with a ASC Corporation such as AmSurg or Novamed. These physicians also
have a 5% or greater ownership share in the ASC. The corporation tends to own the management with 51% share
and the rest is generally owned by physician group(s).
5. Non-Controlling Interest Owners are in a JV with a Hospital Corporation such as HCA or Health South. These
physicians do not own a controlling interest and generally at such ASCs, the corporation tends to
own the management with greater than 51% share and the rest is divided among 40-60 physicians.
6. The variable publications measures the number of articles published by physicians who do not have faculty
appointments.
a
p-value < 0.01
b
p-value < 0.05
c
p-value < 0.10

106
Table C.2: Coefficient Estimates from the Log Odds Regression (Robust Standard Errors)

Dependent Variable : Log Odds of HOPD Practice v/s ASC Specifications1

(1) (2) (3)
Castle
Explanatory Variables Top 30 Honor Roll
Connolly
Physician Ownership & Hospital Safety2
Controlling Interest Owner3 -0.590a -0.599a -0.591a
(0.111) (0.111) (0.112)
Controlling Interest Owner * High-Risk (Nursing) -0.260b -0.233c -0.243b
(0.123) (0.123) (0.123)
Controlling Interest Owner * High-Risk (Surgery) -0.118 -0.123 -0.116
(0.111) (0.113) (0.111)
Controlling Interest Owner (ASC Corp JV)4 -0.741a -0.74a -0.735a
(0.137) (0.137) (0.138)
Controlling Interest Owner (ASC Corp JV)* High-Risk (Nursing) -0.299b -0.286b -0.3b
(0.13) (0.13) (0.131)
Controlling Interest Owner (ASC Corp JV)* High-Risk (Surgery) -0.028 -0.056 -0.033
(0.159) (0.156) (0.16)
Non Controlling Interest Owner (Hosp Corp JV)5 0.120 0.119 0.123
(0.119) (0.118) (0.118)
Non Controlling Interest Owner (Hosp Corp JV)* High-Risk (Nursing) -0.058 -0.047 -0.052
(0.12) (0.12) (0.12)
Non Controlling Interest Owner (Hosp Corp JV)* High-Risk (Surgery) 0.038 0.031 0.035
(0.116) (0.116) (0.117)
High-Risk (Nursing) 0.171b 0.156c 0.166b
(0.082) (0.082) (0.083)
High-Risk (Surgery) 0.036 0.042 0.036
(0.097) (0.097) (0.097)
Physician Quality
‘Star’ Status -0.054 0.128c -0.048
(0.055) (0.078) (0.069)
6 0.001 0.001 0.001
Publications
(0.007) (0.007) (0.007)
Board Certification 0.082 0.084 0.087
(0.057) (0.056) (0.057)
Physician Practice Profile
Inpatient Volume Previous Quarter 0.009a 0.009a 0.009a
(0.001) (0.001) (0.001)
Staff Privileges -0.014 -0.013 -0.014
(0.015) (0.015) (0.015)
Faculty Appointment 0.162c 0.152c 0.16c
(0.086) (0.083) (0.085)
Experience -0.002 -0.002 -0.001
(0.003) (0.003) (0.003)
Distance to the HOPD -0.038a -0.040a -0.038a
(0.008) (0.009) (0.008)
Number of Physicians = 1097

Notes:
1. All specifications include fixed effects for the HOPD and the standard errors are corrected for heteroscedasticity.
2. High-Risk Nursing and Surgery variables are binary indicators of high-risk status of the hospital due to nursing
or surgery complications. A high-risk hospital has a surgery complication rate that is 1 standard deviation greater
than the average.
3. Controlling Interest Owners are physicians who have a 5% or greater ownership share in the ASC. Such ASCs are
either individually owned or by a small group of physicians and many of these ASCs are either single specialty facilities
or small size multi-specialty facilities.
4. Controlling Interest Owners in a JV with a ASC Corporation such as AmSurg or Novamed. These physicians also
have a 5% or greater ownership share in the ASC. The corporation tends to own the management with 51% share
and the rest is generally owned by physician group(s).
5. Non-Controlling Interest Owners are in a JV with a Hospital Corporation such as HCA or Health South. These
physicians do not own a controlling interest and generally at such ASCs, the corporation tends to
own the management with greater than 51% share and the rest is divided among 40-60 physicians.
6. The variable publications measures the number of articles published by physicians who do not have faculty
appointments.
a
p-value < 0.01
b
p-value < 0.05
c
p-value < 0.10

107
 APPENDIX D

HUMAN SUBJECTS APPROVAL

MEMORANDUM

Office of the Vice President For Research Human Subjects Committee Tallahassee, Florida
32306-2742 (850) 644-8673 FAX (850) 644-4392
APPROVAL MEMORANDUM
Date: 6/13/2011
To: Nitin Dua
Address: 2180 Dept.: ECONOMICS
From: Thomas L. Jacobson, Chair
Re: Use of Human Subjects in Research Competition for Outpatient Surgeries: ASC
Entry, Hospital Safety and Physician Ownership
The application that you submitted to this office in regard to the use of human subjects
in the proposal referenced above have been reviewed by the Secretary, the Chair, and one
member of the Human Subjects Committee. Your project is determined to be Expedited
per per 45 CFR 46.110(7) and has been approved by an expedited review process.
The Human Subjects Committee has not evaluated your proposal for scientific merit,
except to weigh the risk to the human participants and the aspects of the proposal related
to potential risk and benefit. This approval does not replace any departmental or other
approvals, which may be required.
If you submitted a proposed consent form with your application, the approved stamped
consent form is attached to this approval notice. Only the stamped version of the consent
form may be used in recruiting research subjects.

108
 If the project has not been completed by 6/11/2012 you must request a renewal of
approval for continuation of the project. As a courtesy, a renewal notice will be sent to you
prior to your expiration date; however, it is your responsibility as the Principal Investigator
to timely request renewal of your approval from the Committee.
You are advised that any change in protocol for this project must be reviewed and ap-
proved by the Committee prior to implementation of the proposed change in the protocol.
A protocol change/amendment form is required to be submitted for approval by the Com-
mittee. In addition, federal regulations require that the Principal Investigator promptly
report, in writing any unanticipated problems or adverse events involving risks to research
subjects or others.
By copy of this memorandum, the Chair of your department and/or your major profes-
sor is reminded that he/she is responsible for being informed concerning research projects
involving human subjects in the department, and should review protocols as often as needed
to insure that the project is being conducted in compliance with our institution and with
DHHS regulations.
This institution has an Assurance on file with the Office for Human Research Protection.
The Assurance Number is FWA00000168/IRB number IRB00000446.
Cc: Gary Fournier, Advisor HSC No. 2011.6550

109
 BIBLIOGRAPHY

AHRQ, “AHRQ Quality Indicators - Guide to Patient Safety Indicators,” Technical Report
AHRQ Pub.03-R203, Version 2.1, Revision 2, AHRQ 2003.

, “Clinical Classifications Software (CCS) for ICD-9-CM,” http://www.hcup-

us.ahrq.gov/toolssoftware/ccs/ccsfactsheet.jsp 2007.

, “National Healthcare Quality Report 2009,” Report to Congress. AHRQ Pub-

lication No. 10-0003, March 2010, Agency for Healthcare Research and Quality,
US Department Health & Human Services, Rockville, MD 20850 March 2010.
www.ahrq.gov/qual/qrdr09.htm.

Barr, Judith K., Shulamit L. Bernard, Shoshanna Sofaer, Tierney E. Giannotti,

Nancy F. Lenfestey, and David J. Miranda, “Physicians’ Views on Public Reporting
of Hospital Quality Data,” Medical Care Research and Review, 2008, 65 (6), 655–673.

Barro, Jason R., Robert S. Huckman, and Daniel P. Kessler, “The effects of cardiac
specialty hospitals on the cost and quality of medical care,” Journal of Health Economics,
July 2006, 25 (4), 702–721.

Bazzoli, Gloria J., Hsueh-Fen Chen, Mei Zhao, and Richard C. Lindrooth, “Hos-
pital Financial Condition and the Quality of Patient Care,” Health Econ, 2008, 17, 977995.

Blendon, Robert J., Catherine M. DesRoches, Mollyann Brodie, John M.

Benson, Allison B. Rosen, Eric Schneider, Drew E. Altman, Kinga Zapert,
Melissa J. Herrmann, and Annie E. Steffenson, “Views of Practicing Physicians
and the Public on Medical Errors,” New England Journal of Medicine, 2002, 347 (24),
1933–1940.

Börsch-Supan, Axel, “On the compatibility of nested logit models with utility maximiza-
tion,” Journal of Econometrics, 1990, 43 (3), 373 – 388.

Brennan, Troyen A., Ralph I. Horwitz, F. Daniel Duffy, Christine K. Cassel,

Leslie D. Goode, and Rebecca S. Lipner, “The Role of Physician Specialty Board

110
 Certification Status in the Quality Movement,” JAMA: The Journal of the American
Medical Association, 2004, 292 (9), 1038–1043.

Brown, Philip H. and Caroline Theoharides, “Health-seeking behavior and hospital

choice in China’s New Cooperative Medical System,” Health Economics, 2009, 18 (S2),
S47–S64.

Burke, Mary, Gary M. Fournier, and Kislaya Prasad, “The Diffusion of a Medical
Innovation: Is Success in the Stars?,” Southern Economic Journal, January 2007, 73 (3),
588–603.

, , and , “The Diffusion of a Medical Innovation: Is Success in the Stars? New

Evidence,” Southern Economic Journal, April 2009, 75 (4).

Burns, L R and D R Wholey, “Factors affecting physician loyalty and exit: a longitu-
dinal analysis of physician-hospital relationships.,” 1992.

Burns, Lawton R., Jeffrey A. Alexander, Stephen M. Shortell, Howard S. Zuck-

erman, Peter P. Budetti, Robin R. Gillies, and Teresa M. Waters, “Physician
Commitment to Organized Delivery Systems,” Medical Care, 2001, 39 (7), pp. I9–I29.

Cameron, Trudy Ann, “A Nested Logit Model of Energy Conservation Activity by Own-
ers of Existing Single Family Dwellings,” The Review of Economics and Statistics, 1985,
67 (2), pp. 205–211.

Carlson, Greg and Hugh Greeley, “Is the Relationship Between Your Hospital and
Your Medical Staff Sustainable?.,” Journal of Healthcare Management, 2010, 55 (3), 158
– 173.

Casalino, Lawrence P., Elizabeth A. November, Robert A. Berenson, and

Hoangmai H. Pham, “Hospital-Physician Relations: Two Tracks And The Decline
Of The Voluntary Medical Staff Model,” Health Affairs, 2008, 27 (5), 1305–1314.

, Kelly J. Devers, and Linda R. Brewster, “Focused Factories? Physician-Owned

Specialty Facilities,” Health Aff, 2003, 22 (6), 56–67.

Castle, John K and John J Connolly, America’s Top Doctors, 9th ed., Castle Connolly
Medical Ltd., 2010.

Chernew, Michael, Gautam Gowrisankaran, and A. Mark Fendrick, “Payer type

and the returns to bypass surgery: evidence from hospital entry behavior,” Journal of
Health Economics, 2002, 21 (3), 451 – 474.

111
CMS.gov, “2011 Physician Quality Reporting System,” Physician Quality Reporting Sys-
tem, at www.cms.gov 2011.

Courtemanche, Charles and Michael Plotzke, “Does competition from ambulatory

surgical centers affect hospital surgical output?,” Journal of Health Economics, 2010, 29
(5), 765–773.

David, Guy and Mark D. Neuman, “Physician division of labor and patient selection
for outpatient procedures,” Journal of Health Economics, 2011, 30 (2), 381 – 391.

Encinosa, William E. and Didem M. Bernard, “Hospital Finances and Patient Safety
Outcomes,” Inquiry, 2005, 42, 60–72.

Escarce, Jos J., Bernard S. Bloom, Alan L. Hillman, Judy A. Shea, and J. San-
ford Schwartz, “Diffusion of Laparoscopic Cholecystectomy among General Surgeons
in the United States,” Medical Care, 1995, 33 (3), pp. 256–271.

Fournier, Gary M. and Yunwei Gai, “What Does Willingness-to-Pay Reveal About
Hospital Market Power in Merger Cases?,” SSRN eLibrary, 2007.

Gabel, Cheryl Fahlman Jon R. et al., “Where Do I Send Thee? Does Physician-
Ownership Affect Referral Patterns To Ambulatory Surgery Centers?,” Health Affairs
Web Exclusive, 2008, pp. w165–174. 1 8 M a r c h 2 0 0 8.

Gardner, Lara and Sharmila Vishwasrao, “Physician Quality and Health Care for the
Poor and Uninsured,” Inquiry, 2010, 47 (1), 62–80.

Gowrisankaran, Gautam and Robert Town, “Competition, payers, and hospital qual-
ity,” Health Services Research, 2003, 38 (6), 1403–1422.

Haugh, Richard, “Outpatients: Here today, Gone Tomorrow?,” Hospital & Health Net-
works Magazine August 2006.

Herriges, Joseph A. and Catherine L. Kling, “Testing the consistency of nested logit
models with utility maximization,” Economics Letters, 1996, 50 (1), 33 – 39.

Hollingsworth, John M., Zaojun Ye, Seth A. Strope, Sarah L. Krein, Ann T.
Hollenbeck, and Brent K. Hollenbeck, “Physician-Ownership Of Ambulatory
Surgery Centers Linked To Higher Volume Of Surgeries,” Health Affairs, 2010, 29 (4),
683–689.

Jones, Andrew M., “Health econometrics,” in A. J. Culyer and J. P. Newhouse, eds.,

Handbook of Health Economics, Vol. 1 of Handbook of Health Economics, Elsevier, 06
2000, chapter 6, pp. 265–344.

112
Kessler, Daniel P. and Jeffrey J. Geppert, “The Effects of Competition on Variation
in the Quality and Cost of Medical Care,” Journal of Economics & Management Strategy,
2005, 14, 575–589(15).

and Mark B. McClellan, “Is Hospital Competition Socially Wasteful?*,” Quarterly

Journal of Economics, 2000, 115 (2), 577–615.

Lynk, William J. and Carina S. Longley, “The Effect Of Physician-Owned Surgicenters

On Hospital Outpatient Surgery,” Health Aff, 2002, 21 (4), 215–221.

Marshall, W. Raffel, K. Raffel Norma, and K. Barsukiewicz Camille, The U.S.

Health System: origins and functions, 5th ed., Delmar Thomson Learning, 2002.

Mechanic, David, Lynn B. Rogut, David C. Colby, and James R. Knickman,

Policy Challenges in Modern Health Care, Rutgers University Press, 2005.

Michael, Gautam Gowrisankaran A. Mark Fendrick Chernew, “Payer type and

the returns to bypass surgery: evidence from hospital entry behavior,” Journal of Health
Economics, 2002, 21, 451–474.

Mitchell, Jean M., “Do Financial Incentives Linked to Ownership of Specialty Hospitals
Affect Physicians Practice Patterns?,” Medical Care, July 2008, 46 (7), 732–737.

Neuman, Mark D., Guy David, Jeffrey H. Silber, J. Sanford Schwartz, and
Lee A. Fleisher, “Changing Access to Emergency Care for Patients Undergoing Outpa-
tient Procedures at Ambulatory Surgery Centers: Evidence From Florida,” Medical Care
Research and Review, 2011, 68 (2), 247–258.

Plotzke, Michael Robert and Charles Courtemanche, “Does procedure profitability

impact whether an outpatient surgery is performed at an ambulatory surgery center or
hospital?,” Health Economics, 2010.

Poole, Ellen L., “Ambulatory surgery:The growth of an industry,” Journal of perianes-

thesia nursing : official journal of the American Society of PeriAnesthesia Nurses /
American Society of PeriAnesthesia Nurses, August 1999, 14 (4), 201–206.

Puig-Junoy, Jaume, Marc Saez, and Esther Martnez-Garca, “Why do patients

prefer hospital emergency visits? A nested multinomial logit analysis for patient-initiated
contacts,” Health Care Management Science, 1998, 1, 39–52. 10.1023/A:1019082232606.

ResidentPhysician.com, “Top Residency Departments By Medical Specialty,” Web

Database, at www.residentphysician.com 2005.

113
Russo, C. Allison, Pamela Owens, Claudia Steiner, and Justin Josephsen, “Am-
bulatory Surgery in U.S. Hospitals, 2003 HCUP Fact Book No. 9.,” AHRQ Publication
07-0007, Agency for Healthcare Research and Quality, Rockville, MD. January 2007.

Sahn, David E., Stephen D. Younger, and Garance Genicot, “The Demand for
Health Care Services in Rural Tanzania,” Oxford Bulletin of Economics and Statistics,
2003, 65 (2), 241–260.

Smallet, John A., “Hospital-Owned Ambulatory Surgery Centers as a Hospital Outpa-

tient Department,” Website article November 2008.

Tay, Abigail, “Assessing Competition in Hospital Care Markets: The Importance of Ac-
counting for Quality Differentiation,” The RAND Journal of Economics, 2003, 34 (4),
pp. 786–814.

Train, Kenneth, Discrete Choice Methods with Simulation, Cambridge University Press,
2003.

Weber, Ellerie, “Measuring Welfare from Ambulatory Surgery Centers: A Spatial Anal-
ysis of Demand for Healthcare Facilities,” The University of Chicago, 2010.

Winter, Ariel, “Comparing The Mix Of Patients In Various Outpatient Surgery Settings,”
Health Affairs, 2003, 22 (6), 68–75.

114
 BIOGRAPHICAL SKETCH

The author was born in New Delhi, India in 1982. He graduated from the University
of Delhi, India in 2004 with a Masters degree in Business Economics. At the University
of Delhi, he received scholarship for academic excellence every year from 1999 to 2002.
After a short stint as a quantitative researcher, he entered the economics Ph.D. program
at Florida State University in 2005. There he received the James H Gapinski and the
Irvin & Peggy Sobel awards for academic achievement in 2006 and 2008, respectively. His
research interests include Health Economics, Industrial Organization, Labor Economics,
Applied Econometrics, and Applied Microeconomics. Starting August 2011, Nitin will
begin employment as a Senior Consultant at Bates White, LLC.

115