INTRODUCTION American hospitals and health systems

“Healthcare in the U.S. is in a state of hyper-turbulence characterized by accumulated

waves of change in payment systems, delivery systems, technology, professional relations,

and societal expectations” – Shortell, Gillies, & Devers, 1995.

American hospitals and health systems merge for a wide variety of reasons but

always with one common, overarching goal: achieve some significant, tangible gain for

the organizations involved (Brooks & Jones, 1997; Haspeslagh & Jemison, 1991). Some

stand-alone hospitals seek the protection and additional resources that a larger system

may bring (Dunn, 2013; Vogel & Hadfield, 2017). A rural hospital may opt to merge to

avoid closure or bankruptcy, thereby ensuring access to care for the remote, vulnerable

population it serves (Noles, Reiter, Boortz-Marx, & Pink, 2015). Two not-for-profit

health systems may merge as a way to preserve and enhance their ability to fulfill their

religious, charitable, or philanthropic mission (Cutler, 2000). Two for-profit, publicly

traded hospital companies may decide to consolidate in an effort to streamline operations,

decrease costs, increase profits, and maximize shareholder value (Cutler, 2000).

Regardless of motivation, senior management and the boards of directors involved with

the decision propose that a merger is the best overall strategic option for the future of the

organizations; combining the assets and resources of the organizations enhances their

ability to meet the health care needs of the patients and communities they serve. It takes

many years to determine whether or not a merger achieved its intended goals and fulfilled

the original promise. In most cases, the legal, financial, accounting, and regulatory work

required to complete the merger transaction is much easier than executing the post-

merger plan and making the merger work (Appelbaum & Gandell, 2002). The benefits of

a merger can accrue to the enterprise, shareholders (for-profit health system), bondholders

(not-for-profit health system), senior management, employees, physicians,

purchasers/payors, communities, and patients. However, a merger usually results in

winners and losers; the advantages of a merger may be enjoyed by a few stakeholders

while disadvantaging others (Blair, Durrance & Sokol, 2016).

The term “Total Net Welfare” (TNW) was developed by Katona and Canoy

(2013) to compute and express the net effect of a hospital or health system merger on a

given regional market. TNW is broadly defined as the sum of the positive and negative

welfare impacts of the merger. For example, a hospital merger may result in improved

quality performance (a positive impact) and higher prices (a negative impact). Katona and

Canoy found that hospital mergers directly and indirectly affect patients, employees

working for the merging organizations, local and regional employers who provide health

insurance for their employees, health insurance companies and health plans, physicians,

and many others. Every hospital merger has advantages and disadvantages.

To the extent the positive impacts outweigh the negative impacts, the merger has a

positive TNW. Conversely, if the merger-related disadvantages exceed the advantages, the

merger has a negative TNW. The TNW model attempts to incorporate all elements

affected by the merger including quality, safety, access, efficiency (costs), pricing, and

service. Positive impacts of a merger may include improvements in quality or safety

performance, better access to care, lower costs for care, increased investments in public

health initiatives, better online access to health information, lower pricing for health

insurance premiums, new services offered, incremental investments in technology, and

better service for patients (i.e., improvements in the patient experience). Negative impacts

involve the opposite: lower quality or safety performance, diminished access, higher

costs, and decreased service.

The vast majority of mergers across all industries and time periods either fail or

do not achieve their goals (Bergh, 1997; Gugler & Yurtoglu, 2008; Montgomery &

Wilson, 1986; Ravenscraft & Scherer, 1997, 1989; Tichy, 2001). Research conducted by

economists, business consultants, sociologists, psychologists, and organizational behavior

experts have identified numerous factors associated with merger failures or poor

postmerger performance. In the vast majority of cases the primary culprit was the

inability of the two organizations to address critical human factors by successfully

combining, blending, and harmonizing culture (Chakravorty, 2013; Ettore, 1999;

Grossman, 1999; Segil, 1998; Tompkins, 1998).

Research conducted by Hellriegel, Slocum, and Woodman (1976/1986) found

even two companies operating in the same industry/market, serving the same region, and

sharing common customers, may evolve and establish two very different and distinct

cultures. Cultural problems and conflict among senior management disrupt, handicap, and

hamper the newly merged organization. The underlying strategic, operational, and

financial rationale is usually sound and fully justifies the merger decision. However, the

lingering human factors and social phenomena that accompany a merger are oftentimes

underestimated or overlooked (Chakravorty, 2013).

Merger and Consolidation Trend

The U.S. health care industry is experiencing a time of uncertainty, volatility,

unprecedented change, and transformation (American Hospital Association, 2016;

American Public Health Association, 2016; Britnell, 2015; Pearl, 2015). The U.S.

Congress and President Trump intend to significantly revise or completely replace the

Affordable Care Act (ACA), which creates more uncertainty and, potentially, more

upheaval within the health care industry. New value-based reimbursement models,

implemented under the ACA, have reformed the way hospitals and physicians coordinate

and deliver care (Deloitte, 2015; Fried & Sherer, 2016). Regional integration efforts are

accelerating as hospitals, physicians, self-insured employers, health insurance companies,

suppliers, and other stakeholders align and connect their activities around a new

paradigm: population health (Franz, Skinner, & Murphy, 2016).

In the 25-year period between 1990 and 2015 there were 2,748 consolidation

transactions (horizontal mergers, acquisitions, and other formal integration/alignment

arrangements) involving U.S. hospitals and health systems (Maslan & Johnston, 2016).

During the 1990s, approximately 900 hospital consolidation transactions occurred,

leading to the formation of many regional health systems (Patrick, 2014). Following this

surge in the 1990s, the pace of consolidation slowed. However, since 2008 the number of

consolidations has increased substantially (American Hospital Association, 2013, 2015;

Levin Associates, 2015; PWC, 2015; Summer, 2010; Vaida & Wess, 2015; Wirtz, 2015)

(see Figures 1-3). Many industry experts call for this trend to continue for the foreseeable

future (Guerin-Calvert & Maki, 2014; O’Connell, 2015; PWC, 2015; Vaida & Wess,

2015). Eighty-two percent (82%) of hospital CEOs surveyed at the end of 2015 stated

that their organization was either “likely” or “very likely” to undergo a consolidation

transaction within the next five years (Illinois State Medical Society, 2016).

Blue bar = number of deals; green bar = number of hospitals

Some health care futurists predict continued consolidation will lead to

“megasystems” that will compete with other mega-systems over large, multi-state regions

(Deloitte, 2014; Ellison, 2016; Evans, 2014; Herman, 2014). Deloitte (2014) forecasts a

50% decrease in the number of health systems over the next 10 years. Morrison, health

care futurist, stated “[w]e are moving pretty rapidly toward a situation where 100 to 200

large, integrated regional health systems” will dominate American health care (GE

Healthcare, 2014, p. 3).

Do Hospital Mergers Create Value?

In light of the merger wave and large-scale consolidation of health care providers,

many researchers, government regulators, elected officials, and consumer groups have

asked these key questions: Do hospital and health system mergers create value? In other

words, do mergers result in an increase in TNW (Total Net Welfare)? How do mergers

affect quality, safety, access, cost, pricing, profitability, and service?

As will be discussed in Chapter 2, hospital mergers have been thoroughly

examined over the past 25 years. Many studies have analyzed the merger impact on

clinical quality and safety performance. Several researchers evaluated the merger effect

on access to care especially for at-risk, vulnerable, under-served, remote, or rural

populations. Numerous economic studies examined the impact of mergers on internal

operating expenses; pricing (i.e., fees charged to insurance companies, health plans and

patients); health insurance premiums; operating margins; and profits. However, there is

no published study that assesses the impact of hospital mergers on customer service in

terms of patient experience or how patients perceive and rate their inpatient care episode.

Statement of the Problem

Due to mergers and acquisitions, American health systems are growing larger in

terms of number of hospitals, hospital beds, and annual revenues. As consolidation

occurs, hospital market concentration has increased substantially. Applying the

Herfindahl-Hirschfield Index (HHI), the majority of metropolitan areas in the United

States are considered “highly concentrated” (National Academy of Social Insurance,

2015). The more highly concentrated a market becomes, the less competitive it becomes

(Archer, 2013).

In light of the growing consolidation and concentration of hospital services across

the United States, several questions arise:

• What incremental value is generated from consolidation?

• How do hospital mergers affect hospital performance from a quality, safety,

access, efficiency, pricing, and service perspective?

• How do hospital mergers affect patients and the communities served?

• What benefit do patients and communities receive?

Significant research has been conducted in the past 25 years to address these

questions. Most of the research treated the merger as the independent variable and

examined the impact of the merger on various performance-related outcomes: quality,

safety, access, internal operating costs, pricing, operating margin, and insurance premium

pricing. Some studies applied a difference-in-differences statistical method while others

compared an intervention group of merged hospitals with a control group of hospitals that

did not merge. While several authors found a positive correlation between mergers and

certain performance outcomes, the vast majority of studies found no effect or a negative

effect (Gaynor & Town, 2012; Universal Health Care Foundation of Connecticut, 2014).

One significant gap exists in this comprehensive body of merger-related research: no

study has examined the relationship between hospital mergers and patient experience.

Patient Experience Background

American hospitals began using formal surveys in the early 1990s to obtain

feedback from patients to improve service and, thereby, increase patient loyalty and

preference (Fottler, Ford, & Heaton, 2002). Starting in 2007, in an attempt to objectively

measure patients’ perspectives on hospital care, the Centers for Medicare and Medicaid

Services (CMS) began requiring all acute care hospitals to report patient experience

scores as a condition of receiving Medicare and Medicaid payments. CMS stipulated the

use of a survey instrument jointly developed by CMS and the Agency for Healthcare

Quality and Research: the Hospital Consumer Assessment of Healthcare Providers and

Systems Survey (HCAHPS). Standardized and publicly reported, HCAHPS provides a

tool for comparing a patient’s perception of care delivery/service (i.e., patient experience)

within and between hospitals. Beginning in 2013, HCAHPS performance became part of

CMS’s Value-Based Purchasing program. Not only is reporting HCAHPS scores a

requirement for receiving any CMS payment, a hospital is subject to additional financial

penalties based upon its HCAHPS performance (CMS, 2017).

Significance of Study

This study provides a unique perspective which adds to the extant research on

hospital mergers. A hospital’s service performance, as measured by HCAHPS ratings and

scores, is a core component of value definition. Patient experience is part of the TNW

equation; it addresses whether or not a merger positively or negatively affects the service

patients (consumers) receive from a hospital.

Mergers are oftentimes controversial events. The increasing size and market

concentration of hospitals and health systems is of particular interest to state and federal

government regulators, the Federal Trade Commission, the U.S. Department of Justice,

consumer watchdog groups, elected officials, physicians and physician group practice

associations, nurses and nursing organizations, policymakers, and health care

administrators. Determining how mergers impact patient experience is an important

contribution to the overall evaluation of the consolidation of health care services.

Statement of Purpose and Research Questions

The purpose of this study was to examine the overall relationship between

hospital mergers and patient experience (measured with HCAHPS) and answer the

following five research questions:

1. Prior to merger, do hospitals pursuing a merger have different HCAHPS scores

compared to similar hospitals that are not pursuing a merger?

2. How does a merger affect HCAHPS performance?

3. Does time play a role in this effect? In other words, as more time passes

postmerger, does any difference in HCAHPS performance disappear?

4. Does the type of transaction (merger-of-equals vs. acquisition) moderate this

relationship?

Summary

In the past 25 years, the vast majority of U.S. hospitals have, through merger or

acquisition, become part of large health systems serving multiple geographic markets

and/or regions of the country. Since 2008, another wave and magnitude of consolidation

has occurred with many multi-hospital systems merging with other multi-hospital

systems forming mega-systems. This significant and unprecedented level of

concentration raises many regulatory and economic concerns about the net value (i.e.,

sum of positives and negatives) of mergers on patients, payors, employers, and the

communities impacted. Understanding the association of hospital mergers with patient

experience is an important component of that impact/value analysis. The aim of this study

was to provide unique insights and help policymakers and health system leaders evaluate

hospital consolidation decisions.

CHAPTER 2

LITERATURE REVIEW, CONCEPTUAL / THEORETICAL FRAMEWORK,

AND STATEMENT OF HYPOTHESES

“The emotions surrounding a merger are much like getting married and divorced in the

same day” – Bruhn, 2001.

This chapter opens with a literature review followed by a presentation of the

conceptual framework and pertinent theories that support and underlie the hypotheses.

The chapter ends with a statement of 12 hypotheses that were empirically tested through

this research study.

The literature review section is comprised of the following four sub-sections:

A) review of empirical research involving the evaluation of patient experience

in the hospital setting;

B) review of research involving the impact of mergers on customer satisfaction;

C) review of research on post-merger hospital performance; and

D) review of research on the post-merger disruption effect.

Literature Review

Sub-Section A – Literature Review on Patient Experience and Satisfaction

Managing Patient Experience Is Critical to Hospitals

Patient perceptions of their inpatient experiences and their overall satisfaction

levels (as measured by the Hospital Consumer Assessment of Healthcare Providers and

Systems, [HCAHPS]) are important strategic considerations for most hospitals. Hospitals

with comparatively high patient experience ratings enjoy a better image and reputation

(Andaleeb, 1988). Consequently, they also may have a distinct competitive advantage in

their market area (Hall, 2008). In addition, several authors have found a positive

relationship between a hospital’s patient experience performance and financial

performance (Charmel & Frampton, 2008; Deloitte, 2016; Hall, 2008; Richter &

Muhlestein, 2017).

One patient’s bad experience, and the resulting negative word-of-mouth, can cost

a hospital between $6,000 and $400,000 in lost lifetime revenues (Strasser, Schweikhart,

Welch, & Burge, 1995). Hospitals must be as aggressive as other service industries at

creating loyalty and customers for life (Carrus, Cordina, Gretz, & Neher, 2015; Spoerl,

2012; Zeithaml & Bitner, 2000). Research conducted by Reichheld and Markey (2011)

found that a hospital’s overall HCAHPS performance had a positive relationship with

repeat and referral business. The increasing role of patients in the selection of their

hospital (i.e., patient preference versus physician choice), coupled with public reporting

of HCAHPS scores, are driving more hospitals to make patient experience a top strategic

priority (Reichheld & Markey, 2011).

Due to the nature of how the CMS value-based purchasing program’s penalty for

HCAHPS is computed, hospitals must continuously improve HCAHPS scores in order to

maintain their relative performance. Avoiding the HCAHPS-based financial penalty

requires increasing expertise, effort, resources, and financial investments (Stanowski,

Simpson, & White, 2015).

Determinants of Patient Experience

Empirical research conducted over the past 20 or more years has identified

numerous factors that influence and determine patient perceptions of their inpatient

experience. Clinical quality (competency of staff and clinical outcomes) and safety

(absence of harm) are two factors but difficult for patients to accurately measure due to

their “inherent intangibility, heterogeneity and inseparability features” (Conway &

Willcocks, 1997, p. 132). Convenient access and availability, assessed by geographic

location and waiting times, are important elements and easily evaluable (Tucker, 2002).

The appearance and condition of the physical facility (exterior and interior), sound levels,

food quality, and overall cleanliness influence patient perceptions, experience, and

satisfaction (Swan, Richardson, & Hutton, 2003; Woodside, Frey & Daly, 1989).

However, the most influential factors affecting the inpatient experience involve

human interaction and service delivery. These factors include communication (explaining

things well and listening), responsiveness, timeliness, courtesy, respect, empathy, concern

about pain management, compassion, friendliness, staff demeanor, and the amount of

interaction time with nurses and physicians (Boshoff & Gray, 2004; Fowdar,

2005; Tucker & Adams, 2001; Ware, Davies-Avery, & Stewart, 1978).

Link between Employee Satisfaction and Patient Experience Performance Many

companies, representing the full gamut of service industries, have found a link between

employee satisfaction and customer satisfaction (Brown, 2005). An extensive meta-

analysis of empirical research on customer service performance in the banking industry

found a strong correlation between employee satisfaction and customer loyalty. This

same analysis found evidence to support causation: happy employees provide better

service leading to higher customer satisfaction (Anastasiou &

Nathanailides, 2015). Research in the hospital setting resulted in similar findings. Larson

(2004) combined the surveys of 50,000 hospital employees from 33 hospitals with 20,000

patient experience surveys. The conclusion was that higher levels of employee satisfaction

were strongly associated with higher levels of patient experience performance.

Research with Patient Experience (HCAHPS) as the Dependent Variable

Mazurenko, Collum, Ferdinand, and Menachemi (2017) conducted a meta-

analysis and systematic literature review focusing on predictors of hospital patient

experience as measured by HCAHPS. Twenty-nine of the 41 studies evaluated involved

hospital-level predictors. Certain organizational and working environment attributes were

positively associated with HCAHPS performance including well-defined nursing

standards (Kutney-Lee et al., 2009), cultural competency (Weech-Maldonado et al.,

2012), and staff perception of a safety culture (Sorra, Khanna, Dyer, Mardon, &

Famolaro, 2014).

Kutney-Lee and colleagues (2009) conducted the first study to explore how nurse

working environment and patient-to-nurse ratios affected HCAHPS scores. Working

environment was measured using a widely recognized and standardized tool, the Practice

Environment Scale of the Nursing Working Index. More than 98,000 hospital-based

nurses responded to the survey. The authors found that the nurse working environment

had significant effects on HCAHPS performance.

Sub-Section B – Literature Review on Merger Effect on Customer Satisfaction

Several studies have examined the impact of mergers on customer satisfaction. None of

these, however, includes analysis of hospital mergers on patient satisfaction. The general

conclusion from these studies of non-health care firms was that mergers have a negative

effect on customer satisfaction. One study found that customer satisfaction declined post-

merger primarily due to an over-emphasis on cost-cutting and efficiencies (Swaminathan,

Groening, Mittal, & Thomaz, 2013). Another report posited that customer satisfaction

declines were a result of the merged organization becoming internally focused, which

caused frontline employees’ concentration on customer service to decrease (Miles &

Rouse, 2011). The last analysis suggested that the merged organization did not give

employees enough time to adjust to the environment before post-merger changes were

implemented (Institute of Management and Administration, 2005).

Sub-Section C – Review of Research on Post-Merger Hospital Performance

The following review summarizes numerous studies of post-merger hospital

performance. The primary question motivating the research: How did the merger of two

hospitals or health systems impact performance? The common conceptual framework

applied in this research assumed the merged entity would have more financial resources

following the merger (due to better pricing and/or lower costs). Therefore, with additional

resources, the merged enterprise would be able to make certain investments in people,

systems, equipment, and facilities in order to improve operations and performance. The

objective of most of this research was to identify how hospital consolidation affected

performance compared to the pre-merger state. Some of the research sought to understand

why actual post-merger performance fell short of the pre-merger goals.

Many researchers focused on the impact of hospital mergers on quality from both

outcome and process standpoints. The general expectation was that quality would

improve post-merger by two mechanisms: (1) consolidation of certain complex services

(e.g., open heart surgery) into one facility, and (2) greater investments in

qualityenhancing activities and programs from the increased profits generated post-

merger (Birkmeyer, Stukel, & Siewers, 2002; Guerin-Calvert & Maki, 2014). However,

results from the 11 studies reviewed were equivocal. Most of the research conducted

found no statistically significant changes in quality post-merger (Bazzoli et al., 2004;

Gaynor &

Town, 2012; Romano & Balan, 2010; Universal Health Care Foundation of Connecticut,

2014). A few found a positive relationship between mergers and quality performance

(Cuellar & Gertler, 2005; Mutter, Romano, & Wong, 2011) while others found a negative

relationship (Encinosa & Bernard, 2005; Hayford, 2011).

Access-related comparative data are very limited (only two studies). Access was

defined in geographic terms and was measured by identifying one of two events: (1)

decision not to terminate a service that was slated for closure pre-merger, and (2) decision

to open a new service post-merger. Guerin-Calvert and Maki (2014) concluded that, in

aggregate, hospital mergers that occurred during the 20-year period studied (early 1990s

to early 2010s) had a slight favorable effect on preventing the shuttering of hospitals and

needed clinical services for at-risk, indigent, and rural populations. Several

researchers focused their merger analysis on internal efficiencies and cost reduction. For

example, mergers were shown to create a high potential for scale efficiency through the

elimination of duplication and the consolidation of services (Ferrier & Valdmanis, 2004;

Harris, Ozgen, & Ozcan, 2000; Harrison, 2011). Cost savings from technical efficiency

took longer to achieve but was still significant (Gross,

Lien, & Su, 2007). Most of these efficiency-examining studies compared the hospital’s

operating costs post-merger to the pre-merger levels. In the limited situations where the

merger generated efficiencies, the post-merger cost savings (computed by dividing total

operating costs by adjusted discharges) ranged from 2.5% to 14% with most of the

savings occurring in the first three years following the merger. After the initial, one-time,

merger-specific expense reductions were achieved, costs increased similar to other

hospitals (Ferrier & Valdmanis, 2004; Harris et al., 2000; Harrison, 2011; Noether &

May, 2017).

Extensive research on hundreds of hospital mergers, led by economist Dranove,

determined that the majority of hospital mergers did not result in decisions to consolidate

clinical services. Consolidating clinical services involves changes that may directly and

adversely affect key constituents including physicians and nurses. High levels of internal

resistance make it extremely difficult to implement clinical consolidation. Consequently,

most of the limited and temporary cost-savings are attributable to the ability of the larger

enterprise to consolidate administrative and support services (e.g., accounting, human

resources, marketing, information technology) and leverage its enhanced purchasing

power to negotiate lower input pricing for certain drugs and supplies (Dranove &

Lindrooth, 2003; Dranove & Shanley, 1995).

The most controversial findings related to the impact of mergers on pricing. Since

government pricing is fixed, these studies evaluated the prices paid by commercial payers

for inpatient services following a merger with their pre-merger levels. All but one of the

10 studies found a positive relationship between mergers and inpatient pricing (Gaynor &

Town, 2012). Nine of the 10 studies (Akosa, Gaynor, & Vogt, 2007; Dafny, 2009;

Dranove, Lindrooth, White, & Zwanziger, 2008; Haas-Wilson & Garmon, 2011; Melnick

& Keeler, 2007; Tenn, 2011; Thompson, 2011; Town, Wholey, Feldman, & Burns, 2006;

Wu, 2009) used regional data (i.e., limited to one or two states) while only one

comparison (Moriya, Vogt, & Gaynor, 2010) used national data. Inpatient prices

increased 20% to 45% on average immediately after the merger but leveled off after three

to five years. The merged entity was able to leverage its enhanced market power—

sometimes monopolistic power—to increase fees.

Higher inpatient prices charged to commercial payers can lead to higher insurance

premiums for employers. A study published in 2015, which analyzed insurance premiums

throughout California, found monthly premiums in highly concentrated hospital markets

were 8% higher compared to competitive markets (Thompson, 2015).

Another notable finding about a merger’s impact on inpatient pricing is that there

is generally no difference between for-profit (FP) hospitals and not-for-profit (NFP)

hospitals involved with a merger. Post-merger performance and conduct were identical

whether evaluating the consolidation of two FP or two NFP hospitals (Farrell, Pautler, &

Vita, 2009; Vita & Sacher, 2001).

One small study (merger of two hospitals) examined nurse reactions to hospital

consolidation (Armstrong-Stassen, Cameron, Mantler, & Horsburgh, 2001). The study

was based upon previous research conducted by Cartwright (1997) that analyzed

employee responses to mergers across many industries, which identified high levels of

post-merger stress and dissatisfaction. Armstrong-Stassen et al. (2001) found a

statistically significant decrease in job satisfaction among nurses during the first two

years following the merger.

Perhaps the most comprehensive meta-analysis of hospital mergers was performed

by Bazzoli, Dynan, Burns, and Yap (2004). These researchers reviewed 38 merger-related

empirical research studies conducted over a 16-year time period. Bazzoli et al. applied a

conceptual model of organizational change based on four dimensions of change: content,

process, context, and outcomes. The authors concluded that most hospital mergers were

largely motivated by financial, strategic, and operational factors and primarily pursued to

increase market power, lower costs, and protect profitability. Merging organizations, in

general, did not fully consider, estimate, or place a high importance on psychological

costs (human factors) associated with consolidation. The human toll and resistance

originating from internal structures and relationships were commonly overlooked.

Consequently, estimates on the net impact of the merger (i.e., projected merger benefit –

projected merger cost = net impact) under-estimated the merger-related costs by not

appreciating the human factors and negative effects associated with the social

phenomenon. Essentially, the merger’s projected net impact

was over-estimated and overstated.

Sub-Section D – Review of Research on the Post-Merger Disruption Effect

There is a substantial body of quantitative and qualitative research identifying,

examining, and explaining the post-merger disruption effect. Marks and Mirvis (1985)

were the first researchers to formally label the social phenomena as the “merger

syndrome” (p. 50). This syndrome encompasses the various aspects of the “culture clash”

associated with the blending of two organizations (Marks & Mirvis, 1992, p. 19). The

symptoms of the syndrome include preoccupation by senior management with internal

power struggles, poor communication to frontline employees, decreased morale,

increased turnover, confusion, resistance to change, personal stress and anxiety,

uncertainty, active and passive conflict, and deteriorating organizational performance

(e.g., revenues not hitting forecasts, higher expenses, falling productivity, declining

profits, quality issues, customer complaints) (Altendorf, 1986; Meeks, 1977; Sinetar,

1981). The short-term consequence of this phenomenon is that the organization’s

attention turns inward in an effort to resolve syndrome-related problems and restore

organizational performance (Marks, 1994; Marks & Mirvis, 1985, 1997; Mirvis & Marks,

1992).

In an effort to view and understand the post-merger environment from an

individual perspective, Shaw (2002) incorporated elements of Hofstede’s (1991)

sixdimensional culture classification model and Stuart’s (1995) personal journey map into

a survey of individual employees involved with the merger of two British hospitals. This

approach provided insights into the various stages or “terrains across which individuals

make their change journey” (Shaw, 2002, p. 218). Although not every person in a

postmerger integration experiences all terrains, the general pathway includes the

following:

• sensing,

• worry work,

• positioning,

• shock,

• hoping and sharing,

• holding on,

• letting go,

• moving, and

• moving on and moving away.

Employees described personal feelings in the post-merger environment by using terms

such as emotional distress, deep pain, sad, tearful, unsettling, resentment, anger, despair,

grief, and negative (Shaw, 2002).

Cartwright and Cooper (1993) examined post-merger disruption at the middle

manager level. Middle managers play an important role in any organization since they (a)

frequently serve as a key intermediary between senior management and frontline

workers; (b) exert significant influence on the job satisfaction and overall perceptions of

their employees; and (c) are oftentimes the group of employees most adversely affected

by merger-related restructuring, consolidation, and rationalization plans.

Research by Cartwright and Cooper involved surveys of 157 middle managers

following the merger of two large professional service organizations (building societies in

England). The two organizations were considered highly culturally compatible prior to

merger based on results from an independent review and a cultural survey of employees.

The two societies served a similar customer base and performed similar services.

However, despite these apparent pre-merger cultural advantages, there was a substantial

level of disruption in the post-merger environment. Cartwright and Cooper used a

questionnaire of middle managers to gain insights into the underlying problems and

source of post-merger disruption.

One set of questions measured stress and mental health using two standardized

scales (Free Floating Anxiety and Hysterical Personality). The survey revealed high

levels of stress and fear among middle managers one year following the merger. These

managers revealed their reluctance to express anxiety believing that it could jeopardize

their career opportunities. The merger appeared initially and overtly to be progressing

well; however, the covert and pernicious effects were occurring beneath the surface

(Cartwright & Cooper, 1993).

Not all mergers trigger or result in the same level of post-merger disruption. Poor

cultural fit between two organizations is a critical factor in post-merger integration and

human resource problems. Cultural differences in the post-merger environment result in

misunderstandings, clashes, conflict, tension, emotional reactions, infighting, and

disruption (Hambrick & Cannella, 1993; Nahavandi & Malekzedah, 1988; Weber &

Schweiger, 1992). Weber’s exploratory research (1996), which focused on 73

acquirertarget mergers in the same industry (homogenous operations with similar

products and a common market of customers), sought to isolate, operationalize, and

measure certain variables (i.e., characteristics) of the two top management teams (TMT),

which are key components of cultural fit.

Weber (1996) hypothesized that the “greater cultural differences between the

combining top management teams, the lower the effectiveness of the integration process

and the lower the financial performance of the buying firm” (p. 1185). Controlling for

several organizational and merger characteristics, Weber found: (a) cultural differences

among TMTs were negatively associated with the effectiveness of the integration process

in acquirer-target situations, and (b) the effectiveness of the integration process affected

post-merger organizational performance and shareholder value (Weber, 1996).

Overall Synthesis of the Literature Review’s Sub-Sections A, B, C, & D

Providing a positive patient experience is a top priority for hospitals. How patients

perceive their inpatient care episode directly influences HCAHPS ratings and patient

preference/loyalty. Reimbursement from CMS and other payors is affected by

HCAHPS performance; therefore, a hospital’s financial well-being is linked to its patient

experience performance. Patient perceptions of inpatient care are predominantly

influenced by their interactions with physicians, nurses, and other staff. The working

environment and overall job satisfaction of physicians, nurses, and staff directly

influences the patient experience and a hospital’s HCAHPS ratings. It is nearly

impossible to deliver a positive patient experience when the key individuals responsible

for creating that experience (physicians, nurses, and general staff) are distracted,

frustrated, anxious, stressed, unhappy, disgruntled, or angry.

Hospital mergers are substantially motivated by a desire to achieve greater scale,

increase market power, and improve financial performance. Scale can be leveraged to

protect or increase pricing. Scale can also be utilized to achieve certain internal operating

efficiencies. In general, hospital mergers do not result in lower hospital fees or health

insurance premiums for the markets served. In most cases, prices and premiums increase

at a higher rate post-merger compared to similar hospitals that did not merge. Viewed as a

whole, merged hospitals do not outperform non-merged hospitals in terms of quality,

safety, and access.

Significant, internal disruptions occur following most mergers. Successfully

blending two organizations is delayed or hampered by culture conflict and TMT clashes.

Organizational performance can be adversely and temporarily affected by this postmerger

disruption effect as resources and attention are focused inward to address and resolve the

social disturbance.

Conceptual Framework

Most efforts to effect significant organizational change fail to meet expectations

(Burke, 2002; Probst & Raisch, 2005). Organizational change-related research, conducted

over a 30-year period involving numerous industries, found that 50% to 70% of major

change initiatives do not achieve their intended goals (Beer & Nohria, 2000; Sturdy &

Gray, 2003). Overall organizational performance oftentimes declines in the near-term

aftermath of core organizational change (March, 1981; Mellert, Scherbaum, Oliveira, &

Wilke, 2015). Significant organizational change also increases the potential for

organizational death (Singh, House, & Tucker, 1986).

Mergers represent a radical change in an organization’s structure, ownership,

governance, and strategic direction (Peus, Frey, Gerkhardt, Fischer, & Traut-Mattausch,

2008). Most mergers—across all industries, time periods and nations—either fail or fall

substantially short of achieving their intended goals (Bergh, 1997; Gugler & Yurtoglu,

2008; Montgomery & Wilson, 1986; Ravenscraft & Scherer, 1987, 1989; Tichy, 2001).

Additionally, Carper (1990) found a negative relationship between mergers and

shareholder wealth. Finally, in a meta-analysis of mergers and acquisitions (M&A), King,

Dalton, Daily, & Covin (2004) found post-M&A financial performance did not improve

and, on average, performance declined.

The U.S. health care industry, including the hospital sector, is no exception to the

challenges associated with mergers. Extensive research has been conducted examining

the impact of hospital mergers on various outcome variables including quality, safety,

access, service development, and expansion, pricing, internal operating costs, and

profitability. The general conclusions from this large body of empirical research suggest

that most hospital mergers do not achieve their intended operational or financial goals, do

not increase shareholder value (for-profit hospitals), and do not enhance consumer/public

welfare (Bazzoli et al., 2004; Gaynor, 2011; Gaynor & Town, 2012; Town et al., 2006).

Figure 4 and Figure 5 provide graphical presentations of the conceptual

framework developed for this research of hospital mergers and their impact on patient

experience. This framework depicts the relationships and social phenomena this study

examined. Figure 4 reflects the general mechanism associated with core, structural, or

second-order organizational change. Numerous research studies on radical change and

mergers identified and analyzed these phenomena (Carroll & Harrison, 2004; Haveman,

1992; Morgeson, Mitchell, & Liu, 2015; Pritchett, 1987; Weber & Camerer, 2003; Zell,

2003). How merging organizations address people and cultural integration directly

influences productivity, efficiency, quality, retention rates, and customer service (Bogen

& Symmers, 2001).

ORGANIZATIONAL PERFORMANCE

Figure 4. General mechanism associated with core, structural, or second-order

organizational change.

Figure 5 combines the concepts associated with core organizational change with

merger-related disruptive effects applicable to the multi-hospital merger situation.

Starting in the upper left corner, hospital or multi-hospital system A merges with hospital

or multi-hospital system B, resulting in multi-hospital system C. The merger is motivated

by offensive (proactive) and/or defensive (reactive) desires, goals, and forces. In addition

to significant investments of time and money, the pre-merger process involves various

strategic, financial, economic, operational, technological, legal, regulatory, governance

and political analyses, issues, and hurdles.

Once those hurdles are addressed, hospital/system A and hospital/system B merge,

becoming system C. Despite the wide-ranging pre-merger planning and preparatory work,

internal issues arise post-merger. Following the formal start of the merged entity’s

operations (Time 0 in Figure 5), system C experiences certain disruptive effects due to the

integration and blending of the two organizations. These cultural and human factor

effects, which will be discussed in Section Two, collectively cause an overall disturbance

Disruptive Effect on

Operations &

Culture

Employee

Resistance &

Emotional

Response

Conflict &

Alienation

Internal

Focus

Radical

Change

NEGATIVE IMPACT ON

(

distraction, decreased morale, and increased turnover

)

in system C’s internal operations. Consistent with findings by Bogen and Symmers

(2001), this social phenomenon, in turn, may adversely impact system C’s

organizational performance and stakeholder value creation.

One important measure of a hospital’s performance is patient satisfaction.

Therefore, a reasonable postulate based on this conceptual framework is that system C’s

patient experience performance will initially decline as a direct result of the mergerrelated

social phenomena and its disruptive effects.

Over time (shown in Figure 5 as three years post-merger or Time +3), the social

issues are fully resolved or, if not, the merger fails (the merger is dissolved with system A

and system B resuming independent operations). Assuming the disruptive effects are

resolved, system C’s operations enter a state of equilibrium as shown in Figure 5. The

focus shifts back to external value creation and, consequently, organizational performance

improves or is restored as measured by the various elements at the bottom of

Figure 5.

Figure 5. Organizational change and merger-related disruption effects.

Pertinent Theories

There are two broad categories of organizational behavior theories (see Figure 6)

applicable to this research: (1) organizational change related theories, and (2) merger

related theories. Each category contains several specific theories that explain the

relationships and general impact mechanisms depicted in the proposed conceptual

framework (see Figures 4 and 5). Two pertinent, specific theories (one theory from each

of the two broad categories) were selected to support this research.

PERTINENT THEORIES

Figure 6. Pertinent theories including organizational change theories and merger theories.

Organizational Change: Structural Inertia Theory

Organizational change has several definitions. The definition developed by Huber,

Sutcliffe, Miller, and Glick (1993) supports the conceptual framework discussed in

Section One. The authors defined the phenomenon as “[c]hange that involves differences

in how an organization functions, who its members and leaders are, what form it takes,

and how it allocates resources” (p. 216). In 1995, Van de Ven and Poole defined

organizational change as a difference in form, quality, or state. Health systems and health

ORGANIZATIONAL

CHANGE THEORIES

MERGER

THEORIES

M Disruption

D Pe

care organizations are under pressure to become more efficient, improve coordination,

take on more risk for clinical performance, and enhance service to patients (Bazzoli et al.,

2004). To address these challenges health care enterprises are changing, sometimes in

radical and discontinuous ways (Shortell, Gillies, Anderson, Mitchell, &

Morgan, 1993).

The organizational change-related theories emanate from comprehensive research

examining how and why organizational change occurs, how employees respond, how

change efforts affect performance, how change efforts often fail, and other factors.

Structural Inertia Theory (SIT) was applied in this study to support the conceptual

framework and hypotheses.

SIT is one theory among a group of organizational change theories labeled as

selection theories (Barnett & Carroll, 1995). Barnett and Carroll classified change

theories into two groups: (1) strategic adaptation theories, and (2) selection theories. The

first group, strategic adaptation theories, explains organizational change as a response to

market and technological pressures with the change progressing in harmony with the

external environment thus leading to positive results and market affirmation. In contrast,

selection theories suggest that organizations “cannot easily change and face substantial

risks when they do” (Bazzoli et al., 2004, p. 252).

Mellert et al. (2015) applied SIT, along with other change theories, in research on

the effectiveness of organizational change initiatives and the impact on financial

performance. Mellert and colleagues (2015) found substantial evidence to support SIT.

There is a high risk of failure associated with organizational change, particularly change

in core structure. Singh et al. (1986) also found that core organizational change increases

the probability of organizational death. In general, employees at all levels of the

organization resist change and respond negatively to change efforts (structural inertia)

(Peus et al., 2008).

The magnitude of change is an important correlate of change commitment: the

more the change impacts organizational operations, the larger the resistance (Fedor,

Caldwell, & Herold, 2006). Internal elements in the organization (e.g., authority

structures, culture, routines) serve as passive or active deterrents to change (Peus et al.,

2008). These phenomena are amplified in merger and acquisition situations. A merger is a

special form of core organizational change and typically results in greater internal

resistance and disruptive effects (Cloodt, Hagedoorn, & van Kranenburg, 2006; Correia,

Cunha, & Scholten, 2013).

In previous research applying SIT, Haveman (1992) found that core organizational

changes (e.g., change in strategic direction, merger, acquisition) reset the

“liability-of-newness clock” (Stinchcombe, 1965, p. 143). The organization must

establish new routines, learn new patterns of communication, build new flows of

information, and integrate new members. Immediately following the implementation of

radical change, the “organization diverts a considerable portion of its resources from

operating to restructuring” (Haveman, 1992, p. 51). In other words, the organization

becomes more inwardly focused and less focused on its customers. Consequently,

operating efficiency declines which is followed by lower performance.

Morgeson et al. (2015) examined organizations as a composition of events within

an industry comprised of a broad pattern of environmental events. The authors

categorized events using three parameters: novelty, disruption, and criticality. All events

have some level of strength (impact), space (origin and scope of impact within the

organization), and time (duration). Merger events are highly novel, disruptive, and

critical. Mergers have high impact, affect all levels of an organization, and have a long

duration (two to three years or longer). Mergers may be a shock, jolt, and turning point

for an organization and are described as a discontinuous, non-routine, and discrepant

event. Consequently, a lingering period of internal confusion, uncertainty, disturbance,

and instability may follow a merger. Disruptive events, such as mergers, shake people out

of their routine thinking, behavior, and response modes. These changes affect all levels of

the organization including the individual, team, department, and business unit. Finally,

the bigger the event, the more time it may take the organization to fully absorb the

changes and return to a state of equilibrium, stability and normalcy (Morgeson et al.,

2015).

Zell’s research (2003) on organizational change applied Kubler-Ross’s (1969) five

stages of grief to understand resistance to major change within professional

bureaucracies. Zell defined professional bureaucracy as “an organization consisting of

highly trained and autonomous professionals such as hospitals and universities” (p. 73).

Mintzberg (1983) identified special dynamics in professional bureaucracies that uniquely

affected organizational change. Resistance to change is higher, the pace of change is

slower, and the success rate of sustainable change is significantly lower (Mintzberg,

1983).

Merton’s (1957) research on bureaucracies and change identified two

dysfunctions related to organizational inertia: (a) bureaucracies are excessively rigid in

their application of rules and regulations, which limits their ability to change; and (b)

bureaucracies are “inherently conservative and resistant to innovation” (p. 48). Physicians

and nurses are members of professional societies and enjoy varying levels of

independence in their practice. Successful and sustainable change in a professional

bureaucracy is highly contingent upon the active support of professional staff. Active or

passive resistance to change by members of the professional bureaucracy creates more

upheaval, emotional turmoil, conflict, and disruption. Due to the special characteristics of

the professional bureaucracy, Zell observed that the final Kubler-Ross stage (acceptance)

may not occur until two to three years following the commencement of the change

process (Zell, 2003).

Merger-Related: Culture Conflict Theory

Several merger-related theories have been developed by economists,

organizational behavior scientists, sociologists, and psychologists to answer one

overarching question: Why do most mergers fail or substantially fall short of their

intended goals? Culture conflict has been found to be the principal culprit. Merger

research has discovered a unique set of behavior factors that affect and shape the

postmerger environment. Human actors’ behavior affects, and is affected by, the distinct

and temporary dynamics that occur post-merger. Certain patterns in this organizational

behavior have been observed and categorized. Culture Conflict Theory (CCT) was used

to explain the social phenomena and support the conceptual framework.

Developed by Weber and Camerer in 2003, CCT is an extension of prior,

mergerrelated empirical research involving culture conflict. Weber and Camerer

conducted laboratory experiments of human subjects in a simulated merger situation and

discovered prospective partners significantly underestimated or ignored cultural hurdles

during the pre-merger due diligence period. Weber and Camerer posited that culture is

rooted in shared meaning and shared meaning is based on common language, beliefs, and

understanding. Study results supported research hypotheses. Performance of the

simulated merger operations was adversely affected by culture conflict among human

subjects (compared to pre-merger performance). Culture conflict was observed and

measured using standardized metrics.

Weber and Camerer constructed a merger-specific culture conflict theory that

includes the following broad components: (a) culture conflict can be mitigated and

potentially avoided if the parties to a merger take appropriate precautions in advance; (b)

culture conflict leads to post-merger performance deterioration; (c) even when the

potential for culture conflict is acknowledged, participants in the experiment significantly

underestimated the performance decline that would occur; and (d) individuals are more

likely than not to assign blame for culture conflict to employees from the other

organization than take responsibility for their contributions to the conflict (Weber &

Camerer, 2003).

CCT is an extension of a large body of research on organizational culture as both

a dependent and independent variable (Deal & Kennedy, 2000; Kilmann, Saxton, &

Serpa, 1985; Kotter & Heskett, 1992; Schein, 1992). Mergers involve acculturation, the

blending of two organizations’ cultures (Carroll & Harrison, 2004). Nahavandi and

Malekzadeh (1988) developed a conceptual framework with four modes of successful

post-merger acculturation: (1) preservation (both parties maintain their existing cultures),

(2) symbiotic (intentional efforts to combine the best cultural elements of the two

organizations), (3) absorption (only the culture of the dominant organization is

maintained), and (4) transformation (a completely new culture is established). Many

organizational culture researchers suggest that acculturation is the single most important

predictor of post-merger success (Carroll & Harrison, 2004; Ellis & Lamont, 2004;

Larsson & Finkelstein, 1999).

Carroll and Harrison (2004) provided insight on the roles culture and

acculturation play in post-merger performance. Evidence showed that the most effective

and expeditious method for resolving post-merger culture conflict consisted of two steps:

(a) clearly establish and communicate the norms and values that define the culture of the

merged entity, and (b) use alienation tactics as a way to achieve conformity or elimination

of those who decide not to conform. This method works best in acquirertarget situations

where a larger entity acquires or takes over a smaller organization (Carroll & Harrison,

2004).

CCT is further supported by research conducted by Allred, Boal, and Holstein

(2005). The authors initially set out to examine why there was such a high failure rate for

mergers. Allred and colleagues found that economic logic, financial factors, and

economic performance post-merger did not explain the failure rate. After evaluating

numerous variables Allred et al. (2005) hypothesized that human integration and

interaction factors were the dominant contributors to merger failures.

Furthermore, the authors noted a close psychological parallel between the

dynamics involved with the merging of two step-families and the integration of two

merged organizations. In a remarriage situation, the two partners’ needs are the focus

often without regard to the needs of the children (Lampard & Peggs, 1999). Allred et al.

(2005) identified three stepfamily models that explain the relationship problems that often

occur: (1) biological discrimination (stress between the stepparent and stepchildren); (2)

incomplete institutionalization (greater stress post-merger combined with a lack of

common rules, which leads to a perception of insiders and outsiders); and

(3) the deficit-comparison (employees from the dominant organization in the merger will

have greater job security and resource allocation compared to employees of the subordinate

organization).

Relatedness and relative size between the two merging organizations may also

limit and influence the level of culture conflict. There are two unique merger types

subject to the related and relative size influence. The first is when a large firm acquires a

small firm (i.e., a takeover or acquirer-target situation). In this case, employees and

managers of the small firm may feel unimportant and trivialized by the buyer. This leads

to feelings of alienation and discontent among the small firm staff. Culture conflict is

limited to the small firm since the culture of the large, acquiring firm is imposed upon the

target firm’s operations (Weber, 1996).

The second situation in which relatedness and relative size moderates culture

conflict is in a full merger-of-equals (MOE) situation involving firms of similar size

operating in the same industry. In this case the two firms are rivals; they know each other

well and have overlapping geographic market areas. There is no acquirer or target and no

clear winner or loser. As a result, MOEs intensify culture conflict with TMT power

struggles, surfacing of hidden agendas, board control clashes, policy disputes, and

political infighting (Devine, Lamont, & Harris, 2016; Lodorfos & Boateng, 2006).

Acculturation is severely impaired by this environment, which results in internal turmoil

and suboptimal organization performance until the culture conflict issues are resolved

(McEntire & Bentley, 1996).

Post-merger culture conflict varies by the type(s) of organizations involved.

Culture conflict may be lower when two manufacturing firms consolidate (Weber, 1996)

and more intense, with a longer duration, within the service industry (Bond & Weisman,

(1997). Among service industry firms, the hospital and health system environment may

be the most challenging of all (Bruhn, 2001).

Composite: Structural Inertia Theory Plus Culture Conflict Theory

The composite theoretical framework is shown in Figure 3. These two theories

(SIT and CCT) provide the supporting empirical research and conceptual foundation for

the primary impact mechanism of interest to this research: significant and predictable

social disruptions directly cause a temporary decrease in post-merger organizational

performance. In other words, post-merger performance is inversely related to the level of

disruption that occurs so that higher disruption equals lower performance.

The disruptive effect is a social phenomenon. People at all levels of authority and

seniority in the merging organizations frequently respond and react in negative ways.

They passively or actively resist change. There is an increase in stress, confusion, fear,

uncertainty, anxiety, betrayal, and feelings of isolation. Job satisfaction and job security

levels decline. Cultural conflict exacerbates the intensity and duration of the disruption.

Human energy and efforts are distracted and inwardly focused. Consequently,

organizational performance is often adversely affected until the integration and social

issues are fully resolved. Eventually, the merged organization achieves equilibrium and

resumes stable operations.

The disruptive effect and resistance may be worse in a hospital setting due to the

entrenched professional bureaucracies of physicians and nurses (Bond & Weisman, 1997;

Zell, 2003). Organizational change is difficult. Mergers, as a radical form of

organizational change, are extremely challenging and risky. Of all business and industry

environments, the unique context of a merger in a hospital or health care system setting

may be the most perplexing organizational change situation (Bruhn, 2001).

Combining the post-merger research with the two pertinent theories further

illuminates, and provides a compelling reinforcement for, the conceptual framework. The

study of hospital mergers, including an examination of the factors that led to the merger

decision and the merged entity’s post-merger performance, reveals several interesting

points. First, the pre-merger activity is primarily focused on the various technical tasks

required to consummate the transaction (i.e., legal, regulatory, financial due diligence).

Second, little, if any, attention is given to the soft or human components of the merger.

Third, senior leadership of the surviving entity initially focuses their post-merger

operations management and strategic execution on maximizing the merged entity’s

market power (i.e., optimize reimbursement from commercial payers and negotiate lower

input pricing). Fourth, the social phenomenon and the various disruptive effects begin to

manifest themselves soon after the merger. Fifth, it takes two to three years following the

merger for the merger disruption effects to fully dissipate.

Statement of 12 Hypotheses

It is expected that post-merger disruption will adversely impact a recently merged

hospital’s overall performance. One measure of a hospital’s performance is patient

experience. Consistent with the literature review, conceptual framework, SIT, and CCT, a

hospital’s patient experience performance will decline in the near-term following a

merger:

Hypothesis 1: A hospital’s overall patient experience ratings will be negatively

associated with having recently completed a merger (within the last 12-24

months). Compared to similar hospitals that did not merge, recently merged

hospitals’ year-over-year change in overall patient experience ratings will be

lower.

Post-merger disruption is a social phenomenon that affects human behavior and

individuals’ emotional and mental states. Patient experience performance is directly

influenced by the attitudes and behaviors of key employees including physicians, nurses,

and other staff. The post-merger disruption effect manifests itself by increased

anxiety/stress, increased turnover, and decreased job satisfaction among employees. A

hospital’s HCAHPS composite scores among each of the three human interaction

domains (physician communication, nurse communication, and staff responsiveness) will

be adversely affected as a consequence of the post-merger disruption.

Hypothesis 2a: A hospital’s physician communications ratings will be negatively

associated with having recently completed a merger (within the last 12-24

months). As a group, recently merged hospitals’ year-over-year change in

physician communications ratings will be lower compared to the physician

communications ratings for similar hospitals that did not merge.

Hypothesis 2b: A hospital’s nurse communications ratings will be negatively

associated with having recently completed a merger (within the last 12-24

months). As a group, recently merged hospitals’ year-over-year change in nurse

communications ratings will be lower compared to the nurse communications

ratings of similar hospitals that did not merge.

Hypothesis 2c: A hospital’s staff responsiveness ratings will be negatively

associated with having recently completed a merger (within the last 12-24

months). As a group, recently merged hospitals’ year-over-year change in staff

responsiveness ratings will be lower compared to the staff responsiveness ratings

of similar hospitals that did not merge.

As discussed previously in this chapter, culture conflict and the merger-related

disruption effect are oftentimes greater and more difficult to resolve in the MOE situation

versus an acquisition.

Hypothesis 3a: Compared to the subgroup of Acquired hospitals, the subgroup of

MOE hospitals will have a lower, year-over-over change in overall patient

experience ratings. In other words, the MOE hospitals subgroup will have a

bigger negative comparative difference in overall patient experience ratings

versus the Acquired hospitals subgroup.

Hypothesis 3b: Compared to the subgroup of Acquired hospitals, the subgroup of

MOE hospitals will have a lower, year-over-year change in physician

communications ratings. In other words, the MOE hospitals subgroup will have a

bigger negative comparative difference in physician communications ratings

versus the Acquired hospitals subgroup.

Hypothesis 3c: Compared to the subgroup of Acquired hospitals, the subgroup of

MOE hospitals will have a lower, year-over-year change in nurse communications

ratings. In other words, the MOE hospitals subgroup will have a bigger negative

comparative difference in nurse communications ratings versus the Acquired

hospitals subgroup.

Hypothesis 3d: Compared to the subgroup of Acquired hospitals, the subgroup of

MOE hospitals will have a lower, year-over-year change in staff responsiveness

ratings. In other words, the MOE hospitals subgroup will have a bigger negative

comparative difference in staff responsiveness ratings versus the Acquired

hospitals subgroup.

Merger disruption is a temporary phenomenon. Based on a vast body of

postmerger research, there is general agreement that the merged operation will, in most

cases, achieve cultural equilibrium and resume normal operations within three years

following the merger. Performance declines, attributable to the disruption effect, will be

reversed. Hypothesis 4a: Merged hospitals’ overall patient experience ratings will return

to normal levels within 36 months post-merger with no performance difference compared

to similar hospitals that did not merge.

In the long-term, the post-merger disruption effect that negatively impacted

physicians, nurses, and other employees will be fully resolved. The merged hospital’s

HCAHPS composite scores among each of the three human interaction domains

(physician communication, nurse communication, and staff responsiveness) will return to

normal levels.

Hypothesis 4b: There will be no difference in physician communications ratings

between merged and similar unmerged hospitals three years following the merger.

Hypothesis 4c: There will be no difference in nurse communications ratings

between merged and similar unmerged hospitals three years following the merger.

Hypothesis 4d: There will be no difference in staff responsiveness ratings

between merged and similar unmerged hospitals three years following the merger.

CHAPTER 3

METHODOLOGY

Chapter 3 describes the research methodology. First, the general research design

is described followed by a definition and explanation of the measures for the dependent

and independent variables. Second, the process is explained for identifying the merged

and unmerged/matched hospitals. Third, data sources for each variable are presented. The

chapter concludes with a summary of the statistical testing that was conducted for each of

the 12 hypotheses.

Research Design

The purpose of this research was to examine the relationship between hospital

mergers and patient experience. Hospitals were the subjects of this study. Consistent with

other hospital consolidation and closure research (Farrell et al., 2009; Ona, Hudoyo, &

Freshwater, 2007; Sinay & Campbell, 2002), this study used a retrospective,

quasiexperimental design of merged hospitals (intervention group or the “Event

hospitals”) compared to unmerged hospitals (synthetic control group or the counterfactual

“Matched

Hospitals”). For purposes of this study, “intervention” will refer to a merger between two

or more hospitals.

Since patient experience was measured both pre- and post-merger, with three

measures for both pre- and post-periods, the design meets the definition of an interrupted

time series study (AHRQ Taxonomy for Defining a Research Design, Appendix C).

Measures

Dependent Variables – Patient Experience Scores

The dependent variables for this study were four HCAHPS composite ratings (see

Appendix B for the complete set of HCAHPS survey questions):

(a) Overall patient experience (Question 21), hereafter referred to as “OVERALL”.

The percentage of patients who rated their overall patient experience as a “9” or a “10”,

on a scale of 0 to 10);

(b) Nurse communication (Questions 1, 2, and 3), hereafter referred to as “NURSE

domain.” The percentage of patients who responded “always” (among the four

response options: “never,” “sometimes,” “usually,” and “always”). These

percentages were averaged across the three nurse communication questions to

create a composite NURSE domain variable;

“PHYSICIAN domain.” The percentage of patients who responded “always” (among the

four response options: “never,” “sometimes,” “usually,” and “always”). These

percentages were then averaged across the three physician communication questions to

create a composite PHYSICIAN domain variable;

(d) Staff responsiveness (Questions 4 and 11), hereafter referred to as “STAFF

domain.” The percentage of patients who responded “always” (among the four

response options: “never,” “sometimes,” “usually,” and “always”). These

percentages were then averaged across the two staff communication questions to

create a composite of the

STAFF domain variable.

HCAHPS measurements (Figure 7) from six 12-month periods were obtained for

each Event hospital and Matched hospital: “T1” corresponded with the period three years

prior to merger, “T2” corresponded with the period two years prior to merger, “T3”

corresponded with the period one year prior to merger, “T4” corresponded with the period

one year following merger, “T5” corresponded with the period two years following

merger, and “T6” corresponded with the period three years following the merger year.

Time Span of the Six Year-Ending HCAHPS Measures

T1 T2 T3 T4 T5 T6

25-36 Months 13-24 Months 1-12 Months 1-12 Months 13-24 Months 25-36 Months Pre-Merger Pre-Merger Pre-Merger Post-Merger

Post-Merger Post-Merger

Figure 7. Time span of the six HCAHPS measures.

Independent Variable – Merged or Unmerged

The independent variable for this study was a dummy variable that reflected the

merger status of a hospital: “event” hospital (hospital that was part of a merger or was

acquired) (1) or unmerged or “matched” hospital (0). The following sections provide the

working definition that was used to identify the Event hospitals and the matching

approach that was utilized to identify the Matched hospitals.

Rationale for the Six Measures and the Timing of Each

The first three measures (T1, T2 and T3) provide a snapshot of the “pre-event”

status. These three measures reflect an Event hospital’s annual HCAHPS performance

levels prior to the potential impact of merger-related disruption effects. The last three

measures (T4, T5 and T6) provide the post-merger HCAHPS performance: one year post,

two years post, and three years post. Applying the theoretical basis concerning merger

disruptions, the hypothesized, adverse impact on HCAHPS performance will manifest

itself within the first 12 to 24 months following the merger event (T4 and T5). The final

measure (taken three years following the merger event) reflects the HCAHPS

performance beyond the merger disruptive effects. In other words, the final measure (T6

in Figure 7) reflects HCAHPS performance after the full resolution of the disruption

effects (i.e., the hospital has resumed normal operations).

Three years post-merger was selected based on several research studies in which

researchers suggested that it takes two to three years for the organizational change and

merger-related negative effects to dissipate (Bazzoli et al., 2004; Habeck, Kroger, &

Tram, 2000; Lim, 2014; Zell, 2003). Individual years were selected in order to detect any

temporal effects, trends and differences for each of the years leading up to the merger

event and for each of the years following the merger event. Looking at a composite rating

for all three years prior to the event (composite of T1, T2 and T3 in one HCAHPS rating)

or post-event could mask a temporal trend. For example, if HCAHPS performance among

the Event hospitals significantly declined in comparison to the Matched hospitals in only

one of the three post-merger years, a single post-merger composite rating (combining T4,

T5 and T6 into one value) would, possibly, mask that performance difference.

Study Population – Inclusion and Exclusion Criteria

Definition of Event Hospitals

The population of Event hospitals consisted of U.S.-based acute care hospitals

(ACH), excluding federal government-owned/operated and Indian Health Service

hospitals, that were part of a merger transaction that closed during the five-year time

period of January 1, 2009 through December 31, 2013. An “Event” was defined as a

transaction between two or more ACHs, or health systems owning/operating two or more

ACHs, involving the combining or consolidation of those hospitals’ operations into one

surviving organization (i.e., one single tax identification number entity) that has one

governing body and one common management team.

This type of organizational consolidation can occur through a Merger-of-Equals

(MOE) or Acquisition. The following working definition was used for an MOE (adapted

from Georgieff & Latsky, 2018, page 61; DePhamphilis, 2014, pages 412-413):

a transaction consisting of two organizations of generally similarly sizes (primarily

using most recent year-end operating revenue to determine size),

• there was no designated “acquirer”,

• post-merger, the two legacy organizations formed a new entity with a

new name and new governance structure (i.e., neither of the two

legacy system’s names survived), and

• all assets held by the two legacy organizations were placed under the

control of the new entity and its governance.

Conversely, an Acquisition transaction involves one larger organization

purchasing the assets or stock of a smaller organization with the larger organization’s

name and governance surviving the transaction. Post-transaction, all of the assets of the

acquired entity are under the control of the acquiring entity. A dummy variable was

created for “merger type” with each Event hospital categorized as an MOE (1) or

Acquisition (0).

The following multi-hospital consolidation transactions were excluded from the

sample: ACHs that entered into other forms of ACH-to-ACH or health system-to-health

system relationships (e.g., joint ventures, partial ownership interests, affiliations, shared

service agreements, contract management); ACHs acquired through bankruptcy

proceedings; merged or acquired ACHs that terminated operations during the study

period; and ACHs that completed one merger followed by another merger or

consolidation transaction within the 2009-2013 time period.

Although the MOE or Acquisition results in a consolidated operation with one tax

identification number entity owning the merged or acquired ACHs, in the vast majority of

cases, each merged ACH continues to operate under its own National Provider Identifier

(NPI). Since each ACH is required by CMS to report its individual HCAHPS survey

results, HCAHPS ratings continued to be available for each merging ACH.

Identifying the Study’s Event Hospitals

Irving Levin Associates (ILA) compiles and publishes an annual report containing

the pending, announced and completed hospital mergers and acquisitions during the

preceding calendar year. To evaluate the status of ILA as a source for hospital mergers

and acquisitions, a literature search, limited to items published between 2000 and 2017,

was conducted using the following key words: “hospital” + “merger” + “Levin.” Over

1,000 published articles and reports were identified, which reference the ILA reports at

their source for hospital merger and acquisition activity. In addition, queries were

submitted to Modern Healthcare, Dixon Hughes Goodman, Becker’s Hospital Review,

and Kaufman-Hall seeking sources of information for tracking hospital mergers in the

United States. All the individuals contacted use the ILA report exclusively. Therefore,

The ILA Hospital Acquisition reports for 2006 through 2016 (11 calendar years) were

purchased. These reports served as the source for identifying the qualifying ACHs that

merged.

The first step involved identifying all qualifying transactions announced during

the five-year time period, 2009 through 2013. This transaction period was chosen to allow

for three-year pre- and post-merger periods of evaluation. Based on the review of the ILA

reports, 381 ACHs were part of a Merger-of-Equals (MOE) or Acquisition announcement

from January 1, 2009 through December 31, 2013 (collectively referred to as “Event”

hospitals). Figure 8 provides a graphical summary of the overall identification and

matching process resulting in the final set of 99 pairs (198 hospitals) for this study.

NOTE: For purposes of this study, all the ACHs involved in an MOE were

included but only the target hospital (i.e., the hospital being acquired) was counted in the

Acquisition event category. The rationale for not including all the acquiring entity’s

hospitals is based upon the theoretical framework and empirical research on the

potentially disruptive effects of mergers and acquisitions. In an MOE situation, both

legacy organizations are subject to the substantial change and social upheaval associated

the blending of the two organizations (Devine, Lamont, & Harris, 2016; Lodorfos &

Boateng, 2006). In an Acquisition situation, the larger, acquiring organization remains

operationally and culturally intact while the “target” organization (for this study, the

hospital that was purchased), potentially, experiences significant change and disruption as

it transitions into a new governance and management structure, new name, new policies

and procedures, and new culture.

The next step involved determining whether each of the pending, announced, or

proposed transactions was consummated. Following the initial listing of a proposed

transaction, the ILA reports did not consistently include a subsequent entry for closed

transactions or previously proposed deals that did not close. Therefore, in addition to the

ILA reports, other sources and methods were used to ascertain the final status for each of

the proposed transactions. Those sources included Modern Healthcare, Becker’s Hospital

Review, and regional newspapers and business periodicals. Finally, in some cases, none of

the published sources contained conclusive information about the final status of a

proposed transaction originally recorded in ILA. In these cases, a Google search was

conducted using a combination of the following key words: name of hospital(s), merged

(or acquired), closed, final and approved.

After applying all sources and methods for verifying which transactions closed,

164 hospitals were eliminated. In all 164 cases, the initially proposed MOE or

Acquisition failed to materialize for a variety of reasons, including one or both parties to

the deal deciding not to pursue the merger, community backlash, medical staff backlash,

disapproval or a threatened challenge from state or federal authorities, lack of financial

resources, or other factors. Of the originally identified 381 hospitals, 217 hospitals

completed the MOE or Acquisition transaction (381 - 164 = 217). When each MOE or

Acquisition was confirmed, the month and year of the transaction’s closing was recorded.

The final step involved applying the previously described inclusion and exclusion

criteria. For example, if a hospital was acquired in 2009 by one health system and the

acquiring health system merged with another health system in 2011, that hospital was

excluded from the study. If a merged or acquired hospital closed within the three-year

period following the transaction, it was excluded. Thirty-eight hospitals were removed

after applying the inclusion and exclusion criteria. A total of 179 acute care hospitals

remained in the study’s population (217 - 38 = 179).

Building the Profile for Each MOE or Acquired Hospital

In preparation for matching, a comprehensive profile was constructed for each of

the 179 hospitals. The AHA Annual Survey was used as the primary source. The year of

the transaction determined which AHA Annual Survey year to use. For example, if an

MOE occurred in 2009, the MOE hospital’s profile was completed using data from the

2009 AHA Annual Survey. The following 10 data points were obtained and recorded

from the AHA reports:

a. hospital name and address,

b. AHA identification number,

c. Medicare certification number,

d. National Provider identification number,

e. geographic area (name of the CBSA or the word “rural” if the hospital was

not located in a defined CBSA),

f. number of licensed beds (continuous variable),

g. annual admissions (continuous variable),

h. annual adjusted admissions (continuous variable),

i. system status (categorical, dummy variable), and

j. profit status (categorical, dummy variable).

In approximately 25% of cases, the NPI number was not included in the AHA Annual

Survey. An alternative source was identified and used to obtain all the missing NPI

numbers (www.NPInumberlookup.org).

Matched Hospitals

Hospitals in the Matched hospital group were identified using pre-determined

matching variables (see next section, Matching Approach). The goal was a 1:1 ratio of

Matched hospitals and Event hospitals with no hospital serving as a match more than

once. All Matched hospitals were part of a multi-hospital system or a single hospital

operation that did not experience a merger event during the same time period (20092013).

Consistent with the theoretical framework, the Matched hospital group excluded hospitals

that were part of a merger or acquisition (including a hospital or health system that

acquired another hospital or health system) that occurred three years prior or three years

after the corresponding Event hospital’s transaction year. This was done to avoid overlap

with the disruptive effects of a merger within the Matched hospital group.

Matching Approach

The methods used for selecting the Matched hospitals were critical components of

the study. Hospital performance is significantly affected by local and regional factors

such as demographics, population growth or decline, economic conditions,

reimbursement rates, payor mix, state Medicaid expansion, presence of certificate-ofneed,

social issues, and more. The goal was to find the ACH that most closely resembled each

merged ACH.

The matching process involved two sequentially applied methods: (1) an

automated approach using Propensity Scoring Matching with greedy matching (PSM),

and (2) manual matching. PSM was conducted first followed by manual matching.

Manual matching was necessary since some of the Event hospitals were not located in a

defined CBSA and since PSM was limited to “within the same CBSA” hospitals.

Initial Matching Using PSM

A hierarchical PSM was used for the initial identification of a Matched hospital

for each of the Event hospitals. PSM used the following categories of matching

covariates: (1) geographic match by name of the Core Based Statistical Area (CBSA) (see

Appendix A for the U.S. map of CBSAs), and (2) size match using licensed beds,

admissions and adjusted admissions. The year of the Event hospital’s transaction

determined which year of AHA Annual Survey to utilize for the matching covariates. For

example, if a hospital was acquired in 2010, PSM utilized data from the 2010 edition of

the AHA Annual Survey for PSM matching.

PSM provided a preliminary match for 152 of the 179 Event hospitals. Several of

the 27 “unmatched” Event hospitals were located outside of a defined CBSA (i.e.,

hospitals located in rural areas). In these cases, the PSM method was unable to locate a

nearby geographic match.

Manual Matching

Following the output of PSM, a manual or non-automated matching approach was

used. This involved a case-by-case review of all hospitals located within the Event

hospital’s CBSA, adjacent CBSA, or nearby rural county (for Event hospitals located

outside of a defined CBSA). Appendix D contains a summary of the manual matching

algorithm. Of the 27 “unmatched” instances from PSM, manual matching located 20

acceptable matches. In total, there were 172 hospitals with preliminary matches, resulting

in the elimination of seven of the 179 Event hospitals.

Examination and Testing of the 172 Preliminarily Matched Hospitals

Since the AHA Annual Survey does not include information indicating that a

hospital has recently merged or been acquired (within the past three years) or was a party

to a future merger or acquisition (within three years into the future), the PSM and manual

matched list needed to be further examined, tested, and, if necessary, revised. For

example, if PSM or manual matching provided a matched hospital that merged or was

acquired in the past three years or within three years following the MOE or Acquisition

transaction, that hospital could not serve as an acceptable counterfactual hospital.

Each Matched hospital was examined to determine whether or not it was part of a

merger or acquisition (M&A) event during the applicable three years pre- or post-event.

The three years pre- or post-event period was determined by the transaction date of the

Event hospital being matched. For example, a matched hospital for an Event hospital that

closed a transaction in 2009 could not have completed a merger or acquisition transaction

three years pre- (2006-2008), in 2009, or three years post- (2010-2012) transaction. When

the examination identified a preliminarily matched hospital that was a party to its own

event during the applicable time period, an alternative match was sought using the

manual matching method.

The ILA reports served as the primary source for the examination and testing of

the Matched hospitals (to determine if a preliminarily matched hospital was a party to a

merger or acquisition in the applicable three-year time period prior to and following the

paired Event hospital’s transaction. For situations in which there was a question or

concern about the accuracy or confirmatory status in the ILA report, other sources were

used including Modern Healthcare, Becker’s Hospital Review, and a Google search. A

verifying note was entered into the database for each of the 172 Matched hospitals.

In some CBSAs (e.g., Chicago and Seattle), M&A activity was intense between

2006 to 2016. Consequently, otherwise suitable matches or alternative matches were

disqualified due to a recent or future M&A. This made it impossible to locate an

acceptable matched hospital for several Event hospitals.

Sixty-four of the preliminarily matched hospitals were disqualified for various

reasons including discovery of a time-overlapping merger or acquisition (52 occurrences),

PSM yielding a Veterans Administration hospital (inclusion/exclusion criteria specified

no government owned and operated hospitals) as a match (four occurrences), matched

hospital terminating operations within three years (six occurrences), and PSM yielding a

rural hospital more than 200 miles away from the Event hospital (two occurrences).

Twenty-five (25) of these hospitals were replaced with an alternative match using another

manual match process. The remaining 39 that could not be matched with the manual

matching process were eliminated, along with the corresponding Event hospital. This

resulted in 133 Event hospitals and 133 Matched hospitals (172 - 39 = 133).

Final Verification of the Event and Matched Hospitals

To ensure the highest level of accuracy, a final verification review was conducted

for all 266 Event and Matched hospitals. This review involved a hospital-by-hospital,

final quality assurance test of the inclusion and exclusion criteria along with a

confirmation of suitable matches. After all final examination and testing, 133 Event and

Matched hospital pairs remained.

Obtaining the HCAHPS Ratings – Elimination of 34 Pairs

HCAHPS data were provided by Press-Ganey (PG). PG is a publicly traded

company with over 3,000 hospital clients in the United States. PG’s primary service is

helping hospitals and health systems measure, manage, and optimize their patient

experience performance. PG is also one of the CMS-approved companies for providing

HCAHPS surveys and intermediating HCAHPS reporting to CMS.

Hospitals report their HCAHPS performance data to CMS on a quarterly basis

with a rolling, cumulative 12-month rating reported for each quarter. In other words, the

quarterly value reported by CMS reflects the hospital’s composite HCAHPS ratings for

the last 12 months.

The principal investigator sent PG a data file containing the name, address, NPI

number, and Medicare number for all 266 hospitals. PG imported the quarterly ratings for

each of the four HCAHPS domains. Thirty-four (34) Event hospitals along with their

counterfactual Matched hospitals were eliminated from the study due to missing

HCAHPS ratings for either one or both hospitals in the pair (68 hospitals removed in

total). In some cases, the Event and Matched hospital were critical access hospitals

(CAH). CAHs are not required to submit HCAHPS data to CMS. In other cases,

particularly for a merger or acquisition event that occurred in 2009, HCAHPS ratings

from 2007 (two years prior to the event) were not reported. In cases involving missing

HCAHPS ratings for a Matched hospital, additional efforts were made to find an

alternative matched hospital. No suitable matches were identified. Missing HCAHPS

ratings for an Event hospital resulted in the elimination of both the affected Event

hospital and the corresponding Matched hospital.

Complete HCAHPS ratings for the four HCAHPS domains (Overall,

PHYSICIAN, NURSE, and STAFF) and for all six time periods were available for 99

Event and 99 Matched hospitals (133 pairs – 34 pairs = 99 pairs). These 198 hospitals

defined the population of hospitals in this study. Appendix E contains the complete list of

the final 198 hospitals included in the analytic sample. Figure 8 provides a graphical

summary of the identification and matching process.

Figure 8. Identification and matching process resulting in the 99 pairs.

Goodness-of-Match and Balance Diagnostics

One of the challenges in studies involving counterfactual subjects as a synthetic

control group is finding suitable matches (Campbell & Harper, 2012). Due to the number

of matching covariates involved, finding an “exact” match was not practical.

In the hierarchical matching algorithm for this study, geographic match was given

the highest priority. All 99 matched pairs in this study contain hospitals operating in the

same CBSA, an adjacent CBSA, or another rural area (non-CBSA) in the same state or

adjacent state within 200 miles of the Event hospital. The mean distance between the

Event and Matched hospital was 36.3 miles. There were only four sets of paired hospitals

located more than 100 miles apart, with all eight hospitals located in rural counties within

the same state or an adjacent state. Therefore, there was a moderate to high level of

geographic matching. As previously described, the rationale for prioritizing geographic

matching was to account for the demographic and cultural characteristics, economic

conditions, and secular factors that vary from one region of the United States to another.

Additionally, since hospitals in the same or a nearby region may compete for the same

market of patients, the HCAHPS performance of those in-region hospitals may exert

more competitive influence on each other versus a hospital from another region of the

United States. In other words, since patients tend to seek hospital care from hospitals near

their residence, the competitive actions and performance of hospitals near each other (i.e.,

in the same or adjacent CBSA or adjacent rural county) may be more influential on

decisions to improve access, quality, or patient service (Noether, 1987; Rivers & Glover,

2008; Zwanziger, Melnick, & Eyre, 1994).

The second most important matching criterion was size. All but four of the pairs

met the stipulated thresholds for size matching. A “size match” was deemed minimally

acceptable if the Event hospital matched on at least one of the three size-related variables

(licensed beds, admissions, and adjusted admissions). A “match” was defined as the

Event hospital being no less than 50% and no more than 200% the size of the Matched

hospital.

An exception to the “50%/200%” rule was made in four cases where size did not

match on at least one of the three size-related continuous variables (beds, admission, and

adjusted admissions). The first exception involved the match for the Elkhart General

Hospital-Indiana University Hospital Goshen (IUGH) pair. IUGH’s adjusted admissions

were within one-half of one percentage point from the 50% threshold. The second

exception involved the match for Memorial Hospital in York, Pennsylvania: York

Hospital. Although York Hospital is significantly larger, it is only seven miles away from

Memorial Hospital and is the only competing hospital in York, Pennsylvania. Both

hospitals matched for system status and profit status. The third exception involved the

match for Northern Michigan Hospital (NMH) in Petoskey, Michigan: Otsego Memorial

Hospital (OMH) in Gaylord, Michigan. OMH was only nine beds below the 50% cut-off

and both hospitals matched for system and profit status. In addition, OMH was the

nearest hospital to NMH with only 38 miles distance between the two and with both

hospitals serving overlapping geographic markets/regions. The final exception involved

the match for Nix Hospital in San Antonio, Texas: University Hospital in San Antonio.

Nix has 173 licensed beds or 16 short of the number needed to equal 50% of University

Hospital’s total licensed beds (378). University Hospital is only 11 miles away from Nix

and both draw their patients from an overlapping subsection of the San Antonio

metropolitan area.

To further test whether the Matched hospitals represented a reasonable

counterfactual comparison group, bivariate tests were conducted to assess whether there

were significant differences between the Event and Matched groups after matching (i.e.,

were the two groups balanced). Table 1 contains a summary of descriptive statistics

comparing the 99 Event hospitals to the 99 Matched hospitals. Paired t-testing was

performed using Stata 13 on the three size-related matching variables (continuous). There

was no significant difference between the Event and Matched hospitals with respect to

licensed beds (t=1.05, p=0.30) and admissions (t=1.55, p=0.13). There was a significant

difference with respect to adjusted admissions (t=9.44, p=0.001), with Matched hospitals

reporting more adjusted admissions (M=22,953; SD=19,177) than Event hospitals

(M=20,334; SD=14,634).

To test for balance between the 99 pairs of Event and Matched hospitals on the

two categorical variables (dummy variables) indicating profit status (for-profit or not-

forprofit) and system status (hospital was part of a system or a standalone independent

hospital), McNemar’s test were performed. The “mcc” command in Stata 13 was used.

Thirteen pairs had differences in profit status; however, the McNemar’s test found no

statistically significant difference between the two groups (McNemar’s Chi2=0.69,

p=0.41). Sixteen pairs had differences in system status. McNemar’s test found these

differences to be statistically significant (McNemar’s Chi2=4.0, p=0.046**).

Table 1

Summary of Information of the 99 Event Hospitals and 99 Matched Hospitals

Goodness of Fit Statistic

Event Matched

(Balance Test) (if Match?

Hospitals Hospitals

applicable)

Number of

99 99 1:1 ratio YES

hospitals

Event Year

2009 7

2010 14

2011 36

2012 20

2013 22

Merger Type

Merger of Equals

44 44 1:1 ratio YES

Hospitals

Acquired Hospitals 55 55 1:1 ratio YES

Number of Unique

CBSAs & Non- 65 65 YES

CBSA (rural areas)

Number of Unique

50 50 YES

CBSAs

Number of Unique

15 15 YES

Rural Areas

Number of U.S.

33 31

States

Same CBSA or 65 65 1:1 ratio

Rural Match

Mean Distance

Between the 99 Pairs 36.3 36.3

of Hospitals

Mean Distance for

76 Pairs in the Same 22.2 22.2

CBSA

Mean Distance for

the 23 Pairs in Rural 83.2 83.2

Areas

Licensed Beds

Mean t=1.05;

237 259 Paired t-test YES

p=0.30

Standard Deviation 190 236

Minimum 18 25

Maximum 1,037 1,516

Annual Admissions

Mean t=1.55;

10,962 11,962 Paired t-test YES

p=0.13

Standard Deviation 9,663 11,342

Minimum 307 137

Maximum 52,583 93,568

Event

Hospitals

Matched

Hospitals

Goodness of Fit

(Balance Test)

Statistic

(if

applicable)

Match?

Adjusted

Admissions

Mean

20,334

22,953

Paired t-test

t=9.44;

p=0.001**

Standard Deviation

14,634

19,177

Minimum

618

137

Maximum

77,928

150,030

Tax Status

McNemar

Chi2=0.69

YES

For Profit

p=0.405

Not-for-Profit

System Status

McNemar

Chi24.0

Part of a System

p=0.046**

Standalone (not part

of a system)

*100% of the Matching hospitals met the criteria for geographic matching.

HCAHPS Performance of the 99 Event and 99 Matched Hospitals

Appendix F contains 99 sets of graphs: one set for each matched pair of Event and

Matched hospitals. Each set contains four graphs: one for each of the four HCAHPS

ratings (OVERALL, PHYSICIAN Domain, NURSE Domain, and STAFF Domain).

Each of the four individual graphs reflects the year-ending ratings for each of the six

years (T1, T2, T3, T4, T5, and T6).

Final Balance Test: T-testing of the HCAHPS Ratings Prior to the M&A Event

To determine pre-event similarity in HCAHPS ratings (i.e., parallel trends assumption),

paired t-testing was conducted using Stata 13 for each HCAHPS domain for each of the

three pre-event time periods (T1, T2 and T3). The results, summarized in Table 2,

indicate no statistically significant difference in any of the pre-Event HCAHPS ratings

between the two groups of hospitals.

Collectively, based on these results, it can be concluded that up until the event

(the MOE or Acquisition), the 99 pairs of Event and Matched hospitals had similar

attributes in terms of geographic location, size, profit status, and HCAHPS performance.

Table 2

Goodness of Fit Testing of Pre-Event HCAHPS Ratings for T1, T2, and T3 Using Paired

T-Testing

Event

Hospitals

Matched

Hospitals

64.71%

Statistic

Match?

Mean Pre-Event HCAHPS

Ratings for T1

OVERALL

64.10%

t=0.71;

p=0.48

YES

PHYSICIAN Domain

78.92%

78.72%

t=0.40;

p=0.69

YES

NURSE Domain

73.38%

73.91%

t=0.23;

p=0.82

YES

STAFF Domain

59.66%

60.70%

t=1.34;

p=0.18

YES

Mean Pre-Event HCAHPS

Ratings for T2

OVERALL

65.47%

65.97%

t=0.59;

p=0.55

YES

PHYSICIAN Domain

78.89%

78.81%

t=0.15;

p=0.88

YES

NURSE Domain

74.75%

75.09%

t=0.68;

p=0.50

YES

STAFF Domain

60.98%

62.00%

t=1.49;

p=0.14

YES

Mean Pre-Event HCAHPS

Ratings for T3

OVERALL

66.81%

67.20%

t=0.46;

p=0.64

YES

PHYSICIAN Domain

78.89%

78.97%

t=0.18;

p=0.86

YES

NURSE Domain

76.04%

76.01%

t=0.06;

p=0.95

YES

STAFF Domain

62.41%

62.90%

t=0.60;

p=0.55

YES

Paired T-Testing of HCAHPS Means Post-Event

Paired t-tests were conducted to provide additional descriptive statistics of the

HCAHPS performance of the Event and Matched hospitals for each of the three years

(T4, T5 and T6) post-event (Table 3). Since a paired t-test is limited to a “snapshot

comparison” of one year’s HCAHPS ratings, this study did not rely upon the results of

individual year paired t-testing to evaluate any of the 12 hypotheses. This study’s aim was

to determine the effect of a merger or acquisition event on HCAHPS performance

“over time” or from “year to year” (i.e., the temporal effects on the rate of change).

Although useful for descriptive purposes, paired t-testing of individual years could not

fully evaluate study’s hypotheses.

Analytic Strategy for Testing Study Hypotheses

One statistical test, fixed effects difference-in-differences (FEDID), was used to

test the 12 hypotheses and answer the temporal differences questions (i.e., the rate of

changes in HCAHPS ratings over time).

FEDID Tests

Cross-sectional FEDID testing was utilized to answer the question of whether

HCAHPS performance of the two groups (Event hospitals and Matched hospitals)

changes differently over time. FEDID is a multivariate modeling approach that is

designed to minimize the potential impact of omitted variable bias, especially (relatively)

stable yet unobserved factors (e.g., organizational culture, management style) that may

differ between two groups of subjects in this study (Event versus Matched hospitals), and

bias the study results (Qui, Kirkeide, & Wang, 2018). By including fixed effects (hospital

dummies), one can account and control for these differences between subjects. The fixed

effects effectively “soak up all the across-group action” leaving only the “within-group

action” (Blumenstock, 2016, p. 7). Time period dummy variables account for temporal

changes in HCAHPS scores.

The 12-month time period immediately preceding the merger/event year (T3) was

chosen as the referent time period for comparison purposes because that year reflects the

HCAHPS ratings for the last 12-month period the Event hospital operated outside the

influence of a merger or acquisition. In other words, T3 reflects HCAHPS performance

prior to the potentially disruptive effects of a merger or acquisition. In addition, with the

exception of a minor difference in the PHYSICIAN domain ratings (0.03 lower in T3 vs.

T1), the HCAHPS ratings of the Event hospitals were at their highest level in T3

compared to T1 or T2. Selecting T1 or T2 as the pre-event referent year could have biased

the results of FEDID testing since a year with lower values would have served as the

baseline.

The FEDID model is specified below:

HCAHPSit= + 1Eventi + 2T4 + 3T5 + 4T6 + 5Eventi*T4 + 6Eventi*T5 +

7Eventi*T6 + εit

Where “i” indexes hospitals; HCAHPS is the hospital’s patient experience rating; Event is

the indicator for whether a hospital merged or not; and “T” indicates the post-merger,

year-ending time period (T4, T5, and T6).

MOE vs. Acquired Hospitals

Given the study’s interest in testing whether a merger event was more disruptive

among MOE hospitals compared to Acquisition hospitals (i.e., H3a-H3d), an additional

set of FEDID models were estimated that included two new dummy variables that

distinguished between the two types of events: MOE (merger-of-equal) hospitals and

Acquired hospitals. T3 was the referent time period and Acquired hospitals served as the

referent group. This FEDID model is specified as:

HCAHPSit= + 1MOEi + 2Acquisitiont + 3MOEi*T4 + 4Acquisitioni*T4 +

5MOEi*T5 + 6Acquisitioni*T5 + 7MOEi*T6 + 8Acquisitioni*T6 + εit

Where “i” indexes hospitals; HCAHPS is the hospital’s patient experience score; MOE is

the indicator for whether a hospital experienced a Merger-of-Equals; Acquisition is the

indicator for whether a hospital experienced an acquisition; and “T” indicates the

postmerger, year-ending time period (T4, T5, and T6).

Levels of Statistical Significance – P-value Thresholds

Due to the relatively modest sample size, two levels of statistical significance

were considered in the study:

• p < 0.05, which was designated with two asterisks (**)

• p < 0.10, which was designated with one asterisk (*).

CHAPTER 4

RESULTS

Chapter 4 summarizes the results of statistical testing conducted to provide

descriptive information and evaluate the 12 hypotheses. The results are organized and

presented in six sections:

• Section One: Paired T-Testing of Individual Year HCAHPS Mean Ratings

(T1-T6)

• Section Two: FEDID Testing of OVERALL Ratings

• Section Three: FEDID Testing of PHYSICIAN Domain Composite Ratings

• Section Four: FEDID Testing of NURSE Domain Composite Ratings

• Section Five: FEDID Testing of STAFF Composite Domain Ratings, and

• Section Six: FEDID Testing to Evaluate HCAHPS Ratings by Merger Event

Type (44 Merger-of-Equals hospitals and 55 Acquired hospitals).

Section One: Paired T-Testing of Individual Year HCAHPS Mean Ratings (T1-T6)

Tables 2 and 3 provide a descriptive summary of the paired t-tests of the six individual

years and the four HCAHPS ratings. Table 2 contains the pre-event information (T1, T2,

and T3) and Table 3 contains the post-event information (T4, T5, and T6). As mentioned

previously, paired t-testing found no statistical difference between the Event hospitals and

Matched hospitals in any of the three pre-event years (T1, T2 and T3). For all three pre-

event years and for all four HCAHPS ratings, there was approximately 1% or less

difference between the two groups. The lowest p-value in Table

2 (p=0.14) was in T2 for the STAFF domain.

The post-event results summarized in Table 3 reflect six significant “within year”

differences between the two groups. The OVERALL, NURSE domain, and STAFF

domain ratings for T5 and T6 were significantly different between Event hospitals and

Matched hospitals. There were no statistically significant differences in the PHYSICIAN

domain ratings in any of the three post-event years. Likewise, there were no differences

in the OVERALL, NURSE domain, and STAFF domain ratings for T4.

Consistent with the study’s method described in Chapter 3, paired t-testing (for

both the pre-event and post-event years) was used for descriptive purposes only.

Table 3

Post-Event HCAHPS Ratings Comparisons for T4, T5, and T6 Using Paired T-Testing

Event

Hospitals

Matched

Hospitals

Statistic

Mean Post-Event HCAHPS

Ratings for T4

OVERALL

67.76%

68.74%

t=1.22;

p=0.23

PHYSICIAN Domain

79.63%

79.29%

t=0.80;

p=0.42

NURSE Domain

76.59%

77.03%

t=0.97;

p=0.33

STAFF Domain

63.70%

64.42%

t=1.05;

p=0.30

Mean Post-Event HCAHPS

Ratings for T5

OVERALL

68.11%

69.80%

t=2.14;

p=0.03**

PHYSICIAN Domain

79.84%

79.82%

t=0.05;

p=0.96

NURSE Domain

77.12%

78.17%

t=2.32;

p=0.02**

STAFF Domain

63.56%

64.97%

t=2.10;

p=0.04**

Mean Post-Event HCAHPS

Ratings for T6

OVERALL

68.56%

70.03%

t=1.84;

p=0.07*

PHYSICIAN Domain

79.80%

79.89%

t=0.19;

p=0.85

NURSE Domain

77.73%

78.72%

t=2.38;

p=0.02**

STAFF Domain

64.35%

65.79%

t=2.39;

p=0.02**

The following sections – organized by HCAHPS domain - provides more details on these

comparisons.

OVERALL Rating

For T4, the OVERALL mean ratings for both groups increased, with the

percentage of patients rating their patient experience as a “9” or “10” (i.e., average top

box score) increasing from 66.81% in T3 to 67.76% in T4 for the 99 Event hospitals, an

increase of 0.95 percentage points. Likewise, mean ratings for the Matched hospitals

increased from 67.20% in T3 to 68.74% in T4. The paired t-test, however, indicated that

the difference between the two groups in average overall experience ratings was not

statistically significant in the time period immediately following the merger (t(98)=1.22,

p=0.23).

For T5, both groups again experienced an increase in OVERALL patient

experience, with a mean rating of 68.11% for the EHs (an increase of 0.35 percentage

points above T4), and mean score of 69.80% for the MHs (an increase of 1.06 percentage

points above T4). A paired t-test found a significant difference between the two groups of

hospitals (t(98)=2.14, p= 0.035), with the EHs reporting lower ratings (M=68.11%,

SD=0.075) than MHs (M=69.80%, SD=0.064).

Both groups continued the trend of increasing OVERALL patient experience

ratings in T6, with a mean ratingof 68.56% for the EHs (an increase of 0.45 above T5),

and mean rating of 70.03% for the MHs (an increase of 0.23 above T5). A paired t-test

found that the difference in ratings between these two groups of hospitals was statistically

significant in the T6 period (t(98)=1.84, p= 0.069), with EHs reporting lower scores

(M=68.56%, SD=0.070) than MHs (M=70.03%, SD=0.064).

PHYSICIAN Domain Composite Rating

For T4, the mean top box ratings (the percentage of patients giving an “always”

rating) for both groups increased with EH mean ratings increasing from 78.89% in T3 to

79.63% in T4. The MHs experienced an absolute percentage point increase of 0.32, from

78.97% in T3 to 79.29% in T4. A paired t-test found no significant difference in the T4

PHYSICIAN domain composite ratings (t(98)=-0.80, p=0.42), with the EHs reporting

higher scores (M=79.63, SD=0.037) than the MHs (M=79.29, SD=0.036).

For T5, both groups experienced an increase in PHYSICIAN domain ratings with

a mean score of 79.84 for the EHs (an increase of 0.20 above T4), and mean score of

79.82 for the MHs (an increase of 0.53 above T4). The paired t-test, however, indicated

no significant difference (t(98)=-0.05, p= 0.96) with the EHs (M=79.84, SD=0.037) and

MHs (M=79.84, SD=0.037) having nearly identical scores in the T5 period.

For T6, both groups experienced an increase in PHYSICIAN domain ratings with

a mean score of 79.80 for the EHs while the mean rating for the MHs increased by 0.07 to

79.89. A paired t-test found no significant difference in the T6 PHYSICIAN rating

(t(98)=0.19, p= 0.85), with the EHs, again, reporting nearly identical scores (M=79.80,

SD=0.035) as the MHs (M=79.89, SD=0.037).

NURSE Domain Composite Rating

For T4, the mean top box ratings (the percentage of patients giving an “always”

rating) for both groups increased with the EH mean ratings increasing from 76.04 in T3 to

76.59 in T4. The MHs experienced an absolute increase of 1.02 in T4, from 76.01 in T3

to 77.03 in T4. A paired t-test indicated no significant difference in the T4 NURSE ratings

(t(98)=0.19, p= 0.85), with the EHs (M=76.59, SD=0.042) reporting lower scores than the

MHs (M=77.03, SD=0.040).

For T5, both groups experienced an increase in NURSE ratings with a mean score

of 77.12 for the EHs (an increase of 1.08 above T4), and mean score of 78.17 for the

MHs (an increase of 1.14 above T4). The paired t-test found a significant difference in the

T5 NURSE ratings (t(98)=2.32, p= 0.022), with the EHs reporting lower ratings

(M=77.12%, SD=0.041) than the MHs (M=78.12%, SD=0.041).

For T6, both groups experienced an increase in NURSE ratings with a mean rating

of 77.73% for the EHs (an increase of 0.61 above T5), and a mean rating of 78.72% for

the MHs (an increase of 0.55 above T5). A paired t-test found a significant difference in

the T6 NURSE ratings (t(98)=2.38, p= 0.02), with the EHs reporting lower ratings

(M=77.73%, SD=0.038) than MHs (M=78.72%, SD=0.038).

STAFF Domain Composite Rating

For T4, the mean top box ratings (the percentage of patients giving an “always”

rating) for both groups increased with the EH mean ratings increasing from 62.41% in T3

to 63.70% in T4. The MHs experienced an absolute increase of 1.52 percentage points in

T4. However, a paired t-test found no significant difference between the two hospital

groups for the T4 period (t(98)=1.05, p=0.30), with the EHs reporting lower ratings

(M=63.70%, SD=0.069) than MHs (M=64.42%, SD=0.067).

For T5, the EHs’ mean STAFF rating was 63.56% (a decrease of 0.14 compared to

T4). The MHs had a mean rating of 64.97% (an increase of 0.55 above T5). The paired t-

test found a significant difference in the T5 STAFF rating (t(98)=2.10, p=0.04), with the

EHs reporting lower ratings (M=63.56%, SD=0.066) than the MHs (M=64.97%,

SD=0.066).

For T6, both groups experienced an increase in STAFF HCAHPS ratings with a

mean rating of 64.35% for the EHs (an increase of 0.79 above the prior year), and mean

rating of 65.79% for the MHs (an increase of 0.82 above the prior year). A paired t-test

found a significant difference in the T6 STAFF ratings (t(98)=2.39, p=0.02), with the EHs

reporting lower scores (M=64.35%, SD=0.063) than the MHs (M=65.79%,

SD=0.063).

Section Two: FEDID Testing of the OVERALL Rating

This section summarizes the results of FEDID testing performed to evaluate

Hypothesis 1 and 4a. Table 4 contains a summary of the FEDID tests of the OVERALL

rating for the three post-event years (T4, T5, and T6) with T3 serving as the referent year.

Table 4

Post-Event OVERALL Rating

Variable

Mean Rating (%)

Beta (SE)

P-Value

T3 Event Hospitals

66.81

T4 Event Hospitals

67.76

T5 Event Hospitals

68.11

T6 Event Hospitals

68.56

T3 Matched Hospital

67.20

T4 Matched Hospital

68.74

T5 Matched Hospital

69.80

T6 Matched Hospital

70.03

Interaction Terms

Event*T4

-0.006 (0.006)

0.34

Event*T5

-0.013 (0.007)

0.07*

Event*T6

0.011 (0.008)

0.17

Non-Interaction Terms

Event Hospitals

-0.004 (0.010)

0.70

1st Post-Event (T4)

0.015 (0.004)

0.00**

2nd Post-Event (T5)

0.026 (0.004)

0.00**

3rd Post-Event (T6)

0.028 (0.005)

0.00**

Referent Group: Matched hospitals

Referent Year: T3 (12-month period prior to Event)

Controlling for relatively stable, hospital-level characteristics, the increase in

OVERALL ratings from T3 to T4 was smaller for EHs, on average, compared to their

matched counterparts, although the difference was not statistically significant (b= -0.006,

p=0.34). The increase in OVERALL ratings from T3 to T5 for Event hospitals was again

smaller, on average, compared to Matched hospitals, and in this case the difference was

statistically significant (b= -0.013, p=.07). The results for the three-year post-merger time

period (T6) were similar to those for the time period immediately following merger (T4).

Specifically, the increase in Overall HCAHP ratings from T3 to T6 was smaller for Event

hospitals, on average, compared to Matched hospitals, although the difference was, once

again, not statistically significant (b= -0.011, p=0.17).

In sum, the results of the analysis supported Hypothesis 1: a hospital’s overall

patient experience ratings will be negatively affected by having recently completed a

merger (within the last 12-24 months). The overall patient experience ratings of recently

merged hospitals will be lower compared to the overall patient experience ratings of

similar hospitals that did not merge. The negative relationship between a merger and a

hospital’s OVERALL ratings occurred in T5, the second year (months 13-24) following

the merger event.

The results also supported Hypothesis 4a: Event hospitals’ overall patient

experience ratings will return to normal levels within 36 months post-merger with no

performance difference compared to similar hospitals that did not merge. For T6 (three

years post-event), there was no difference in OVERALL ratings between the Event

hospitals and Matched hospitals. The statistically significant, negative relationship found

in T5 did not continue in T6.

Section Three: FEDID Testing of the PHYSICIAN Domain Composite Ratings

This section summarizes the results of FEDID testing performed to evaluate hypotheses

H2a and H4b. Table 5 contains a summary of the PHYSICIAN domain HCAHPS ratings

and FEDID tests for each post-merger year (T4, T5, and T6) with T3 serving as the

referent year.

Table 5

Post-Event PHYSICIAN Domain Composite Ratings

Variable T3

Event Hospitals

Mean Rating (%)

78.89

Beta (SE)

P-Value

T4 Event Hospitals

79.63

T5 Event Hospitals

79.64

T6 Event Hospitals

79.80

T3 Matched Hospital

78.97

T4 Matched Hospital

79.29

T5 Matched Hospital

79.82

T6 Matched Hospital

79.89

Interaction Terms

Event*T4

0.004 (0.004)

0.27

Event*T5

0.001 (0.004)

0.82

Event*T6

0.001 (0.004)

0.80

Non-Interaction Terms

Event Hospitals

-0.001 (0.005)

0.88

1st Post-Event (T4)

0.003 (0.003)

0.22

2nd Post-Event (T5)

0.008 (0.003)

0.001**

3rd Post-Event (T6)

0.009 (0.003)

0.001**

Referent Group: Matched hospitals

Referent Year: T3 (12-month period prior to Event)

The mean PHYSICIAN domain HCAHPS ratings of the EHs increased 0.74 from

T3 (12 months prior to the merger) to T4 (first year post-merger). During this same time

interval, the Matched hospitals’ mean PHYSICIAN domain ratings increased by 0.32.

Although the increase for EHs was greater than for MHs, the difference was not

significant (b= 0.004, p=0.27).

The mean PHYSICIAN domain ratings of the EHs increased 0.95 from T3 (12

months prior to the merger) to T5 (second year post-merger). During this same time

interval, the MHs’ mean PHYSICIAN domain ratings increased by 0.85. The change in

ratings between T5 and T3, however, was not significantly different for EHs and MHs

(b= 0.001, p=0.816).

The mean PHYSICIAN domain ratings of the EHs increased 0.91 from T3 (12

months prior to the merger) to T6 (third year post-merger). During this same time

interval, the Matched hospitals’ mean PHYSICIAN domain ratings increased by 0.91.

Once again, the change between T6 and T3 did not differ significantly for EHs and MHs

(b= 0.001, p=0.802).

In sum, the analysis did not support Hypothesis 2a: a hospital’s physician

communications ratings will be negatively affected by having recently completed a

merger (within the last 12-24 months). Patient ratings of physician communications for

recently merged hospitals will be lower compared to the ratings of physician

communications for similar hospitals that did not merge. There was no post-merger effect

on PHYSICIAN domain ratings.

There was support for Hypothesis 4b: there will be no difference in physician

communications ratings between merged and similar unmerged hospitals three years

following the merger. However, the finding of “no difference” in T6 was not due to a

rebound in the PHYSICIAN Domain ratings for the EHs from T4 and T5. PHYSICIAN

Domain ratings between the EHs and MHs remained statistically similar for all six years

tested using both paired t-tests and FEDID testing.

Section Four: FEDID Testing of the NURSE Domain Composite Ratings

This section summarizes the results of the statistical testing performed to evaluate

hypotheses H2b and H4c. Table 6 contains a summary of the NURSE domain HCAHPS

ratings and FEDID results each of the three post-merger years (T4, T5, and T6).

Table 6

Post-Event NURSE Domain Composite Ratings

Variable

Mean Rating (%)

Beta (SE)

P-Value

T3 Event Hospitals

76.04

T4 Event Hospitals

76.59

T5 Event Hospitals

77.12

T6 Event Hospitals

77.73

T3 Matched Hospital

76.01

T4 Matched Hospital

77.03

T5 Matched Hospital

78.17

T6 Matched Hospital

78.82

Interaction Terms

Event*T4

-0.005 (0.004)

0.24

Event*T5

-0.108 (0.005)

0.02**

Event*T6

-0.010 (0.005)

0.055*

Non-Interaction Terms

Event Hospitals

0.003 (0.006)

0.96

1st Post-Event (T4)

0.010 (0.003)

0.00**

2nd Post-Event (T5)

0.023 (0.003)

0.00**

3rd Post-Event (T6)

0.027 (0.003)

0.00**

Referent Group: Matched hospitals

Referent Year: T3 (12-month period prior to Event)

The mean, NURSE domain top box ratings of the EHs increased 0.55 from T3 (12

months prior to the merger) to T4 (first year post-merger). During this same time interval,

the Matched hospitals’ mean NURSE domain top box ratings increased by 1.02.

Although the increase for EHs was smaller than for Matched hospitals, after controlling

for other factors, the change in NURSE domain ratings between T3 and T4 for EHs was

not significantly different than MHs (b=-0.005, p=0.24).

The mean, NURSE domain ratings of the EHs increased 1.08 percentage points

from T3 (12 months prior to the merger) to T5 (second year post-merger). During this

same time interval, the Matched hospitals’ mean NURSE domain ratings increased by

2.16. The change in NURSE domain scores between T5 and T3 was significantly smaller

for EHs compared to MHs (b=-0.108, p=0.02).

The mean NURSE domain ratings of the EHs increased 1.69 percentage points

from T3 (12 months prior to the merger) to T6 (third year post-merger). During this same

time interval, the Matched hospitals’ mean NURSE domain ratings increased by 2.71

percentage points. Similar to T5, the increase in NURSE domain ratings fromT6 to T3

was significantly smaller for EHs compared to MHs (b=-0.010, p=0.06).

In sum, the results of the analysis supported Hypothesis 2b: a hospital’s nurse

communications ratings will be negatively affected by having recently completed a

merger (within the last 12-24 months). Patient ratings of nurse communications for

recently merged hospitals will be lower compared to the ratings of nurse communications

for similar hospitals that did not merge. The NURSE domain ratings of the EHs in T5

(second year following the merger) were significantly lower compared to the MHs, which

was consistent with H2b.

However, the results did not support Hypothesis 4c: there will be no difference in

nurse communications ratings between merged and similar unmerged hospitals three

years following the merger. FEDID tests found a statistically significant difference in the

Nurse Domain ratings between EHs and MHs for T6 (three years post-event), indicating

the NURSE domain performance did not rebound in T6.

Section Five: FEDID Testing of the STAFF Domain Composite Ratings

This section provides the results of the statistical testing performed to evaluate

hypotheses H2c and H4d. Table 7 contains a summary of the results from the FEDID

testing on the STAFF domain rating each post-event year (T4, T5, and T6).

Table 7

Post-Event STAFF Domain Composite Ratings

Variable

Mean Rating (%)

Beta (SE)

P-Value

T3 Event Hospitals

62.41

T4 Event Hospitals

63.70

T5 Event Hospitals

63.56

T6 Event Hospitals

64.35

T3 Matched Hospital

62.90

T4 Matched Hospital

64.42

T5 Matched Hospital

64.97

T6 Matched Hospital

65.79

Interaction Terms

Event*T4

-0.002 (0.006)

0.69

Event*T5

-0.009 (0.007)

0.21

Event*T6

-0.009 (0.008)

0.22

Non-Interaction Terms

Event Hospitals

0.005 (0.010)

0.61

1st Post-Event (T4)

0.015 (0.004)

0.00**

2nd Post-Event (T5)

0.021 (0.004)

0.00**

3rd Post-Event (T6)

0.029 (0.004)

0.00**

Referent Group: Matched hospitals

Referent Year: T3 (12-month period prior to Event)

The mean STAFF domain ratings of the EHs increased 1.29 percentage points

from T3 (12 months prior to the merger) to T4 (first year post-merger). During this same

time interval, the Matched hospitals’ mean STAFF domain ratings increased by 1.52

percentage points. Although the increase for EHs was smaller compared to Matched

hospitals, the difference was not statistically significant (b= -0.002, p=0.69).

The mean STAFF domain ratings of the EHs increased 1.15 percentage points

from T3 (12 months prior to the merger) to T5 (second year post-merger). During this

same time interval, the MHs’ mean STAFF domain ratings increased by 2.07 percentage

points. The change in STAFF ratings from T5 to T3 was not significantly different

between EHs and MHs (b= -0.009, p=0.21).

The mean STAFF domain ratings of the EHs increased 1.94 percentage points

from T3 (12 months prior to the merger) to T6 (third year post-merger). During this same

time interval, the MHs’ mean STAFF domain ratings increased by 2.89 percentage points.

Once again, the difference between EHs and MHs was not statistically significant (b= -

0.0009, p=0.22).

In sum, the results of the analysis did not support Hypothesis 2c: a hospital’s staff

responsiveness ratings will be negatively affected by having recently completed a merger

(within the last 12-24 months). Patient ratings of staff responsiveness for recently merged

hospitals will be lower compared to the ratings of nurse communications for similar

hospitals that did not merge.

The results did support Hypothesis 4d: There will be no difference in staff

responsiveness ratings between merged and similar unmerged hospitals three years

following the merger. However, the finding of “no difference” in T6 was not the result of

performance improvement following a period of post-merger performance decline.

Rather, the analysis found no difference between EHs and MHs in STAFF domain ratings

for any of the three post-merger years.

Section Six: FEDID Testing of the HCAHPS Ratings of the 44 MOE Hospitals

Compared to the 55 Acquired Hospitals

This final section of Chapter 4 summarizes the results of the analysis used to

evaluate hypotheses H3a (OVERALL rating), H3b (PHYSICIAN Domain), H3c

(NURSE Domain), and H3d (STAFF Domain). These four hypotheses were based upon

supporting theories and empirical research that suggests greater post-merger disruption

and performance decline among organizations coming together via an MOE versus an

Acquisition (Devine, Lamont, & Harris, 2016; Lodorfos & Boateng, 2006). Applied to

this study on hospital mergers, these four hypotheses posited a negative relationship

between MOE hospitals and HCAHPS ratings compared to Acquired hospitals and

HCAHPS ratings.

Table 8 provides a summary of the four categories of HCAHPS ratings, by year,

for each of the two subgroups of Event hospitals (55 MOE hospitals and 44 Acquired

hospitals). Tables 9 through 12 summarize the FEDID testing, specifically the beta

coefficients, for each of the three post-event years and for each of the four categories of

(

)

(

)

79.00

73.80

58.55

HCAHPS ratings (OVERALL, PHYSICIAN Domain, NURSE Domain, and STAFF

Domain).

Table 8

HCAHPS Ratings by Merger Type

OVERALL PHYSICIAN NURSE Domain STAFF Domain

Domain

Acquire MOE Acquired MOE Acquire MOE Acquired MOE d (%) (%) (%) d (%) (%)

T1 63.29 65.11 78.85 73.76 60.55

64.13

67.16

79.11

78.75

79.11

78.75

60.95

61.02

65.78

68.09

79.29

78.39

79.29

78.39

63.38

61.20

66.58

69.23

79.51

79.77

79.51

79.77

64.00

63.32

66.36

70.30

79.71

80.00

79.71

80.00

63.38

63.77

66.78

70.77

79.64

80.00

79.64

80.00

64.55

64.11

Table 9

FEDID Testing: Post-Event OVERALL Rating by Merger Type

Variable

Beta (SE)

P-Value

Interaction Terms

MOE*T4

0.003 (0.009)

0.72

MOE*T5

0.016 (0.011)

0.14

MOE*T6

0.017 (0.012)

0.15

Non-Interaction Terms

MOE Hospitals

0.023 (0.015)

0.12

1st Post-Event (T4)

0.008 (0.007)

0.27

2nd Post-Event (T5)

0.005 (0.008)

0.48

3rd Post-Event (T6)

0.010 (0.009)

0.28

Referent Group: Acquired hospitals

Referent Year: T3 (12-month period prior to Event)

Table 10

FEDID Testing: Post-Event PHYSICIAN Domain Composite Ratings by Merger Type

Variable

Beta (SE)

P-Value

Interaction Terms

MOE*T4

0.012 (0.005)

0.03**

MOE*T5

0.012 (0.007)

0.09*

MOE*T6

0.011 (0.007)

0.13

Non-Interaction Terms

MOE Hospitals

-0.009 (0.008)

0.26

1st Post-Event (T4)

0.002 (0.003)

0.52

2nd Post-Event (T5)

0.004 (0.005)

0.38

3rd Post-Event (T6)

0.005 (0.005)

0.24

Referent Group: Acquired hospitals

Referent Year: T3 (12-month period prior to Event)

Table 11

FEDID Testing: Post-Event NURSE Domain Composite Ratings by Merger Type

Variable

Beta (SE)

P-Value

Interaction Terms

MOE*T4

0.012 (0.006)

0.03**

MOE*T5

0.016 (0.007)

0.03**

MOE*T6

0.018 (0.008)

0.03**

Non-Interaction Terms

MOE Hospitals

-0.005 (0.008)

0.51

1st Post-Event (T4)

-1.35 (0.004)

1.00

2nd Post-Event (T5)

0.003 (0.006)

0.52

3rd Post-Event (T6)

0.009 (0.007)

0.16

Referent Group: Acquired hospitals

Referent Year: T3 (12-month period prior to Event)

Table 12

FEDID Testing: Post-Event STAFF Domain Composite Ratings by Merger Type

Variable

Beta (SE)

P-Value

Interaction Terms

MOE*T4

0.015 (0.009)

0.098*

MOE*T5

0.016 (0.007)

0.03**

MOE*T6

0.018 (0.008)

0.03**

Non-Interaction Terms

MOE Hospitals

-0.005 (0.008)

0.51

1st Post-Event (T4)

-1.35 (0.004)

1.00

2nd Post-Event (T5)

0.003 (0.006)

0.52

3rd Post-Event (T6)

0.009 (0.007)

0.16

Referent Group: Acquired hospitals

Referent Year: T3 (12-month period prior to Event)

FEDID Testing of the OVERALL Ratings for the Three Post-Merger Years (T4, T5 &

T6)

The mean OVERALL rating of the 44 MOE hospitals increased 1.14 percentage

points from T3 (12 months prior to the merger) to T4 (first year post-merger). During this

same time interval, the mean OVERALL rating of Acquired hospitals increased by 0.80

percentage points. MOE hospitals increased by 2.20 percentage points in T5 compared to

T3, and by 2.68 percentage points in T6 compared to T3. In comparison, the Acquired

hospitals’ OVERALL ratings increased by 0.58 percentage points in T5 compared to T3,

and by 1.00 percentage points in T6 compared to T3.

Although the comparative increases in the OVERALL ratings for MOE hospitals

were larger than for Acquired hospitals, FEDID testing found that the changes for MOE

hospitals were not significantly different than the Acquired hospitals for T4, T5 and T6

compared to T3 (respectively, b= 0.003, p=0.719; b=0.016, p=0.14; b=0.017, p=0.15).

In sum, although the MOE hospitals’ year-over-year increases were greater than

the corresponding increases among the Acquired hospitals, the differences were not

significantly different. Thus, there was no support for H3a.

FEDID Testing of the PHYSICIAN Domain Composite Ratings for the Three

PostMerger Years (T4, T5 & T6)

The mean PHYSICIAN domain rating of the 44 MOE hospitals increased 1.39

percentage points from T3 (12 months prior to the merger) to T4 (first year post-merger).

During this same time interval, the mean PHYSICIAN rating of Acquired hospitals

increased by 0.22 percentage points. MOE hospitals increased by 1.61 percentage points

in T5 compared to T3, and by 1.61 in T6 compared to T3. The Acquired hospitals, on

average, increased by 0.42 percentage points in T5 compared to T3, and by 0.35 in T6

compared to T3.

FEDID testing found the positive changes in PHYSICIAN ratings between T3 and

T4 and T5 to be significantly greater for MOE hospitals compared to Acquired hospitals

(respectively, b= 0.012, p=0.03; b=0.012, p=0.09). The positive changes in PHYSICIAN

ratings between T3 and T6, however, were not significantly different between MOE

hospitals and Acquired hospitals (b=0.011, p=0.13).

Collectively, this pattern of results is the opposite relationship than what H3b

posited, thus, the findings support the opposite relationship with MOE hospitals

performing better than Acquired hospitals; therefore, H3b was not supported.

FEDID Testing of the NURSE Domain Composite Ratings for the Three PostMerger

Years (T4, T5 & T6)

The mean NURSE ratings of the 44 MOE hospitals increased 1.23 percentage

points from T3 (12 months prior to the merger) to T4 (first year post-merger). During this

same time interval, the mean NURSE ratings of Acquired hospitals stayed the same (no

increase or decrease). MOE hospitals increased by 1.98 percentage points in T5 compared

to T3, and by 2.66 percentage points in T6 compared to T3. The Acquired hospitals

increased by 0.36 percentage points in T5 compared to T3, and by 0.91 in T6 compared to

T3. FEDID testing found the changes in NURSE rating for all three postmerger years

were significantly different, on average, between MOE hospitals and Acquired hospitals

(T4, T5, and T6, respectively: b= 0.012, p=0.03; b=0.02, p=0.03; b=0.018, p=0.03).

In sum, instead of Acquired hospitals performing comparatively better

postmerger, MOE hospitals’ performance was significantly better than the Acquired

hospitals. Thus, the findings did not support H3c.

FEDID Testing of the STAFF Domain Composite Ratings for the Three PostMerger

Years (T4, T5 & T6)

The mean STAFF ratings of the 44 MOE hospitals increased 2.11 percentage

points from T3 (12 months prior to the merger) to T4 (first year post-merger). During this

same time interval, the mean STAFF ratings of Acquired hospitals increased by 0.62

percentage points. MOE hospitals increased by 2.57 percentage points in T5 compared to

T3, and by 2.91 percentage points in T6 compared to T3. The Acquired hospitals’ mean

ratings declined by 0.62 percentage points in T5 compared to T4. Acquired hospitals’

STAFF ratings rebounded in T6 with a comparative increase of 1.16 percentage points

over T3.

FEDID testing found the changes in STAFF ratings were significantly different

between MOE hospitals and Acquired hospitals for T4 and T5 (first two years following

the merger or acquisition event, respectively, b= 0.015, p=0.098; b=0.026, p=0.03).

In sum, the findings did not support H3d.

Summary of Results for the Statistical Testing of All 12 Hypotheses

Table 13 provides a summary of testing results. The results supported five of the

12 hypotheses. Seven of the 12 hypotheses were not supported with three of those seven

involving contrary findings: a significant finding but in the opposite direction of what

was posited. Notably, two of the hypotheses (H4b and H4d) were supported by the

findings but not because of any improvement in HCAHPS ratings in the third year

postmerger, but rather because there was no initial impact on ratings after the event.

Table 13

Summary of FEDID Testing Results for the 12 Hypotheses

Hypothesis

Domain

Supported

OVERALL

Yes

H2a

PHYSICIAN

H2b

NURSE

Yes

H2c

STAFF

H3a

OVERALL

H3b

PHYSICIAN

No*

H3c

NURSE

No*

H3d

STAFF

No*

H4a

OVERALL

Yes

H4b

PHYSICIAN

Yes**

H4c

NURSE

H4d

STAFF

Yes**

*The opposite relationship was found with MOE hospitals performing significantly better

than Acquired hospitals.

**H4b and H4d were supported but not because of an underlying improvement in

HCAHPS performance in the third year post-merger.

The average rate of change for the OVERALL and NURSE domain ratings

among the Event hospitals compared to the Matched hospitals was significantly lower in

the first 12-24 months post-event. That negative difference persisted in the third year

post-merger for the NURSE domain ratings. However, the rate of change in the

OVERALL ratings among the EHs and MHs in T6 were no longer significantly different.

The EHs did not experience a comparative decline in rate of change in the PHYSICIAN

and STAFF domain ratings during the first 12-24 months following the merger or

acquisition event.

In terms of the hypothesized difference between MOE hospitals and Acquired

hospitals, the opposite relationship was found for three of the four HCAHPS domains

(PHYSICIAN, NURSE, and STAFF). In these three domains, the MOE hospitals

performed significantly better post-merger compared to the Acquired hospitals. There

was no difference between the MOE hospitals and Acquired hospitals in the post-merger

OVERALL ratings.

CHAPTER 5

DISCUSSION

This chapter begins with a restatement of the purpose and background of this

study. Next, there is a general discussion of the major findings, organized around the four

research questions. The third section describes the strengths, weaknesses, and limitations

of this study. The chapter concludes with a review of the implications of this study for

future research, health services managers, policy-makers, and regulators.

Restatement of Study’s Purpose and Theoretical Framework

The U.S. hospital industry has experienced several consolidation waves during the

past 25 years (Maslan & Johnson, 2016). Consequently, most hospital markets in the

United States are considered highly concentrated (National Academy of Social Insurance,

2015). Many economists and health services researchers have examined the impact of

hospital mergers from variety of perspectives, including quality, safety, access, internal

operating costs, pricing, operating margin, and more. No study, however, has evaluated

the effect of hospital mergers on patient experience. This study attempted to fill that gap

by answering the following questions:

1. Prior to merger, do hospitals pursuing a merger have different HCAHPS

ratings compared to similar hospitals that are not pursuing a merger?

2. How does a merger affect HCAHPS performance?

3. Does time play a role in this effect? Does any negative, “merger-related”

effect on HCAHPS ratings dissipate over time? In other words, do HCAHPS

ratings rebound resulting in no difference compared to similar hospitals that

did not complete a merger?

4. Does the type of transaction (merger-of-equals vs. acquisition) moderate this

relationship?

The extant literature on hospital mergers provides strong evidence that

patients/consumers do not receive a quality or safety benefit following a hospital merger

(Bazzoli et al., 2004; Gaynor & Town, 2012; Romano & Balan, 2010; Universal Health

Care Foundation of Connecticut, 2014). Due to increased market power, hospital mergers

oftentimes lead to higher prices for services and higher health insurance premiums

(Gaynor & Town, 2012; Moriya, Vogt, & Gaynor, 2010). The overwhelming body of

research supports the conclusion that hospital mergers are not associated with improved

or enhanced consumer welfare (Bazzoli et al., 2004; Gaynor, 2011; Gaynor & Town,

2012; Town et al., 2006).

Structural Inertia Theory (SIT) and Culture Conflict Theory (CCT) were utilized

to relate the internal, socially disruptive, and distractive impact of a merger on a

hospital’s performance (Figures 4 & 5). Prior research on mergers in other, nonhealthcare

service industries found customer satisfaction scores decline following a merger (Miles &

Rouse, 2011; Swaminathan, Groening, Mittal, & Thomaz, 2013). Research on the

determinants of patient experience found the most influential factors affecting a patient’s

satisfaction with an inpatient episode involved the interaction with physicians, nurses,

and other staff (Boshoff & Gray, 2004; Fowdar, 2005; Tucker & Adams, 2001; Ware et

al., 1978). By integrating SIT and CCT, along with the aforementioned empirical

research, 12 hypotheses were developed that, in general, predicted a negative relationship

between a hospital merger and HCAHPS performance in the 12 to 24 month period

following the merger event. Consistent with the two theories and prior research, merger

disruption and its adverse impact on a hospital’s performance would be temporary (12-24

months) with an expectation of no difference in HCAHPS performance between hospitals

that merged and similar hospitals that did not merge three years after the merger event.

The study also posited that merger type would affect HCAHPS performance and

moderate the merger-related effect, with an expectation that Acquired hospitals would

perform better than MOE hospitals. These hypotheses were based on research on

postmerger acculturation suggesting that imposed acculturation through absorption,

which is the predominant approach used in an acquisition, is more effective than

symbiotic or transformative acculturation methods, which are associated with MOE

situations (Carroll & Harrison, 2004, Ellis & Lamont, 2004; Larsson & Finkelstein,

1999). Additional research on MOEs found the blending of two, previously rival,

organizations operating in the same industry and in the same geographic region triggers

intense post-merger control clashes, power struggles, and political infighting (Devine,

Lamont, & Harris, 2016).

Discussion of the Major Findings – Responses to the Four Research Questions

Question 1: Prior to merger, do hospitals pursuing a merger have different

HCAHPS ratings compared to similar hospitals that are not pursuing a merger?

The HCAHPS ratings of the 99 Event hospitals and the 99 Matched hospitals

were similar and not significantly different for any of the three years leading up to

the merger event (T1, T2 and T3). Consistent with national trends in HCAHPS

ratings, both groups of hospitals increased their mean HCAHPS ratings year-over-

year and at a similar rate during the three-year period leading up to the merger

event. This was expected since the 99 hospital pairs in this study operated in the

same or nearby geographic areas. In most cases, the hospitals in the matched pairs

were directly competing for market-share and patient loyalty from within the same

CBSA or adjacent market. This finding also suggests any strategic, operational, or

financial factors motivating a hospital to pursue a merger did not affect that

hospital’s HCAHPS performance for the three-year time period prior to merger. This

finding also validates the “goodness-of-match” between the 99 Event hospitals and

the 99 Matched hospitals.

Question 2: How does a merger affect HCAHPS performance?

With the exception of the PHYSICIAN Domain and STAFF Domain

composite ratings, the Event hospitals experienced a significantly smaller increase

in their HCAHPS performance in the 12 to 24 month period following the merger

event, when compared to the Matched hospitals’ HCAHPS performance for the

same time period. The year-over-year improvement of HCAHPS ratings for the

OVERALL question and NURSE Domain questions was negatively associated with

Event hospitals.

This finding is supported by this study’s theoretical framework and accompanying

empirical evidence. As depicted in Figure 5, a merger is a radical change within an

organization (Peus, Frey, Gerkhardt, Fischer, & TrautMattausch, 2008). In general,

mergers trigger operational distraction, culture conflict, internal disruption, increased

stress, social upheaval, decreased morale, and anxiety (Marks & Mirvis, 1992). Merger

disruption oftentimes adversely affects organizational performance (Altendorf, 1986;

Bogen & Symmers, 2001; Meeks, 1977; Sinetar, 1981). Job satisfaction levels of hospital

nurses may decline post-merger (Armstrong-Stassen et al., 2001).

Numerous research studies found human interaction (i.e., patient interaction

with nurses, physicians and staff) is the single most influential and predictive

variable on a hospital’s patient experience ratings (Boshoff & Gray, 2004; Fowdar,

2005: Tucker & Adams, 2001; Ware et al., 1978). Furthermore, a large-scale study

involving 98,000 nurses found that nurses’ perceptions of the hospital working

environment had significant effects on HCAHPS performance.

This study’s findings of a merger’s negative effect on the NURSE Domain

composite ratings may be explained by the unique impact that merger disruption

has on the nursing staff and/or the unique response nurses may have to the

organizational disruption. Unlike physicians working within a hospital setting—

who are predominantly independent and/or employed by a separate organization

not under the hospital’s direct control—all hospital-based nurses are

employed/contracted by the hospital with direct supervision and control from

hospital management. They are, therefore, not independent. A hospital-based

nurse’s working environment is determined by the hospital or health system. Since

2010, hospitals in the United States have experienced an increasing

shortage of nurses (Snavely, 2016). Nurses in a hospital going through the social

upheaval and aftermath of a merger may be more willing to express concerns

about the merger knowing they have more job options. Based upon the research

mentioned earlier, how patients perceive their interaction with the nursing staff

significantly affects patients’ perception of their hospital experience (Boshoff &

Gray, 2004; Fowdar, 2005; Tucker & Adams, 2001; Ware et al., 1978). This

study’s findings suggest that nurses experience and respond to merger disruption

differently compared to physicians and non-nursing staff. Consequently, the Event

hospitals’ HCAHPS performance within the NURSE Domain and the OVERALL

rating were adversely affected.

The finding of no difference within the PHYSICIAN Domain composite

ratings may indicate physician behavior and performance was not affected by the

merger. There are varying categories of physicians working within the hospital

setting including independent (private physicians who are solo practitioners or part

of a traditional physician group practice), contracted hospitalists (physicians who

are 100% hospital-based and working for a separate company under contract with

the hospital to provide inpatient care), and employed hospitalists (physicians who

are W-2 employees of the hospital and assigned to provide inpatient care).

This study’s profile of the Event and Matched hospitals did not consider the status of

physicians working within the hospital setting. Therefore, the PHYSICIAN

Domain results cannot be tested by the prevailing physician category

(independent, contracted, and employed) in each hospital.

However, some conjectures can be formed. The PHYSICIAN Domain results

suggest: (1) physicians as a group did not experience a significant level of merger-

related disruption on their work environment, (2) to the extent many physicians

working in a hospital are not employed by or directed by hospital administration,

they are less likely to incur or be the subject of merger disruption, and/or (3)

physicians affected by merger disruption were resilient, thus they were better able to

manage and respond compared to nursing staff.

The finding of no statistically significant difference within the STAFF

Domain is more challenging to explain. All things being equal, one would expect

the post-merger experience of non-nursing staff to be similar to nursing staff.

Although not statistically significant, the STAFF Domain FEDID beta coefficients

for all three post-merger years were negative (T4= -0.002, T5= 0.009, and T6= -

0.009), reflecting a consistently negative gap. The difference between the NURSE

Domain and STAFF Domain findings could be due to the differences in

professional status between the two employee groups. Research conducted by Zell

(2003), Mintzberg (1983), and Merton (1957) found professional bureaucracies to

be especially resistant to organizational change. Nurses are members of a

profession and professional bureaucracy with a certain level of autonomy in their

daily work. To the extent a merger results in organizational disruption and a

specific strain on the merged hospital’s nursing staff, nurses individually and

collectively may perceive a loss of autonomy and, therefore, respond negatively

by resisting the imposed change. Consequently, their job satisfaction levels may

decrease, which, in turn, may affect their job performance and interaction with

patients. Non-nursing staff, many of whom do not hold a professional license,

may not experience the same level of negative response to the merger due to their

lower status and, potentially, greater likelihood of simply accepting organizational

changes. Non-nursing staff may also not have as many alternative job

opportunities as their nurse colleagues; therefore, nonnursing staff may be

reluctant to express concerns or allow their concerns to affect their interactions

with patients. In general, non-professional employees may be more accepting of

merger-related changes and may, therefore, respond in a way that does not affect

patient perception of staff responsiveness.

The significant, negative difference in HCAHPS ratings between Event and

Matched hospitals appeared in T5 or the second year post-merger. FEDID testing

found no difference in HCAHPS performance for T4 or the first year postevent.

This suggests merger-related disruption and social upheaval may not occur

immediately following the merger or acquisition. This may be explained by an

initial “honeymoon effect” as the two organizations celebrate, both internally and

publicly, the merger or acquisition event. After the honeymoon period ends, the

difficult work of blending the two organizations together starts. Operational

changes occur and the reality of the merger and its disruptive effects are felt by

frontline staff.

Question 3: Does time play a role in this effect? In other words, as more times

passes post-merger, do any initial post-merger differences in HCAHPS performance

dissipate over time?

Results of the study suggest dissipation does occur for three of the four

HCAHPS performance metrics. Except for NURSE Domain ratings, the HCAHPS

performance difference between the two groups of hospitals found in the second

post-merger year dissipated by the third post-merger year. Notably, however, there

was no initial (12 to 24 month) post-merger difference in the PHYSICIAN and

STAFF Domain performance between the Event hospitals and Matched hospitals.

However, NURSE Domain ratings did not improve with a finding of

continued significant difference between the Event hospitals and Matched

hospitals in the third year post-merger. This may indicate merger disruption effects

are stronger and/or persist for a longer period among the nursing staff compared to

non-clinical employees. As discussed previously in the answer to Question 2, these

findings suggest the negative impact on nurses and nurses’ experience/response to

the merger disruption effects are worse compared to physicians and the non-

nursing staff. There are several possible explanations.

First, there may some correlation between the level and significance of the

merger disruption effect and the time it takes for that disruptive effect to be resolved

(i.e., the greater the disruption the greater amount of time to dissipate). The second

reason could be tied to how nursing staff experience significant organizational

change nurse and their resilience in dealing with that change. Research on nursing

work environments found a significant, negative relationship between workplace,

environmental stress and nursing performance (i.e., increased stress in the working

environment resulted in decreased nursing performance) (Clegg, 2001). Mergers

increase stress and anxiety levels. Nurses’ response to merger-related disruption and

accompanying stress may be more intense and last longer. One possible explanation

is that nurses, as a group, may not have the same level of resilience and coping

abilities compared to physicians. Note: This study’s post-merger time horizon was

limited to three years, therefore, whether the merger related negative effect on

nurses persisted beyond year three is unknown. Question 4: Does the type of

transaction (Merger-of-Equals vs. Acquired) moderate this relationship?

The findings suggest the effects of a merger on HCAHPS scores does vary

based upon the type of transaction, however, but not in the direction hypothesized.

Except for OVERALL ratings, the study found MOE hospitals performed

significantly better than Acquired hospitals following a merger event.

Why did MOE hospitals perform better? One potential explanation is tied

to the employees’ perception of loss. In an acquisition, employees may interpret

the change as “we have been taken over.” Whereas in an MOE situation, the

perception and prevailing narrative may be more about growth, expansion, and a

positive/promising change. Being acquired involves relinquishing control to the

acquiring entity and a shift in culture. Another factor is how culture is affected and

how acculturation is achieved. In an MOE, there may be an intentional effort to

blend the cultures of the two organizations or establish a new culture with a new

organizational identity. Conversely, in an acquisition, employees of the target

hospital (the hospital being acquired) may perceive the culture change as imposed

or forced. They must give up their culture and organizational identity. This could

lead to more resistance and resentment among the employees.

Weber’s (1996) research sheds light on this phenomenon. In situations

involving an acquisition or takeover (one large firm acquires and completely

absorbs a smaller firm), employees and managers working in the smaller, target

firm may feel unimportant, marginalized, and trivialized. Their feelings of

alienation and discontent may lead them to openly complain to each other and to

customers about the new environment and how things are not the same after being

acquired. In some situations, culture conflict and resentment linger for several

years.

Research on post-merger acculturation methods may offer some insight

into the finding that MOE hospitals performed better post-merger than Acquired

hospitals. The conceptual framework developed by Nahavandi and Malekzadeh

(1988) identified four approaches for the successful blending of two organizations.

Two of those four are commonly used in MOE situations:

symbiotic (intentional efforts to combine the best cultural elements of the two

organizations) and transformation (establishing a completely new culture). This study’s

findings may indicate senior leaders and boards involved with the 44

MOE hospitals made acculturation an important element of pre-merger planning. This

study did not evaluate or classify post-merger acculturation. However, the findings

suggest the 44 MOE hospitals may have done a better job developing and

implementing a post-merger culture-blending plan as compared to the 55

Acquired hospitals.

Another interesting takeaway from the MOE vs. Acquired comparison is the

difference in the PHYSICIAN Domain, with the MOE hospitals having

significantly better post-merger PHYSICIAN Domain ratings compared to the

Acquired hospitals. Note that no difference was found in the PHYSICIAN

Domain ratings in the aggregate groups: the 99 Event hospitals compared to the

99 Matched hospitals. Post-merger, the PHYSICIAN Domain ratings of the

Acquired hospitals was unchanged, meaning no improvement (Table 8). However,

the PHYSICIAN Domain ratings of the MOE hospitals increased significantly in

the first year post-merger.

There are several possible explanations. MOE situations may be more

likely to involve physicians in both the merger planning and post-merger

execution. Since an MOE is an agreement among two hospitals or health systems

to combine their assets into a new entity, leaders of the two organizations may

take more deliberate steps to engage physicians earlier in the process and, thereby,

achieve a higher level of physician buy-in. The opposite may occur in an

acquisition situation. Leadership of the target hospital may be less inclined to open

the process and negotiations to the medical staff. Physicians in an acquisition

situation may not know about the deal until late in the process or until the deal is

finalized. This low engagement and lack of transparency, both before and after the

transaction is concluded, could trigger a significant negative reaction among the

medical staff.

Another factor involves loss of control. Prominent physicians in the target hospital

may have enjoyed substantial control and influence in the hospital operations prior to the

acquisition. Post-acquisition, the new owner may change leadership at the acquired

hospital and centralize power and decision-making, thus significantly reducing the local

physicians’ influence and control. This loss of control, coupled with the low engagement

and lack of transparency, could cause overt opposition and animosity between the

medical staff and the acquiring entity. The combination of resentment, anger, and loss

would likely, adversely affect the hospital environment, physician commitment and

performance, and how patients perceive their interactions with physicians.

Strengths and Weaknesses/Limitations

This study has several strengths. First, the sample size of 198 hospitals (99 Event

hospitals and 99 Matched hospitals) enhanced the power of the study. The majority of the

published, empirical research on hospital mergers conducted in the past 30 years involved

a significantly smaller sample size of hospitals. Next, the study included hospitals from

diverse geographic areas, including 50 CBSAs, 15 rural areas, and 33 states. Third, the

study included mergers that occurred during a five-year time period (2009 through 2013),

which was one of the more active hospital merger waves in the past 30 years. Fourth, the

study had access to a standardized (and consistently used) tool for measuring and

reporting patient experience (i.e., the HCAHPS assessment). Fifth, the 99 Event hospitals

and the 99 Matched hospitals had equivalent HCAHPS ratings for the three-year time

period leading up to the merger event, which enhances the ability to correlate differences

in the study’s outcome variable (post-event HCAHPS performance) to the predictor

variable (merger status of the 198 hospitals in the study). Last, with a few exceptions, the

Event hospitals and the Matched hospitals were well-matched geographically and by size

with an acceptable level of equivalence (Table 1).

The study also has some notable weaknesses and limitations. First, although the

99 pairwise matches were, in aggregate, equivalent, some of the individual pairs were

weaker matches. As discussed in the Methods chapter, a few discretionary exceptions

were made to allow several pairs to remain in the study, despite some geographic, size, or

other differences. The fact that numerous geographic markets experienced a high level of

mergers during the study’s time period (2009-2013) limited the ability to locate suitable

matches for many Event hospitals, resulting in the removal of those hospitals from this

study. Similarly, many of the merger events that occurred in 2009 were eliminated from

the study due the lack of HCAHPS ratings for three years prior (in 2006). These

necessary exclusions could have affected the results of this study.

Financial performance, both pre- and post-merger, could have affected HCAHPS

performance. This study’s results are limited by not controlling for a hospital’s financial

performance (i.e., operating margin). Although most not-for-profit health systems and

publicly traded for-profit hospital companies publicly report their financial statements on

a consolidated basis, operating margins by individual NPI# hospital are not publicly

available. Since individual hospitals were the subject of this study, it was not possible to

control for financial performance. Therefore, it is unknown whether or not an individual

hospital’s financial performance impacted its HCAHPS ratings.

The MOE vs. Acquired hospital analysis and findings were limited by the fact that

a several large health system mergers represented most of the MOE hospitals. For

example, 25% (11 of 44) of the MOE hospitals in this study were involved with one

Texas merger: Baylor’s merger with Scott & White. This one merger could skew the

MOE hospitals’ collective performance. Consequently, interpretation of the MOE vs.

Acquired hospital FEDID testing results should take this into consideration.

The findings support the general hypothesis of a negative association between

hospital mergers and HCAHPS ratings; however, no causal relationship could be

concluded. Although some significant negative relationships were found, those

relationships may be due to mediating and/or moderating variables that were not included

in this study.

Implications for Future Research

There are several important implications for future research. Since the theoretical

framework and literature review focused on a merger’s impact on the internal working

environment and culture, future researchers should consider quantifying pre- and

postmerger culture and acculturation factors as possible mediators of post-merger

disruption. This research could increase understanding of how physicians, nurses, and

other hospital staff respond to different cultures and acculturation techniques post-merger.

Similarly, future research could be conducted to examine the relationship between

hospital mergers and physician satisfaction, nurse satisfaction, and general staff

satisfaction. As noted in the discussion about this study’s weaknesses, it is unknown

whether the post-merger difference in HCAHPS performance between the Event and

Matched hospitals involved changes in physician, nurse, and staff satisfaction.

Another implication for future research involves the discordant findings with the

PHYSICIAN Domain and NURSE Domain ratings. Future research could focus on

physicians in a merger environment to better understand how physicians experience and

respond to a merger event. For the NURSE Domain, future research could examine why

these ratings did not improve three years following the merger as was found with the

OVERALL ratings and STAFF Domain ratings.

How a hospital merger affects patient expectations is another potential area for

future research. In many cases, a hospital merger involves community input and support.

Board members, physicians, and community leaders interested in preserving access to

care may engage the public for support of a proposed merger. Is it possible that

community engagement and public support for a hospital merger changes patient

expectation? Do patients expect more from the hospital following a merger? Do patients

associate a merger with better hospital performance? Future research could examine the

relationship between a hospital merger and the effect on patient expectations.

Contrary to previous research involving non-healthcare mergers, and opposite of

the four applicable hypotheses, Acquired hospitals performed worse than MOE hospitals.

Future qualitative and quantitative research could examine these merger types to better

explain this study’s findings. Are there certain pre-event characteristics (e.g., market size,

scope of services, employee satisfaction, financial performance) that are associated with

Acquired hospitals? Are there differences in how post-merger acculturation is done

between Acquired and MOE hospitals? What is the role of physicians in the pre-event and

post-event planning and decision-making? Do hospitals that are acquired have more

reduction in services post-event compared to MOE hospitals? Do hospitals that are

acquired experience more job losses compared to MOE hospitals? These and many other

questions could be explored in future research on how hospital merger type affects

postevent performance.

Finally, future research could extend the post-merger time horizon beyond three

years to five or more years to determine when, if ever, the merger-related, negative

effects on HCAHPS performance fully dissipate (i.e., there is no difference in HCAHPS

performance between Event hospitals and Matched hospitals).

Implications for Health Services Leaders, Hospital Boards, Policy-Makers &

Regulators

The findings from this research provide a new perspective and additional answers

to the question: What value does a hospital merger generate for the patients and

communities it serves? For executives and board members considering a merger or

acquisition, the results of this study can be used to guide pre-merger discussions and post-

merger implementation. Knowing that a merger can cause social upheaval and, therefore,

may have a negative effect on HCAHPS performance, leaders may consider

implementing additional initiatives to moderate post-merger culture conflict and

mergerrelated disruptions among physicians, nurses, and general staff. Hospitals and

health systems pursuing a merger should also pay special attention to and make additional

investments in the nursing staff.

Healthcare policy-makers and regulators, at both the federal and state level, could

use this study to develop targeted requirements for hospitals and health systems seeking

government approval for a merger or acquisition. Specifically, the findings from this

study could open a constructive dialogue about the importance of patient experience and

HCAHPS performance as a critical part of the evaluation of the merger’s impact on

consumer welfare. To that end, regulators should consider holding hospitals and health

systems accountable for achieving and maintaining minimum levels of post-merger

HCAHPS performance. Furthermore, additional oversight and monitoring may be needed

for acquired hospitals.

Conclusion

This quasi-experimental study aimed to examine and evaluate the relationship

between hospital mergers and patient experience. This was the first study on hospital

mergers with patient experience as the variable of interest. The study sought to extend a

large body of prior research on hospital mergers, which found little to no improvement in

consumer welfare following a merger event. In general, the evidence on hospital mergers

suggests neutral to negative effects on quality, safety, and pricing.

This study found that hospital mergers do not result in improved patient

experience performance. To the contrary, compared to similar hospitals that did not

merge, the rate of change of OVERALL ratings and NURSE Domain ratings of merged

hospitals declined compared to similar hospitals that did not merge. This study also found

a significant difference in HCAHPS performance based upon merger type: MOE

hospitals compared to Acquired hospitals.

To optimize consumer welfare and deliver more benefit to patients, health services

researchers, healthcare leaders, health policy-makers, and government regulators should

consider this study’s findings in light of the continued consolidation of hospitals and

health systems in the United States.

Students also viewed