Does Private Equity Hurt or Improve Healthcare Value? New Evidence and Mechanisms
Abstract
What is the impact of private equity (PE) ownership on healthcare value? Does PE ownership reduce healthcare value, and if so, can its effect be mitigated using health information technologies (IT)? Given the significant investments by PE firms in the healthcare sector in recent years, it is increasingly important for stakeholders, including policymakers, care providers, and patients, to understand their likely impact and whether PE ownership is aligned with their interests. Because PE firms seek to improve efficiency and generate short-term financial returns, the cost-cutting strategies used to achieve these goals may come at the expense of care quality, leading to competing hypotheses on the impact of PE ownership on healthcare value. We argue that IT-enabled health information sharing capabilities can help PE-owned hospitals to better manage this cost-quality tension and achieve higher healthcare value. Using a staggered difference-in-differences approach and data from U.S. hospitals from 2008 to 2020, we estimate changes in healthcare value following PE acquisition and ownership. Overall, we find that PE ownership of hospitals reduces healthcare value. This decline is partially driven by reduction in clinical staffing following PE ownership. We further find that preexisting hospital health information sharing capabilities mitigate the negative impact of PE ownership on healthcare value. Specifically, such capabilities can strengthen the effect of PE ownership on cost reduction and, at the same time, mitigate PE-induced care quality deterioration. Our results highlight the critical role of common data standards needed to promote IT-enabled information sharing between healthcare providers, which, in turn, can align the incentives of PE firms with the goals of value-based care.
History: Sean Xu, Senior Editor; Ling Xue, Associate Editor.
Funding: I. Bardhan is grateful for the support of the Charles and Elizabeth Prothro Regents Chair in Healthcare Management at the McCombs School of Business. The authors gratefully acknowledge the Dean’s Excellence Research Grant for funding the data acquisition.
Supplemental Material: The online appendix is available at https://doi.org/10.1287/isre.2025.2332.
1. Introduction
In recent years, the healthcare sector has witnessed a significant surge in private equity (PE) investments. The total deal value of PE investments in the global healthcare sector exceeded $151 billion in 2021, up from $30 billion in 2011 (Jain et al. 2023). This trend has sparked considerable debate about the impact of PE ownership on the quality of care delivered. Proponents argue that PE firms can provide much-needed capital, improve operational efficiency, and unlock profitability (Borsa et al. 2023). However, critics have raised concerns about the potential negative impact of PE, manifested in the form of poor patient experience and quality of patient care (Bruch et al. 2021). Healthcare policymakers, as well as the Federal Trade Commission (FTC) and Antitrust Division of the Department of Justice (DOJ), are concerned that PE may prioritize short-term cost savings over quality of patient care and patient well-being (Hans et al. 2024).
The literature presents mixed evidence on the impact of PE ownership in healthcare. Some studies suggest that PE can lead to improvements in healthcare quality in hospitals and nursing homes (Bruch et al. 2020). However, other studies highlight potential risks, suggesting that hospital care quality declines following PE acquisition (Offodile et al. 2021, Gupta et al. 2024). This divergence in findings creates significant challenges for policymakers, as it complicates the task of understanding and evaluating the overall impact of PE on healthcare.
The impact of PE on healthcare can be multifaceted, and improvements in one area may lead to tradeoffs in another. These interdependencies among different dimensions of healthcare performance suggest that a comprehensive evaluation is necessary to fully understand the overall effect of PE ownership, which represents a governance shift that fundamentally alters hospitals’ operational priorities, resource allocation, and coordination capacity (Kaplan and Stromberg 2009). In this study, our first research question is stated as follows: What is the impact of PE ownership on healthcare value? Following prior literature, healthcare value, often referred to as care value, is defined as the degree to which healthcare providers expend clinical resources effectively to deliver services that impact patient health outcomes (Porter 2010, Bardhan et al. 2023).
We believe that a focus on healthcare value is especially important in the PE context, as it is aligned with the recent movement toward value-based care (VBC), which ties reimbursements to performance on both cost and quality dimensions, and rewards hospitals that deliver better outcomes more efficiently. Unlike a fee-for-service model, where providers are reimbursed based on the volume of services, the VBC model seeks to optimize patient outcomes while minimizing resource expenditures (Tsevat and Moriates 2018). Because PE firms are motivated to improve financial performance, gains in care value can potentially translate into higher margins and greater exit valuations (Powers et al. 2021). Hence, PE firms have strong incentives to implement initiatives that can improve care value within their acquired hospitals.
Further, prior studies have highlighted two important benefits of information technology (IT)-enabled health information sharing, namely, its effects on improved efficiency and enhanced care coordination (Lammers et al. 2014, Ayabakan et al. 2017, Janakiraman et al. 2023). Because of the tradeoffs faced by PE-owned hospitals with respect to balancing the tension between operating cost reduction and quality, an important question is whether health information sharing can potentially help PE-owned hospitals to better handle such tradeoffs. Thus, our second research question is: How does IT-enabled health information sharing moderate the impact of PE ownership on care value?
We argue that although PE firms seek to improve efficiency and generate short-term financial returns, the cost-cutting strategies used to achieve these goals may come at the expense of care quality. This tension gives rise to competing hypotheses on how PE ownership could shape healthcare value. We posit that IT-enabled health information sharing capabilities may help PE-owned hospitals to manage this tension. Specifically, health information sharing can strengthen the impact of PE ownership on cost reduction and, at the same time, mitigate the effect of PE-induced care quality deterioration. Given such dual effects, health information sharing capabilities may positively moderate the relationship between PE ownership and care value.
To test these hypotheses, we combined multiple data sources to construct a panel data of 428 hospitals in the United States from 2008 to 2020. Data on PE deals were collected from the SDC Platinum database. We manually matched these PE deals with the affected hospitals through public records. Our empirical analysis reveals that hospitals exhibit lower care value following PE ownership. Further analyses suggest that staff reduction serves as an important mechanism through which PE ownership affects both hospital operating costs and care quality, thereby shaping care value. Specifically, PE ownership leads to lower clinical staffing levels and staff-to-patient ratios which, in turn, reduce operating costs but also weaken hospitals’ capacity to maintain care quality.
Next, we observe that hospital health information sharing capabilities can mitigate the negative impact of PE ownership on care value. Specifically, among hospitals with strong preexisting information sharing capabilities (i.e., prior to PE acquisition), PE firms can leverage these capabilities to reduce costs more effectively. At the same time, different types of information sharing may help PE-owned hospitals to improve different dimensions of care quality. Ancillary information sharing mitigates PE’s negative impact on experiential quality by reducing the coordination burden faced by staff under leaner operating conditions, whereas clinical information sharing mitigates PE’s adverse impact on readmissions by reducing information gaps and supporting diagnosis and treatment decisions.
Our study contributes to the ongoing debate on the role of PE ownership in healthcare organizations. We adopt a holistic approach by examining the impact of PE ownership through the lens of care value. Our findings indicate that, although PE ownership reduces hospital operating costs, such cost savings are obtained at the expense of lower care value. Further, we contribute to the literature on the economic and clinical impact of health information sharing between care providers (Ayabakan et al. 2017, Adjerid et al. 2018, Janakiraman et al. 2023). To the best of our knowledge, this is one of the first studies to demonstrate that health information sharing can help hospitals mitigate the negative impact of PE ownership on care value through both operating cost reduction and care quality preservation. This has important implications for advancing the goals of VBC by demonstrating how IT-enabled capabilities can bridge the gap between financial sustainability and care quality.
Our work has important implications for health policy and health IT adoption and use. First, policymakers may find it challenging to design effective regulatory responses without a complete understanding of the overall consequences of PE ownership. By emphasizing a VBC framework, our study highlights the need for a comprehensive assessment of the benefits and risks associated with PE ownership. Second, policymakers should consider developing relevant policy frameworks to ensure greater balance between financial objectives and care value. This could include mandating transparency in postacquisition quality reporting and enforcing regulations that mandate minimum staffing levels in critical areas. Third, our research highlights the need for administrators and policymakers to prioritize IT investments and adopt health data interoperability standards that facilitate effective health information sharing, to safeguard against a potential decline in care value after PE ownership.
2. Related Literature
2.1. Impact of PE Ownership in Healthcare
PE firms often bring substantial financial resources and management expertise to healthcare organizations, enabling expansion in service offerings, investments in advanced technologies, and upgrades to facilities, which offer the potential to enhance patient care (Robbins et al. 2008). However, PE deals are typically characterized by relatively short investment horizons, usually between three and seven years. This short-term focus may incentivize strategies aimed at maximizing short-term financial gains rather than long-term stability (Richards and Whaley 2024). The growing trend of PE ownership in healthcare has sparked debate and scholarly interest regarding its implications for healthcare delivery. Researchers have examined the impact of PE ownership in various settings, such as nursing homes (Braun et al. 2021), hospitals (Bruch et al. 2020, Offodile et al. 2021, Richards and Whaley 2024), and physician practices (Singh et al. 2022). Despite this literature, the overall impact of PE ownership on healthcare outcomes remains ambiguous. For instance, Cerullo et al. (2022) reported a reduction in mortality rate after PE acquisition, whereas Liu (2022) found no significant change. Conversely, Liu (2022) reported a decline in readmission rate, whereas Cerullo et al. (2022) found no effect.
Extant research mostly focuses on the effects of PE ownership on a single dimension of care delivery such as quality or financial metrics. A notable gap is a lack of understanding on how PE ownership influences overall performance, given the complexity of care delivery. In this research, we address this gap by evaluating the holistic impact of PE ownership from a care value perspective.
2.2. Healthcare Value
Healthcare value is a critical concept in the context of measuring healthcare performance. It generally refers to the health outcomes achieved per dollar spent (Porter 2010). This concept emphasizes not just the cost of care but the quality and effectiveness of care delivered. Porter and Teisberg (2006) introduced the concept of “value-based competition,” emphasizing the importance of measuring outcomes that matter to patients. The Institute for Healthcare Improvement advocates for the “Triple Aim” framework, which focuses on improving patient experience, improving population health, and reducing per capita costs (Berwick et al. 2008). It highlights the need to balance multiple interrelated goals, ensuring that improvements in one area do not come at the expense of another (Bardhan et al. 2023).
Accordingly, we define hospital care value as the degree to which a hospital efficiently utilizes its input resources to deliver healthcare services that enhance patient-centered care, that is, cost-adjusted care quality. Under a value-based framework, both costs and care quality are inherently multidimensional. In particular, costs include both labor and nonlabor costs. Care quality includes dimensions related to patient experience with care delivery, such as staff responsiveness, communication, and perceived coordination, and dimensions related to clinical outcomes, such as readmission rates and mortality (Manary et al. 2013). Thus, our focus on care value provides a comprehensive assessment of both the financial and clinical performance impact of PE ownership.
2.3. IT-Enabled Health Information Sharing
The business value of IT is well-documented, with prior studies highlighting its potential benefits, including greater flexibility, innovation, quality improvement, cost reduction, productivity enhancement, and profitability (Melville et al. 2004, Kleis et al. 2012, Mithas et al. 2017, Pye et al. 2024). In healthcare, information sharing enables healthcare providers to share patient health data across hospitals, physicians, and laboratories (Adjerid et al. 2018). The literature has highlighted two major benefits of health information sharing. First, health information sharing can reduce operating costs through both labor and nonlabor channels. On the labor side, by reducing the need for manual record retrieval, repeated information collection, and other labor-intensive coordination tasks, it can substitute some routine staff tasks and improve staff’s work efficiency (Goldschmidt 2005, Walker 2018). On the nonlabor side, by providing timely access to comprehensive patient records, including prior diagnoses, medications, and test results, it reduces unnecessary duplication of tests and procedures, thereby lowering the use of clinical supplies, facilities, and nonlabor resources (Yaraghi 2015, Ayabakan et al. 2017). Second, by reducing information gaps and enhancing care coordination across providers, health information sharing can improve providers’ ability to make well-informed treatment decisions, identify medication risks, and prevent adverse events (Lammers et al. 2014, Janakiraman et al. 2023). Therefore, health information sharing can also play an important role in helping hospitals to improve care quality.
Importantly, sharing different types of health information may be relevant for improving different dimensions of care quality. One dimension is ancillary information sharing, which refers to exchange of information including patient demographics, laboratory results, and radiology reports (Davidson and Chismar 2007, American Medical Association 2021). This information is episode-specific and usually consolidated by medical staff for physician use. Hence, sharing such information across providers could reduce the staff coordination burden and allow staff to provide greater attention to tasks that rely on interpersonal interaction, such as patient communication, responding to patient concerns, and coordinating complex care needs (Kurhila and Lehtimaja 2019).
A second dimension is clinical information sharing, which refers to the exchange of information more directly related to patient diagnosis, treatment planning, and care management, such as clinical records and medication history. This type of information includes more longitudinal information that describes the patient’s clinical history and ongoing care management and therefore plays an important role in reducing information gaps and improving information continuity across providers (Cho et al. 2024). By providing timely access to relevant clinical information from other providers, clinical information sharing may support physicians in diagnosis, treatment planning, and care management (Janakiraman et al. 2023).
Despite these established benefits of health information sharing, how health information sharing shapes the impact of PE ownership on care value has not been previously explored. This gap is important because PE-owned hospitals may face conflicting objectives across stakeholders: improving short-term financial performance while maintaining care quality and patient-centered outcomes. Health information sharing may mitigate these tradeoffs by supporting PE firms’ cost reduction strategies while allowing them to better preserve care quality, which in turn can improve healthcare value. Our research represents an important step toward answering this question.1
3. Hypothesis Development
PE ownership represents a distinct form of ownership change, one that differs substantively from both nonprofit and publicly traded for-profit governance in the healthcare sector. Nonprofit hospitals are often community-based and focus on meeting specific social or mission-driven objectives (Horwitz and Nichols 2022). Publicly traded for-profit hospitals tend to balance shareholder returns with ongoing obligations to long-term reputational concerns. As a result, they generally maintain steady reinvestment in operations, medical equipment purchases, and facilities while carrying lower debt levels (Aghamolla et al. 2025).
In contrast, PE firms typically operate under shorter investment horizons (typically three to seven years) with a mandate to generate high returns upon exit through resale or IPO (Offodile et al. 2021, Richards and Whaley 2024). This creates stronger incentives for rapid restructuring, labor optimization, and financial engineering, compared with for-profit hospitals. Prior studies also show that PE-owned hospitals report higher operating margins following acquisition (Richards and Whaley 2024). This is consistent with the short-term value extraction priorities of PE ownership rather than reinvestment in operations.
Together, these characteristics position PE ownership as a unique governance structure that introduces a shift toward stronger emphasis on cost efficiency and profit maximization, which may lead to tensions with competing priorities related to patient care. In this section, we will hypothesize the impact of PE ownership on hospital care value and how IT-enabled health information sharing can help PE firms to manage these tradeoffs and thereby shape healthcare value.
3.1. Impact of PE Ownership on Care Value
Resource-based view (RBV) theory emphasizes that firms create value or maintain sustainable competitive advantage by acquiring, developing, and effectively deploying valuable, rare, inimitable, and nonsubstitutable resources and capabilities (Peteraf 1993). These resources and capabilities can be tangible, such as physical assets and human capital, or intangible, including managerial experience, organizational routines, and reputation. PE firms can provide external managerial expertise, performance monitoring systems, and financial oversight that allow strategic resource allocation and operational restructuring (Meuleman et al. 2009). These capabilities may help acquired hospitals identify inefficiencies, optimize workflows, improve accountability, and deploy existing resources effectively. Such efficiency gains may arise from operational changes, including improved hospital capacity utilization, reduced administrative waste, and efficient labor management (Shrank et al. 2019).
Specifically, staffing is an important area where such resource reconfiguration may occur. Hospitals are labor-intensive organizations, and labor represents a substantial share of hospital operating expenses (American Hospital Association 2024). Prior literature suggests that although sufficient staffing is essential for maintaining care quality, excess staffing or inefficient use of clinical labor can also create operational slack and increase avoidable costs (Valdmanis et al. 2008, Shrank et al. 2019). Under PE ownership, hospitals may need to reduce labor-related slack and pursue leaner staffing models as part of their broader efforts to streamline operations and improve resource utilization (Davis et al. 2014). If staffing adjustments can reduce inefficiency without compromising care quality, PE-owned hospitals may be able to lower operating costs and achieve higher care value, that is, cost-adjusted care quality.
Taken together, we argue that PE ownership may improve care value by bringing managerial expertise and reducing operating costs through leaner staffing and more efficient resource deployment.
PE-owned hospitals exhibit higher care value compared with non-PE-owned hospitals.
However, the efficiency-enhancing pathway may not always translate into higher care value. PE firms are primarily accountable to investors and operate under incentives designed to maximize returns within a relatively short timeframe (Kaplan and Stromberg 2009). In addition, because PE acquisitions are often structured as leveraged buyouts (LBOs), debt service obligations can constrain operating budgets and limit resources available for staffing and clinical operations (Gupta et al. 2024). However, reducing operating costs while maintaining care quality is inherently challenging. Without careful redesign of clinical workflows and operational processes, cost-cutting initiatives such as reducing staffing can increase the workload on existing staff, weaken coordination across care teams, and ultimately compromise care quality.
In particular, staff reduction is an important pathway through which PE ownership may weaken care quality, especially on quality dimensions related to patient care experience. Although reducing staffing can generate short-term cost savings, clinical staff play essential roles in supporting care delivery. They gather patient health information, monitor patient status, conduct diagnostic and laboratory tests, and respond to patient needs (Leape et al. 1999, Needleman et al. 2002). When staffing declines without corresponding improvements in workflow design and reconfiguration, the resulting workload can reduce staff responsiveness, delay care processes, and weaken coordination across clinical teams (Aiken et al. 2002). Consequently, patients may experience less timely communication and longer wait times, leading to lower experiential quality (Braun et al. 2021, Bruch et al. 2021).
Hence, although PE ownership may reduce operating costs, these cost savings may be offset by deterioration in care quality. In other words, PE ownership may reduce hospitals’ abilities to improve patient outcomes, thereby diminishing healthcare value. Hence, we propose a competing hypothesis.
PE-owned hospitals exhibit lower care value compared with non-PE-owned hospitals.
3.2. Moderating Role of Health Information Sharing
Stakeholder theory highlights that businesses operate within a diverse ecosystem, where different stakeholders have distinct goals and expectations, and that long-term success depends on the well-being of all stakeholders (Donaldson and Preston 1995, Freeman and McVea 2001). In hospitals, these stakeholder concerns include delivering high-quality patient care, maintaining adequate staffing and capacity, and supporting clinicians’ ability to provide timely and patient-centered care. Although cutting costs may improve margins, doing so without a clear understanding of care processes can weaken the slack resources that hospitals rely on to ensure care quality (Meliones 2000). Further, many care providers may lack clear understanding of operating costs, making it difficult to work with PE firms to identify opportunities to improve cost efficiency and eliminate non-value-adding activities (Porter and Lawrence 2011). From a stakeholder theory perspective, PE’s cost reduction goals may be misaligned with hospitals’ goals for providing patient-centered care.
We argue that strong health information sharing capabilities allow PE-owned hospitals to be better positioned to pursue cost reduction while preserving care quality goals, thereby positively moderating the relationship between PE ownership and care value. Specifically, when PE-owned hospitals are equipped with strong capabilities for health information sharing with other providers, there are more opportunities to eliminate duplicate tests and procedures, and reduce redundant clinical resources such as diagnostic equipment, supplies, and facilities. Further, staffing reduction may not necessarily translate into lower care quality, as the remaining staff can sustain similar service quality through more efficient workflows. Such hospitals can better leverage timely patient information and support seamless communication across care teams. These changes may allow staff to devote the same amount of time to patient interactions and clinical care despite operating with leaner staffing levels.
In contrast, when PE-owned hospitals have weaker health information sharing capabilities, it is harder to cut costs, as doing so is likely to conflict with resources that hospitals consider essential to maintain care quality, leading to misalignment among stakeholders’ goals. At the same time, staffing reduction following PE acquisition is likely to increase the workload on the remaining staff, because there is less opportunity for PE-owned hospitals to reconfigure existing resources and workflows using health IT. Hence, hospital staff may spend substantial time in obtaining medical records, reconciling missing information, and communicating with other providers (Adjerid et al. 2018, Bardhan et al. 2023). As a result, they may devote less attention to tasks that require interpersonal interaction such as patient communication, leading to lower care quality.
In sum, we argue that health information sharing can strengthen the positive effect of PE ownership on cost reduction and, at the same time, mitigate PE-induced care quality deterioration. Given these dual effects, we expect health information sharing to shift the effect of PE ownership on care value in a positive direction, such that the difference in care value between PE-owned and non-PE-owned hospitals becomes more positive (or less negative) when health information sharing capabilities are stronger. Accordingly, we propose the following hypothesis.
Health information sharing capabilities positively moderate the relationship between PE ownership and care value.
4. Data and Variables
4.1. PE Deals and Affected Hospitals
We focus on hospitals involved in LBOs, a common form of PE deals. Our primary data on LBO deals are obtained from the PE and M&A databases of SDC Platinum. Instances where PE firms provide debt financing or purchase minority stakes are excluded, because they do not provide the same level of operational and strategic control as LBO deals. To identify PE transactions in the healthcare sector from the PE database, we followed the following approach: (1) select the “Investee Company Nation” as the United States and “Investment Date” between 2000 and 2024, (2) restrict “Investee SIC” to general medical and surgical hospitals, and (3) require the “Investee Company Long Business Description” to contain the word “hospital.” We supplement the main LBO deals with M&A deals in our sample where the acquirer is a PE company or PE fund. To compile a list of PE transactions in the healthcare sector from the M&A database, we adopted the following procedures: (1) select the “Target Nation” as the United States and “Date Announced” between 2000 and 2024, (2) restrict “Deal Status” to completed and “Acquisition Techniques” to leveraged buyout, and (3) restrict “Target SIC” as general medical and surgical hospitals. To focus specifically on PE deals, we require “Acquiror Immediate Parent Public Status” and “Acquiror Ultimate Parent Public Status” to be private.
After identifying PE deals in U.S. hospitals, we used archived web pages from the Wayback Machine to confirm the hospital name, location, ownership structure, and affiliation at the time of acquisition, for deals at the level of a hospital system. When the web page information was unavailable in the Wayback Machine, we supplemented it by conducting targeted searches of press releases announcing the acquisition or reviewing the hospital system’s annual financial filings (e.g., 10-K) for the corresponding year of the deal. These reports provide organizational structure and hospital lists, allowing us to confirm whether individual hospitals were included in each system-level transaction. Overall, the manual verification process resulted in complete ownership and location information for all hospitals in the PE ownership group, which led to the final sample of 115 PE-owned hospitals. We then merged these data with hospital-level data from the Centers for Medicare and Medicaid Services (CMS) Cost Reports, staffing levels from the American Hospital Association (AHA) survey, and quality data from the CMS Hospital Compare database to measure other variables, as discussed in greater detail.
If a hospital system experienced multiple PE deals, we only considered the first deal because PE ownership is an absorbing state, that is, PE ownership fundamentally changes hospitals’ operational priorities, resource allocation, and governance (Kaplan and Stromberg 2009). We also verified that none of the PE-owned hospitals in our sample reverted to non-PE ownership through PE exits.
4.2. Healthcare Value
Quantifying healthcare value is challenging, as it involves measuring multiple inputs and outputs that are often interdependent. To address this challenge, we adopt a Data Envelopment Analysis (DEA) approach. It is a nonparametric optimization method which evaluates how efficiently a decision-making unit (DMU) converts its input resources into clinical outputs. DEA has been widely applied to measure hospital efficiency because it accommodates multiple inputs and outputs without requiring strong parametric assumptions (Tiemann and Schreyögg 2012, Bardhan et al. 2023).
DEA constructs a Pareto-efficient frontier across all hospitals and computes a radial efficiency score θ for each hospital. Hospitals on this frontier receive a DEA score of one, indicating optimal efficiency. Hospitals below the frontier receive scores between zero and one, reflecting the focal hospital’s proportional distance from the frontier and its efficiency compared with the most efficient hospitals. Care Value is thus measured based on the DEA score θ, with higher values indicating higher relative efficiency, that is, greater healthcare value. Following Bardhan et al. (2023), we apply an input-oriented Banker-Charnes-Cooper (BCC) model (Banker et al. 1984) as specified in Equation (1).
X and Y denote the input and output vectors of the focal hospital, respectively, and θ represents the DEA measure of care value. The constraint, reflects that this model allows for variable returns to scale, as hospitals typically exert greater control over their input resources than over outputs.2
To evaluate hospital healthcare value, our DEA model utilizes operating expenses, capital expenses, and number of full-time equivalent (FTE) labor as inputs, and patient experiential quality, mortality rate, and readmission rate as outputs. Experiential quality is measured based on an average score of five survey questions on patient satisfaction with communication and responsiveness of hospital staff (Senot et al. 2016).3 Mortality rate is computed as the weighted average of individual mortality rates for heart attack, heart failure, and pneumonia, with the weight based on the number of patients with each condition. Similarly, hospital readmission rate is calculated as the weighted average of readmission rates for these three conditions, respectively. To account for differences in patient volume across hospitals, we normalize input resources based on the number of encounters, resulting in inputs measured on a per-encounter basis. We use the inverse of mortality and readmission rates as output variables in the DEA model, because lower rates of mortality and readmission indicate higher care quality.
4.3. Health Information Sharing and Health IT Capabilities
We measure hospital health information sharing capability (denoted as InfoShare) using a set of questions from the AHA IT supplement survey. These questions evaluate the extent to which hospitals electronically exchange or share different types of patient health data with healthcare providers. The survey measures health information sharing across five categories of patient health data: (a) demographics, (b) clinical care records, (c) laboratory results, (d) medication history, and (e) radiology reports. For each category, the survey asks whether the hospital shares data with four types of providers: (a) hospitals within the same health system, (b) hospitals outside the health system, (c) ambulatory providers within the system, and (d) ambulatory providers outside the system. This yields a total of 20 survey responses as described in Online Appendix A. InfoShare is calculated as the percentage of “Yes” responses across all questions at the hospital-year level. A higher value reflects a focal hospital possessing stronger capability in sharing health information with other care providers.
In addition to understanding the overall moderating role of InfoShare, we also investigate the roles of ancillary and clinical information sharing because of their distinct roles in care delivery processes, as discussed in Section 2.3. Based on the prior literature, we measure a focal hospital’s capability in sharing ancillary information (Ancillary InfoShare) using the percentage of “Yes” responses related to the sharing of patient demographics, laboratory results, and radiology reports (Davidson and Chismar 2007). We measure a focal hospital’s capability in sharing clinical information (Clinical InfoShare) using the percentage of “Yes” responses to sharing of patient clinical records and medication histories.
To account for differences in the overall health IT infrastructure between hospitals affected by PE ownership and those that are not, we follow the literature on IT capabilities (Rai and Tang 2010, Pye et al. 2024) and construct two variables, Intrahospital IT and Interhospital IT, for matching purposes as discussed in our empirical approach. Intrahospital IT refers to a hospital’s internal ability to collect, process, and use clinical and administrative information across departments within the same facility for care delivery. It is measured based on a set of questions that assess the focal hospital’s IT capabilities in several areas: (a) electronic clinical documentation, (b) results viewing, (c) computerized provider order entry (CPOE), (d) clinical decision support, and (e) other functionalities of the electronic health record (EHR) system. Each of these areas is comprised of multiple questions, as shown in Section 1 of Table A.1 in Online Appendix A. Intrahospital IT is calculated as the percentage of responses indicating whether the corresponding capability is fully implemented either in at least one unit or across all units within the focal hospital.
Interhospital IT is defined as the broader IT capability to digitally share patient information with external entities, including both care providers and patients. We use two sets of questions in the AHA IT supplement survey to measure this construct. The first set of questions is the same questions used to measure InfoShare. The second set of questions assesses whether patients at the focal hospital can perform various tasks online, as shown in Online Appendix Table A.1. Interhospital IT is calculated as the percentage of “Yes” responses to the two sets of questions at the hospital-year level.
4.4. Control Variables
To mitigate concerns related to omitted variable bias, we control for a set of hospital-specific characteristics that may influence both the likelihood of being a target for PE ownership and hospital performance. Specifically, we include Medicare patient load, case mix index, operating outlier adjustment factor, capital outlier adjustment factor, total number of beds, intrahospital IT capability, resident-to-bed ratio, cost of living factor, and wage index. These variables account for patient complexity and severity, hospital size, internal health IT infrastructure, and other hospital characteristics.
5. Empirical Approach
5.1. Baseline Model Specification
To test whether PE ownership improves or hurts care value (Hypothesis 1a and Hypothesis 1b), we utilize a staggered difference-in-differences (DID) approach. Hospitals affected by PE ownership are classified in the treatment group. Because the timing of PE ownership varies across hospitals, the posttreatment period differs for each hospital in the treatment (i.e., PE ownership) group. Hospitals that are not affected by PE ownership are classified in the control group. The baseline model is specified in Equation (2), with the unit of analysis at the hospital (i)-year (t) level.4
The dependent variable represents the care value of hospital i in year t. It is a continuous variable ranging from zero to one, where higher values indicate greater care value. The independent variable indicates the PE ownership status of hospital i in year t, taking a value of one if hospital i experienced PE ownership in year t and zero otherwise. In other words, PE_Ownership indicates whether hospital i belongs in the PE-owned group in year t. indicates whether year t belongs in the posttreatment period for both PE-owned and control hospitals. The vector includes a set of time-varying control variables that represent hospital i’s patient complexity and severity in year t. Further, we control for hospital characteristics such as number of staffed beds, resident-to-bed ratio, and wage index for focal hospital i in year t. We also control for overall IT capabilities, measured by the level of Intrahospital IT in hospital i in year t. We include both hospital and year fixed effects, denoted as and to account for unobserved, time-invariant differences at the hospital and general time trends, respectively. represents the error term.
5.2. Mediating Effect of Clinical Staffing
As discussed in the development of Hypotheses 1a and 1b, PE-owned hospitals may have incentives to reduce staffing levels to lower operating expenses. However, without effective resource reconfiguration and workflow redesign, staff reductions may also erode care quality, particularly for patient experience with staff. We therefore investigate whether staff reduction represents an important mediation mechanism through which PE ownership affects both operating costs and care quality.
To examine how PE ownership influences operating costs through staff reduction, we use the following model specification represented in Equations (3)–(6).
In this mediation model, Number of Clinical Staff represents the mediator, and is measured as the total number of FTE clinical staff employed by the hospital, including physicians, nurses, radiology staff, laboratory technicians, and pharmacy technicians. Other variables such as , , , , and follow the same definition as in Equation (2).
We first estimate the direct effect of PE ownership on clinical staffing as shown in Equation (3), followed by the estimation of the relationship between clinical staffing and operating costs as shown in Equation (4). Equation (5) estimates the overall effect of PE ownership on operating costs. Finally, we include both PE_Ownership and Number of Clinical Staff in a single model in Equation (6). If PE ownership partially reduces operating costs by reducing the number of clinical staff, we would expect the coefficients and , in Equations (3) and (5), respectively, to be significantly negative, and in Equation (4) to be significantly positive. Further, we observe partial mediation if the coefficient in Equation (6) remains positive.
To examine how PE ownership influences care quality (especially patient experiential quality), through staff reduction, we use Clinical Staff-to-Patient Ratio as the mediator. This variable is defined as the number of FTE clinical staff per admission. Estimation of the mediation effect is based on similar models as shown in Equations (3)–(6), replacing the Number of Clinical Staff with Clinical Staff-to-Patient Ratio and replacing log(Operating Costs) with Experiential Quality as the outcome variables. We use different measures of clinical staffing for the two mediation analyses because they represent the aspects of staffing most relevant to the respective outcomes. For operating costs, we use the number of clinical staff because total staffing is directly related to labor costs. For care quality, we use the clinical staff-to-patient ratio because care quality depends on the level of staffing available relative to patient volume, rather than the absolute number of staff.
5.3. Moderating Effect of Health Information Sharing
To explore the moderating role of health information sharing on the relationship between PE ownership and care value as stated in Hypothesis 2, we use the specification in Equation (7).
In this model, we include interaction terms , , and while holding other variables the same as in Equation (2). The coefficient of primary interest is , which estimates whether the effect of PE ownership on care value differs between hospitals with high and low health information sharing capability. For easy interpretation of the coefficients and marginal effects, we use a binary variable to measure the information sharing capability, denoted as , which is equal to one if the focal hospital’s health information sharing capability is above the sample mean, and zero otherwise.
We measure using a focal hospital’s preexisting health information sharing capability level (i.e., one year prior to PE ownership) for several reasons as follows. First, as suggested in our theoretical arguments, PE firms may improve financial performance by restructuring acquired hospitals and leveraging their existing resources and capabilities. Health information sharing capabilities are organizational capabilities that typically require substantial investments in technology, workflow redesign, and organizational learning, and therefore are unlikely to be created immediately following acquisition. Instead, PE firms are more likely to deploy and leverage such capabilities already present in the acquired hospitals to support resource reconfiguration and workflow redesign. Thus, using a preexisting measure would be appropriate to test how PE firms utilize focal hospitals’ available resources in shaping care value. Second, using a preexisting measure of information sharing capability can also help address reverse-causality concerns, as PE ownership may influence the subsequent information sharing capability of a focal hospital. Accordingly, our model estimate should be interpreted as whether hospitals with stronger existing information sharing capabilities are better positioned to pursue cost reduction and maintain care quality under PE ownership.
5.4. Matching
Although our analysis includes several controls, unobserved time-varying confounding factors may lead to potential endogeneity concerns. To strengthen the validity of our DID analysis and address potential selection bias, we implement a one-to-three propensity score matching (PSM) approach to construct a comparable control group. This method allows us to account for pretreatment differences and ensure parallel trends in care value between PE-owned hospitals and hospitals in the control group (Heckman et al. 1998). Specifically, we match each treated hospital with three nearest neighbor control hospitals without replacement, based on the average values of key hospital characteristics in the pretreatment period. These matching variables include the number of physicians and nurses, which represents the scale and flexibility of the focal hospital’s clinical workforce; experiential quality, readmission rate, mortality rate, and care value; and the level of intra- and interhospital IT infrastructure, which reflects hospital digital capability that may, in turn, influence PE ownership. This yields 115 hospitals in the treatment (i.e., PE ownership) group and 312 hospitals in the control group, resulting in 5,077 hospital-year observations during the study period from 2008 to 2020.5
Table 1 presents descriptive statistics for the matched sample, along with balance test results comparing treatment and control hospitals prior to PE acquisition. The table is organized into two panels. Panel A (main variables) lists key variables used in our empirical analyses. Panel B (control variables) includes the covariates in vector . According to the p-values in the last column, most of the listed variables do not differ significantly between PE-owned hospitals and hospitals in the control group. The results indicate that the matching procedure achieves balance not only for variables used in the PSM algorithm, but also among covariates not explicitly included in matching. This enhances the credibility of our identification strategy in that observed posttreatment effects can be more reliably attributed to the effects of PE ownership rather than to preexisting differences between the two groups. We use this matched sample for all analyses presented in the paper.
|
Table 1. Descriptive Statistics of Matched Sample
| Period: 2008–2020 | Pretreatment period | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Matched sample | Control | Treatment (PE ownership) | Balance test | ||||||
| Variables | Number of observations | Mean | Standard deviation | Number of hospitals | Mean | Number of hospitals | Mean | Mean Diff | p-value |
| Panel A: Main variables | |||||||||
| Operating costs (in million $) | 5,077 | 136.709 | 162.349 | 312 | 126.661 | 115 | 108.976 | −0.051 | 0.617 |
| Number of clinical staff (in 100s) | 5,077 | 3.148 | 3.216 | 312 | 2.983 | 115 | 2.803 | 0.180 | 0.560 |
| Clinical staff-to-patient ratio | 5,077 | 0.054 | 0.024 | 312 | 0.052 | 115 | 0.047 | 0.005 | 0.014 |
| Experiential quality (in %) | 5,077 | 73.059 | 4.982 | 312 | 72.473 | 115 | 72.238 | 0.235 | 0.629 |
| Staff responsiveness | 5,077 | 64.110 | 7.921 | 312 | 63.854 | 115 | 62.401 | 1.454 | 0.060 |
| Mortality (in %) | 5,077 | 13.145 | 1.978 | 312 | 12.606 | 115 | 12.636 | −0.030 | 0.857 |
| Readmission rate (in %) | 5,077 | 19.793 | 1.980 | 312 | 20.344 | 115 | 20.309 | 0.034 | 0.851 |
| Care value | 5,077 | 0.312 | 0.199 | 312 | 0.304 | 115 | 0.308 | −0.005 | 0.782 |
| InfoShare | 4,915 | 0.597 | 0.284 | 304 | 0.534 | 109 | 0.535 | −0.000 | 0.985 |
| Ancillary InfoShare | 4,915 | 0.613 | 0.287 | 304 | 0.556 | 109 | 0.551 | 0.005 | 0.844 |
| Clinical InfoShare | 4,867 | 0.576 | 0.301 | 301 | 0.506 | 107 | 0.513 | −0.008 | 0.764 |
| Panel B: Control variables | |||||||||
| Intrahospital IT | 5,077 | 0.773 | 0.244 | 312 | 0.716 | 115 | 0.701 | 0.015 | 0.451 |
| Bed total | 5,077 | 158.924 | 119.239 | 312 | 157.254 | 115 | 159.457 | −2.203 | 0.865 |
| Wage index | 5,077 | 0.962 | 0.184 | 312 | 0.045 | 115 | 0.024 | 0.021 | 0.138 |
| Cost of living factor | 5,077 | 1.002 | 0.020 | 312 | 1.002 | 115 | 1.000 | 0.002 | 0.297 |
| Case mix index | 5,077 | 1.370 | 0.264 | 312 | 0.968 | 115 | 0.940 | 0.028 | 0.130 |
| Resident-to-bed ratio | 5,077 | 0.042 | 0.138 | 312 | 1.312 | 115 | 1.360 | −0.048 | 0.071 |
| Operating outlier factor | 5,077 | 0.025 | 0.038 | 312 | 0.028 | 115 | 0.022 | 0.006 | 0.090 |
| Capital outlier factor | 5,077 | 0.035 | 0.064 | 312 | 0.040 | 115 | 0.034 | 0.005 | 0.465 |
| Medicare patient load | 5,077 | 0.458 | 0.148 | 312 | 0.489 | 115 | 0.471 | 0.018 | 0.197 |
Notes. The descriptive statistics are reported based on the matched panel. Main outcome and independent variables are listed in panel A. Variables used as controls are listed in panel B. We performed one-to-three PSM based on mean value of eight hospital characteristics including number of physicians and nurses, experiential quality, readmission rate, mortality rate, measured healthcare value, and strength of intra- and interhospital IT infrastructure. The balance test is conducted by comparing the mean values of the above variables for the years preceding the PE acquisition. Diff, difference.
5.5. Endogeneity and Selection Bias
As noted above, an important concern about our model identification strategy is that PE ownership could be endogenous. We utilize PSM to address endogeneity concerns arising from observable covariates that can influence both the likelihood of PE ownership and hospital outcomes. However, concerns may still exist because of potential endogeneity arising from complex interactions or nonlinear relationships among covariates, which may not be fully captured by parametric models typically used in PSM. To address these concerns, we employ double machine learning (DML), a nonparametric machine learning approach that can address endogeneity issues arising from high-dimensional confounders or their effects on treatment selection and outcomes, which cannot be satisfactorily modeled using parametric methods (Chernozhukov et al. 2018). Additional details are discussed in Online Appendix B. Finally, to address potential endogeneity concerns related to the level of health information sharing in Equation (7), we apply an instrumental variable (IV) approach to test the robustness of the results, as discussed in Section 7.3.
6. Results
6.1. Impact of PE Ownership on Care Value
Table 2 shows the baseline estimate of the impact of PE ownership on hospital care value. Using care value as the dependent variable, the results based on Equation (2) are reported in column (1). The coefficient estimate of Post × PE_Ownership suggests that hospitals exhibit significantly lower care value (i.e., 13% lower) following PE acquisition, which implies that PE ownership can hurt care value.
|
Table 2. Impact of PE Ownership on Care Value
| Dependent variable | (1) | (2) | (3) | (4) | (5) |
|---|---|---|---|---|---|
| Care value | Log(operating costs) | Experiential quality | Mortality rate | Readmission rate | |
| Post | 0.029** | −0.002 | 0.245 | −0.108 | 0.041 |
| (0.012) | (0.016) | (0.187) | (0.104) | (0.086) | |
| Post × PE_Ownership | −0.040*** | −0.067*** | −0.760** | 0.293* | 0.012 |
| (0.014) | (0.025) | (0.298) | (0.156) | (0.123) | |
| Observations | 5,077 | 5,077 | 5,077 | 5,077 | 5,077 |
| Number of hospitals | 427 | 427 | 427 | 427 | 427 |
| Adjusted R2 | 0.035 | 0.215 | 0.310 | 0.502 | 0.551 |
| Controls | Yes | Yes | Yes | Yes | Yes |
| Hospital FE | Yes | Yes | Yes | Yes | Yes |
| Year FE | Yes | Yes | Yes | Yes | Yes |
Notes. Robust standard errors clustered at hospital level are in parentheses. FE, fixed effects.
***p < 0.01; **p < 0.05; *p < 0.1.
As discussed in Section 2.2, care value measures how efficiently the focal hospital uses its resources to deliver high-quality care. Hence, to study the extent to which the negative impact of PE ownership on care value observed in column (1) is driven by changes in costs versus care quality, we replace Care Value in Equation (2) with operating costs and different measures of care quality as outcome variables.6 The corresponding results are reported in columns (2)–(5) in Table 2.
The result in column (2) suggests that hospitals exhibit significantly lower operating costs following PE ownership, which is consistent with the argument that PE firms tend to implement aggressive cost-cutting strategies in acquired hospitals. However, as observed in columns (3) and (4), PE-owned hospitals experience significant decline in care quality, especially with respect to patient experiential quality. The result in column (5) suggests that, on average, readmissions are not significantly affected by PE ownership. A plausible explanation is that readmission rates are closely monitored and penalized under the CMS Hospital Readmissions Reduction Program. Hence, as they implement cost-cutting strategies, PE-owned hospitals may be motivated not to increase readmissions to avoid financial penalties.
Overall, the findings demonstrate a critical challenge for PE firms to manage their acquired hospitals: although aggressive operating cost reduction may improve efficiency and short-term financial performance, such efforts may also erode care quality significantly. As a result, the overall care value could decline following PE ownership.
In our hypothesis development, we argue that if PE firms can potentially bring their valuable managerial expertise to help hospitals reduce inefficiencies and improve resource utilization, the focal hospitals may be better positioned to navigate the challenges of balancing cost reduction and maintaining care quality. This argument implies that PE firms’ prior experience may play an important role in moderating the impact of PE ownership on hospital care value. Motivated by this observation, we measure a PE firm’s experience as the number of years between its founding year and the focal deal year. PE firms are then classified as high-experience (i.e., High-Exper PE equal to one) versus low-experience PE (i.e., High-Exper PE equal to zero) based on the sample median of PE firm experience. We then add the interaction term Post × × High-Exper PE to Equation (2) and rerun the analysis. The results are reported in Table C.1 in the Online Appendix.
Overall, we find that hospitals acquired by more-experienced PE firms exhibit relatively higher care value and experiential quality than hospitals acquired by less experienced PE firms. However, these hospitals also have higher operating costs. This pattern suggests that experienced PE firms may be less likely to rely solely on aggressive cost-cutting strategies and may instead preserve greater operational capacity to maintain care quality. In contrast, less experienced PE firms may pursue aggressive cost reduction, but savings are likely to come at the expense of lower care value and patient experience.
Next, we run a range of robustness checks to address endogeneity concerns and selection bias related to our baseline empirical approach. First, we run the DML analysis based on the approach discussed earlier. As shown in Table B.1 in Online Appendix B, we observe a significant decline in care value following PE ownership, which is consistent with the baseline result. Second, we run the timing falsification exercise to validate the parallel trend assumption in our baseline approach. To assess whether treated and control hospitals exhibit parallel trends in care value prior to PE acquisition, we employed a leads/lags model (Autor et al. 2003), as specified in Equation (8).7
Our sample comprises hospitals with data that span 10 pre- and/or post-PE ownership years. For the sake of space and clarity, we group every two years to construct the and variables, where τ represents the relative time period (in two-year intervals) before or after the PE acquisition for each hospital.8 Specifically, for periods prior to PE ownership (), the variable represents an indicator variable that equals one if year t is or years before the PE acquisition year. Similarly, for periods after PE ownership (), represents an indicator variable that equals one if year t is or years after the PE acquisition year for hospital i. For example, indicates the focal year t is either one or two years after PE ownership of hospital i. indicates year t is the PE ownership start year for hospital i. For hospitals with more than six years post-PE acquisition (or preacquisition), we collapse the subsequent (or prior) years into a period labeled as “3+.” For example, we combine “” “” and “” to form a new dummy, denoted as “” To ensure consistency across periods, we designate and the interaction term (i.e., one or two years before PE ownership) as the comparison period. All other time periods, both pre- and post-PE acquisition, are evaluated against this period.
The parallel trend plot of care value using the leads/lags model is shown in Figure C.1 in the Online Appendix. We find no evidence of difference in pretreatment trends between the PE-owned and control groups. The coefficient estimates for all periods prior to PE acquisition are small and statistically insignificant. In addition, we report the leads/lags regression results using care value, logged operating cost, and care quality outcomes as dependent variables in Table C.2 in Online Appendix C. Across columns, the coefficients corresponding to the pre-PE ownership periods are statistically insignificant, indicating no significant differences on care value, operating cost, or various care quality metrics between PE-owned and control hospitals, prior to PE ownership. This result supports the parallel trend assumption. In contrast, the coefficients for the post-PE ownership periods are consistent with our baseline results. Notably, the magnitudes of the coefficient estimates generally increase over time, suggesting that the effects intensify following PE ownership as operational changes introduced by PE firms take time to materialize.
Third, we conduct placebo tests of spurious trends by randomly assigning treated hospitals. We randomly assign hospitals to form the PE ownership group by selecting 115 (out of 427) hospitals from our main sample. We then estimate the DID regression on the randomly assigned PE ownership group and repeat this process 1,000 times, resulting in a total of 1,000 experiments and 1,000 regression results. We graph a kernel density plot of the results as shown in Figure C.2 in the Online Appendix. Specifically, the x-axis represents the estimated treatment effects and the y-axis represents the probability densities and respective p-values associated with the coefficient estimates in each experiment. The distribution is centered around zero, because the treatment was randomly assigned. This serves as a validity check, ensuring that the DID estimator does not systematically generate spurious treatment effects. We also report a rejection rate of 0.047 for the null hypothesis of no treatment effect at a significance level of 5%. This rate is close to the expected 5% threshold, confirming the validity of our DID analysis.
Fourth, the econometrics literature suggests that, in settings with staggered PE ownership, the two-way fixed effects (TWFE) model may yield biased estimates. This bias arises from invalid comparisons, particularly when early-treated units are used as controls for later-treated units (Baker et al. 2022). To address this concern, we implement two alternative estimators that are robust to treatment effect heterogeneity: the two-stage DID estimator proposed by Gardner (2022) and the group-time average treatment effect estimator proposed by Callaway and Sant’Anna (2021). More details of the estimation process and the corresponding results are reported in Online Appendix D. Overall, the results are qualitatively consistent with our baseline findings.
Lastly, a potential concern is that PE-owned hospitals may differ from non-PE hospitals systematically in unobserved characteristics, such as managerial quality, market reputation, or latent financial health, which could bias our estimates. To address this concern, we implement a two-step Heckman selection model (Heckman 1979) to explicitly account for potential selection bias in PE acquisition decisions. In the first stage, we estimated a probit model of PE acquisition using all exogenous variables from the main specification, along with an instrumental variable that measures the number of successful IPOs in the prior year in a focal hospital’s Core-Based Statistical Area (CBSA). This variable reflects local private capital activity which may influence PE acquisition likelihood but is plausibly exogenous to individual hospital performance. We then compute the inverse Mills ratio (IMR) from the first stage and include it as a correction term in the second-stage regression with hospital and year fixed effects (Wooldridge 2010). Inclusion of the IMR allows us to adjust for unobserved factors influencing both PE selection and hospital performance, thereby mitigating potential endogeneity concerns. The second-stage results, as reported in Table C.3 in the Online Appendix, indicate a significant relationship between PE ownership and decline in care value, consistent with our baseline DID estimates.
6.2. Mediating Role of Clinical Staffing
Our theoretical arguments suggest that clinical staffing may be an area where PE firms reconfigure resources to improve the efficiency and financial performance of acquired hospitals. In other words, staff reduction can serve as one mechanism through which PE ownership affects both operating costs and care quality, thereby shaping care value. This section reports the results of the mediation analyses to test this proposed mechanism.
Table 3 shows the results based on the model specifications in Equations (3)–(6). Column (1) shows that PE-owned hospitals exhibit a statistically significant reduction in the number of clinical staff (coefficient = −0.284, p < 0.05), suggesting that PE firms push these hospitals to pursue leaner staff models. As shown in column (2), the number of clinical staff is positively related to operating costs (coefficient = 0.028, p < 0.01). The effect of PE ownership on operating costs, as shown in column (3), is negative and significant (coefficient = −0.067, p < 0.01) and decreases after controlling for number of clinical staff (coefficient = −0.060, p < 0.05). Meanwhile, the coefficient estimate of Number of Clinical Staff remains significantly positive. The bootstrapped mediation tests further show a significant indirect effect (indirect effect = −0.008, p = 0.040), implying that clinical staff reduction partially explains the post-PE decline in operating cost.
|
Table 3. Mediating Effect of Clinical Staffing Levels on Operating Costs
| Dependent variable | (1) | (2) | (3) | (4) |
|---|---|---|---|---|
| Number of clinical staff | Log(operating costs) | Log(operating costs) | Log(operating costs) | |
| Post | 0.033 | −0.002 | −0.003 | |
| (0.078) | (0.016) | (0.016) | ||
| Post × PE_Ownership | −0.284** | −0.067*** | −0.060** | |
| (0.121) | (0.025) | (0.025) | ||
| Number of clinical staff | 0.028*** | 0.027*** | ||
| (0.006) | (0.006) | |||
| Observations | 5,077 | 5,077 | 5,077 | 5,077 |
| Number of hospitals | 427 | 427 | 427 | 427 |
| Adjusted R2 | 0.099 | 0.224 | 0.215 | 0.228 |
| Controls | Yes | Yes | Yes | Yes |
| Hospital FE | Yes | Yes | Yes | Yes |
| Year FE | Yes | Yes | Yes | Yes |
Notes. Number of clinical staff is measured at the total number of FTE clinical staff, including physicians, nurses, radiology staff, laboratory technicians, and pharmacy technicians. The bootstrapped mediation tests further show a significant indirect effect (indirect effect = −0.008, p = 0.040). Robust standard errors clustered at hospital level are in parentheses.
***p < 0.01; **p < 0.05.
Similarly, Table 4 provides some evidence that staff reduction also represents one mechanism through which PE ownership affects care quality, as measured by experiential quality. In particular, the results in column (1) suggest a significant reduction in clinical staff-to-patient ratio following PE ownership (coefficient = −0.004, p < 0.01). Clinical staff-to-patient ratio is, in turn, strongly positively associated with experiential quality (coefficient = 16.399, p < 0.01). Column (3) shows the overall negative effect of PE ownership on experiential quality (coefficient = −0.760, p < 0.05). When clinical staff-to-patient ratio is included (column (4)), the magnitude of Post × PE_Ownership is reduced, whereas the coefficient of staff-to-patient ratio remains statistically significant. The bootstrapped mediation tests further show a significant indirect effect (indirect effect = −0.068, p = 0.038). Overall, these results suggest that clinical staff-to-patient ratio partially mediates the effect of PE ownership on experiential quality.
|
Table 4. Mediating Effect of Clinical Staff-to-Patient Ratio on Experiential Quality
| Dependent variable | (1) | (2) | (3) | (4) |
|---|---|---|---|---|
| Clinical staff-to-patient ratio | Experiential quality | Experiential quality | Experiential quality | |
| Post | −0.001 | 0.245 | 0.253 | |
| (0.001) | (0.187) | (0.187) | ||
| Post × PE_Ownership | −0.004*** | −0.760** | −0.692** | |
| (0.002) | (0.298) | (0.294) | ||
| Clinical staff-to-patient ratio | 16.399*** | 15.849*** | ||
| (5.019) | (5.009) | |||
| Observations | 5,077 | 5,077 | 5,077 | 5,077 |
| Number of hospitals | 427 | 427 | 427 | 427 |
| Adjusted R2 | 0.246 | 0.313 | 0.310 | 0.315 |
| Controls | Yes | Yes | Yes | Yes |
| Hospital FE | Yes | Yes | Yes | Yes |
| Year FE | Yes | Yes | Yes | Yes |
Notes. Clinical staff-to-patient ratio is measured as number of clinical staff per patient admission. The bootstrapped mediation tests further show a significant indirect effect (indirect effect = −0.068, p = 0.038). Robust standard errors clustered at hospital level are in parentheses.
***p < 0.01; **p < 0.05.
Taken together, our results suggest that PE ownership reduces both clinical staffing and clinical staff-to-patient ratios. Although staffing reduction helps lower operating costs, it also weakens hospitals’ capacity to maintain care quality, consistent with Hypothesis 1b. However, because clinical staffing only partially mediates the effects of PE ownership on operating costs and experiential quality, other mechanisms may also play a role. For example, PE firms may reduce costs through procurement consolidation, supplier renegotiation, and more efficient resource utilization. Care quality may similarly be affected through pathways not fully captured by staffing levels, such as changes in workforce composition, increased work intensity, and reduction in resources supporting care coordination and patient services. Although our data do not allow us to directly examine these additional mechanisms, they represent important directions for future research.
6.3. Moderating Role of Health Information Sharing Capabilities
In Table 5, we report our estimation results based on the model specified in Equation (7).9 The coefficient of Post × PE_Ownership × InfoShare_High suggests that preexisting health information sharing capability positively moderates the relationship between PE ownership and care value (coefficient = 0.057, p < 0.05). The last few rows report the marginal effects of PE at high and low levels of health information sharing capability. Specifically, the marginal effect of PE ownership in hospitals with low health information sharing is −0.068, equivalent to a 17.8% decrease in care value. In contrast, the marginal effect of −0.011 in hospitals with high information sharing is substantially smaller in magnitude and not statistically significant.
|
Table 5. Moderating Impact of Health Information Sharing on Care Value
| Dependent variable | (1) |
|---|---|
| Care value | |
| Post | 0.041** |
| (0.016) | |
| Post × InfoShare_High | −0.030* |
| (0.017) | |
| Post × PE_Ownership | −0.068*** |
| (0.021) | |
| Post × PE_Ownership × InfoShare_High | 0.057** |
| (0.028) | |
| Observations | 4,691 |
| Number of hospitals | 375 |
| Adjusted R2 | 0.037 |
| Controls | Yes |
| Hospital FE | Yes |
| Year FE | Yes |
| Marginal effect of PE at low InfoShare | −0.068*** |
| (0.021) | |
| Marginal effect of PE at high InfoShare | −0.011 |
| (0.018) | |
| Difference between high and low InfoShare (p-value) | 0.043 |
Note. Robust standard errors clustered at hospital level are in parentheses.
***p < 0.01; **p < 0.05; *p < 0.1.
Consistent with Hypothesis 2, the results suggest that PE firms may leverage strong information sharing capabilities in their acquired hospitals to reduce costs through resource reconfiguration and workflow redesign while maintaining care quality. In doing so, there is greater alignment among goals of multiple stakeholders, including PE firms, which focus on cost efficiency and financial performance, and hospitals, which focus on patient-centered outcomes such as care quality. In contrast, it would be harder for PE firms to do so in hospitals with weak information sharing capabilities.
In our theoretical argument, the key moderating role of health information sharing lies in its dual impact on strengthening the positive effect of PE ownership on cost reduction and allowing hospitals to be better positioned to mitigate PE-induced care quality deterioration. Next, we seek to provide further empirical evidence that specifically tests these dual effects.
6.3.1. How Information Sharing Moderates the Impact of PE Ownership on Operating Costs.
As discussed in Section 3.2, strong information sharing capabilities allow PE firms to reduce the use of clinical resources and pursue leaner staff models for information-intensive coordination tasks. Therefore, we expect health information sharing could strengthen PE-owned hospitals’ ability to reduce operating costs, including reduction in nonlabor operating costs and clinical staff. We therefore estimate Equation (7) using the dependent variable(s) operating costs, number of clinical staff, and nonlabor operating costs, separately.
As reported in column (1) of Table 6, the coefficient of Post × × InfoShare_High suggests that the relationship between PE ownership and reduction in operating costs is more pronounced in hospitals characterized by high preexisting information sharing capabilities. Columns (2) and (3) further suggest that this cost reduction operates through both staff reduction and nonlabor cost reduction channels. As indicated in the last few rows of Table 6, among hospitals with strong information sharing capabilities, the marginal effect of PE ownership indicates significant reduction in operating cost, including staff and other nonlabor cost. In contrast, among hospitals with weak information sharing capabilities, there is no significant relationship. Taken together, these findings support our argument on the critical role of health information sharing in enabling PE-owned hospitals to achieve cost reduction through more efficient clinical staffing and nonlabor resource utilization.
|
Table 6. Moderating Impact of Health Information Sharing on Operating Costs
| Dependent variable | (1) | (2) | (3) |
|---|---|---|---|
| Log(operating costs) | Number of staff | Log(nonlabor operating costs) | |
| Post | −0.029 | −0.179** | −0.023 |
| (0.022) | (0.082) | (0.024) | |
| Post × InfoShare_High | 0.046* | 0.457** | 0.047 |
| (0.027) | (0.182) | (0.029) | |
| Post × PE_Ownership | −0.019 | 0.014 | −0.011 |
| (0.039) | (0.136) | (0.042) | |
| Post × PE_Ownership × InfoShare_High | −0.093* | −0.593** | −0.108* |
| (0.052) | (0.244) | (0.056) | |
| Observations | 4,691 | 4,691 | 4,691 |
| Number of hospitals | 375 | 375 | 375 |
| Adjusted R2 | 0.223 | 0.108 | 0.238 |
| Controls | Yes | Yes | Yes |
| Hospital FE | Yes | Yes | Yes |
| Year FE | Yes | Yes | Yes |
| Marginal effect of PE at low InfoShare | −0.019 | 0.014 | −0.011 |
| (0.039) | (0.136) | (0.042) | |
| Marginal effect of PE at high InfoShare | −0.112*** | −0.579** | −0.119*** |
| (0.035) | (0.204) | (0.036) | |
| Difference between high and low InfoShare (p-value) | 0.076 | 0.016 | 0.054 |
Note. Robust standard errors clustered at hospital level are in parentheses.
***p < 0.01; **p < 0.05; *p < 0.1.
6.3.2. How Information Sharing Moderates the Impact of PE Ownership on Care Quality.
Besides the role of health information sharing in strengthening the positive effect of PE ownership on cost reduction, we argue that it can help hospitals better absorb the negative effects of PE ownership on care quality. As discussed in Section 2.3, ancillary information refers to relatively discrete and episode-specific information generated through care processes, and it is typically collected and retrieved by nurses and medical assistants for physician use. Thus, hospitals with strong ancillary information sharing capabilities are better positioned to reduce the coordination burden borne by frontline staff. On the other hand, clinical information refers to longitudinal and integrative information which describes patients’ clinical histories and ongoing care management, enabling physicians to make more consistent treatment, discharge, and follow-up decisions. In other words, its role is to improve the completeness and continuity of clinical decision making. Given these distinctions, we expect these two types of information sharing may moderate PE’s impact on different care quality dimensions.
When hospitals possess strong capabilities for sharing ancillary information, they may be better able to absorb PE-related staffing pressures without deterioration in experiential quality. For example, they can leverage such capabilities to redesign workflows and enable frontline staff to spend less time on retrieving patient information and care coordination (Unertl et al. 2012). As a result, the remaining staff may still be able to support patient interactions with care providers. Therefore, we posit that stronger ancillary information sharing capability can attenuate the adverse effect of PE ownership on care quality dimensions related to patient experience with frontline staff.
Second, when hospitals possess strong capabilities in sharing clinical information, they can provide clinicians with more complete and timely information despite staffing constraints. This may allow physicians to continue to make diagnosis and treatment decisions based on a complete representation of patients (Smith et al. 2005, Walker et al. 2005, Chiedi 2019, Bao and Bardhan 2022, Janakiraman et al. 2023). Hence, clinical information sharing may be particularly effective in mitigating the negative impact of PE ownership on care quality dimensions that depend on information continuity across transitions in patient care, such as readmissions.
Motivated by these observations, we construct two binary indicators, Ancillary InfoShare_High and Clinical InfoShare_High. Ancillary InfoShare_High equals one if the focal hospital’s preexisting ancillary information sharing capability (i.e., one year prior to PE ownership) is above the sample mean, and zero otherwise. Similarly, Clinical InfoShare_High equals one if the hospital’s preexisting clinical information sharing capability is above the sample mean, and zero otherwise. We then use similar model specifications as in Equation (7), where is replaced by or , and the outcome variable is replaced by various dimensions of care quality.
Table 7 reports the estimation results based on these models. Column (1) indicates that, among hospitals with weak ancillary information sharing capabilities, PE ownership leads to a significant decline in experiential quality (coefficient = −1.260, p < 0.01). However, the estimated marginal effect of PE ownership is substantially smaller and statistically insignificant among hospitals with strong ancillary information sharing capabilities (coefficient = −0.238). Column (2) further shows that the difference between hospitals with weak versus strong ancillary information sharing capabilities becomes more pronounced when we focus on staff responsiveness as the outcome. This is plausible because unlike other measures of patient experience that proxy for the quality of staff-patient communication, staff responsiveness is a measure of timeliness. If PE firms leverage hospitals’ ancillary information sharing capabilities to reduce the time that staff spend on retrieving information and coordinating care, the benefits should be stronger for response time than communication quality. On the other hand, ancillary information sharing capabilities do not moderate the negative impact of PE ownership on clinical quality dimensions such as mortality rate and readmission rate, as shown in columns (3) and (4). This is consistent with the role of ancillary information sharing, which is particularly effective in supporting coordination-intensive work typically performed by frontline nurses and medical assistants.
|
Table 7. Moderating Impact of Ancillary and Clinical Information Sharing on Care Quality
| Dependent variable | (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) |
|---|---|---|---|---|---|---|---|---|
| Experiential quality | Staff responsiveness | Mortality rate | Readmission rate | Experiential quality | Staff responsiveness | Mortality rate | Readmission rate | |
| Post | 0.652** | 1.224** | 0.013 | 0.044 | 0.414 | 1.009** | −0.062 | −0.057 |
| (0.272) | (0.475) | (0.144) | (0.118) | (0.267) | (0.467) | (0.135) | (0.113) | |
| Post × PE_Ownership | −1.260*** | −1.598* | 0.169 | 0.199 | −1.031** | −1.398 | 0.327 | 0.416** |
| (0.471) | (0.856) | (0.251) | (0.200) | (0.489) | (0.884) | (0.242) | (0.177) | |
| Post × Ancillary InfoShare High | −0.801** | −1.644*** | −0.153 | −0.034 | ||||
| (0.335) | (0.596) | (0.176) | (0.149) | |||||
| Post × PE_Ownership × Ancillary InfoShare High | 1.023* | 2.301** | 0.216 | −0.286 | ||||
| (0.617) | (1.122) | (0.324) | (0.252) | |||||
| Post × Clinical InfoShare High | −0.372 | −1.212** | −0.039 | 0.196 | ||||
| (0.333) | (0.593) | (0.175) | (0.147) | |||||
| Post × PE_Ownership × Clinical InfoShare High | 0.676 | 1.957* | −0.070 | −0.653*** | ||||
| (0.627) | (1.139) | (0.324) | (0.240) | |||||
| Observations | 4,691 | 4,691 | 4,691 | 4,691 | 4,643 | 4,643 | 4,643 | 4,643 |
| Number of hospitals | 375 | 375 | 375 | 375 | 370 | 370 | 370 | 370 |
| Adjusted R2 | 0.314 | 0.179 | 0.499 | 0.562 | 0.317 | 0.179 | 0.500 | 0.563 |
| Controls | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Hospital FE | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Year FE | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Marginal effect of PE at low ancillary InfoShare | −1.260*** | −1.598* | 0.169 | 0.199 | ||||
| (0.471) | (0.856) | (0.251) | (0.200) | |||||
| Marginal effect of PE at high ancillary InfoShare | −0.238 | 0.702 | 0.386* | −0.088 | ||||
| (0.401) | (0.726) | (0.206) | (0.152) | |||||
| Difference between high and low ancillary InfoShare (p-value) | 0.099 | 0.041 | 0.505 | 0.256 | ||||
| Marginal effect of PE at low clinical InfoShare | −1.031** | −1.398 | 0.327 | 0.416** | ||||
| (0.489) | (0.884) | (0.242) | (0.177) | |||||
| Marginal effect of PE at high clinical InfoShare | −0.355 | 0.559 | 0.257 | −0.238 | ||||
| (0.394) | (0.718) | (0.218) | (0.160) | |||||
| Difference between high and low clinical InfoShare (p-value) | 0.282 | 0.086 | 0.828 | 0.007 |
Notes. Number of observations decreases from 4,691 to 4,643 in columns (5)–(8) because of missing values of variable Clinical InfoShare High. Robust standard errors clustered at hospital level are in parentheses.
***p < 0.01; **p < 0.05; *p < 0.01.
On the other hand, clinical information sharing plays an important role in mitigating the negative effect of PE ownership on clinical quality outcomes. In particular, column (8) shows that PE ownership has a significant, positive impact on readmission rates in hospitals with weak clinical information sharing capability (coefficient = 0.416, p < 0.05). However, in hospitals with strong clinical information sharing capability, the estimated marginal effect of PE ownership becomes statistically insignificant (coefficient = −0.238, p > 0.10). The difference is also negative and statistically significant (coefficient = −0.653, p < 0.01), suggesting that clinical information sharing capabilities are especially effective in supporting the knowledge-intensive work typically performed by physicians.10
7. Robustness Checks
7.1. Alternate DEA Inputs and Outputs
The baseline DEA model includes capital-related expenses associated with the acquisition, maintenance, and use of property, plant, and equipment dedicated to patient care. To ensure the robustness of our findings with respect to input selection, we also estimate an alternative DEA model using the number of staffed beds as a proxy for capital input (Kohl et al. 2019, Bardhan et al. 2023). Specifically, this specification includes operating expenses, number of staffed beds, and labor (including physicians, nurses, and other clinicians) as inputs. The results based on this alternate model are reported in Table C.4 in the Online Appendix. It shows consistent results with respect to the impact of PE ownership on care value, as well as on operating costs and quality-related outcomes.
7.2. Alternate Care Value Measures
Care value can be interpreted as a measure of cost-adjusted care quality. As a robustness check and alternative to DEA, we construct a simplified measure of care value based on standardized care quality outcomes and costs.11 Specifically, we define care value as standardized care quality minus standardized cost, where standardized cost is measured as the average of standardized operating costs and standardized capital costs. Next, we construct Cost-Adjusted Experiential Quality as standardized experiential quality minus standardized cost, with higher values indicating greater care value. Because higher mortality rates reflect poorer care quality, we define Cost-Adjusted Quality (Mortality) as the standardized inverse of the mortality rate minus standardized cost. Similarly, we construct Cost-Adjusted Quality (Readmission) as the standardized inverse of the readmission rate minus standardized cost. We then define an overall Cost-Adjusted Quality measure as the average of these three components. The results based on these alternative care value measures are reported in Table C.5 in the Online Appendix and are consistent with our main findings regarding the negative effect of PE ownership on care value. Further, as reported in Table C.6 in Online Appendix C, we observe consistent results regarding the moderating role of health information sharing in mitigating the negative impact of PE ownership on care value.
7.3. Instrumental Variable Estimation
A critical concern about model identification strategy in estimating Equation (7) is that omitted variables may be correlated with both hospital care value and the health information sharing capability of the focal hospital. For instance, hospitals that are owned by larger health systems may have better capability to share health information and be more likely to deliver higher care value. To address this endogeneity concern, we follow the approach adopted in the literature and use the extant level of information sharing of neighboring hospitals as the IV for (Lu et al. 2018). Specifically, based on the AHA data, we first identify the neighboring hospitals located in the same hospital referral region (HRR) for each focal hospital. We calculate the average level of health information sharing across neighboring hospitals, one year prior to PE ownership, denoted as , using a similar approach as described in Section 4.3. Then we use this average value as the IV for .
Miller and Tucker (2009) suggest a network effect where the adoption of health IT by one hospital is influenced by the adoption decisions of other hospitals within the same region. Dranove et al. (2014) highlight the role of shared local IT services in shaping health IT adoption patterns within local markets. Building on these insights, we argue that greater health information sharing at neighboring hospitals increases the likelihood of the focal hospital to also participate in health information sharing. In other words, this IV is valid and satisfies the relevance condition.
Furthermore, we argue that the exclusion restriction is satisfied for the following reasons. First, hospital care value is primarily influenced by its internal operations, including staffing, clinical processes, and resource utilization. Although neighboring hospitals’ adoption of health information sharing may influence broader regional trends, it is unlikely to directly affect patient flow or care delivery processes at the focal hospital. In other words, information sharing at neighboring hospitals is unlikely to have a direct impact on patient care at the focal hospital.
We conduct several falsification tests to examine the association between our IV and care value, patient flow, and staffing levels at the focal hospital. The results of exogeneity tests are reported in Table C.7 of Online Appendix C. We observe that the IV is not directly associated with care value, number of inpatient visits, inpatient days, or labor at the focal hospital. This suggests that information sharing of neighboring hospitals is unlikely to be correlated with changes in the focal hospital’s operations that may impact care value. Overall, our IV satisfies both relevance and exogeneity conditions.
Based on this IV, we implement a two-stage least-square (2SLS) regression to estimate the heterogeneous effect of PE ownership on care value for hospitals with different levels of InfoShare. The result is consistent with our baseline and reported in column (1) of Table C.8 in the Online Appendix. To further test the robustness of our results, we also employ an alternative IV to estimate the moderating effect of health information sharing. Specifically, we use the strength of interhospital IT capability of neighboring hospitals in the same HRR, denoted as Neighbor Interhospital IT, as an alternate IV. The results, as reported in column (2) of Table C.8, are also consistent with our baseline.
7.4. Moderating Impact of Intrahospital IT
A potential concern is that the moderating effect attributed to health information sharing may instead reflect the intrahospital IT capabilities of PE-owned hospitals. Intrahospital IT systems can improve hospital performance by automating workflows, standardizing internal processes, and supporting clinical documentation (Atasoy et al. 2019). These capabilities may help hospitals operate more efficiently and coordinate care effectively. Further, intrahospital IT systems, such as EHRs, often serve as the foundation for health information sharing, making the two capabilities potentially correlated. However, their direct benefits are primarily realized within the focal hospital (Walker et al. 2005, Atasoy et al. 2019). Intrahospital IT alone does not fully resolve information gaps and reduce coordination gaps that arise when patients receive fragmented care from multiple providers (Smith et al. 2005). As a result, intrahospital IT capabilities may not play similarly dual roles in strengthening PE’s positive effect on cost reduction and mitigating PE’s negative effect on care quality.
To empirically support this argument, we examine whether intrahospital IT plays a similar moderating role in shaping PE’s impact on care value. As reported in Table C.9 in the Online Appendix, the interaction term Post × PE_Ownership × Intrahospital IT High is statistically insignificant across care value, operating cost, experiential quality, mortality, and readmission outcomes. The results suggest that the moderating role documented in our main analyses is not simply driven by stronger intrahospital IT capabilities.
7.5. Alternative Explanation: Potential Changes in Patient Volume
Another potential concern is that the observed changes in operating costs following PE ownership in hospitals may be confounded by changes in patient volume. Specifically, PE-owned hospitals may admit fewer patients to improve outcomes. To assess this possibility, we examine changes in patient flow by using the number of inpatient visits and inpatient days as dependent variables in Equations (2) and (7). Table C.10 in the Online Appendix first reports the estimation results based on Equation (2), followed by the results for Equation (7) in columns (3) and (4). The coefficient estimate for Post × PE_Ownership is statistically insignificant using both log of total inpatient visits and log of total inpatient days as the dependent variable. This suggests that PE ownership does not seem to influence the overall patient flow or inpatient resources expended. It is consistent with prior research which shows that hospital choice is predominantly influenced by geographic proximity and insurance coverage, rather than ownership status (Victoor et al. 2012, Smith et al. 2018). The estimates of Post × PE_Ownership × InfoShare_High in columns (3) and (4) also indicate that the moderating effect of health information sharing is insignificant.
Table C.11 in the Online Appendix provides a summary of potential concerns and analyses addressed by our robustness checks.
8. Discussion
We study the impact of PE ownership on the value of care delivered by U.S. hospitals. Further, we evaluate the role of health information sharing in moderating such a relationship. Our findings make several contributions to the literature. First, our research contributes to the literature on the impact of PE in healthcare by shifting the focus from isolated outcomes to a holistic evaluation metric, that is, care value—a concept that emphasizes the delivery of high-quality and cost-effective care. This approach unites stakeholders’ interests and is especially relevant given the current shift toward VBC models, where the emphasis is on delivering efficient and effective care (Porter 2010, Bardhan et al. 2023). Our findings indicate that although restructuring efforts following PE ownership may yield short-term cost savings through reduction in clinical staffing, they often come at the expense of reduced care quality, leading to lower healthcare value.
Second, we add to the information systems literature by highlighting the unique role of IT-enabled health information sharing in mitigating the decline in care value faced by PE-owned hospitals. We show that PE-owned hospitals with strong preexisting, health information sharing capabilities could reduce operating costs to a greater extent while maintaining care quality, compared with PE-owned hospitals with weak information sharing capabilities. Further, although previous studies have examined the role of information sharing in improving healthcare efficiency and quality (Bao and Bardhan 2022, Janakiraman et al. 2023), how different types of information sharing might moderate the impact of PE on care value remains an underexplored area of research. We address this gap by differentiating the roles of ancillary and clinical information sharing in the context of PE ownership. Specifically, we find that, among hospitals with strong ancillary information sharing capabilities, PE firms can better preserve experiential quality by maintaining frontline coordination and responsiveness with leaner staffing levels. In contrast, clinical information sharing mitigates PE’s negative impact on care quality dimensions that depend on information continuity, such as readmission rates.
Third, our research is related to the broader literature that studies how structural changes affect hospital operations and patient outcomes. Prior studies have analyzed the impact of regulatory mandates (such as hospital price transparency rules) and market coordination mechanisms (such as referral and interorganizational care coordination) on patient choice and care patterns (Christensen et al. 2020, Eftekhari et al. 2023). Our study adds to this discussion by highlighting how PE ownership, as a unique type of organizational governance change, interacts with hospital IT-enabled information sharing capabilities to shape the efficiency and quality dimensions of hospital performance.
Our research also has important implications for various stakeholders. For healthcare policymakers, our research highlights the importance of a holistic approach when evaluating the impact of PE ownership in healthcare. For hospital administrators, our findings suggest the importance of health information sharing capabilities in mitigating the potential negative effect of PE ownership. Administrators in PE-owned hospitals may benefit from proactively strengthening such capabilities, by adopting health IT data standards such as Fast Healthcare Interoperability Resources (FHIR), to support operational efficiency and clinical care coordination.
For PE firms, maintaining high care quality can strengthen their portfolio hospitals’ market share because patients are sensitive to hospital care quality indicators. This is consistent with their objective to increase the value of portfolio hospitals before pursuing an exit (Richards and Whaley 2024). Our research suggests that investing in hospitals with robust health information sharing capabilities enables PE firms to implement effective cost reduction while preserving care quality. This alignment between financial and clinical outcomes can enhance both short-term performance and long-term value creation.
For patients, our research offers evidence that PE ownership does not inherently diminish care value, particularly when supported by strong health information sharing capabilities. This implies that receiving care at PE-owned hospitals with such capabilities may lead to better care outcomes compared with those without such capabilities. At the same time, these insights highlight the importance of continued oversight to ensure that PE restructuring strategies do not compromise patient-centered care.
9. Conclusions
What is the impact of PE ownership on care value, and how can health IT mitigate the potential negative effects of PE ownership? Our work is among the first to answer these questions, focusing specifically on the role of IT-enabled health information sharing. We observe that PE ownership of U.S. hospitals leads to lower care value, as measured by the degree to which a hospital utilizes its resources efficiently in the production of healthcare services that enhance patient health outcomes. We identify reduction in clinical staffing as one important mechanism that explains reduction in hospital operating costs and care quality following PE acquisition. We also observe that health information sharing across providers can moderate this relationship. Specifically, strong capabilities for health information sharing (with other care providers) attenuate the negative impact of PE ownership on care value. Further, we demonstrate the robustness of our results using a range of methods, such as instrumental variable estimation, falsification tests, and alternative measures of care value.
Our research highlights the importance of information sharing as a strategic capability in healthcare. Hospitals that invest in stronger information sharing infrastructure are better positioned to improve care value, even in the face of ownership changes and financial pressures associated with PE ownership. Nevertheless, our study has several limitations. First, our analysis focuses exclusively on PE ownership in general and acute care hospitals. Although hospitals represent a significant portion of PE activity in the healthcare sector, these investments extend far beyond hospital buyouts and acquisitions. PE firms are also increasingly active in other types of care delivery organizations, such as urgent care clinics, nursing homes, rehabilitation centers, specialty hospitals, and dental service organizations (Borsa et al. 2023). These settings represent unique operational challenges, regulatory environments, and patient demographics, which may influence the impact of PE in different ways. Future research can evaluate the effects of PE ownership across various care settings to provide a more comprehensive understanding across the healthcare continuum. Meanwhile, our findings should be interpreted within the context of majority-control, LBO-style PE acquisitions under typical PE investment horizons. Because the mechanisms we theorize rely on the governance change associated with majority ownership, our results may not generalize to other forms of PE involvement, such as minority investments. An important direction for future research is to examine whether and how different ownership structures and governance arrangements influence the impact of PE on healthcare value.
Second, significant barriers remain in achieving high levels of information sharing across healthcare organizations, such as technology incompatibility, lack of common data standards, and concerns over data privacy and security (Li et al. 2023). Future research could explore other dimensions of information sharing, such as adoption of interoperability standards, which have the potential to drive high-quality care and mitigate the negative effects of PE ownership.
Third, our findings regarding the heterogenous effects associated with PE firm experience also provide important insights into the longer-term implications of PE ownership. We find that less experienced PE firms achieve greater cost reduction, but these reductions come with larger declines in care quality and care value. In contrast, more-experienced PE firms exhibit smaller cost reductions and less deterioration in care quality. One possible interpretation is that, as PE firms gain experience, they learn the limits of aggressive cost cutting and may become more selective in reducing resources to preserve care quality. Thus, aggressive cost reduction may not represent a sustainable operating approach as PE firms gain experience. However, because our analysis compares firms with different levels of experience rather than tracking how the same firms change their strategies as they gain experience, future research is needed to examine this learning process and the long-term sustainability of PE-induced cost reductions. PE firms may also differ in their strategic priorities, such as AI adoption, digital transformation, or service-line expansion. Future research can further investigate their heterogeneous effects on care value based on the organizational characteristics of PE firms and acquired hospitals. Meanwhile, to ensure greater focus in the scope of our study, we restrict our locus of attention to the impact of PE ownership on care value. An interesting direction for future research is to explore how PE ownership shapes IT adoption decisions in the acquired hospitals.
Considering the observational nature of our data, the endogeneity issues cannot be fully addressed. Additional studies, particularly using alternative data sources or non-U.S. healthcare settings, are necessary to validate and extend our findings. Despite these limitations, our research represents an important step toward evaluating the impact of PE ownership on care value and proposing potential solutions to mitigate its potential negative impact. To the best of our knowledge, ours is one of the first studies to evaluate the role of health IT in the context of PE ownership and its impact on value-based care delivery in hospitals. Our study can help safeguard the quality of care delivered, ultimately benefiting patients, providers, and investors alike. Our research can thereby serve as a catalyst to open new avenues for future research at the intersection of healthcare finance and information technology.
The authors gratefully acknowledge the constructive feedback from the senior editor, associate editor, and reviewers. The authors thank seminar participants at UT Austin, University of Miami, University of Virginia, National University of Singapore, and Florida International University for their helpful feedback. Earlier versions were presented at the Conference on Health IT and Analytics, INFORMS Annual Meetings, International Conference on Information Systems, and the Workshop on Information Systems and Economics.
1 Hospital IT capabilities also include internal IT capabilities that support within-hospital functions such as electronic clinical documentation and computerized provider order entry. However, the benefits of internal IT are often constrained if information cannot be shared across organizational boundaries (Walker et al. 2005, Atasoy et al. 2019). Hence, in our empirical analyses, our focus is on the impact of interhospital information sharing after controlling for hospitals’ internal IT capabilities.
2 We test the robustness of the results using bootstrapped DEA, which yields consistent estimates (Simar and Wilson 1998).
3 Patient satisfaction data are drawn from the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey in the Hospital Compare database, and include measures such as nurse communication, doctor communication, staff responsiveness, communication about medications, and communication about postdischarge instructions.
4 Although we include the term in Equation (2), its effect is absorbed by the hospital fixed effect . As a result, it is automatically dropped from the regression, and we are unable to estimate .
5 We also implemented 1:1 coarsened exact matching (CEM) using the same set of matching variables (Iacus et al. 2012), resulting in 109 treated hospitals matched to 109 controls. The estimated effects of PE ownership on care value and its underlying effects on cost and care quality are consistent with our baseline results.
6 We use log transformation of operating costs to reduce its skewness in all analyses.
7 As in Equation (2), the term is dropped because of the hospital fixed effect .
8 The results based on yearly measurement of and are consistent and available upon request.
9 The terms , and in Equation (7) are dropped because of the hospital fixed effect . Therefore, , or in Equation (7) is not reported.
10 We note that these two pathways are not outcome-exclusive. Ancillary information sharing may also influence clinical outcomes, and clinical information sharing can affect patient experience as demonstrated by its marginally positive association with staff responsiveness. Rather, we identify the outcomes for which each information sharing capability provides the most effective buffering mechanism under PE-induced staffing pressure.
11 Each standardized variable z is calculated as , where is the original value, is the sample mean, and is the sample standard deviation.
References
- (2018) Reducing Medicare spending through electronic health information exchange: The role of incentives and exchange maturity. Inform. Systems Res. 29(2):341–361.Link, Google Scholar
- (2025) The (going) public option: Equity market financing in the hospital industry. Preprint, submitted April 2, https://doi.org/10.2139/ssrn.5177973.Google Scholar
- (2002) Hospital nurse staffing and patient mortality, nurse burnout, and job dissatisfaction. J. Amer. Medical Assoc. 288(16):1987–1993.Crossref, Google Scholar
American Hospital Association (2024) America’s hospitals and health systems continue to face escalating operational costs and economic pressures as they care for patients and communities. Accessed October 12, 2026, https://www.aha.org/costsofcaring.Google ScholarAmerican Medical Association (2021) Key considerations in providing ancillary services in your physician practice. Accessed October 11, 2026, https://www.ama-assn.org/system/files/private-practice-checklist-ancillary-services.pdf.Google Scholar- (2019) The digitization of patient care: A review of the effects of electronic health records on health care quality and utilization. Annual Rev. Public Health 40:487–500.Crossref, Google Scholar
- (2003) The skill content of recent technological change: An empirical exploration. Quart. J. Econom. 118(4):1279–1333. Google Scholar
- (2017) The impact of health information sharing on duplicate testing. MIS Quart. 41(4):1083–1104.Crossref, Google Scholar
- (2022) How much should we trust staggered difference-in-differences estimates? J. Financial Econom. 144(2):370–395.Crossref, Google Scholar
- (1984) Some models for estimating technical and scale inefficiencies in data envelopment analysis. Management Sci. 30(9):1078–1092.Link, Google Scholar
- (2022) Performance of accountable care organizations: Health information technology and quality–efficiency trade-offs. Inform. Systems Res. 33(2):697–717.Link, Google Scholar
- (2023) Value implications of sourcing electronic health records: The role of physician practice integration. Inform. Systems Res. 34(3):1169–1190.Link, Google Scholar
- (2008) The triple aim: Care, health, and cost. Health Affairs 27(3):759–769.Crossref, Google Scholar
- (2023) Evaluating trends in private equity ownership and impacts on health outcomes, costs, and quality: Systematic review. BMJ 382:1–17.Crossref, Google Scholar
- (2021) Association of private equity investment in US nursing homes with the quality and cost of care for long-stay residents. JAMA Health Forum 2(11):1–13.Crossref, Google Scholar
- (2020) Changes in hospital income, use, and quality associated with private equity acquisition. JAMA Internal Medicine 180(11):1428–1435.Crossref, Google Scholar
- (2021) Characteristics of private equity-owned hospitals in 2018. Ann. Internal Medicine 174(2):277–279.Crossref, Google Scholar
- (2021) Difference-in-differences with multiple time periods. J. Econometrics 225(2):200–230.Crossref, Google Scholar
- (2022) Association between hospital private equity acquisition and outcomes of acute medical conditions among Medicare beneficiaries. JAMA Network Open 5(4):1–13.Crossref, Google Scholar
- (2018) Double/debiased machine learning for treatment and structural parameters. Econometrics J. 21(1):C1–C68.Crossref, Google Scholar
- (2019) ACOs’ strategies for transitioning to value-based care: Lessons from the Medicare shared savings program. Office of Inspector General, U.S. Department of Health and Human Services, Washington, DC.Google Scholar
- (2024) Impact of a nationwide medication history sharing program on the care process and end-user experience in a tertiary teaching hospital: Cohort study and cross-sectional study. JMIR Medical Informatics 12(1):1–10.Google Scholar
- (2020) The only prescription is transparency: The effect of charge-price-transparency regulation on healthcare prices. Management Sci. 66(7):2861–2882.Link, Google Scholar
- (2007) The interaction of institutionally triggered and technology-triggered social structure change: An investigation of computerized physician order entry. MIS Quart. 31(4):739–758.Crossref, Google Scholar
- (2014) Private equity, jobs, and productivity. Amer. Econom. Rev. 104(12):3956–3990.Crossref, Google Scholar
- (1995) The stakeholder theory of the corporation: Concepts, evidence, and implications. Acad. Management Rev. 20(1):65–91.Crossref, Google Scholar
- (2014) The trillion dollar conundrum: Complementarities and health information technology. Amer. Econom. J. Econom. Policy 6(4):239–270.Crossref, Google Scholar
- (2023) Impact of health information exchange adoption on referral patterns. Management Sci. 69(3):1615–1638.Link, Google Scholar
- (2001) A stakeholder approach to strategic management. Preprint, submitted March 16, https://doi.org/10.2139/ssrn.263511.Google Scholar
- (2022) Two-stage differences in differences. Preprint, submitted July 13, https://arxiv.org/abs/2207.05943.Google Scholar
- (2005) HIT and MIS: Implications of health information technology and medical information systems. Comm. ACM 48(10):68–74.Crossref, Google Scholar
- (2024) Owner incentives and performance in healthcare: Private equity investment in nursing homes. Rev. Financial Stud. 37(4):1029–1077.Crossref, Google Scholar
- (2024) FTC and DOJ signal greatly increased scrutiny of private equity firms’ acquisitions in health care. Polsinelli, Accessed October 13, 2026, https://www.polsinelli.com/publications/ftc-and-doj-signal-greatly-increased-scrutiny-of-private-equity-firms-acquisitions-in-health-care.Google Scholar
- (1979) Sample selection bias as a specification error. Econometrica 47(1):153–161.Crossref, Google Scholar
- (1998) Matching as an econometric evaluation estimator. Rev. Econom. Stud. 65(2):261–294.Crossref, Google Scholar
- (2022) Hospital service offerings still differ substantially by ownership type. Health Affairs 41(3):331–340.Crossref, Google Scholar
- (2012) Causal inference without balance checking: Coarsened exact matching. Political Anal. 20(1):1–24. Crossref, Google Scholar
- (2023) Healthcare private equity market 2022: The year in review. Bain & Company, Accessed October 12, 2026, https://www.bain.com/insights/year-in-review-global-healthcare-private-equity-and-ma-report-2023/.Google Scholar
- (2023) The effects of health information exchange access on healthcare quality and efficiency: An empirical investigation. Management Sci. 69(2):791–811.Link, Google Scholar
- (2009) Leveraged buyouts and private equity. J. Econom. Perspect. 23(1):121–146.Crossref, Google Scholar
- (2012) Information technology and intangible output: The impact of IT investment on innovation productivity. Inform. Systems Res. 23(1):42–59.Link, Google Scholar
- (2019) The use of Data Envelopment Analysis (DEA) in healthcare with a focus on hospitals. Health Care Management Sci. 22(2):245–286.Crossref, Google Scholar
- (2019) Dealing with numbers: Nurses informing doctors and patients about test results. Discourse Stud. 21(2):180–198.Crossref, Google Scholar
- (2014) Does health information exchange reduce redundant imaging? Evidence from emergency departments. Medical Care 52(3):227–234.Crossref, Google Scholar
- (1999) Pharmacist participation on physician rounds and adverse drug events in the intensive care unit. J. Amer. Medical Assoc. 282(3):267–270.Crossref, Google Scholar
- (2023) Successes and barriers of health information exchange participation across hospitals in South Carolina from 2014 to 2020: Longitudinal observational study. JMIR Medical Informatics 11(1):1–11.Google Scholar
- (2022) Bargaining with private equity: Implications for hospital prices and patient welfare. Preprint, submitted November 1, https://doi.org/10.2139/ssrn.3896410.Google Scholar
- (2018) Does technology substitute for nurses? Staffing decisions in nursing homes. Management Sci. 64(4):1842–1859.Link, Google Scholar
- (2013) The patient experience and health outcomes. New England J. Medicine 368(3):201–203.Crossref, Google Scholar
- (2000) Saving money, saving lives. Harvard Bus. Rev. 78(6):57–67.Google Scholar
- (2004) Review: Information technology and organizational performance: An integrative model of IT business value. MIS Quart. 28(2):283–322.Crossref, Google Scholar
- (2009) Agency, strategic entrepreneurship, and the performance of private equity-backed buyouts. Entrepreneurship Theory Practice 33(1):213–239.Crossref, Google Scholar
- (2009) Privacy protection and technology diffusion: The case of electronic medical records. Management Sci. 55(7):1077–1093.Link, Google Scholar
- (2017) Information technology, revenues, and profits: Exploring the role of foreign and domestic operations. Inform. Systems Res. 28(2):430–444.Link, Google Scholar
- (2002) Nurse-staffing levels and the quality of care in hospitals. New England J. Medicine 346(22):1715–1722.Crossref, Google Scholar
- (2021) Private equity investments in health care: An overview of hospital and health system leveraged buyouts, 2003–17. Health Affairs 40(5):719–726.Crossref, Google Scholar
- (1993) The cornerstones of competitive advantage: A resource-based view. Strategic Management J. 14(3):179–191.Crossref, Google Scholar
- (2010) What is value in health care? New England J. Medicine 363(26):2477–2481.Crossref, Google Scholar
- (2011) How to solve the cost crisis in health care. Harvard Bus. Rev. 89(9):46–52.Google Scholar
- (2006) Redefining Health Care: Creating Value-Based Competition on Results (Harvard Business Press, Boston).Google Scholar
- (2021) Private equity and health care delivery: Value-based payment as a guardrail? J. Amer. Medical Assoc. 326(10):907–908.Crossref, Google Scholar
- (2024) Business value of information technology capabilities: An institutional governance perspective. Inform. Systems Res. 35(1):28–44.Link, Google Scholar
- (2010) Leveraging IT capabilities and competitive process capabilities for the management of interorganizational relationship portfolios. Inform. Systems Res. 21(3):516–542.Link, Google Scholar
- (2024) Hospital behavior over the private equity life cycle. J. Health Econom. 97:1–27.Crossref, Google Scholar
- (2008) Private equity investment in health care services. Health Affairs 27(5):1389–1398.Crossref, Google Scholar
- (2016) The impact of combining conformance and experiential quality on hospitals’ readmissions and cost performance. Management Sci. 62(3):829–848.Link, Google Scholar
- (2019) Waste in the US health care system: Estimated costs and potential for savings. J. Amer. Medical Assoc. 322(15):1501–1509.Crossref, Google Scholar
- (1998) Sensitivity analysis of efficiency scores: How to bootstrap in nonparametric frontier models. Management Sci. 44(1):49–61.Link, Google Scholar
- (2022) Association of private equity acquisition of physician practices with changes in health care spending and utilization. JAMA Health Forum 3(9):1–12.Crossref, Google Scholar
- (2018) Patient choice modelling: How do patients choose their hospitals? Health Care Management Sci. 21(2):259–268.Crossref, Google Scholar
- (2005) Missing clinical information during primary care visits. J. Amer. Medical Assoc. 293(5):565–571.Crossref, Google Scholar
- (2012) Changes in hospital efficiency after privatization. Health Care Management Sci. 15(4):310–326.Crossref, Google Scholar
- (2018) Value-based health care meets cost-effectiveness analysis. Ann. Internal Medicine 169(5):329–332.Crossref, Google Scholar
- (2012) Health information exchange technology on the front lines of healthcare: Workflow factors and patterns of use. J. Amer. Medical Informatics Assoc. 19(3):392–400.Crossref, Google Scholar
- (2008) Hospital quality, efficiency, and input slack differentials. Health Services Res. 43(5p2):1830–1848.Crossref, Google Scholar
- (2012) Determinants of patient choice of healthcare providers: A scoping review. BMC Health Services Res. 12(1):1–16.Crossref, Google Scholar
- (2018) Does participation in health information exchange improve hospital efficiency? Health Care Management Sci. 21(3):426–438.Crossref, Google Scholar
- (2005) The value of health care information exchange and interoperability. Health Affairs 24(Suppl1):W5-10–W5-18.Crossref, Google Scholar
- (2010) Econometric Analysis of Cross Section and Panel Data, 2nd ed. (MIT Press, Cambridge, MA).Google Scholar
- (2015) An empirical analysis of the financial benefits of health information exchange in emergency departments. J. Amer. Medical Informatics Assoc. 22(6):1169–1172.Crossref, Google Scholar
Minghong Yuan is a PhD candidate in the Department of Information, Risk, and Operations Management at the McCombs School of Business, the University of Texas at Austin. His research lies at the intersection of information systems, AI, and healthcare, with a particular interest in how digital technologies and AI influence physician decision making, hospital performance, and healthcare value. He expects to receive his PhD in 2027.
Wen Wen is an associate professor in the Department of Information, Risk, and Operations Management at the McCombs School of Business at the University of Texas at Austin. She received her PhD from Georgia Institute of Technology. Her research has been published in leading academic journals, including Information Systems Research, MIS Quarterly, and Management Science.
Indranil Bardhan is the Charles and Elizabeth Prothro Regents Chair in Healthcare Management in the McCombs School of Business at the University of Texas at Austin. His research focuses on the economic and clinical impact of information technologies, and development of novel analytics solutions to improve health outcomes and value. Dr. Bardhan holds a PhD in management science and information systems from the McCombs School of Business at the University of Texas at Austin.

