The ownership relationships required to meet our definition of “vertically integrated” were identified using a range of data sources, including AHA data, practice surveys, integrated delivery/health system “top lists,” and Medicare or other health plan claims. Studies evaluated a heterogeneous set of systems, patients, and outcomes, which prohibited meta-analysis. Instead, we produced a descriptive summary of studies by outcome domains. Given the volume of results, we considered p < .05 to be statistically significant unless otherwise noted. When considering the evidence for an outcome, we considered the precision of the findings (including p > .05) together with the magnitude and consistency of the associations within and across studies.
Most study designs were assessed as “low quality” because they reported associations between health systems and performance indicators without addressing the selection of hospitals or practices into integrated configurations. Other studies were assessed as “moderate quality” because they examined within-hospital variation or used other quasiexperimental techniques such as a difference-in-differences. Four studies were downgraded because of “indirectness of evidence.” In most cases, these studies assessed vertical integration at the market level (Baker, Bundorf, & Kessler, 2014 ; Herrel et al., 2017 ; Neprash, Chernew, Hicks, Gibson, & McWilliams, 2015); one study measured vertical integration as the percentage of health plan care delivered by group or staff model systems (Gillies, Chenok, Shortell, Pawlson, & Wimbush, 2006).
In addition, one study was downgraded due to failure to adequately control for confounding factors (Liepert, Bledsoe, Stevens, & Cochran, 2014), and two studies were downgraded because they used proxy measures for our study definition of vertical integration (Crespin, Christianson, McCullough, & Finch, 2016 ; Weeks et al., 2010). Specifically, Weeks et al. (2010) evaluated the performance of the Council of Accountable Physician Practices (CAPP), of which the majority but not all practices met our definition of vertical integration, and Crespin et al. (2016) evaluated the performance of acquired clinics but addressed the dual issues of being acquired and changes to their size postacquisition (many of the acquired clinics came from smaller health systems).
Studies Examining Quality of Care
Ten studies included in this review examined the relationship between vertical integration and one or more quality-of-care measures (Table 2). Six of these studies showed that, compared with nonintegrated systems, vertically integrated health systems had a favorable relationship with quality of care (Crespin et al., 2016 ; Gillies et al., 2006 ; Herrel et al., 2017 ; Leibert, 2011 ; Rhoads, Patel, Ma, & Schmidt, 2015 ; Weeks et al., 2010). Studies used a range of measures to assess quality, with most studies examining optimal care for specific conditions or cancer screening. Other quality measures included rates of immunization and smoking cessation. In general, studies examined quality of care among Medicare beneficiaries in ambulatory care settings.
Six studies showed a favorable relationship between vertical integration and optimal care for specific conditions (often diabetes) in a range of health system configurations that met our inclusion criteria (Crespin et al., 2016 ; Gillies et al., 2006 ; Herrel et al., 2017 ; Leibert, 2011 ; Rhoads et al., 2015 ; Weeks et al., 2010). Crespin et al. (2016) examined 100 clinics acquired by health systems mostly in Minnesota between 2007 and 2013 and assessed their diabetes care performance (a composite measure of five treatment goals, including hemoglobin A1c, blood pressure, and low-density lipoprotein cholesterol). The authors found no significant difference in overall diabetes care between acquired and nonacquired clinics until the third year after acquisition. By their fifth year, acquired clinics scored 3.6 percentage points higher in diabetes care than if they had never been acquired.
Among the other studies, Leibert (2011) examined quality at vertically integrated health systems defined as “flagship hospitals” in a top 50 list of integrated health systems, comparing these hospitals with 50 matched control hospitals. The author found higher disease-specific clinical quality for patients with heart attack, heart failure, and pneumonia in vertically integrated systems. Rhoads et al. (2015) examined the delivery of evidence-based care to colorectal cancer patients treated at health systems that were “integrated,” as defined by the Berkeley Forum to Improve California’s Healthcare Delivery System; under this definition, “one organization is responsible for all services, including delivery of care, payment, and risk management” (p. 855). The authors found that treatment at vertically integrated health systems was associated with better adherence to the National Comprehensive Cancer Network’s guidelines on two of three quality measures for colon cancer (no difference for the third measure).
Two studies used indirect measures to define vertical integration (Gillies et al., 2006 ; Herrel et al., 2017), and one examined group practices, in which most but not all the practices met our definition (Weeks et al., 2010). Gillies et al. (2006) found that health plans served by a higher percentage of group or staff models were associated with higher-quality diabetes care. Herrel et al. (2017) analyzed adherence to 12 guidelines for prostate cancer treatment in hospital referral regions that had at least 90% of discharges from a top integrated delivery system, based on Becker’s Hospital Review. Nationally, Herrel et al. (2017) identified three markets that were vertically integrated and found that patients treated in those markets were more likely to completely adhere to their suite of prostate cancer quality measures than those in markets with no vertical integration (odds ratio: 1.38). Weeks et al. (2010) found that Medicare patients assigned to CAPP-affiliated physicians (most of which belong to vertically integrated health systems) received more evidence-based care than those in the same market who were not assigned to CAPP physicians (relative risks [RRs] ranged from 1.05 to 1.08 for hemoglobin A1c, lipid tests, and retinal exams for diabetic patients).
Three studies reported a mixed, small negative, or not statistically significant relationship between vertical integration and optimal care for specific conditions (Kralewski, Dowd, Knutson, Tong, & Savage, 2015 ; Kralewski, Dowd, Savage, & Tong, 2014 ; McWilliams, Chernew, Zasalavsky, Hamed, & Landon, 2013). McWilliams et al. (2013) compared the performance of hospital-based groups to medium and large independent provider groups. Authors defined groups as “hospital based” if 10% or more of a group’s practice sites were at hospitals, which included 94% of groups with at least one hospital location. Compared with small independent groups, hospital-based groups had lower rates of lipid testing for patients with cardiovascular disease (by 0.7 percentage points) but higher rates of three diabetes care services (by 0.8 percentage points). Kralewski et al. (2015) examined differences between physician-owned and hospital-owned practices and found that patients at hospital-owned practices had slightly lower rates of hemoglobin A1c screening for patients with diabetes and lipid screening for patients with cardiovascular disease. The results in Kralewski et al. (2014) were similar.
Six studies examined the impact of vertical integration on cancer screening rates, with varying results (Carlin, Dowd, & Feldman, 2015 ; Gillies et al., 2006 ; Kralewski et al., 2014, 2015 ; McWilliams et al., 2013 ; Weeks et al., 2010). Three of the studies showed a favorable relationship between vertical integration and cancer screening, but two used proxy or indirect measures of vertical integration (Gillies et al., 2006 ; Weeks et al., 2010). McWilliams et al. (2013) found a 2.8 percentage point higher rate of breast cancer screening for hospital-based groups compared with small independent physician groups. Similarly, Weeks et al. (2010) found higher rates of breast cancer screening for CAPP-affiliated physicians (RR = 1.12), and Gillies et al. (2006) found the percentage of health plans served by group or staff models was associated with higher-quality women’s health screening (a composite measure of breast cancer, cervical cancer, and chlamydia screening).
Carlin et al. (2015) examined changes in cancer screening rates after three stand-alone clinic systems were acquired by two local integrated delivery systems in Minnesota. After the acquisition, two of the clinics had higher rates of screening for colorectal and cervical cancer, but changes in breast cancer screening were mixed (one clinic improved, one declined, and one saw no statistically significant changes). Kralewski et al. (2015) found slightly lower rates of colon cancer screening (p < .10) for hospital-owned versus physician-owned practices. No statistically significant differences were found by Kralewski et al. (2014) for cervical cancer screening.
Studies Examining Efficiency
Twenty-four studies in the review examined the relationship between vertical integration and one or more measures of efficiency (Table 2). Most of these studies (n = 20) showed that vertical integration was associated with lower efficiency or no statistically significant differences in efficiency. Utilization measures focused on ambulatory care-sensitive (ACS) admissions, rates of emergency department (ED) use, and readmissions. Other studies focused on spending, cost inefficiency, or prices. Studies that examined prices generally focused on discounted patient charges (actual charges multiplied by the yearly average discount rate that each hospital grants to private paying patients; Ciliberto & Dranove, 2006). Thirteen studies focused on practices or facilities nationally; the remaining articles focused on smaller regional areas. Whereas quality of care was often measured in ambulatory care settings, efficiency was often measured in hospitals.
Six studies concluded that vertically integrated health systems were associated with lower efficiency, as measured by higher health care utilization (Casalino et al., 2014 ; Henke et al., 2018 ; Koch, Wendling, & Wilson, 2017 ; Kralewski, Dowd, Knutson, Savage, & Tong, 2013 ; Madison, 2004 ; McWilliams et al., 2013). Koch et al. (2017) examined 27 large physician groups that were acquired by hospital systems and found vertical integration was associated with significant changes in acquired physicians’ behavior, with shifts toward physicians delivering more care in their acquiring system’s hospital. On average, the authors found a net increase in utilization for acquiring hospitals after the mergers (the increased utilization of acquiring hospitals’ outpatient departments by acquired physicians outweighed the reductions in utilization by other clinicians; p < .10). McWilliams et al. (2013) found that, compared with smaller independent groups, hospital-based groups were associated with a 1.3 percentage point higher 30-day readmission rate. Casalino et al. (2014) examined hospital-owned versus physician-owned practices and found that hospital-owned practices had higher rates of ACS admissions (5.3 vs. 4.6 ACS admissions per 100 patients) when the authors controlled for a variety of attributes (including pay for performance, public reporting, and patient-centered medical home scores). Kralewski et al. (2013) found a similar relationship; hospital-owned practices had higher utilization rates for nonemergent ED use, ED use for conditions treatable in primary care, and ACS admissions compared with physician-owned practices. Madison (2004) compared hospitals with and without integrated salary models, finding that heart attack patients admitted to hospitals with integrated salary models (a form of hospital–physician affiliation) had higher procedure rates. Finally, Henke et al. (2018) used the AHA health system cluster and found that CHSs and CPIHSs had slightly longer lengths of stay (4.09 days vs. 4.08 days) and higher cesarean section probability (30.2% vs. 29.6%) compared with noncentralized system hospitals.
Seven studies produced findings on vertical integration and health care utilization that were mixed or not statistically significant (Baker et al., 2014 ; Carlin et al., 2015 ; Cuellar & Gertler, 2006 ; Kralewski, Dowd, & Xu, 2012 ; Kralewski et al., 2015 ; Scott et al., 2017). Scott et al. (2017) examined hospitals that switched to an employment relationship (foundation, equity, or group staff models) with their privileged physicians; the authors found that, of eight measures of readmissions and length of stay, seven of the associations were not statistically significant, and one was statistically significant but small (switched hospitals had a 0.6 percentage point lower readmission rate for pneumonia). Carlin et al. (2015) found mixed findings on appropriate health care use and general utilization among three clinics acquired by an integrated delivery system. Cuellar and Gertler (2006) found lower rates of ACS admissions for managed care patients, but not for indemnity patients. Kralewski et al. (2012) found slightly higher rates of avoidable hospital admissions but not in ED visits for conditions treatable in primary care. Other research did not identify statistically significant differences in utilization in vertically integrated versus unintegrated systems (Baker et al., 2014 ; Kralewski et al., 2014, 2015).
Three studies showed that vertical integration was associated with higher efficiency as measured by lower health care utilization (Al-Amin, 2016 ; Liepert et al., 2014 ; Weeks et al., 2010). Al-Amin (2016) found vertical integration was associated with small reductions in 30-day all-cause readmissions. Liepert et al. (2014) studied duplicated computerized tomography scans among 481 trauma patients transferred to Intermountain Medical Center (which is vertically integrated) and University of Utah Medical Center (which is not vertically integrated) and found lower rates of duplicate scans in the vertically integrated system. Weeks et al. (2010) found that patients with CAPP-affiliated physicians had lower ACS admission rates compared with patients with non-CAPP physicians (RR = 0.9).
Nine studies showed that vertical integration was associated with higher spending (Baker et al., 2014 ; Chukmaitov, Harless, Bazzoli, Carretta, & Siangphoe, 2015 ; Henke et al., 2018 ; Kralewski et al., 2000, 2012, 2014 ; McWilliams et al., 2013 ; Neprash et al., 2015 ; Robinson & Miller, 2014). Robinson and Miller (2014) compared physician-owned and hospital-owned practices that served 4.5 million patients in a commercial health maintenance organization in California; they found that, compared with physician-owned organizations, local hospital-owned physician organizations had expenditures that were 10.3% higher per patient (adjusted difference = $435), and multihospital systems had expenditures that were 19.8% higher (adjusted difference = $704). McWilliams et al. (2013) found that total spending per Medicare patient was $849 higher for hospital-based groups than for small independent physician groups. This type of finding was replicated in three articles by Kralewski and colleagues (2000, 2012, 2014) that looked at costs per member per year and in two articles that used the AHA health system cluster (Chukmaitov et al., 2015 ; Henke et al., 2018). Henke et al. (2018) examined hospital inpatient expenditures using a large sample of hospitals and nearly 100 million discharges in 44 states; they found higher total costs per discharge for centralized systems, but the impact was small ($11,416 vs. $11,314).
Two studies involved a market-level analysis of vertical integration and spending, both showing that vertical integration was associated with lower efficiency (Baker et al., 2014 ; Neprash et al., 2015). Neprash and colleagues (2015) examined the share of physicians in metropolitan statistical areas who billed as “hospital outpatient departments”. Hospital outpatient department is a place-of-service code in the Medicare outpatient payment services; it can be billed when the ambulatory care site for physician visits and other professional services is owned by a hospital. The authors found that an increase in physician–hospital integration equivalent to the 75th percentile of changes experienced by metropolitan statistical areas was associated with a mean increase of $75 per enrollee in annual outpatient spending—a 3.1% increase relative to mean outpatient spending in 2012. The authors also found that increases in outpatient spending were driven almost entirely by price increases because associated changes in utilization were minimal. Baker et al. (2014) looked at the share of the market served by hospitals that are vertically integrated and found that an increase of 1 standard deviation in the market share of vertically integrated organizations was associated with a 2.4% increase in county-level spending for hospital services.
Despite some evidence suggesting vertical integration was associated with higher spending, not all articles we reviewed came to this conclusion (Kralewski et al., 2015 ; Leibert et al., 2011 ; Weeks et al., 2010). In contrast to earlier studies by Kralewski and colleagues (2000, 2012), a later study did not show a statistically significant relationship between hospital-owned practices and higher per-member-per-year costs (Kralewski et al., 2015), though the relationship trended toward higher spending. Leibert et al. (2011) examined the hospital cost index to gauge the efficiency of flagship hospitals (determined based on a list of top integrated hospitals) and found no statistically significant difference in cost efficiency between highly integrated and nonsystem hospitals. Furthermore, Weeks et al. (2010) found that patients with CAPP-affiliated physicians had lower physician spending (adjusted difference = −$176), lower inpatient spending (adjusted difference = −$103), and lower total Medicare payments (adjusted difference = −$272) compared with patients who did not have CAPP-affiliated physicians.
Two articles examined the relationship between the AHA health system taxonomy and cost inefficiency, in which inefficiency was defined as the deviation from the minimum feasible cost, controlling for output quantity and for level of quality (Carey, 2003 ; Rosko, Proenca, Zinn, & Bazzoli, 2007). Carey (2003) found no statistically significant relationship between CHSs or CPIHSs and inefficiency. Rosko et al. (2007) examined a national sample of 1,144 short-term urban general hospitals and found that CPIHSs had the lowest inefficiency rates (4.1%; i.e., the highest efficiency among system types) but that CHSs had no statistically significant differences from the overall mean inefficiency estimate (8.8% vs. 8.4%).
Three articles focused on the relationship between vertical integration and prices (Baker et al., 2014 ; Ciliberto & Dranove, 2006 ; Cuellar & Gertler, 2006). Studies that examined prices generally focused on discounted patient charges (actual charges multiplied by the yearly average discount rate that each hospital grants to private paying patients). Ciliberto and Dranove (2006) investigated prices among privately insured patients at 320 short-term general hospitals in California between 1994 and 2000. The authors concluded that hospitals have lower prices when they are vertically integrated; however, considerable variation in price effects and the small sample of vertically integrated systems rendered the authors’ estimates imprecise and not statistically significant in the main model specification. Cuellar and Gertler (2006) examined the relationship between vertical integration and prices and found no statistically significant relationship. The authors did find higher prices for some forms of hospital–physician affiliations, but not for vertically integrated forms that met our definition. Baker et al. (2014) found that an increase of 1 standard deviation in the market share of hospitals that are vertically integrated was associated with a 3.2% increase in county-level indices of prices. In this article, price referred to the actual transaction payments under health plans, including payments made by patients and insurers (the allowed amount, not charges or functions of charges).
Studies Examining Patient-Centered Outcomes
Nine studies in our review examined the relationship between vertical integration and one or more patient-centered outcomes (Table 2). Most of the patient-centered outcomes centered on measures of mortality (n = 7). Two studies examined patient satisfaction with care, and one study examined surgical complication rates. Although mortality was a common measure, it was defined in a variety of ways (e.g., disease-specific mortality, hospital-wide mortality, and 30-day mortality), and the results were mixed across studies and, at times, within studies. Rhoads et al. (2015) used a more stringent definition of vertical integration (requiring systems to provide delivery of care, payment, and risk management) and found that the vertical integration was associated with better survival for patients with colorectal cancer (hazard ratio = 0.87). Three studies used the AHA health system cluster to examine mortality rates. Chukmaitov et al. (2015) found that CHSs and hospitals that became more centralized during the study period had lower all-cause 30-day mortality than independent hospital systems or freestanding hospitals. Henke et al. (2018) found centralized system hospitals had lower mortality from acute myocardial infarction but higher stroke mortality compared with noncentralized system hospitals. The authors did not find any statistically significant differences for heart failure mortality or pneumonia mortality.
Other studies showed no statistically significant relationship between vertical integration and mortality outcomes (Chukmaitov et al., 2009 ; Cuellar & Gertler, 2006 ; Madison, 2004 ; Scott et al., 2017). Two studies examined patient satisfaction with care using the Hospital Consumer Assessment of Healthcare Providers and Systems measures (Gillies et al., 2006 ; Leibert, 2011). Neither study identified a statistically significant relationship between vertical integration and patient satisfaction with care. Gillies et al. (2006) examined the percentage of a health plan served by group or staff models and found no statistically significant relationship with satisfaction with medical group care, physician care, or health plan care. Similarly, Liebert (2011) found no statistically significant difference between the mean Hospital Consumer Assessment of Healthcare Providers and Systems scores of flagship hospitals (as determined by a list of top integrated hospitals) and nonsystem hospitals.
Vertical integration likely has both positive and negative effects on policy-relevant outcomes. The studies we identified in this review suggest that vertical integration may be associated with improvements in some quality-of-care measures—but with no differences or lower efficiency as measured by avoidable utilization, spending, and prices. Impacts varied by measure and context. Higher performance on quality-of-care measures among vertically integrated health systems might reflect both actual improvements in quality through better coordination as well as better perceived quality through health IT and improved documentation of processes of care. At the same time, decreases in efficiencies (as measured by prices) might reflect the increased market power of larger systems. Furthermore, increases in hospital utilization and spending might reflect differences in clinical service use, which favor hospital-based services. These findings are consistent with the hypothesis that physician behavior is influenced by the Centers for Medicare & Medicaid Services’ location-based billing policy, which provides higher compensation for care delivered in hospital settings than in doctors’ offices (Koch et al., 2017).
There are a variety of factors that could explain the large number of mixed findings both within and between studies. For example, even within our narrow definition of health systems, included studies used a range of data sources and approaches to identify health systems. Also, system attributes and market factors might impact the outcomes of vertical integration. For example, integration with a health plan, duration of provider integration, or existing levels of market consolidation. Thus, more attention on the role of moderating factors is warranted in future research. Furthermore, the cross-sectional design of many of the included studies restricts our ability to say whether or not the outcome is a result of the integration itself or predates when a provider joined or formed a system.
This review did not identify many articles that evaluated a patient-centered outcome. Of the studies that did, most evaluated the relationship between vertical integration and mortality and did not identify any relationship. Future studies should consider additional patient-centered outcomes, particularly given the limitations of mortality as an outcome measure. Specifically, mortality has been shown to have a weak link to quality of care (Shahian, Iezzoni, Meyer, Kirle, & Normand, 2012), particularly for patients with multiple chronic conditions near the end of life (Holloway & Quill, 2007). Only two studies in our review examined the relationship between vertical integration and patient satisfaction with care. Neither of these studies identified any association; however, past studies that examined large-sized or staff/group model health maintenance organizations suggest that large systems sometimes struggle to perform as well on patient satisfaction measures as smaller physician practices (Safran et al., 2002 ; Seibert, Strohmeyer, & Carey, 1996).
Included studies had several common limitations. Most studies were observational and did not address the issue of selection bias—and thus were considered “low quality” according to the GRADE approach. Vertical integration of hospitals and physician practices is not a randomly determined attribute but is a strategic choice by providers, and it is likely that unobservable factors affecting the decision to vertically integrate also affect relevant outcomes. For example, physicians or hospitals with a greater interest in the use of integrated data systems for care coordination or quality improvement might be more likely to join or form systems. Moreover, patients of different characteristics may choose to receive care from different providers based on attributes related to vertical integration. For example, patients with multiple complex conditions might be more likely to seek care from health systems because of their specialty composition and research expertise (Retchin, 1998). Certain studies were better equipped to deal with these selection issues, including studies that focused on changes in the performance of acquired clinics using a difference-in-differences approach or studies that used panel data to examine within-hospital changes for hospitals that switched affiliations during the study period. In addition, many studies put hospitals that have any ownership relationships with physician practices into the same group (without indicating how widespread these relationships were). Despite these limitations, many studies evaluated large samples that were nationally representative, making them well powered to detect small changes in health system performance and to offer generalizable conclusions.
It bears repeating that our definition of vertically integrated health systems focused on form (e.g., practices owned by hospitals), not on function (e.g., degree of clinical integration), and it does not explicitly require a relationship with postacute care providers or health plans. But other definitions of health care systems emphasize key functional aspects, such as the provision of the continuum of health care services. For example, Enthoven (2009) defined an integrated delivery system as:
…an organized, coordinated, and collaborative network that (1) links various healthcare providers, via common ownership or contract, across three domains of integration—economic, noneconomic, and clinical—to provide a coordinated, vertical continuum of services to a particular patient population or community and (2) is accountable, both clinically and fiscally, for the clinical outcomes and health status of the population or community served, and has systems in place to manage and improve them. (p. S285)
This definition emphasizes coordinated care and integration across several dimensions. Burns and Muller (2008) found a weak and inconsistent relationship between clinical integration and economic integration, including the type we investigated here. Achieving better patient care via vertical integration might require hospitals and practices to ensure care is actually integrated from the patient’s perspective.
Given the focus of our review, it was difficult to distinguish the effect of vertical integration per se from the effects of other health system characteristics and the markets in which they develop. Vertically integrated health systems identified in this review vary with respect to size, health IT strategy, or clinical integration, among other health system and market attributes. One attribute that may be particularly relevant is whether systems have a strong payer relationship with non-fee-for-service incentives. Health system attributes and market factors were not always explicitly or clearly documented in the literature we reviewed, and their effects were not easily teased apart. Extending the analysis and synthesis in this article to examine the specific aspects of systems that affect outcomes is an important area for future research.
Our findings should be considered alongside literature on vertically integrated health systems that did not meet the criteria for this review but is nonetheless relevant to understanding how integration influences outcomes. For example, several studies analyze the performance of vertically integrated health systems on measures of financial performance (Burns, Gimm, & Nicholson, 2005) and operating costs (Burns et al., 2015). These outcomes are important to understanding hospital and physician organizations’ decisions to integrate. Other studies consider the relationship between vertical integration and patients’ choice of hospital (Baker et al., 2016), referral patterns (Carlin, Feldman, & Dowd, 2016), adoption of health IT (Everson et al., 2016 ; Lammers, 2013), and care coordination or care management processes (Casalino et al., 2013 ; Rittenhouse, Grumbach, O’Neil, Dower, & Bindman, 2004 ; Rodriguez et al., 2016). These studies might help explain the functional mechanism for the findings we identified in this review. In addition, there is research on the performance of other definitions of integration, such as horizontal integration of hospitals (Ho & Hamilton, 2000) or of physician practices (Mehrotra et al., 2006 ; Rodriguez, Von Glahn, Rogers, & Safran, 2009), as well as literature that examines specific health system attributes, such as volume, teaching status, or nonprofit status (Thornlow & Stukenborg, 2006). This literature is relevant because forms of integration are not mutually exclusive, and other definitions of integration may explain modifiable characteristics of health systems that influence key outcomes.
Vertical integration will likely continue to gain momentum, in part through the effects of implementing the MACRA of 2015. This legislation creates strong incentives for physicians and other clinicians paid under Medicare’s physician fee schedule to participate in alternative payment models to earn enhanced Medicare reimbursement. Although the literature debates the nature of the relationship between alternative payment models and vertical integration, recent evidence suggests the trend toward vertical integration will likely continue as providers respond to changing payment models and market factors (Neprash, Chernew, & McWilliams, 2017).
This changing landscape of physician and hospital affiliations has complex implications for payers, professional associations, and policymakers. The mixed findings on utilization and spending found in this review suggest that physicians and practice managers preparing for more population-oriented Medicare payment reforms under MACRA should be prudent when considering the potential benefits of integrating with hospital systems. Specifically, hospital-dominated systems may be organized to enhance system Medicare revenues through provision of inpatient as well as hospital outpatient department services. However, this arrangement may not be preferable to provider entities at financial risk under population-based payment models if it is associated with increased spending. Thus, more information is needed to identify the health system attributes and market factors that contribute to improved outcomes, as well as which practice and policy levers can minimize anticompetitive effects and maximize the benefits of these affiliations. By these means, payers and policymakers can shape more targeted incentives to encourage the development of health systems that deliver higher-value care.
Al-Amin M. (2016). Hospital characteristics and 30-day all-cause readmission rates. Journal of Hospital Medicine
, 11(10), 682–687.
Baker L. C., Bundorf M. K., Kessler D. P. (2014). Vertical integration
: Hospital ownership of physician practices is associated with higher prices and spending. Health Affairs
, 33(5), 756–763.
Baker L. C., Bundorf M. K., Kessler D. P. (2016). The effect of hospital/physician integration on hospital choice. Journal of Health Economics
, 50, 1–8.
Bazzoli G. J., Shortell S. M., Dubbs N., Chan C., Kralovec P. (1999). A taxonomy of health networks and systems: Bringing order out of chaos. Health Services Research
, 33(6), 1683.
Burns L. R., Gimm G., Nicholson S. (2005). The financial performance of integrated health organizations. Journal of Healthcare Management
, 50(3), 191.
Burns L. R., Goldsmith J. C., Sen A. (2013). Horizontal and vertical integration
of physicians: A tale of two tails. Advances in Health Care Management
, 15, 39.
Burns L. R., McCullough J. S., Wholey D. R., Kruse G., Kralovec P., Muller R. (2015). Is the system really the solution? Operating costs in hospital systems. Medical Care Research and Review
, 72(3), 247–272.
Burns L. R., Muller R. W. (2008). Hospital-physician collaboration: Landscape of economic integration and impact on clinical integration. Milbank Quarterly
, 86(3), 375–434.
Burns L. R., Wholey D. R., McCullough J. S., Kralovec P., Muller R. (2012). The changing configuration of hospital systems: Centralization, federalization, or fragmentation? In Friedman L. H., Goes J., Savage G. T. (Eds.), Annual review of health care management: Strategy and policy perspectives on reforming health systems
(pp. 189–232). Bingley, United Kingdom: Emerald Group Publishing Limited.
Carey K. (2003). Hospital cost efficiency and system membership. Inquiry
, 40(1), 25–38.
Carlin C. S., Dowd B., Feldman R. (2015). Changes in quality of health care delivery after vertical integration
. Health Services Research
, 50(4), 1043–1068.
Carlin C. S., Feldman R., Dowd B. (2016). The impact of hospital acquisition of physician practices on referral patterns. Health Economics
, 25(4), 439–454.
Casalino L. P. (2014). Categorizing accountable care organizations: Moving toward patient-centered outcomes research that compares health care delivery systems. Health Services Research
, 49(6), 1875–1882.
Casalino L. P., Pesko M. F., Ryan A. M., Mendelsohn J. L., Copeland K. R., Ramsay P. P., Shortell S. M. (2014). Small primary care physician practices have low rates of preventable hospital admissions. Health Affairs
, 33(9), 1680–1688.
Casalino L. P., Wu F. M., Ryan A. M., Copeland K., Rittenhouse D. R., Ramsay P. P., Shortell S. M. (2013). Independent practice associations and physician–hospital organizations can improve care management for smaller practices. Health Affairs
, 32(8), 1376–1382.
Chukmaitov A. S., Bazzoli G. J., Harless D. W., Hurley R. E., Devers K. J., Zhao M. (2009). Variations in inpatient mortality among hospitals in different system types, 1995 to 2000. Medical Care
, 47(4), 466–473.
Chukmaitov A. S., Harless D. W., Bazzoli G. J., Carretta H. J., Siangphoe U. (2015). Delivery system
characteristics and their association with quality and costs of care: Implications for accountable care organizations. Health Care Management Review
, 40(2), 92–103.
Ciliberto F., Dranove D. (2006). The effect of physician–hospital affiliations on hospital prices in California. Journal of Health Economics
, 25(1), 29–38.
Crespin D. J., Christianson J. B., McCullough J. S., Finch M. D. (2016). Health system
consolidation and diabetes care performance at ambulatory clinics. Health Services Research
, 51(5), 1772–1795.
Cuellar A. E., Gertler P. J. (2006). Strategic integration of hospitals and physicians. Journal of Health Economics
, 25(1), 1–28.
Cutler D. M., Scott Morton F. (2013). Hospitals, market share, and consolidation. Journal of the American Medical Association
, 310(18), 1964–1970.
Enthoven A. C. (2009). Integrated delivery systems: The cure for fragmentation. American Journal of Managed Care
, 15(12), S284.
Everson J., Lee S. D., Adler-Milstein J. (2016). Achieving adherence to evidence-based practices: Are health IT and hospital–physician integration complementary or substitutive strategies? Medical Care Research and Review
, 73(6), 724–751.
Gillies R. R., Chenok K. E., Shortell S. M., Pawlson G., Wimbush J. J. (2006). The impact of health plan delivery system
organization on clinical quality and patient satisfaction. Health Services Research
, 41(4 Pt. 1), 1181–1199.
Henke R. M., Karaca Z., Moore B., Cutler E., Liu H., Marder W. D., Wong H. S. (2018). Impact of health system
affiliation on hospital resource use intensity and quality of care. Health Services Research
, 53(1), 63–86. doi:10.1111/1475-6773.12631
Herrel L. A., Kaufman S. R., Yan P., Miller D. C., Schroeck F. R., Skolarus T. A., Hollenbeck B. K. (2017). Health care integration and quality among men with prostate cancer. The Journal of Urology
, 197(1), 55–60.
Ho V., Hamilton B. H. (2000). Hospital mergers and acquisitions: Does market consolidation harm patients? Journal of Health Economics
, 19(5), 767–791.
Holloway R. G., Quill T. E. (2007). Mortality as a measure of quality: Implications for palliative and end-of-life care. Journal of the American Medical Association
, 298(7), 802–804.
Hwang W., Chang J., LaClair M., Paz H. (2013). Effects of integrated delivery system
on cost and quality. American Journal of Managed Care
, 19(5), e175–e184.
Koch T. G., Wendling B. W., Wilson N. E. (2017). How vertical integration
affects the quantity and cost of care for Medicare beneficiaries. Journal of Health Economics
, 52, 19–32.
Kralewski J., Dowd B., Knutson D., Savage M., Tong J. (2013). Medical group practice characteristics influencing inappropriate emergency department and avoidable hospitalization rates. Journal of Ambulatory Care Management
, 36(4), 286–291.
Kralewski J., Dowd B., Knutson D., Tong J., Savage M. (2015). The relationships of physician practice characteristics to quality of care and costs. Health Services Research
, 50(3), 710–729.
Kralewski J., Dowd B., Savage M., Tong J. (2014). Do integrated health care systems provide lower-cost, higher-quality care? Physician Executive
, 40(2), 14–18.
Kralewski J. E., Dowd B. E., Xu Y. W. (2012). Medical groups can reduce costs by investing in improved quality of care for patients with diabetes. Health Affairs
, 31(8), 1830–1835.
Kralewski J. E., Rich E. C., Feldman R., Dowd B. E., Bernhardt T., Johnson C., Gold W. (2000). The effects of medical group practice and physician payment methods on costs of care. Health Services Research
, 35(3), 591–613.
Lammers E. (2013). The effect of hospital–physician integration on health information technology adoption. Health Economics
, 22(10), 1215–1229.
Leibert M. (2011). Performance of integrated delivery systems: Quality, service, and cost implications. Leadership in Health Services
, 24(3), 196–206.
Liepert A. E., Bledsoe J., Stevens M. H., Cochran A. (2014). Protecting trauma patients from duplicated computed tomography scans: The relevance of integrated care systems. American Journal of Surgery
, 208(4), 511–516.
Madison K. (2004). Hospital-physician affiliations and patient treatments, expenditures, and outcomes. Health Services Research
, 39(2), 257–278.
Maeda J. L., Lee K. M., Horberg M. (2014). Comparative health systems research among Kaiser Permanente and other integrated delivery systems: A systematic literature review. Permanente Journal
, 18(3), 66–77.
McWilliams J. M., Chernew M. E., Zaslavsky A. M., Hamed P., Landon B. E. (2013). Delivery system
integration and health care spending and quality for Medicare beneficiaries. JAMA Internal Medicine
, 173(15), 1447–1456.
Mehrotra A., Epstein A. M., Rosenthal M. B. (2006). Do integrated medical groups provide higher-quality medical care than individual practice associations? Annals of Internal Medicine
, 145(11), 826–833.
Neprash H. T., Chernew M. E., Hicks A. L., Gibson T., McWilliams J. M. (2015). Association of financial integration between physicians and hospitals with commercial health care prices. JAMA Internal Medicine
, 175(12), 1932–1939.
Neprash H. T., Chernew M. E., McWilliams J. M. (2017). Little evidence exists to support the expectation that providers would consolidate to enter new payment models. Health Affairs
, 36(2), 346–354.
Reschovsky J. D., Rich E. (2015). Hospital acquisition of physician groups: On the road to value-based or higher-priced care? JAMA Internal Medicine
, 175(12), 1939–1941.
Reschovsky J. D., Rich E. C., Lake T. K. (2015). Factors contributing to variations in physicians’ use of evidence at the point of care: A conceptual model. Journal of General Internal Medicine
, 30(3), 555–561.
Retchin S. M. (1998). Adverse selection at academic health centers. Clinical Performance and Quality Health Care
, 6(1), 38–43.
Rhoads K. F., Patel M. I., Ma Y., Schmidt L. A. (2015). How do integrated health care systems address racial and ethnic disparities in colon cancer? Journal of Clinical Oncology
, 33(8), 854–860.
Rittenhouse D. R., Grumbach K., O’Neil E. H., Dower C., Bindman A. (2004). Physician organization and care management in California: From cottage to Kaiser. Health Affairs
, 23(6), 51–62.
Robinson J. C., Miller K. (2014). Total expenditures per patient in hospital-owned and physician-owned physician organizations in California. Journal of the American Medical Association
, 312(16), 1663–1669.
Rodriguez H. P., McClellan S. R., Bibi S., Casalino L. P., Ramsay P. P., Shortell S. M. (2016). Increased use of care management processes and expanded health information technology functions by practice ownership and Medicaid revenue. Medical Care Research and Review
, 73(3), 308–328.
Rodriguez H. P., Von Glahn T., Rogers W. H., Safran D. G. (2009). Organizational and market influences on physician performance on patient experience measures. Health Services Research
, 44(3), 880–901.
Rosko M. D., Proenca J., Zinn J. S., Bazzoli G. J. (2007). Hospital inefficiency: What is the impact of membership in different types of systems? Inquiry
, 44(3), 335–349.
Safran D. G., Wilson I. B., Rogers W. H., Montgomery J. E., Chang H. (2002). Primary care quality in the Medicare Program: Comparing the performance of Medicare health maintenance organizations and traditional fee-for-service Medicare. Archives of Internal Medicine
, 162(7), 757–765.
Schünemann H., Brożek J., Guyatt G., Oxman A. (2014). GRADE handbook for grading quality of evidence and strength of recommendations
Updated October 2013. The GRADE Working Group. Retrieved from https://gradepro.org/cite.html
Scott K. W., Orav E. J., Cutler D. M., Jha A. K. (2017). Changes in hospital–physician affiliations
in U.S. hospitals and their effect on quality of care. Annals of Internal Medicine
, 166(1), 1–8.
Seibert J. H., Strohmeyer J. M., Carey R. G. (1996). Evaluating the physician office visit: In pursuit of a valid and reliable measure of quality improvement efforts. The Journal of Ambulatory Care Management
, 19(1), 17–37.
Shahian D. M., Iezzoni L. I., Meyer G. S., Kirle L., Normand S. L. (2012). Hospital-wide mortality as a quality metric: Conceptual and methodological challenges. American Journal of Medical Quality
, 27(2), 112–123.
Thornlow D. K., Stukenborg G. J. (2006). The association between hospital characteristics and rates of preventable complications and adverse events. Medical Care
, 44(3), 265–269.
Walston S. L., Kimberly J. R., Burns L. R. (1996). Owned vertical integration
and health care: Promise and performance. Health Care Management Review
, 21(1), 83–92.
Weeks W. B., Gottlieb D. J., Nyweide D. E., Sutherland J. M., Bynum J., Casalino L. P., Fisher E. S. (2010). Higher health care quality and bigger savings found at large multispecialty medical groups. Health Affairs
, 29(5), 991–999.
delivery system; health system; hospital–physician affiliations; vertical integration
Supplemental Digital Content
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved