Price Effects of Vertical Integration and Joint Contracting between Physicians and Hospitals

Abstract

Vertical integration in health care has recently garnered scrutiny by antitrust authorities and state regulators. We examined trends, geographic variation, and price effects of vertical integration and joint contracting between physicians and hospitals using physician affiliations and all-payer claims data from Massachusetts in 2013–2017. Vertical integration and joint contracting with small and medium systems rose from 19.5% to 32.8% for primary care physicians and from 26.1% to 37.8% for specialists. Vertical integration and joint contracting with large systems slightly declined while geographic variation in these physician affiliations rose. We found that vertical integration and joint contracting led to price increases of 2.1%–12.0% for primary care physicians and 0.7%–6.0% for specialists, with the greatest increases in large systems. These findings can inform policymakers seeking to limit growth in health care prices.

Introduction and Background

Consolidation between physicians and hospitals has become more prevalent in recent years. These relationships include hospital ownership of physician practices, hospitals’ direct employment of physicians, and joint contracting between physician practices and hospitals. Various economic and contractual relationships between physicians and hospitals have waxed and waned for decades.¹ From 2012 to 2018, the share of physicians in practices with at least some hospital ownership or who were direct hospital employees rose from 29.0% to 34.7%.²

Alongside recent increases in ownership, joint contracting between physicians and hospitals has risen in prominence, largely in the form of Clinically Integrated Networks (CINs). CINs can vary substantially in structure including a mix of ownership and affiliation as well as exclusive and non-exclusive arrangements involving physicians, hospitals, and other providers.³ CINs may also reduce horizontal competition. When a physician practice is acquired or included in joint contracting with the hospitals in a system, it often joins other medical groups and physician networks, thereby potentially affecting both efficiency and pricing leverage solely due to horizontal consolidation.

Like Physician-Hospital Organizations (PHOs), CINs exist for the purpose of contracting with payers and typically have a stated focus on improving care coordination, quality, and efficiency. CINs vary in their approaches to supporting quality improvement and cost control, but little is known about their impact on these outcomes or on prices.³ Recent increases in vertical integration and joint contracting may in part be a response to providers’ needs to manage global risk contracts such as accountable care organizations (ACOs) created by the Affordable Care Act.^1,4 Consolidation may continue as providers respond to decreased revenue due to the COVID-19 pandemic.⁵

Vertical integration between physicians and hospitals may have several potential benefits including improved care coordination, improved alignment of provider incentives through the “internalization” of externalities, less duplication of services, and economies of scale for administrative functions such as deployment of electronic health records.^6–14 These benefits could result in cost savings that lower prices for physician services. Several health care organizations, such as Kaiser Permanente, Intermountain Health Care, and the Mayo Clinic, have been touted as models of integrated health delivery and are highly regarded for providing high-quality and efficient health care.^1,15,16

While some have argued that health reforms such as “accountable health plans” in the 1990s and ACOs in the 2010s were motivated by an intention to foster the creation of Kaiser-like integrated health systems,^1,17 in practice it has been difficult for these organizations to enter new markets¹⁸ or for other health care providers to replicate this success.¹⁹ Moreover, federal antitrust authorities are increasingly examining vertical mergers, a change in policy reflecting concern that some vertical mergers may raise prices or have other anticompetitive effects.^20,21 In June 2020, the Department of Justice and Federal Trade Commission issued joint guidelines for evaluating vertical mergers, the first major revision to guidance on vertical mergers since 1984;²² the Federal Trade Commission subsequently withdrew its approval of these guidelines in September 2021.²³

There are two primary concerns about vertical mergers in health care. First, newly integrated physicians might redirect referrals and thereby foreclose rivals. Second, they might leverage enhanced market power from being part of a health system to raise prices. Joint contracting between physicians and hospitals through PHOs or CINs theoretically may have similar benefits and risks as vertical mergers, depending on factors such as the extent of investments in quality improvement initiatives and whether the arrangements are exclusive.²⁴ Joint contracting may also reduce horizontal competition.

Prior studies found that vertical integration led to higher hospital prices in Arizona, Florida, and Wisconsin,²⁵ and nationally,²⁶ but not in California²⁷; higher hospital prices for physician-hospital organizations²⁵; and no change in hospital prices for hospitals that employ physicians as salaried employees.²⁵ An emerging literature has shown that integration between physicians and hospitals is usually associated with an increase in physician prices^28–31 as well as hospital prices.³² These studies have used hospital-level measures from the American Hospital Association^25–27 or physician measures aggregated by geography;^28,29 some of those that used physician-level measures have focused on a subset of physicians or patients in a market.^33–35

In this paper, we provide new evidence on the impact of vertical integration and joint contracting on outpatient commercial insurance prices in Massachusetts. We examine affiliations between physicians and health systems along with explicit ownership relationships, both of which usually employ joint negotiation of provider prices in our setting.³⁶ Our analyses capture vertical and horizontal consolidation, or consolidation between physicians and hospitals or among physicians. This is a broader analysis than those that have only examined ownership relationships and builds on previous research describing PHOs and CINs and their effects on physician practices.^3,14

We note that we cannot fully disentangle the effects of new vertical ties from related horizontal consolidation. The health systems we study in Massachusetts are predominantly large, complex entities with multiple affiliates and corporate members. However, we partially address this limitation by including a sensitivity analysis in which we measure vertical integration as legal ownership for a subset of physicians. We revisit this issue in the discussion section.

Study Data and Methods

DATA

We used four datasets in this analysis. First, we obtained the Massachusetts Provider Database (MPD) from Massachusetts Health Quality Partners (MHQP) for 2013, 2015, and 2017.³⁷ The MPD reports affiliations of physicians with practice sites, medical groups, and health systems. Our main analysis sample included all physicians at practice sites in Massachusetts who were not members of community health centers and who were classified as primary care, specialist, or hospitalist physicians (Supplemental Exhibit 1).

Second, we used Massachusetts hospital reports from 2013, 2015, and 2017 from the Center for Health Information and Analysis (CHIA) that contain hospital affiliations and total discharges. Third, we used 2013–2017 commercial paid medical claims data from the Massachusetts All-Payer Claims Database (MA APCD). Finally, we used information from the Massachusetts Health Policy Commission’s Registered Provider Organization (RPO) database.

OUTCOMES AND MAIN MEASURES

We used the MHQP MPD to create a physician-level indicator of vertical integration or joint contracting with a health system, which captured both ownership and affiliation. We were unable to distinguish between ownership and affiliation in the MHQP data.³⁶ Approximately 20% of physicians experienced a change in vertical integration or joint contracting status during our study period as a result of joining or leaving a health system (Supplemental Exhibit 2). We measured a second physician-level indicator for physician organizations legally owned by a health system (“vertical integration”) using the RPO data. Approximately 9% of physicians changed vertical integration status according to the RPO data.

We used the CHIA hospital reports to identify the hospitals in each health system. We focused on the 10 health systems in both the MHQP and CHIA data sources, which were each affiliated with at least 2 and up to 8 hospitals (Supplemental Exhibit 3). The health systems varied in the share of hospital discharges within the health system’s geographic market based on hospital referral region (HRR)³⁸ in 2017 (Supplemental Exhibit 3). We classified health systems as large if the share of hospital discharges was at least 20%; as medium if the share was at least 10% but less than 20%; and otherwise, as small. We also separately examined the health system with the largest share of state-wide hospital discharges (the “largest” system).

We measured outpatient commercial physician prices using the MA APCD for 2013–2017. We constructed two summary price measures: the implied price (a physician’s average price relative to the statewide average price for the basket of procedures that she delivered) and a price index (for a common “basket” of services—those with the highest expenditure shares within a physician’s specialty). The implied price and price index were highly correlated, with a correlation of 0.766. We measured the facility fees in conjunction with a physician service in three ways: the sum of all outpatient facility paid claims for the same patient and date for which the physician provided a service; and the sum of facility fees occurring within 1 week or 1 month.

We included a Herfindahl-Hirschman Index (HHI)—an index of concentration equal to the sum of squared spending shares in a defined market)—in all our analyses to capture the effects of vertical integration and joint contracting on market concentration. More vertical integration and joint contracting not only combines hospitals and physicians, but also combines newly added physician groups with a system’s existing groups; that horizontal consolidation can have a separate impact on prices. We followed Capps, Dranove, and Ody (2018)³⁰ by computing HHIs (for each specialty and year combination) for patient 3-digit ZIP codes and then constructing a weighted-average HHI for each physician using firm-level weights; this served as a flexible measure that did not rely on fixed market boundaries.

ANALYSES

We examined trends in the percent of physicians who were independent versus affiliated with systems. To measure trends unaffected by entry and exit, we used the sample of physicians continuously present in all 3 years for which we observed affiliations in the MHQP data. Results were similar when we used the full sample of physicians (Supplemental Exhibit 4) or weighted observations by outpatient spending.

We examined variation in physician affiliations across hospital service areas (HSAs)³⁸ and HRRs. We measured statewide geographic variation using the coefficient of variation for the HSA-level percent of physicians vertically integrated or joint contracting with large systems, equal to the mean divided by the standard deviation.

Our main outcome was implied price, which had a mean of 1.009. We linked each physician-level indicator of vertical integration or joint contracting by year to the implied price measure using the National Provider Identifier. We linked price measures from the concurrent year and following year, except for 2017 when we only used the concurrent year. We constructed event study figures for the implied price, which show separate estimates for each year relative to the year of a physician’s change in vertical integration or joint contracting status (Supplemental Exhibits 5–6). We found similar pre-event trends for physicians who changed status and those who did not, providing support for our empirical strategy.

We estimated price effects of vertical integration and joint contracting for small, medium, large, and the largest systems using multivariate regression. All models included physician fixed effects, which allowed us to control for unobserved time-invariant differences across physicians. We controlled for secular trends using year fixed effects and for physician market concentration.³⁰ We estimated effects separately for primary care physicians (PCPs) and specialists. We adjusted for correlations in prices within medical groups by clustering standard errors at the medical group level. We also estimated effects on facility fees.

In sensitivity analyses, we obtained similar results when we weighted observations by spending, used the price index as the dependent variable, controlled for alternative measures of HHI, and excluded manually attributed physician affiliations (Supplemental Exhibits 7–10). In a sensitivity analysis defining vertical integration as legal ownership using the RPO data, we estimated price effects that were similar but slightly larger in magnitude (Supplemental Exhibits 11–12).

LIMITATIONS

Our primary measure of vertical consolidation reflects a combination of ownership relationships and affiliations between physicians and health systems. These affiliations also affect horizontal consolidation, which can have a separate impact on prices. We are also unable to measure changes in physicians’ direct contractual relationships with hospitals.

The analysis includes only one state over a 5-year period, which may limit generalizability of the findings to other parts of the US and to longer time periods. However, Massachusetts features several very large health systems competing alongside smaller health systems and independent providers; many large metropolitan areas around the country have a similar configuration of providers.³⁹ The exact time horizon when we would expect to see effects on prices is not entirely clear, since it depends on the unobservable frequency of provider-payer negotiations and adjustments to provider payment rates.

Although we use within-physician changes in vertical integration and joint contracting to estimate its effects, we cannot rule out that our estimated effects could be driven by other market-wide changes that affect the same physicians who happen to be undergoing vertical integration. In addition, the 2016 US Supreme Court ruling in Gobeille v. Liberty Mutual Insurance Company made submitting claims to the all-payer claims database optional for self-insured employers.⁴⁰ The APCD data used to construct physician prices lack some of these self-insured claims in 2016–2017, though we still observe a large number of claims for most insurer-provider pairs. Finally, we do not examine effects on total medical spending, which may differ from price effects.

Study Results

SAMPLE CHARACTERISTICS

Our main analysis sample, all physicians in the MHQP MPD, includes 56,929 physician-year observations (Exhibit 1); 34.0% are PCPs, 62.6% are specialists, and 3.4% are hospitalists. PCPs are slightly overrepresented in Massachusetts compared to nationwide, where approximately 29.9% of physicians were PCPs in 2018.⁴¹ The median practice site size (across physicians) was 13, the median medical group size was 475, and the median health system size was 4,037. There were 24.9% of physicians who were independent (e.g., not vertically integrated or involved in joint contracting), 11.9% in a small system, 19.2% in a medium system, and 44.0% in a large system.

Exhibit 1:

Sample Characteristics

	2013 (1)	2015 (2)	2017 (3)	2013–2017 (4)
# Physicians	17,982	18,531	20,416	24,603
# Practice Sites	3,935	3,965	4,007	5,498
# Medical Groups	121	131	138	158
# Systems	9	10	10	10
% Physicians
Primary Care	34.6	34.0	33.6	34.0
Specialist	62.5	62.6	62.6	62.6
Hospitalist	2.9	3.4	3.8	3.4
Practice Site Size
Mean	36.7	42.6	60.8	47.3
Median	12	12	15	13
Medical Group Size
Mean	756.2	812.4	904.5	827.7
Median	469	475	503	475
System Size
Mean	4140.0	3475.4	3575.6	3721.2
Median	5,628	4,037	4,504	4,037
% Physicians
Small Medical Group (1–99)	12.0	13.7	12.8	12.8
Medium Medical Group (100–499)	43.2	40.5	36.4	39.9
Large Medical Group (500+)	44.8	45.9	50.8	47.3
% Physicians
Independent	31.3	21.8	22.1	24.9
Small System	5.7	15.4	14.1	11.9
Medium System	18.2	19.0	20.4	19.2
Large System	44.8	43.8	43.5	44.0

SOURCE Authors’ analysis of Massachusetts Health Quality Partners (MHQP) provider data. NOTES The table shows sample characteristics for the main analysis sample, including the number of physicians, practice sites, medical groups, and systems; sizes (in terms of number of physicians) of practice sites, medical groups, and systems; and the percent of physicians by specialty category, medical group size, and system type.

TRENDS

Among PCPs who were continuously present during 2013–2017, the percent who were independent fell from 42.0% to 31.5%; the percent vertically integrated or joint contracting with a small system rose from 3.8% to 14.6%, with a medium system from 15.7% to 18.2%; while the percent vertically integrated or joint contracting with a large system fell from 38.5% to 35.7% (Exhibit 2). Specialists were more likely to be vertically integrated or joint contracting than PCPs but had similar trends. Among specialists, the percent who were independent fell from 26.4% to 17.4%; the percent vertically integrated or joint contracting with a small system rose from 7.1% to 17.3%; with a medium system rose from 19.0% to 20.5%; and with a large system fell from 47.5% to 44.8%.

EXHIBIT 2.

Trends in Vertical Integration and Joint Contracting in Massachusetts

SOURCE: Authors’ analysis of Massachusetts Health Quality Partners (MHQP) provider data. NOTES: The figure shows the percent of physicians who are vertically integrated or joint contracting with health systems of various sizes. The sample consists of physicians who are continuously present in all 3 years of the study period (2013, 2015, and 2017). The left side shows results for primary care physicians (N = 4,936) and the right side shows results for specialists (N = 8,649). The percent of physicians is unweighted.

GEOGRAPHIC VARIATION

There was substantial geographic variation in vertical integration or joint contracting with a medium or large system across HSAs (Exhibit 3); as well as with a large system (Supplemental Exhibit 13). Massachusetts HSAs are contained within three HRRs. In the Boston HRR, 35.8% of physicians were vertically integrated or joint contracting with a large system in 2017, compared to 84.8% in the Springfield HRR and 74.7% in the Worcester HRR (Supplemental Exhibit 14). For the statewide HSA-level percent of physicians vertically integrated or joint contracting with a large system, the coefficient of variation rose from 107.1% to 121.7%, an increase of 13.6% in overall geographic variation.

EXHIBIT 3.

Geographic Variation in Vertical Integration and Joint Contracting with a Medium or Large System

SOURCE: Authors’ analysis of Massachusetts Health Quality Partners (MHQP) provider data and Massachusetts All-Payer Claims Database (APCD) medical claims data. NOTES: The figure shows geographic variation in the percent of physicians vertically integrated or joint contracting with a medium or large system in 2017. The map shows hospital service areas (HSAs) shaded according to the percent of physicians who are vertically integrated or joint contracting. The figure also shows boundaries for the Boston, Springfield, and Worcester hospital referral regions (HRRs) as well as the headquarters for each medium or large system.

EFFECTS ON PHYSICIAN PRICES

Vertical integration and joint contracting with a small system did not significantly impact prices, but vertical integration and joint contracting with a medium system led to a price increase of 7.7% for PCPs and 2.4% for specialists (Exhibit 4). Vertical integration and joint contracting with a large system led to a price increase of 12.0% for PCPs and 6.0% for specialists, and with the largest system a price increase of 15.7% for PCPs and 7.0% for specialists. HHI was not significantly associated with outpatient physician prices for any of the HHI measures (Supplemental Exhibits 15 and 9).

EXHIBIT 4.

Effect of Vertical Integration and Joint Contracting on Physician Prices

SOURCE: Authors’ analysis of Massachusetts Health Quality Partners (MHQP) provider data. NOTES: The figure shows estimates of the effect of vertical integration and joint contracting between physicians and health systems on physician prices. The left side shows results for primary care physicians and the right side shows results for specialists. Observations are at the physician-year level and are unweighted. The outcome is the natural logarithm of the implied price, which is the price of a physician’s basket of procedures relative to average state-wide procedure prices. For the “small system” coefficient, a physician is defined as vertically integrated or joint contracting if she is part of a small, medium, or large health system. For the “medium system” coefficient, a physician is defined as vertically integrated or joint contracting if she is part of a medium or large health system. For the “large system” coefficient, a physician is defined as vertically integrated or joint contracting if she is part of a large health system. For the “largest system” coefficient, a physician is defined as vertically integrated or joint contracting if she is part of the largest health system. Each specification includes year fixed effects, physician fixed effects, and the physician market Herfindahl-Hirschman Index (HHI). Robust standard errors are clustered by medical group and 95% confidence intervals are shown as gray bars.

For specialists integrating or joint contracting with the largest system, facility fees within 1 week following the provision of patient services increased by $136 (Supplemental Exhibit 16), or 4.9% relative to the mean. We found a similar increase for specialists’ facility fees occurring on the same date as the service (Supplemental Exhibit 17), though this was not statistically significant. There was no statistically significant effect of vertical integration and joint contracting on facility fees in our other models, including those for facility fees occurring within 1 month.

Discussion

This study, which uses detailed data on physician affiliations linked to an all-payer claims database in Massachusetts from 2013–2017, has three main findings. First, the percent of physicians who were vertically integrated or joint contracting with small and medium systems rose sharply for PCPs and for specialists. In contrast, the percent vertically integrated or joint contracting with large systems fell slightly. Second, there was considerable geographic variation in the percent of physicians vertically integrated or joint contracting with large systems, which rose during the study period. Third, vertical integration and joint contracting with systems led to price increases that increased monotonically with the size of the system.

Our analyses capture both the vertical and horizontal components of vertical integration and joint contracting–in other words, the economic consequences of consolidation among physicians and between physicians and hospitals. While imperfect, we tried to disentangle the two phenomena in two ways that suggest vertical relationships matter. First, we controlled for physician HHI in our specifications—the horizontal component of vertical integration and joint contracting—and it did not significantly alter our findings. Second, we implemented a sensitivity analysis for a subset of physicians using a legal ownership definition of vertical integration and found similar price effects.

In contrast to some prior work examining the effects of hospital acquisitions of physician practices, which found substantial increases in facility fees,^28,30 we found little effect on facility fees except for a modest increase for specialists affiliated with the largest system. This may be because effects on facility fees are smaller in our setting or because our measure of vertical integration is less strongly associated with facility fee increases than hospital acquisitions of physician practices. The larger effect on facility fees for the largest system may be because our measure of vertical integration was more likely to reflect ownership for that system (Supplemental Exhibit 11).

Our primary measure of consolidation between physicians and health systems includes both ownership and contractual affiliations. As the documentation for the RPO data clarifies, “contracting affiliates are entities that the provider organization negotiates contracts on behalf of but does not own or control.” This is in contrast to other studies that focus on hospital acquisitions of physician practices^{28–30,33–35} or integration based solely on contractual networks.^8,42 Our sensitivity analysis using a legal ownership definition of vertical integration yielded similar but slightly larger effects. Because this sensitivity analysis was only possible for a subset of physicians and was subject to measurement error, we cannot determine whether the differing results were due to different effects of vertical integration compared to joint contracting, a different effect for the subset of physicians included in the sensitivity analysis, or attenuation bias due to measurement error. In any case, our finding indicates that a substantial proportion of the price effects for the main analysis sample are driven by changes in vertical integration defined as ownership.

Our finding of higher physician prices with vertical integration could result in higher or lower spending, the latter if integrated systems can coordinate and manage care to reduce quantities provided or channel patients to lower-cost sites of care. Prior studies did not find evidence that vertical integration of health systems with physician groups improved quality of care,^25,35 though more research is needed on this topic.

Policy Implications

In light of the apparent trend toward vertical integration and joint contracting among health care providers,⁴³ our findings suggest that policymakers may need to consider taking steps to mitigate price effects with a particular focus on large systems. Such steps could involve antitrust enforcement, increased oversight or regulation of commercial insurance contracts, and laws that promote competition.⁴⁴

While federal antitrust enforcement agencies and the Centers for Medicare and Medicaid Services hold some of the levers that affect provider competition, recent research has underscored the importance of state-specific factors as drivers of relative prices, pointing to the importance of state policy. Models for strengthening health care provider competition policy at the state level can be found in every region of the country.⁴⁴ In 2012 Massachusetts enacted one of the nation’s most comprehensive cost control laws, with a focus on price competition.⁴⁵ The law resulted in the establishment of the Massachusetts Health Policy Commission, which has the authority to conduct cost and market impact reviews of proposed mergers between physician practices and health systems, the results of which are included in recommendations to the Attorney General. The independent Commission consists of 11 members, none of whom may represent health care entities under its purview.

While the Health Policy Commission cannot block mergers, per se, its cost and market impact reviews appear to be influential. For example, the Hallmark physician group, with nearly 400 physicians, sought to be acquired by Partners in 2012 but ultimately called off this plan in 2015 due to antitrust concerns. Instead, Hallmark joined Wellforce, a much smaller health system.⁴⁶ Our results suggest that this type of oversight can have quantitatively important effects, especially in local markets where mergers are typically too small to invite scrutiny from federal antitrust authorities. Similar review of joint contracting arrangements may also be warranted. Relatedly, we would be remiss if we did not highlight the need for more detailed public data collection to better distinguish contractual affiliations from true vertical integration.

In addition to proactive oversight of mergers, acquisitions and joint contracting, the actions of policymakers, insurers, and employers to empower health care consumers with information and incentives to choose lower-cost providers may help mitigate the price effects of consolidation. To this end, employers and health plans have increasingly offered enrollees access to cost transparency tools and benefit designs that include tiered copayments, reference pricing, and incentives to seek care at centers of excellence. Such “steering” mechanisms have been shown to lower costs and put downward pressure on prices.^47–52 State policymakers can influence the adoption of steering mechanisms in several ways including the prohibition of “anti-steering” clauses in commercial insurance contracts and the promotion of steering tools in plans offered on state health insurance exchanges.

Conclusion

We find strong evidence that vertical integration and joint contracting between physicians and large health systems leads to large increases in physician prices, particularly for PCPs. Our findings complement a literature that has found increases in physician prices associated with hospital acquisitions of physician practices.^{28–30,33–35} More research is needed to assess potential benefits of vertical integration and joint contracting, particularly those that may improve patient outcomes, such as better care coordination or alignment of provider incentives. This body of research can inform policymakers seeking to foster improvements in quality of care while limiting growth in health care prices.