Increases in Medicare Spending and Use after Private Equity Acquisition of Retina Practices

Abstract

Purpose:

Private equity (PE) firms increasingly are acquiring physician practices in the United States, particularly within procedural-based specialties such as ophthalmology including retina. To date, the potential impact of ophthalmology practice acquisitions remains unknown. We evaluated the association between PE acquisition and Medicare spending and use for common retina services.

Design:

Retrospective difference-in-differences analysis using the 20% Medicare fee-for-service claims dataset from January 1, 2015, through December 31, 2019.

Participants:

Eighty-two practices acquired by PE during the study period and matched control practices.

Methods:

We used novel data on PE acquisitions of retina practices linked to the 20% sample Medicare claims data. Retina practices acquired by PE between 2016 and 2019 were matched to up to 3 non-PE (control) practices based on characteristics before acquisition. Private equity-acquired practices were compared with matched control practices through 6 quarters after acquisition using a difference-in-differences event study design. Data analyses were performed between August 2022 and April 2023.

Main Outcome Measures:

Medicare spending and use of common retina services.

Results:

Relative to control practices, PE-acquired retina practices increased the use of higher-priced antievascular endothelial growth factor (VEGF) agents including aflibercept, which differentially increased by 6.5 injections (95% confidence interval, 0.4–12.5; P = 0.03) per practice-quarter, or 22% from baseline. As a result, Medicare spending on aflibercept differentially increased by $13 028 per practice-quarter, or 21%. No statistically significant differences were found in use or spending for evaluation and management visits or diagnostic imaging.

Conclusions:

Private equity acquisition of retina practices are associated with modest increases in the use of higher-priced anti-VEGF drugs like aflibercept, leading to higher Medicare spending. This finding highlights the need to monitor the influence of PE firms’ financial incentives over clinician decision-making and the appropriateness of care, which could be swayed by strong economic incentives.

Financial Disclosure(s):

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Acquisitions of medical practices by private equity (PE) firms have become increasingly prevalent in recent years. This trend particularly has affected ophthalmology, which is among the most commonly targeted office-based specialties for PE acquisition, ranking second only to dermatology in 2018.^1,2 Private equity acquisitions in ophthalmology tripled between 2016 and 2019, with more than 10% of ophthalmologists estimated to be working in PE-acquired settings as of 2019.^2–5 Although most PE investments within ophthalmology have focused on large multispecialty practices, acquisitions of subspecialty retina practices also accelerated between 2017 and 2019.^2,4,5

Prior research showed that physician practices acquired by PE increased health care spending among commercially insured populations both through higher patient volume and higher payments per service relative to non-PE practices.^5–8 Private equity ownership also has been associated with changes to the clinical workforce structure.⁹ In procedural specialties such as ophthalmology, concern has emerged that the growth of PE investment may alter care delivery patterns, shifting care toward more lucrative procedures or billing practices to drive investor returns. Specifically, within retina care, PE’s profit potential may be related directly to the volume of intravitreal injections administered and payment to retina practices as a percentage of the sales price of increasingly expensive drugs. Although evidence continues to emerge about PE acquisitions of physician practices,^2–5 no reports have analyzed the implications of such investments within subspecialties of ophthalmology, particularly retina care. The entry of PE investment into the field has had particular impact, generating intense interest from vitreoretinal surgeons, trainees, and professional societies.

Private equity can leverage a few specific mechanisms within the clinical setting of retina practices to drive profitability. First, by consolidating smaller practices through add-on acquisitions, PE improves negotiating power for favorable volume-based rebate contracts from pharmaceutical manufacturers. Second, PE can leverage Medicare reimbursement policies for intravitreal injections, particularly those involving anti–vascular endothelial growth factor (VEGF) agents such as ranibizumab and aflibercept. These drugs account for nearly 95% of all ophthalmology drug reimbursements.^2,10,11 Although anti-VEGF agents such as bevacizumab, aflibercept, and ranibizumab each are used commonly to treat macular degeneration, diabetic retinopathy, diabetic macular edema, and retinal vein occlusion,^11–14 prices differ substantially; average per-dose prices for aflibercept and ranibizumab are more than 28 times that of bevacizumab.¹⁵ In traditional Medicare, providers are reimbursed for Part B drugs based on the average sales price of the drug plus a 4.3% add-on payment¹⁶ to cover expenses related to purchasing variability, shipping fees, patient monitoring, and storage and handling requirements. Add-on payments increase reimbursement for high-cost drugs and may incentivize the prescription of higher-cost drugs even when a similarly effective lower-cost option exists.^15,17–20 Coupled with PE’s focus on generating 20% annual returns over short investment horizons, this payment structure may increase Medicare spending on higher-cost anti-VEGF agents, either through switching existing patients from lower-priced to more expensive agents or through prescribing higher-priced agents to new patients.

This study contributes new knowledge on PE in health care by examining the unique case of retina practices. Specifically, we examined changes in Medicare spending and use after PE acquisitions of retina practices. We focused on 4 common categories of retina services: retina procedures; commonly injected anti-VEGF agents, including bevacizumab (Avastin; Genentech), aflibercept (Eylea; Regeneron), and ranibizumab (Lucentis; Genentech); diagnostic imaging services; and evaluation and management visits.

Given the higher prevalence of age-related retina disease in older adults, we examined the Medicare population. Linking novel data on PE acquisitions to the 20% sample Medicare fee-for-service claims data from 2015 through 2019, we used a difference-in-differences event study design to examine changes in practice-level Medicare spending and use after PE acquisition.

Methods

Study Design

This retrospective difference-in-differences analysis compared spending and use for common retina services at practices acquired by PE versus practices not acquired by PE. We combined several data sources to identify physicians affiliated with PE-acquired practices. We constructed our analytic sample in multiple steps, first identifying PE acquisitions of physician practices, then identifying multiple practice sites and physicians associated with each acquisition, and finally linking this information to outcome measures constructed using the 20% random sample of Medicare fee-for service claims. This dataset includes a 20% representative, random sample of all Medicare beneficiaries 65 years of age and older with fee-for-service coverage of Medicare Parts A, B, and D. Data analyses were conducted from August 2022 through April 2023. The study was deemed exempt by Oregon Health & Science University’s Institutional Review Board (identifier, 22582). All research adhered to the tenets of the Declaration of Helsinki. The requirement for informed consent was waived because of the retrospective nature of the study.

Data Sources

Identifying Private Equity Acquisitions.

First, to identify PE acquisitions of ophthalmology practices from 2016 through 2019, we used proprietary data from PitchBook, Inc, a financial database that has been used by other studies examining PE in health care.^21,22 We manually verified and expanded this list using a combination of press releases, industry reports, and physician practice websites.

Linking Acquisition Deals to Practices.

To identify individual physicians (MD and DO degree holders) affiliated with both acquired and nonacquired practices, we used 2 databases from IQVIA, a health care data vendor: the 2016 SK&A Office Based Physicians database and the 2019 OneKey database, which use the same approaches to identify and verify affiliations.²³ Both datasets are verified independently with clinician-level information (e.g., age, location, specialty, clinician credentials, National Provider Identifiers) and practice-level information, including ownership and corporate affiliations, for 9.7 million health professionals in the United States.^24,25 We used probabilistic record linkage algorithms to link exact and nonexact records of practice names, addresses, and ownership information (e.g., parent organization) in the 2019 OneKey data to reported acquisitions.

Identifying Physicians Affiliated with Practices.

Next, to facilitate linkages to claims data, we identified physicians located at acquired and nonacquired practice sites. Because the linkages between physicians and their practice sites were available only in 2016 and 2019, we limited our analytic sample to physicians who remained at the same practice before and after acquisition, in 2016 and 2019, respectively. Further details on assembling this acquisition data are described in an earlier study.⁵ Using provider National Provider Identifiers, we linked our sample of physicians to the 20% sample Medicare dataset from January 1, 2015, through December 31, 2019. Following prior research, we considered all physicians submitting claims under a common federal Tax Identification Number to belong to the same practice.^26–28

Identifying Vitreoretinal Specialists

Our primary specification included claims billed by retina specialists. To identify surgical and medical retina specialists, we used an algorithm that relied on physicians’ billing history (summarized in Fig S1, available at www.aaojournal.org). First, for each ophthalmologist (physician specialty code 18) over the study period, we calculated the share of retina procedures among the provider’s total procedural volume. Using the distribution of each ophthalmologist’s share of procedures that were retina-specific procedures, we classified physicians with an above-median share to be medical retina specialists.

Next, we identified surgical retinal specialists as physicians who performed a vitrectomy for retinal detachment repair (Current Procedural Terminology code 67108) at least once per year. Using this approach outlined above, from among 19 529 ophthalmologists identified in the study period, we flagged 2703 physicians as medical retina specialists (14%) and 1448 physicians as surgical retina specialists (7%), which is consistent with estimates from the literature.²⁹

Study Variables

Primary outcomes at the practice-quarter level included total Medicare spending and services billed by retina specialists. We defined outcomes separately for 4 service categories using the following Current Procedural Terminology and Healthcare Common Procedure Coding System codes: retina procedures (66720, 66761, 66821, 67031, 67015, 67025, 67028, 67101, 67105, 67110, 67141, 67145, 67208, 67210, 67220, 67227, 67505, 67228, 67515, 65800, 67820, 67221, 67225, 67036, 67039, 67040, 67228, 67108, 67113, 67042, 67041, 66850, 66986, and 66682); injections of commonly used anti-VEGF agents, including bevacizumab (Avastin, J7999 and J9035), aflibercept (Eylea, J0178), and ranibizumab (Lucentis, J2778); diagnostic imaging services (76511–2, 76529, 92020, 92083, 92201–2, 92235, 92250, 92283, 92240, 92132–134, 92136, and 92286–7); and Evaluation and Management visits (92004–92014, 99202–99205, and 99212–99215; Table S1, available at www.aaojournal.org).

Statistical Analysis

We used an event study approach with difference-in-differences estimation to compare outcomes in PE-acquired practices with those of matched control practices. Private equity-acquired practices were matched with control practices in 2015, the year before any PE acquisition in our sample. Nonacquired control practices were identified using 3:1 caliper matching without replacement that matches within 1 standard deviation for the total number of surgical retina specialists in a practice. Given regional variation in use of on-label drugs, we also required matched control practices to be in the same state as the acquired practice, defined as the state where the largest number of billed claims originated.³⁰ Overall, 82 PE-acquired practices were matched to 218 control practices (89% match rate), with 66 PE practices (72%) matching to a full set of 3 matched control practices. We compared practice characteristics before acquisition for acquired practices and matched control practices. We also examined trends in outcomes before acquisition between PE and non-PE practices.

In the event study analyses, event time 0 denoted the quarter of acquisition. We lined up all practices (acquired practices and their matched control practices) and their outcome data at event time 0, the quarter of acquisition. We used data from up to 6 quarters before acquisition (event time −6, . . . , −1) through up to 6 quarters after acquisition (event time +1, +2, . . . , +6), with the quarter of acquisition as the reference period. The unit of analysis was the practice-quarter. A linear difference-in-differences regression model compared changes in outcomes in PE-acquired practices relative to those in matched control practices from before to after acquisition. In difference-in-differences analyses, the quarter before and after acquisition were excluded from analyses as a washout period.

Adjusted differential changes in percentage terms for outcomes of interest were derived by dividing estimates from event study models by the unadjusted mean of the outcomes among PE practices before acquisition. All regressions included practice and time fixed effects to account for time-invariant practice-specific trends as well as secular time trends in the outcomes. To mitigate concerns that results were driven by differences in practice size, for example, variation in number of physicians, all regressions were weighted by the average practice size (number of physicians) over the study period. Standard errors were clustered at the level of the practice. Several robustness checks assessed the sensitivity of results to: (1) use of regression weights, (2) approach to identifying retina specialists, (3) unit of analysis, (4) alternate matching methodology, and (5) differential timing of acquisition and heterogeneous treatment effects.³¹

Results

Our final analytic sample included a total of 82 PE-acquired practices and 218 matched control practices. Table 2 reports summary statistics for PE-acquired practices and their matched control practices at baseline in 2015 before any acquisition, and Table 3 summarizes the increase in number of acquisitions after 2016.

Table 2.

Characteristics of Practices Acquired by Private Equity and Those Not Acquired by Private Equity before and after Matching, 2015

	Before Matching			After Matching
Characteristic	Non–Private Equity (n = 1177)	Private Equity (n = 92)	P Value	Non–Private Equity (n = 218)	Private Equity (n = 82)	P Value
Practice size and composition
Surgical retina specialists	0.48 ± 0.76	1.18 ± 1.70	< 0.001	0.66 ± 0.61	0.77 ± 0.73	0.18
Medical retina specialists	0.89 ± 0.66	1.29 ± 1.17	< 0.001	0.79 ± 0.67	1.33 ± 1.24	< 0.001
No. of physicians	3.45 ± 3.42	7.62 ± 5.84	< 0.001	3.59 ± 3.37	7.33 ± 5.99	< 0.001
Use (no. of services per practice)
Evaluation and management	258.42 ± 413.05	586.00 ± 909.32	< 0.001	266.25 ± 418.59	551.66 ± 916.88	< 0.001
Imaging services	553.57 ± 606.35	1085.48 ± 1410.19	< 0.001	676.36 ± 677.92	868.99 ± 1070.11	0.07
Retina procedures	207.47 ± 277.49	405.20 ± 676.02	< 0.001	236.48 ± 271.08	271.82 ± 343.52	0.36

Data are presented as mean ± standard deviation unless otherwise indicated. Practices not acquired by private equity represent independently owned physician practices identified using 3:1 caliper matching without replacement. Matching algorithm requires an exact match on state and within 1 standard deviation for practice size (total number of surgical retina specialists in a practice).

Table 3.

Acquisitions by Year, 2016–2019

Year	No. of Identified Acquisitions
2015	0
2016	7
2017	16
2018	33
2019	26
Overall	82

Counts of practices represent the number of practices identified as being acquired by private equity in the study period. For private equity acquisitions of ophthalmology practices from 2016 through 2019 we used proprietary data from PitchBook, Inc, a financial database that has been used by other studies examining private equity in health care. We manually verified and expanded this list using a combination of press releases, industry reports, and physician practice websites. Next, practices with retina specialists were identified using a multistep process to identify retina specialists described in the text. The number of identified acquisitions in this table represents the number of practices with retina specialists in the Medicare data, using the federal Tax Identification Number associated with a claim as a proxy for practice boundaries.

At baseline before any acquisition, PE-acquired practices were larger than matched control practices, as measured by the number of ophthalmologists (5.23 in PE-acquired practices compared with 2.15 in nonacquired practices; P < 0.001). However, no statistically significant differences were found in the number of surgical retina specialists per practice (0.77 vs. 0.66; P = 0.18). Similarly, although PE-acquired practices and matched control practices differed on the basis of volume of evaluation and management visits and diagnostic imaging services before acquisition, no statistically significant differences were found in the volume of retina procedures performed at PE-acquired practices versus at matched control practices (271.82 procedures vs. 236.48 procedures, respectively; P = 0.36). Despite differences in means before acquisition, trends before acquisition were parallel among primary outcomes, supporting the identifying assumption that group differences would remain constant in the absence of acquisition (Fig S2, available at www.aaojournal.org).

In adjusted event study estimates, Figure 3 shows spending and use of common retina services before and after PE acquisition among acquired and control practices. Before acquisition, some differences in Medicare spending and use were found between PE-acquired practices and matched control practices (Table S4, available at www.aaojournal.org, summarizes joint F tests). After PE acquisition, acquired practices demonstrated a consistent differential increase in use and spending for retina procedures and aflibercept (Fig 3, Table 5).

Figure 3.

A–L, Graphs showing changes in Medicare spending and use associated with private equity (PE) acquisition, by quarter: (A) use, retina procedures; (B) total spending, retina procedures; (C) use, aflibercept; (d) total spending, aflibercept; (E) use, bevacizumab; (F) total spending, bevacizumab; (G) use, ranibizumab; (H) total spending, ranibizumab; (I) use, evaluation and management (E&M); (J) total spending, evaluation and management; (K) use, imaging; and (L) total spending, imaging. The figures present event study coefficients corresponding to PE acquisition of physician practices. In the event study regression, we compared outcomes in PE-acquired practices with those of matched control practices. Each point in the respective figure panels represents the coefficient obtained by estimating an event study regression that includes practice and quarter fixed effects and with the unit of analysis at the practice-quarter level. Error bars represent 95% confidence intervals. Outcomes represent total number of services and total spending per practice for services provided by physicians identified as retina specialists. The vertical dash line represents the quarter of acquisition that serves as the reference period. Event time 0 denoted the quarter of acquisition. We looked back up to 6 quarters before acquisition (event time −6, . . . , −1) and 6 quarters after (event time +1, +2, . . . , +6). Standard errors are clustered at the level of the practice.

Table 5.

Event Study Differential Change in Medicare Spending and Use for Private Equity and Non–Private Equity Practices

		Differential Change in Each Quarter after Acquisition, Coefficient (Standard Error)
Variable	Before Acquisition Baseline	Quarter 1	Quarter 2	Quarter 3	Quarter 4	Quarter 5	Quarter 6
Procedures
Spending	16 861 ± 19 365	458.9 (569.1)	1217.4 (569.1)	1131.5 (593.2)	2140.2 (676.5)	2117.5 (809.3)	1289.8 (744.5)
Use	89 ± 95	3.6 (2.1)	8.2 (2.7)	10.2 (11.9)	11.9 (4.1)	11.8 (4.31)	12.1 (4.8)
Aflibercept
Spending	68 299 ± 106 932	4824.1 (3344.7)	14829.9 (4018.3)	17 487.2 (4018.3)	24 115.3 (6320.0)	27 080.8 (7488.6)	34017.5 (10 906.6)
Use	33 ± 49	2.6 (1.7)	7.2 (1.8)	9.0 (2.2)	11.7 (2.9)	12.8 (3.4)	15.1 (4.7)
Bevacizumab
Spending	1897 ± 2583	89.3 (140.1)	87.6 (140.1)	293.3 (183.3)	278.3 (169.7)	203.1 (180.7)	140.2 (211.9)
Use	25 ± 34	0.6 (1.3)	0.2 (1.7)	1.6 (2.1)	1.7 (2.2)	0.5 (2.3)	−0.9 (3.2)
Ranibizumab
Spending	35 306 ± 50 231	4339.6 (2486.7)	3154.1 (2787.5)	3833.9 (4115.2)	−570.2 (3354.2)	-2329.8 (4584.7)	−1907.1 (5645.7)
Use	18 ± 26	1.9 (1.1)	2.2 (1.6)	2.8 (2.7)	0.9 (2.1)	−0.8 (2.5)	0.1 (3.1)
Evaluation and management
Spending	17 893 ± 27 042	29.7 (314.6)	−477.6 (328.3)	−511.1 (370.7)	−662.3 (357.1)	−532.4 (431.9)	−520.3 (444.7)
Use	164 ± 263	−0.9 (2.8)	−5.6 (2.9)	−5.8 (3.5)	−7.8 (3.6)	−6.3 (4.3)	−6.6 (4.4)
Imaging
Spending	15 991 ± 19 675	−144.8 (170.1)	−184 (236.1)	−93.8 (296.5)	−291.7 (316.1)	−309.3 (356.5)	−455.5 (426.4)
Use	290 ± 325	−0.01 (2.7)	1.6 (3.7)	5.1 (4.8)	3.3 (5.6)	5.2 (6.4)	4.8 (7.13)

Data are presented as mean ± standard deviation. Event study coefficients are estimated using a linear difference-in-differences model that includes practice and time fixed effects and are weighted by the total number of physicians per practice over the study period. Standard errors are clustered at the level of the practice. Outcomes represent total number of claims and total Medicare spending per practice, billed by physicians identified as retina specialists (surgical or medical). Event time 1 denotes the first quarter after acquisition. Spending is measured as total Medicare spending in United States dollars per practice-quarter, and use is measured as the number of services billed to Medicare per practice-quarter.

Table 6 shows the means before and after acquisition, along with differential changes in outcomes for PE-acquired practices and matched control practices. Across all quarters after acquisition, Medicare use of aflibercept differentially increased by 6.45 injections per practice-quarter or 22.2% (95% confidence interval, 0.4%–12.5%; P = 0.03) in PE-acquired practices relative to matched control practices. As a result, Medicare spending on aflibercept differentially increased by $13 028 per practice-quarter (95% confidence interval, $860–$25 196; P = 0.03), or 21.1% percent relative to matched control practices.

Table 6.

Differential Change in Medicare Spending and Use for Private Equity and Non–Private Equity Practices

	Private Equity Practices		Control Practices		Differential Change
Variable	Before	After	Before	After	Unadjusted	Adjusted Difference-in-Differences (95% CI)	%	P Value
Procedures
Spending	15 966 (18 460)	17 493 (21 876)	11 724 (12 600)	12 035 (13 511)	1216	676.4 (−1008, 2360)	4.2	0.43
Use	81 (88)	93 (105)	66 (75)	73 (86)	5	6.6 (−1.3, 14.5)	8.1	0.1
Aflibercept
Spending	61 848 (99 434)	80 678 (110 292)	44 480 (66 203)	65 011 (87 444)	−1701	13028 (860, 25 196)	21.1	0.03
Use	29 (46)	39 (51)	22 (32)	31 (41)	1	6.5 (0.4, 12.5)	22.2	0.03
Bevacizumab
Spending	1423 (2071)	2033 (2895)	2037 (3378)	2700 (4289)	−53	331.2 (−292.6, −954.9)	23.3	0.29
Use	20 (27)	25 (34)	28 (44)	32 (49)	1	1.9 (−5.2, 8.9)	9.4	0.6
Ranibizumab
Spending	31 497 (44 722)	39 819 (53 117)	25 751 (49 400)	32 351 (57 963)	1722	8745.3 (−10 324, 27 815)	27.8	0.38
Use	16 (22)	22 (29)	13 (23)	17 (29)	2	5.4 (−4.5, 15.3)	33.6	0.28
Evaluation and management
Spending	16 950 (26 427)	18 095 (27 385)	7476 (9675)	7269 (9116)	1562	−598.1 (−1951, −754)	−3.52	0.38
Use	157 (258)	163 (260)	69 (102)	64 (88)	11	−7.6 (−22.3, 7.1)	−4.80	0.31
Imaging
Spending	16 180 (20 062)	15 963 (23 224)	10 569 (10 331)	9450 (8806)	902	−160.2 (−1332, 1011)	−0.10	0.78
Use	273 (301)	313 (415)	183 (170)	195 (181)	28	11.8 (−4.9, 28.6)	4.33	0.16

Unadjusted and adjusted differential changes in outcome variables averaged at the practice-quarter level for private equity practices and matched control practices. Adjusted regression coefficients are estimated using a linear difference-in-differences model that includes practice and time fixed effects and a washout period including the quarter before and after acquisition. Regressions are weighed by the total number of physicians per practice. “Before” and “after” estimates are presented with standard errors clustered at the level of the practice. Adjusted percentage differential change is calculated by dividing the adjusted differential change obtained from the difference-in-differences regression by the mean for private equity–acquired practices before acquisition. Spending is measured as total Medicare spending in United States dollars per practice-quarter, and use is measured as the number of services billed to Medicare per practice-quarter.

Although Medicare spending and use for retina procedures increased by 8% and 4%, respectively, this was not statistically significant. In addition, no statistically significant differential changes were found in the spending or use associated with other anti-VEGF agents, specifically, bevacizumab and ranibizumab, as well as spending or use associated with evaluation and management visits or diagnostic imaging.

Our findings were robust to (1) unweighted regression specifications (Fig S4, available at www.aaojournal.org), (2) an alternate approach to identifying retina specialists (Fig S5, available at www.aaojournal.org), (3) a physician-quarter unit of analysis (Fig S6, available at www.aaojournal.org), and (4) an alternate matching approach that improves the baseline balance in practice characteristics before acquisition (Table S7, Fig S7, available at www.aaojournal.org). We also performed an analysis to determine the sensitivity of our results to differential timing of acquisition and heterogeneous treatment effects using the Callaway and Sant’Anna (2021) estimator. Although results are qualitatively similar using this specification (Fig S8, available at www.aaojournal.org), they are not statistically significant.

Discussion

Using novel data of PE transactions linked to a 20% random sample of Medicare fee-for-service claims, we found that PE acquisitions increased the use of higher-priced anti-VEGF agents like aflibercept, leading to increases in Medicare spending. Extrapolating our quarter-level findings to the entire fee-for-service population translates into annual increases in Medicare spending by more than $260 000 per acquired practice per year (e.g., $13 000 multiplied by 5 for the total population, and by 4 quarters for an annual estimate). Although this is likely to be an underestimate, it is modest compared with annual Medicare drug expenditure on aflibercept of roughly $1.4 billion on average.³² We did not find any significant change in the prescribing patterns for lower-cost anti-VEGF agents that would be indicative of patient switching. However, this might be the result of the relatively small sample of acquired practices, the recency of these acquisitions, and a limited follow-up period. Therefore, we consider these novel findings still to be preliminary. Higher use of aflibercept may be appropriate clinically and may suggest increased access to care; yet, it could also signal potential overuse of higher-priced drugs or lower-value spending imposed on Medicare, consistent with other studies examining PE acquisitions of physician practices.^5–8

Our findings suggest that PE firms, or the new management that oversees the acquired practices, may influence retina specialists’ clinical choices. However, whether the PE incentives to pursue rapid returns on investment are the dominant explanation remains unclear. Other factors such as patient selection, patient or physician preferences, and Food and Drug Administration approval status also matter. For example, some patient populations with worse initial visual acuity may benefit differentially from aflibercept versus bevacizumab, suggesting a role for nuanced clinical considerations.³³ Future research should examine the extent to which the increased use of higher-cost anti-VEGF agents like aflibercept are driven by the financial incentives of PE firms relative to other considerations. To the extent that practice management and ownership indeed affect physicians’ prescribing and practice behaviors, it is important to understand the downstream effects of these changes on quality of care and patient well-being.

We did not find any statistically significant differential change in the volume or spending associated with evaluation and management visits or diagnostic imaging service billed by retina specialists. One possible explanation is that these practices are seeing patients at capacity and are unable to increase visit volume in the short run. Another potential explanation may be that under PE ownership, retina specialists are more likely to practice at the top of their license (e.g., performing intravitreal injections) and may be discouraged from performing routine services (e.g., eye examinations) that also can be delivered by general ophthalmologists and optometrists. Examining potential shifts in the identity and composition of clinical providers who perform various services, although beyond the scope of this analysis, is necessary to understanding the managerial and staffing strategies that PE firms may pursue in retina practices to grow profitability.

Our findings contribute to the growing body of literature evaluating the rise of private equity in physician practices. As evidence continues to emerge about how PE acquisitions can contribute to rising health care spending with poorly understood implications for patient quality and access, various policy levers may help to align the incentives of PE firms with that of clinicians, patients, and taxpayers. Greater transparency of health care provider ownership data or transaction disclosures could facilitate closer monitoring not only of PE ownership of medical practices, but also of other corporate and health system ownership. Revisiting corporate practice of medicine doctrines–which currently are active in some, but not all, states–may help to solidify guardrails against undue corporate influence on clinical decision-making, as can restrictions against fee splitting for physician services with nonclinical entities. Given our findings specific to costly injectable drugs, changes to Medicare payment structure for Part B drugs, which generally are a higher-revenue service, also may be worth examining. For example, where competition can be enhanced, payment policies that incentivize physicians to choose the lower-priced option, when clinically appropriate, may discourage potentially wasteful spending.¹⁵ Other gatekeeping tools, such as prior authorization or step edits, also can be used as a tool to manage use of expensive Part B drugs^34,35; however, this involves potential tradeoffs between drug spending, administrative burden, and patient access.

Although this study fills an important gap in the understanding of PE and retina care in traditional Medicare, several limitations are worth noting. First, our analyses use the 20% sample Medicare claims dataset, which underestimates the aggregate effect of PE acquisitions on the Medicare program. Extrapolating our quarter-level findings to annual estimates for the entire Medicare fee-for-service population sheds additional light on the likely magnitude of demonstrated changes,³⁶ although we are unable to generalize these findings to Medicare Advantage or other commercially insured populations. Second, given the relative recency of these practice buyouts and limited data from these acquisitions, both our estimates and their statistical significance likely are limited by the sample size and narrow follow-up period. With additional years of data, future work can examine whether PE ownership results in differential use patterns among other high-cost drugs, including recently approved drugs such as faricimab (Vabysmo; Genentech) and pegcetacoplan (Syfovre; Apellis Pharmaceuticals). Third, given the lack of reporting and disclosure requirements, some PE acquisitions may not have been observed in the data. Fourth, practices targeted by PE firms may differ from other physician practices in unobserved ways; some of these differences persist even after matching acquired practices with nonacquired practices at baseline. Fifth, our findings may have limited generalizability to other settings within ophthalmology, such as multispecialty or optometry practices, which may have different service lines and revenue-generating mechanisms. Other potential changes to physician prescribing behavior, including changes to frequency of drug use, use of drugs for off-label purposes, or changes in doses administered, remain key areas for future research. In addition, our claims-based approach to classifying physicians to practices based on the Tax Identification Numbers may result in misclassification at the practice level, although our results are consistent using a physician-level, rather than practice-level, unit of analysis. Nevertheless, these findings raise important concerns about PE’s potential contribution to overuse and unnecessary spending that has been documented in the United States health care system.

Conclusions

This study provides preliminary evidence that PE acquisitions of retina practices modestly increase the use of higher-priced anti-VEGF drugs in the Medicare population, with increased spending for the Medicare program. Despite our relatively small sample size and limited follow-up period, findings are consistent with other studies that have high-lighted PE’s potential for overuse and unnecessary spending.

Supplementary Material

Supplemental material available at www.aaojournal.org.

Abbreviations and Acronyms:

PE

private equity

VEGF

vascular endothelial growth factor