Physicians’ Participation in Medicaid Increased Only Slightly Following Expansion

Abstract

While most primary care physicians treated at least 1 Medicaid patient in 2013, Medicaid was a small share of their payer mix. From 2013 to 2015, most physicians maintained or slightly increased their Medicaid participation, with greater increases observed in states that expanded Medicaid eligibility. Medicaid patients remained concentrated among relatively few physicians.

Policymakers and researchers have long been worried about access to physician services for Medicaid recipients.¹ Existing estimates suggest that 75–95 percent of physicians accept Medicaid patients,^2–4 though Medicaid represents a small share of their payer mix⁵ and physicians frequently report not accepting any new Medicaid patients.^6,7 Little is known about how physicians responded to the recent expansion of Medicaid eligibility, with one report of a 9 percent increase in primary care appointment availability for new Medicaid patients in 10 states.⁸ To our knowledge, no research has studied changes in the distribution of Medicaid patients across primary care physicians (PCPs).

This report presents the first claims-based, nationwide estimates of PCP participation in Medicaid – before and after the Medicaid expansion. The data come from Healthcare View, a joint project of the Robert Wood Johnson Foundation and athenahealth, a cloud-based healthcare information technology company. Exhibit 1 shows a modest increase in Medicaid participation among PCPs from 2013 to 2015. This was driven by PCPs in states that expanded Medicaid eligibility. On average, PCPs in non-expansion states maintained their Medicaid participation.

EXHIBIT 1. Share of Primary Care Physicians’ Panel from Adult Medicaid Patients, 2013 to 2015.

SOURCE Authors’ analysis of data from Healthcare View.

NOTES Estimates derive from regressions (with separate analyses for all states, expansion states, and non-expansion states) of the share of each PCP’s panel comprising adult Medicaid patients on an indicator for year (2013 or 2015) and state fixed effects. Standard errors were clustered at the state level.

* Indicates statistical significance that the P<.05 level.

** Indicates statistical significance at the P<.01 level.

Study Data and Methods

This analysis compared PCP participation in Medicaid, between 2013 and 2015. The PCPs in this study came from practices that contracted with athenahealth to provide medical billing, electronic health record, and practice management services. Our sample included physicians with primary care specialties, including internal medicine, family practice, and general practice. We excluded pediatricians, as the Medicaid eligibility expansions predominantly affected adults and any response to the expansion should thus be concentrated within physicians with predominantly adult patient panels.

We further restricted our sample to PCPs with stable patient panels, defined as treating at least 50 patients in 2013 and 2015. This restriction avoided confounding changes in payer mix that might result from a physician entering or exiting the athenahealth platform. Finally, we excluded physicians in states that expanded Medicaid eligibility before⁹ or after 2014¹⁰ to facilitate a clean comparison of Medicaid participation among PCPs who were or were not affected by a 2014 Medicaid expansion in their state (see Appendix Exhibit 1A for sample inclusion criteria).¹¹

Our final sample included 3,820 PCPs, 52 percent of whom practiced in 18 states that expanded Medicaid in 2014. For each physician, we calculated their Medicaid participation as the annual count of distinct adult (aged 18–64) patients with Medicaid as their primary insurer, divided by the annual count of all distinct patients treated – as observed in the submitted claims for services rendered to patients of all payers. We computed descriptive statistics for the Medicaid participation measure and used regression analysis to examine changes over time and between Medicaid expansion and non-expansion states. To address time-invariant differences between states, we included state fixed effects when comparing changes over time and indicators for whether a state continued or let the Medicaid primary care payment increase expire in 2015¹², when comparing baseline Medicaid participation in expansion versus non-expansion states.

To assess the national representativeness of our sample, we compared PCPs in our sample to those in the Centers for Medicare and Medicaid Physician Compare database.¹³ Our sample resembled PCPs nationwide on observable characteristics including gender and practice size (Appendix Exhibit 2A).¹¹ However, our sample over-represented younger PCPs, PCPs practicing in the South, PCPs with specialty designations of family or general practice, and doctors of osteopathic medicine.

Our study had several limitations. First, our analyses were descriptive and did not support causal conclusions about the effect of the Medicaid eligibility expansion on physicians’ Medicaid participation decisions. Second, the Affordable Care Act-mandated increase of Medicaid primary care reimbursement expired during our study period, resulting in a sizeable payment cut for many PCPs, which may also have affected their Medicaid participation decisions. Third, our results derived from a convenience sample of PCPs and may not generalize to PCPs nationwide.

Study Results

BEFORE THE MEDICAID EXPANSION

Exhibit 2 shows that roughly nine in ten PCPs treated at least one adult Medicaid patient in 2013. However, adult Medicaid patients represented less than 10 percent of the patient panel for half of the PCPs in our sample. PCPs at the 90^th percentile of Medicaid participation had a patient panel comprising 20 percent adult Medicaid patients.

EXHIBIT 2. 2013 Share of Primary Care Physicians’ Panel from Adult Medicaid Patients.

SOURCE Authors’ analysis of data from Healthcare View.

NOTES We defined ‘0 Medicaid Patients’ as any physician who billed zero claims for Medicaid patients in that year. Estimates may not sum to 100 due to rounding error. Estimates derive from linear probability models that regressed an indicator for participation level (i.e., <5%) on Medicaid expansion status, controlling for whether the state continued the Medicaid primary care payment increase beyond 2014.

** Indicates statistical significance that the P<.01 level.

Medicaid participation in 2013 varied between PCPs in states that did or did not expand Medicaid eligibility the following year. More than 60 percent of PCPs in non-expansion states saw few (<5 percent of their patient panel) or no adult Medicaid patients, compared to 38 percent of PCPs in expansion states. Conversely, only 4 percent of PCPs in non-expansion states saw a large share of adult Medicaid patients (>=20 percent of their patient panel), compared to 16 percent of PCPs in expansion states.

AFTER THE MEDICAID EXPANSION

Exhibit 1 shows that PCPs increased their Medicaid participation by 1.6 percentage points on average, with adult Medicaid patients comprising 7.8 percent of the average PCP’s panel in 2013 and 9.4 percent in 2015. This overall increase was driven by changes in Medicaid participation among PCPs in expansion states, where the average PCP’s panel comprised 10.2 percent adult Medicaid patients in 2013 and 13.6 percent in 2015. We found no change in the Medicaid participation decisions of PCPs practicing in non-expansion states (see Appendix Exhibit 3A for full regression results).

Exhibit 3 shows how changes in Medicaid participation varied by level of participation in 2013, categorizing physicians in all states by the share of their patient panel comprising adult Medicaid patients in 2013 and displaying changes in Medicaid participation in 2015. Most physicians (65 percent) maintained their 2013 level of Medicaid participation. Of those who changed categories of Medicaid participation, more increased than decreased their adult Medicaid share of patient panel.

EXHIBIT 3.

Change in Primary Care Physicians’ Share of Patient Panel from Medicaid, by Participation Category, 2013 to 2015

2013 Adult Medicaid Share of Patient Panel	2015 Adult Medicaid Share of Patient Panel
	Lower Category	Same Category	Higher Category	P-value
		All States
0 Medicaid Patients	-	71%	29%	<.001
<5%	6%	71%	23%
5–9%	11%	49%	40%
10–19%	16%	58%	26%
>=20%	16%	84%	-
OVERALL	9%	65%	26%	<.001
		Expansion States
0 Medicaid Patients	-	69%	31%	<.001
<5%	2%	60%	38%
5–9%	4%	35%	61%
10–19%	5%	59%	37%
>=20%	9%	91%	-
OVERALL	4%	59%	37%	<.001
		Non-Expansion States
0 Medicaid Patients	-	72%	28%	<.001
<5%	9%	79%	12%
5–9%	21%	66%	14%
10–19%	37%	57%	6%
>=20%	46%	54%	-
OVERALL	14%	71%	15%	<.001

SOURCE Authors’ analysis of data from Healthcare View

NOTES Estimates may not sum to 100 due to rounding error. P-values are from Pearson chi-square tests.

The second and third panels of Exhibit 3 present changes in Medicaid participation, separating PCPs into those who practiced in Medicaid expansion and non-expansion states. In non-expansion states, 72 percent of physicians maintained their 2013 Medicaid participation level, while the remainder were slightly more likely to decrease their participation. In expansion states, 60 percent of physicians maintained their 2013 Medicaid participation level. PCPs in expansion states who changed their Medicaid participation between 2013 and 2015 were nine times more likely to increase than decrease their participation. Increased participation was most likely among PCPs with positive, but low Medicaid participation in 2013. Physicians who had treated zero Medicaid patients in 2013 were least likely to increase their Medicaid participation.

We repeated our analyses for pediatricians (Appendix Exhibits 4A-6A) and found that Medicaid patients represented a much larger share of pediatricians’ panels, with few initial differences or changes in Medicaid participation by expansion status. We also repeated our analyses for PCPs in states that expanded Medicaid early (prior to 2014) and late (after 2014), finding that PCPs in early-expansion states showed an increase in Medicaid participation similar to those in states that expanded Medicaid in 2014, while PCPs in late-expanding states showed a smaller, but significant increase (Appendix Exhibit 7A). Finally, we stratified our analyses by whether the PCP practiced in a state that retained the Medicaid primary care payment increase beyond 2014, finding the largest increases in Medicaid participation in expansion states with permanent payment increases. We found no significant change in Medicaid participation among PCPs in non-expansion states, regardless of whether the payment increases persisted beyond 2014 (Appendix Exhibit 8A).

DISTRIBUTION OF MEDICAID PATIENTS

From 2013 to 2015, the number of adult Medicaid patients seen by athenahealth physicians increased by roughly 20 percent (data not shown). In both years, adult Medicaid patients were concentrated among a small number of PCPs, with roughly 60 percent of adult Medicaid patients treated by 20 percent of all PCPs who participated in Medicaid (Exhibit 4).

EXHIBIT 4. Cumulative Primary Care Physician Share of Adult Medicaid Patients, 2015.

SOURCE Authors’ analysis of data from Healthcare View.

NOTES While this presents 2015 data, the same analysis for 2013 looks virtually indistinguishable.

Discussion

From 2013 to 2015, 9 out of 10 physicians in our sample maintained or slightly increased their Medicaid participation. Adult Medicaid patients continued to represent less than 10 percent of the average PCP’s patient panel. As such, adult Medicaid patients remained concentrated among a small number of PCPs, with roughly 20 percent of PCPs treating approximately 60 percent of adult Medicaid patients in 2013 and 2015. This suggests that greater demand for primary care among newly-insured Medicaid patients was accommodated by a small group of high-Medicaid physicians.

Patterns of Medicaid participation differed between Medicaid expansion and non-expansion states. PCPs in expansion states had higher initial levels of Medicaid participation than PCPs in non-expansion states. From 2013 to 2015, most PCPs maintained their Medicaid participation. Among those who changed their payer mix, PCPs in expansion states were much more likely to increase than decrease their Medicaid participation, compared to PCPs in non-expansion states, who were slightly more likely do decrease their participation. Increases in Medicaid participation in expansion states were most likely among PCPs with low to moderate Medicaid participation in 2013.

Conclusion

More than 15 million individuals gained coverage through the Medicaid expansion, making Medicaid the insurer for roughly one in seven adults. Despite the large gains in coverage, we found that most PCPs made no change or small increases in the share of their panel comprising adult Medicaid patients. This may be due to low reimbursement rates,¹⁴ administrative complexities,¹⁵ or other factors that make Medicaid less attractive relative to other payers.¹⁶ Additionally, PCPs’ Medicaid participation decisions may also respond to recent changes in health care provider organization, including increasingly narrow provider networks within commercial insurance, the shift to managed care for state Medicaid programs, and the ongoing trend of financial integration between hospitals and physician groups.¹⁷

We find that large changes in Medicaid program eligibility were associated with only modest realignment of PCPs’ payer mix towards Medicaid. As such, adult Medicaid patients remained concentrated among a small number of PCPs. To the extent that PCPs expanded their Medicaid participation, the largest increases occurred among physicians who were already participating in Medicaid prior to the eligibility expansion. These findings suggest that policy efforts to increase access to primary care services for Medicaid recipients may be most successful by targeting existing Medicaid providers.