When Michigan expanded Medicaid eligibility to low-income adults with incomes up to 138% of the federal poverty line via the “Healthy Michigan Plan” (HMP), the state’s uninsured rate fell from 15% in 2013 to 6% by the end of 2017,1,2 and increased accessibility to primary care providers (PCPs) on par with national trends.3,4 Although the Affordable Care Act’s (ACA) coverage expansions largely removed many financial barriers to seeking medical care and spurred increased health care utilization,5–11 Under Medicaid expansion, uninsurance rapidly fell due to rising Medicaid enrollments, especially in rural settings areas likely with limited appointment availability before expansion.12,13 However, Medicaid coverage alone may be insufficient to increase access to care for new enrollees due to preexisting provider shortages and constraints on provider capacity.14–16 Geographical context is an important nonfinancial factor influencing access to care,17,18 but little is known about how preexisting geographic access to primary care moderated the impacts of the Medicaid expansion.
The limited capacity of existing health care supply in these areas and other markets with chronic health professional shortages can be a substantial barrier to access, despite increases in insurance coverage. Our study aims to assess the extent to which local provider supply influences providers’ acceptance of new Medicaid patients following Michigan’s 2014 Medicaid expansion. Michigan, like other states, has substantial within-state variation in the relative size of its health care workforce. Accommodating newly insured Medicaid patients could be easier in provider-dense markets where the increase in primary care visits are more evenly distributed across practices. Furthermore, provider-dense markets may also have a larger concentration of safety net providers (eg, community health centers focused on caring low-income patients), which may absorb the increased demand for primary care.
Michigan Simulated Patient File (MSPF)
We used data gathered from a simulated patient (secret shopper) study conducted in Michigan from 2014 to 2015 known as “MSPF.”3 The MSPF is a representative sample of Michigan primary care practices, with sampling stratified by practice size and percentage of preACA Medicaid-eligible uninsured individuals in the practice’s county. In addition to the characteristics of 295 clinics, the MSPF includes appointment availability for Medicaid patients. Trained research staff posing as prospective patients to request an appointment interviewed primary care clinics at 1 month before and at 4-month intervals after the state’s Medicaid expansion. This provides a single preexpansion and 3 postexpansion periods. Callers requested the earliest appointment with any provider in the practice, and appointments were considered available if callers were offered a specific appointment date and time.
Physician Compare National Downloadable File (PCND)
Provider density is derived using the 2013 Physician Compare National Downloadable files19—a repository of medical care providers participating in the Medicare program. PCND reports on several provider characteristics including provider type [eg, physicians, physician assistant (PA), nurse practitioner (NP)] and addresses of their principal practice location. We restricted providers in the PCND to the following: (1) primary care physicians with primary specialties in family practice, internal medicine, geriatrics, obstetrics-gynecology or acting as general practitioners, and (2) nonphysician providers (ie, PAs and NPs) with a NPI (See Appendix for additional details, Supplemental Digital Content 1, http://links.lww.com/MLR/B756).
All analyses were completed using Stata version 15.1. The ZIP code centroids of the principal practice locations of providers in the PCND were geocoded using the Census Bureau’s Geocoder.20 Using “geonear”—a user-written program for Stata—we identified the number of unique PCP located within a 10-mile radius of every ZIP code in Michigan.21 We then scaled this total per 10,000 population within a 10-mile radius of each ZIP code centroid using data from the Census Bureau’s American Fact Finder and linked to the MSPF.22 This PCP supply measure approximates the total primary care supply within and around the ZIP code of where the practices in the MSPF are located. Our use of the 10-mile radius mirrors an earlier study of Michigan primary care markets, which define “local” providers as those within 10 miles of patients.23
To determine if the density of local PCPs within a geographic market has any moderating effects on the physician’s willingness to accept an additional patient, we use an event-history analysis embedded within a difference-in-differences framework:
The dependent variable of interest (Yizct) is appointment availability among simulated Medicaid patients at clinic i in zip code z of county c at period the audit call took place t. We measure supply (Supplyz) as the natural log of the number of PCPs per 10,000 population to account for the skewness in its distribution. For simplicity and as a sensitivity test, we also measure supply as a binary variable equaling one where provider supply in the practice’s 10-mile radius exceeded the median (92 PCPs per 10,000 population). ZIP codes with more at least 92 PCPs per 10,000 are treated as “provider rich” areas, while ZIP codes with fewer than 92 PCPs are treated as “provider poor.”
Key policy parameters—δ1, δ2, and δ3—identify distinctive trends for clinics at each wave of the MSPF. A statistically, and substantively, significant deviation in the trends in appointment availability for clinics in resource poor areas indicates that providers in low-density areas were not adequately able to respond to the increased demands for primary care, in Michigan’s first year under Medicaid expansion. The 3 primary policy parameters (ie, the interactive effect between the density of providers within the local area and time since Michigan’s expansion) discern if any differential effects were constant across time, or if providers adjusted their capacity to take on the increased demand.
Clinic fixed effects captured in θi allow us to control for unobservable time-invariant heterogeneity between the clinics. Period fixed effects (σt) control variation based on the timing of the audit call. Regression results are presented using linear probability models with county-level clustered heteroscedasticity robust SEs.
Table 1 presents the preexpansion characteristics of clinics in low-supply versus high-supply areas. At baseline, appointments were almost universally made with physicians (99%) in the areas with higher densities of providers. In areas with lower densities of providers, nonphysicians saw a larger share (15%) of the visits than clinics in areas with higher densities of providers (1.3%). The clinics in the study were similar in terms of other practice-level characteristics. Clinics in areas with low provider concentration were more likely to be in rural markets, markets with lower median household income, areas of lower unemployment, and areas with lower shares of racial/ethnic minorities.
Figure 1 presents trends in clinic-level Medicaid acceptance rates, physician appointment availability, and nonphysician appointment availability by market provider density. In low provider density areas, wait times for scheduled primary care visit increased during the first 4 months after the expansion. After this initial increase in wait times, the trend decreased to levels comparable with higher provider densities in the following periods (Fig. 1B).
Table 2 contains our regression results. Panel A compares trends for provider-dense areas against providers in areas with limited provider supply (ie, fewer than 92 PCPs per 10,000 population). In Panel B, PCP supply is expressed continuously in natural logs and statistical significance on the interaction terms between supply and period indicates the strength of the interaction between baseline supply and appointment availability. In concordance with previous studies27–29 we find the share of clinics accepting new Medicaid patients increased following the expansion. We find that the increased acceptance of Medicaid patients potentially exposed patients in low-supply areas to longer wait times in the first 4 months following expansion. However, both our graphical and regression analyses suggest this initial increase in wait times for new-patient appointments subsided afterwards.
Relative to providers in areas limited provider supply, we find no statistically significant differences in the propensity to accept new Medicaid patients by providers in settings with a higher concentration of PCPs. This trend was consistent at all points throughout the first year of Michigan’s Medicaid expansion across both specifications of our regression model (Table 2).
We observe early indication of reduced wait times among providers in more supply-rich areas. At four months postexpansion, Medicaid patients in supply-rich areas had marginally shorter wait times (−0.45, P<0.10) than their counterparts in low-supply areas [Table 2, panel A, column (2)]. This finding would correspond to about 10 days shorter in the time for an initial scheduled appointment. In the following months, this difference largely subsided, and allowing the provider supply measure to enter the model linearly (Table 2, panel B), does little to change the inferences from our results.
Our primary conclusion based these results suggest: (1) the increase in appointment availability was common across areas regardless of relative provider supply densities, and (2) trends in acceptability of new Medicaid patients and appointment availability were generally similar between high and low provider density markets throughout the study period. In addition, our conclusions do not change even after sensitivity analyses using different geographical boundaries (See Appendix Tables A1 and A2, Supplemental Digital Content 1, http://links.lww.com/MLR/B756).
Despite an actual rise in potential patient volume—and almost 497,000 newly enrolled Medicaid beneficiaries from the expansion24,25—our work suggests the Medicaid expansion did not create new barriers in access to primary care. Provider density appeared to have little impact on short-run changes in primary care appointment availability, implying that primary care practices in provider shortage and nonshortage areas were able to adapt to increased demand. Some PCPs could have adapted to changing market dynamics through increased utilization and adoption of technology to improve work flow,26 but Michigan’s generous scope of practice laws may have facilitated the primary care sector’s accommodation of newly insured patients.27 Nonphysician providers (ie, NPs and PAs) provided care at a higher rate following the expansion.
Observing only through Michigan’s first year under expansion, these short-run changes in appointment availability are not generalizable to long-term changes. We also cannot determine how practices adjusted to accommodate changes in patient insurance mix and volume. Observing changes in staffing, for example, we could assess providers’ perceptions of how to adapt to the short-run and long-run effects on provider demand due to the expansion. In spite of this limitation, we argue our findings are noteworthy because they occur during a period with multiple outside forces that could also contribute to potential delays in primary care access.
Our study period overlaps with Michigan’s temporary Medicaid fee “bump” (January 1, 2013–December 31, 2014),16 and related work suggests much of the increased access to care among the Medicaid traditional and expansion population was spurred by the temporary increase in Medicaid payment parity.4,28,29 Medicaid historically reimburses below rates for Medicare and private coverage30–32; though Polsky et al28 find substantive improvements in appointment availability following Medicaid expansion.33,34 In 2012, the Michigan’s Medicaid-to-Medicare fee index was 0.51, and the ACA-supported fee bump increased Medicaid primary care reimbursement fees by 125% between 2012 and 2013 and an additional 35 percent from 2014 to 2016.35,36 Furthermore, the Michigan’s expansion plan (approved by a Section 1115 waiver) included financial incentives for new beneficiaries to complete a health risk assessment and a visit to a PCP within 90 days of enrollment.
Before expansion, practices in areas with higher densities of providers were about 12% more likely to accept new Medicaid patients than were practices in areas with relatively fewer providers. This finding implies markets with higher densities of health care supply more easily absorbed increased demand for medical services following a large uptake of health coverage. Despite increases in wait times immediately following the expansion, initial differences in wait times between high-supply and low-supply areas dissipated over time. Therefore, we do not find that patients in potentially “undersupplied” areas were disadvantaged in realizing increased access to primary care following Michigan’s coverage expansions in the long run. Providers in both adequately served and potentially underserved areas adapted to the increased demand in ways that should be further examined in future inquiry.
3. Tipirneni R, Rhodes KV, Hayward RA, et al. Primary care appointment availability for New medicaid patients increased after Medicaid expansion
in Michigan. Health Aff. 2015;34:1399–1406.
4. Tipirneni R, Rhodes KV, Hayward RA, et al. Primary care appointment availability and nonphysician providers one year after Medicaid expansion
. Am J Manag Care. 2016;22:427–431.
5. Simon KI, Soni A, Cawley J. The impact of health insurance on preventive care and health behaviors: evidence from the First Two Years of the ACA medicaid expansions. J Policy Anal Manage. 2017;36:390–417.
6. Courtemanche CJ, Marton J, Ukert B, et al. Effects of the affordable care act on health care access and self-assessed health after 3 years. Inq J Health Care Organ Provision, Financ. 2018;55:1–10.
7. Decker SL, Lipton BJ, Sommers BD. Medicaid expansion
coverage effects grew in 2015 with continued improvements in coverage quality. Health Aff. 2017;36:819–825.
8. Benitez JA, Adams EK, Seiber EE. Did health care reform help kentucky address disparities in coverage and access to care among the poor? Health Serv Res. 2018;53:1387–1406.
9. Lipton BJ, Decker SL, Sommers BD. The Affordable Care Act appears to have narrowed racial and ethnic disparities in insurance coverage and access to care among young adults. Med Care Res Rev. 2019;76:32–35.
11. McKenna RM, Langellier BA, Alcalá HE, et al. The Affordable Care Act attenuates financial strain according to poverty level. Inq J Health Care Organ Provision, Financ. 2018;55:1–14.
12. Tipirneni R, Rhodes KV, Hayward RA, et al. Geographic variation in medicaid acceptance across michigan primary care practices in the era of the Affordable Care Act. Med Care Res Rev. 2017;75:633–650.
13. Benitez JA, Seiber EE. US health care reform and rural America: results from the ACA’s Medicaid expansions. J Rural Health. 2017;34:213–222.
17. Kullgren JT, McLaughlin CG. Beyond affordability: the impact of nonfinancial barriers on access for uninsured adults in three diverse communities. J Community Health. 2010;35:240–248.
18. Kullgren JT, McLaughlin CG, Mitra N, et al. Nonfinancial Barriers and Access to Care for US adults. Health Serv Res. 2012;47(pt 2):462–485.
23. Billi JE, Pai C-W, Spahlinger DA. The effect of distance to primary care physician on health care utilization and disease burden. Health Care Manage Rev. 2007;32:22–29.
26. Green LV, Savin S, Lu Y. Primary care physician shortages could be eliminated through use of teams, nonphysicians, and electronic communication. Health Aff. 2013;32:11–19.
27. Martsolf G, Kandrack R. Expanding Nurse Practitioner Scope of Practice in Michigan: Effects on Health Care Delivery. Santa Monica, CA: RAND Corporation; 2016.
28. Polsky D, Candon M, Saloner B, et al. Changes in primary care access between 2012 and 2016 for new patients with medicaid and private coverage. JAMA Intern Med. 2017;177:588–590.
29. Candon M, Zuckerman S, Wissoker D, et al. Declining medicaid fees and primary care appointment availability for new medicaid patients. JAMA Intern Med. 2017;178:145–146.
30. Cunningham PJ, Hadley J. Effects of changes in incomes and practice circumstances on physicians’ decisions to treat charity and Medicaid patients. Milbank Q. 2008;86:91–123.
31. Zuckerman S, McFeeters J, Cunningham PJ, et al. Changes in medicaid physician fees, 1998-2003: implications for physician participation. Health Aff (Millwood). 2004;23(Suppl 1):W4-374–W4-384.
32. Lindrooth RC, Bazzoli GJ, Needleman J, et al. The effect of changes in hospital reimbursement on nurse staffing decisions at safety net and nonsafety net hospitals. Health Serv Res. 2006;41(pt 1):701–720.
33. Polsky D, Richards M, Basseyn S, et al. Appointment availability after increases in Medicaid payments for primary care. N Engl J Med. 2015;372:537–545.
34. Richards MR, Saloner B, Kenney GM, et al. Availability of new Medicaid patient appointments and the role of rural health clinics. Health Serv Res. 2016;51:570–591.