Primary Care Appointment Availability and Nonphysician Providers One Year After Medicaid Expansion
Renuka Tipirneni, MD, MSc; Karin V. Rhodes, MD, MS; Rodney A. Hayward, MD; Richard L. Lichtenstein, PhD; HwaJung Choi, PhD; Elyse N. Reamer, BS; and Matthew M. Davis, MD, MAPP
Access to care depends, in large part, on states’ unique healthcare environments, which have varying population demographics and healthcare workforces. After the 2012 Supreme Court decision made Medicaid expansion under the Affordable Care Act (ACA) optional, 19 states decided not to expand Medicaid and the remaining 31 states expanded Medicaid in a variety of ways.1
For this reason, ongoing state-level evaluations are critically important to evaluate the consequences of Medicaid expansions in order to inform other states facing similar expansion challenges.
Although most primary care providers were accepting new patients on the eve of ACA-sponsored coverage expansions in 2012 and 2013,2
there was widespread concern that primary care capacity would be insufficient to meet the growing demand for care by newly insured individuals3,4
—many of whom are establishing care with a primary care provider for the first time. Higher than expected enrollment in Medicaid5-7
has heightened this concern.8
In order to provide timely estimates of changes in primary care access in the era of healthcare reform, we previously reported preliminary results from a simulated patient study examining primary care appointment availability and wait times for new patients with Medicaid and private insurance before and 4 months after implementation of Medicaid expansion in Michigan on April 1, 2014.9
We found an initial increase in primary care appointment availability for new Medicaid patients with no lengthening of wait times 4 months after expansion.
Since then, the number of enrollees in the state’s Medicaid expansion program—or Healthy Michigan Plan—has nearly doubled, from approximately 350,000 to 600,000.10
Each new Healthy Michigan Plan beneficiary is required to obtain a primary care appointment within 90 days of enrollment. With the expanding number of enrollees and the mandate to schedule early appointments, there was increasing uncertainty as to whether the initial increase in primary care appointment access for new Medicaid patients would endure following continued coverage expansion. To monitor ongoing trends in appointment accessibility, we now present longitudinal data through the first 12 months after Medicaid expansion, regarding primary care appointment availability and wait times for new patients with Medicaid and private insurance.
We conducted a simulated patient (“secret shopper”) study, in which trained research staff called a stratified proportionate random sample of primary care practices (N = 295) to request a new patient appointment before (March 17-31, 2014) and, serially, after Medicaid expansion implementation in Michigan on April 1, 2014 (July 21-August 11, 2014; November 17-December 12, 2014; and March 20-April 15, 2015). Details of our study methodology, including sampling, practice classification, and simulated patient call protocols, were reported in our previous article.9
Briefly, we conducted a telephone survey of Michigan primary care clinics caring for adult patients February 24 to March 7, 2014, to confirm study eligibility and acceptance of new Medicaid patients, as well as other practice characteristics. Given the study’s focus on assessing appointment accessibility for new Medicaid patients, we placed simulated patient calls to clinics that accepted Medicaid in order to assess appointment wait times.
Trained research assistants called primary care clinics as simulated prospective patients with Medicaid or private coverage every 4 months up to 1 year after Medicaid expansion (see eAppendix A for a timeline of study calls and enrollment [eAppendices
available at www.ajmc.com
]). Callers requested the next available routine new-patient appointment for a patient with well-controlled hypertension. Calls with each insurance scenario were separated by at least 1 week and appointments were canceled at the end of each call. The University of Michigan Institutional Review Board considered the study “not regulated,” as we did not collect information about individual patients or clinic staff.
Our main outcome was availability of appointments to simulated patients, which we defined as a specific appointment date and time offered during the call. For clinics that accepted Medicaid and offered appointments, we calculated wait times as the difference in calendar days between the call date and appointment date. We also noted whether the appointment was made with a physician or nonphysician provider (eg, nurse practitioner, physician assistant).
We compared changes in appointment availability across the same clinics between pre- and postexpansion periods using logistic regression with time indicators. To compare changes in the skewed distribution of wait times over time and between Medicaid and private insurance, we conducted paired testing with the Wilcoxon signed rank-sum test. In sensitivity analyses, we used linear regression with time indicators after log-transforming the wait time variable, but results did not differ. Because our analyses were based on clinic-level longitudinal data, each clinic could contribute multiple observations to each regression analysis over the 4 waves of data collection. We therefore clustered the residual structure at the clinic level to obtain robust standard errors.11,12
To assess differences across varying categories of clinics (eg, urban vs rural practices) at a given time period, we used χ2
testing for appointment availability and the Wilcoxon rank sum test for wait time differences. We considered a 2-sided P
<.05 statistically significant. All analyses were performed using Stata version 13 (StataCorp, College Station, Texas).
Full details of clinic characteristics have been published previously.9
We found in the current study that approximately half of the 295 study clinics were small practices with 1 to 2 providers (eAppendix B) and that 42% of clinics employed nonphysician providers, such as nurse practitioners or physician assistants. Although only 9% were traditionally defined as safety net clinics, 74% had established patients with Medicaid coverage. Clinics were located mostly in urban locations. During the study period, approximately 600,000 individuals enrolled in the state’s Medicaid expansion program10
and 340,000 enrolled in Marketplace plans,13
together representing approximately 10% of the state’s total population and 90% of the previously uninsured nonelderly adult population (eAppendix A).
Availability of Appointments
The proportion of clinics with appointments available to new Medicaid patients increased from 49% before expansion to 55% by 12 months after expansion (+6 percentage points; 95% CI, 1.6-11.1) (Table 1
). Appointment availability for new privately insured patients decreased from 88% of clinics to 86% by 12 months after expansion (–2 percentage points; 95% CI, –0.5 to –3.8). Changes in appointment availability for both Medicaid and privately insured groups at 12 months post expansion remained stable compared with the 4-month postexpansion findings.
We also examined the proportion of appointments scheduled with nonphysician providers versus physicians, and found that there was an increasing proportion of appointments scheduled with nonphysician providers over the study period (Table 2
). For Medicaid appointments, 8% were scheduled with nonphysician providers before expansion, and this increased to 21% by 12 months post expansion (+13 percentage points; 95% CI, 5.6-20.2). For private insurance appointments, 11% were scheduled with nonphysician providers before expansion, and this increased to 19% by 12 months post expansion (+8 percentage points; 95% CI, 1.3-14.1).
In clinics that accepted patients with Medicaid, median wait times for new Medicaid patients remained stable over the 12-month period (Table 3
). In these same clinics, median wait times for new privately insured patients increased slightly by 12 months post expansion, from 7 to 10 days (P
= .02). However, there was no significant difference between wait times for new Medicaid and new privately insured patients throughout the study period. Furthermore, median new patient wait times remained within 2 weeks, with more than 95% falling within the 90-day requirement of the Healthy Michigan law.
Variation by Practice Characteristics
Larger practices (≥3 providers) were more likely to accept new Medicaid patients pre-expansion and initially experienced significantly increased appointment availability 4 and 8 months post expansion (P
= .002 and P
= .02 , respectively) (eAppendix C); however, this increase was no longer significant by 12 months post expansion (P
= .14). Safety net clinics were much more likely than non–safety net clinics to accept new Medicaid patients at baseline; however, only non–safety net clinics had significantly increased appointment availability after expansion (P
= .005 at 12 months post expansion). Clinics in urban locations were less likely to accept new Medicaid patients than clinics in nonurban locations at baseline, but only urban clinics had increased Medicaid appointment availability post expansion (P
= .006 at 12 months post expansion).
In this extended longitudinal follow-up of a simulated patient (“secret shopper”) study, we found that the proportion of primary care clinics accepting new Medicaid patients increased 12 months post-Medicaid expansion in Michigan. During the study period, there was also a steady increase in the proportion of appointments scheduled with nonphysician providers. Despite the large number of newly insured individuals, median new patient wait times remained within 2 weeks and were similar for Medicaid and privately insured patients. Although there was a statistically significant increase in wait times for new privately insured patients 12 months post expansion, the 3-day difference in wait times is likely not clinically meaningful for a patient scheduling an initial primary care appointment.
Notably, although the ACA-associated rate bump in Medicaid reimbursement to primary care providers expired at the end of 2014, the increased availability of appointments to new Medicaid patients persisted into 2015 despite Medicaid payments returning to pre-ACA levels. Prior literature suggested that states with larger increases in reimbursement during the rate bump had larger increases in Medicaid appointment availability.14
Consequently, many feared that primary care appointment access would decline soon after the rates dropped. Our observation of stable increases in Medicaid appointment availability suggests that this has not happened. Nevertheless, continuous careful attention to this issue is warranted, as primary care practice policies on acceptance of new Medicaid patients may continue to change over time.
The unexpected increase in appointment capacity, despite a fixed clinic sample and declining reimbursement, may be partially explained by the increasingly prominent role of nonphysician primary care providers.15
Given the increasing proportion of appointments scheduled with nonphysician providers during the study period, we speculate that clinics may have either hired additional nurse practitioners or physician assistants, or transferred the work of new patient intakes to existing nonphysician providers. Prior literature also suggests that nonphysician providers are more likely than physicians to care for Medicaid patients.16,17
In spite of lower Medicaid reimbursement in 2015, it is possible that practices were able to continue accepting new patients, as such providers often receive lower salaries than physicians.17
The study design has certain limitations, which were also discussed in our previous article.9
These limitations include incomplete calls; however, these only occurred with less than 5% of sampled clinics. Additionally, the study was limited by our focus on accessibility of appointments to new, rather than established, patients and the focus on a single large Midwestern state. Due to our emphasis on evaluating Healthy Michigan’s appointment rule, we examined wait times only in clinics that accepted Medicaid. Appointment availability and wait times may differ for established patients, patients in clinics that do not accept Medicaid, and patients in other states.
One year following Medicaid expansion in Michigan, appointment availability for new Medicaid patients increased. This is perhaps attributable to increasing proportions of appointments scheduled with nonphysician providers. Future research should examine the evolving role of nonphysician providers and other team-based approaches to expanding primary care capacity and improving access for the newly insured.18
The authors would like to thank research assistants Jessica Landgraf, MA, Robyn Power, BS, Jessica Stallworth, MA, Lisa Fouladbash, MS, Jessica Chee, BS, and Rafael D. Nieves, MHSA, of the University of Michigan Robert Wood Johnson Foundation Clinical Scholars Program; and Sarah Mott, MPH, of the University of Michigan School of Public Health for their dedication to data collection, and Rekar Taymour, of the University of Michigan School of Public Health, for assistance with literature review. They are also grateful to the analytic support of Sonya DeMonner, MPH, of the University of Michigan Robert Wood Johnson Foundation Clinical Scholars Program, and Joshua Fangmeier, MPP, of the Center for Healthcare Research & Transformation. Lastly, they thank Lauren S. Hughes, of the Pennsylvania Department of Health, and Michelle H. Moniz, of the University of Michigan Department of Obstetrics and Gynecology, for thoughtful comments on earlier drafts of this manuscript.