BACKGROUND AND OBJECTIVES: Whether the Medicaid primary care payment increase of 2013 to 2014 changed physician participation remains unanswered amid conflicting evidence. In this study, we assess national and state-level changes in Medicaid participation by office-based primary care pediatricians before and after the payment increase.
METHODS: Using bivariate statistical analysis, we compared survey data collected from 2011 to 2012 and 2015 to 2016 by the American Academy of Pediatrics from state-stratified random samples of pediatrician members.
RESULTS: By 4 of 5 indicators, Medicaid participation increased nationally from 2011 and 2012 to 2015 and 2016 (n = 10 395). Those accepting at least some new patients insured by Medicaid increased 3.0 percentage points (ppts) to 77.4%. Those accepting all new patients insured by Medicaid increased 5.9 ppts to 43.3%, and those accepting these patients at least as often as new privately insured patients increased 5.7 ppts to 55.6%. The average percent of patients insured by Medicaid per provider panel increased 6.0 ppts to 31.3%. Nonparticipants dropped 2.1 ppts to 14.6%. Of the 27 studied states, 16 gained in participation by 1 or more indicators, 11 gained by 2 or more, and 3 gained by all 5.
CONCLUSIONS: Office-based primary care pediatricians increased their Medicaid participation after the payment increase, in large part by expanding their Medicaid panel percentage. Continued monitoring of physician participation in Medicaid at the national and state levels is vital for guiding policy to optimize timely access to appropriate health care for >37 million children insured by Medicaid.
- AAP —
- American Academy of Pediatrics
- CHIP —
- Children’s Health Insurance Program
- MACPAC —
- Medicaid and Children’s Health Insurance Program Payment and Access Commission
- NAMCS —
- National Ambulatory Medical Care Survey
- PMPPPP —
- percentage of Medicaid patients per provider panel
- ppts —
- percentage points
What’s Known on This Subject:
State Medicaid officials identified few new physician participants in Medicaid after the implementation of the payment bump and reported little impact. In contrast, the researchers of a secret shopper study identified increased primary care appointment availability for new Medicaid patients.
What This Study Adds:
National and state-level changes in Medicaid participation were assessed for office-based primary care pediatricians before and after the Medicaid payment increase. Results suggest participation gains and increased percentage of Medicaid patients per provider panel nationally and in most studied states.
The federally funded Medicaid primary care payment increase, legislated as part of the Patient Protection and Affordable Care Act, raised Medicaid payments to Medicare levels for primary care services for 2013 and 2014. Before this increase, Medicaid fees for primary care services for all age groups averaged 59% of Medicare and varied widely by state.1 The Patient Protection and Affordable Care Act intended that the higher payments would stimulate additional primary care physician participation in the Medicaid program in anticipation of a concurrent expansion of Medicaid eligibility.2
Whether this transient payment increase succeeded in encouraging physician participation in the Medicaid program remains an important and unanswered question amid conflicting evidence.3 On the 1 hand, stakeholders commonly viewed a temporary payment increase lasting 2 years as insufficient for providers to invest in additional staff and infrastructure to start or increase participation. Widely reported implementation issues that slowed provider attestations of eligibility, a prolonged claims reprocessing process, and delayed initiation of parity payments also frustrated physicians and increased provider skepticism.4 The Medicaid and Children’s Health Insurance Program Payment and Access Commission (MACPAC) reported to Congress that stakeholders interviewed from 8 states had found few new participants, little effect on participation rate, and no change in primary care service utilization.5 On the other hand, a secret shopper study found that primary care physician appointment availability for new Medicaid patients had increased between 59% and 66% after the payment increase.6 In this study, we present new data based on American Academy of Pediatrics (AAP) member surveys that assess national- and state-level changes in Medicaid participation by office-based primary care pediatricians before and after the Medicaid payment increase.
Pediatrician participation in Medicaid is particularly important because an estimated 36%7 to 48% (Table 1) of US children are insured by Medicaid and they comprise 53% of Medicaid enrollees.8 Meanwhile, Medicaid payment for pediatric primary care services averaged 64% of Medicare and 53% of private insurance payment before 2013.9 Changes in Medicaid payments therefore influenced pediatricians more than adult specialties and are likely to elicit larger adjustments in Medicaid capacity among pediatricians. We focused on office-based primary care pediatricians practicing in solo, 2-physician, and pediatric or multispecialty group practices for 3 reasons. First, they had historically participated in Medicaid less than their counterparts in pediatric subspecialties or their colleagues in safety-net and institutional settings.10–12 Second, they had consistently cited low payment as a top participation barrier.12,13 Finally, they have more autonomy to make or change new patient acceptance policies in their practice than their colleagues in safety-net and institutional settings.
To measure the various dimensions along which participation may change, we used 1 new and 3 conventional participation indicators. The 3 conventional indicators are (1) nonparticipants, a measure frequently referred to by state Medicaid officials2 and recently cited by federal lawmakers and administrators14; (2) participants who accepted at least some (or all) new Medicaid patients, an indicator commonly reported in the literature15–17 and used by MACPAC in its 2011 report to Congress18; and (3) participants who accepted all new Medicaid patients, another frequently reported measure10,19 also included in a US Government Accountability Office report to Congress.20 The new measure we developed for this study counts participants who accepted new Medicaid patients at least as often as new privately insured patients. This fourth measure, which we will refer to as “nonrestrictive” Medicaid participants, counters potential measurement bias inherent in the second and third measures if practice capacity changes between measurements or varies by comparison group. For example, practices that accepted high volumes of new Medicaid patients before and during the 2013 to 2014 period may have run overcapacity such that they could no longer accept any or as many new patients, Medicaid or otherwise, after 2014. These practices, despite serving more Medicaid patients at Time 2, are nonetheless deemed lesser participants than at Time 1 by the second and third measures because of the counting method. Similarly, those in physician shortage areas with high Medicaid enrollment may carry large Medicaid panels while not being able to accept as many new patients. Predicating acceptance of new Medicaid on each respondent’s acceptance of new private patients, our new measure provides a “standalone” indicator to measure physician willingness to serve Medicaid patients unbiased by available practice capacity.
We included the percentage of Medicaid patients per provider panel (PMPPPP) as a fifth measure. Although not a participation indicator per se, it provides key information for interpreting the significance of potential participation change as measured by the first 4 indicators.
The AAP periodically surveys the participation of its members in the Medicaid program. Each survey draws state-stratified random samples of civilian, nonretired, and postresidency pediatricians from the AAP’s membership file. We compared participation before and after Medicaid parity based on 2 such surveys, 1 fielded during 2011 to 2012 (Time 1, baseline) and the other in 2015 to 2016 (Time 2, postpayment increase). The initial sample size was 20 133 for the 2011 to 2012 survey, and 20 958 for the 2015 to 2016 survey. All survey recipients were informed that survey participation was voluntary and anonymous and that only aggregate statistics would be reported.
The 2011 to 2012 survey was mailed initially in April of 2011, and data collection was completed March 2012 after 3 follow-up mailings. Ten thousand one hundred and thirty-eight completed surveys were returned. The final adjusted response rate was 50.8%. The 2015 to 2016 survey was mailed initially in June of 2015, and data collection was completed May 2016 after 4 follow-up mailings. Ten thousand two hundred and twenty-one completed surveys were returned. The final adjusted response rate was 50.1%. Response rates for each survey are provided by state (Supplemental Table 5). All data used in this study were weighted to adjust for survey nonresponse.
Figure 1 shows the survey question on acceptance of new Medicaid and privately-insured patients. Figure 2 shows the question on PMPPPP. Medicaid program names were identified by state in both surveys. Respondents were instructed to report participation and patient load for all Medicaid patients, excluding only those covered under a separate Children’s Health Insurance Program (CHIP).
The 5 indicators we used to measure Medicaid participation change are defined as follows:
Nonparticipants: survey respondents who reported not having any Medicaid patients and not accepting any new Medicaid patients;
Open practice participants: respondents who reported having a Medicaid panel and accepting at least some (or all) new Medicaid patients;
Full participants: respondents who reported having a Medicaid panel and accepting all new Medicaid patients;
Nonrestrictive participants: respondents who reported having a Medicaid panel and accepting new Medicaid patients at least as often as new privately-insured patients; and
PMPPPP: Medicaid patients as a proportion of all patients in the respondent’s practice.
This study includes survey respondents who (1) had at least 50% effort in direct patient care in general pediatrics, (2) had no effort in pediatric subspecialty or other specialty patient care, and (3) practiced primarily in a solo or 2-physician practice or a pediatric or multispecialty group practice. Table 2 shows (1) the distribution of the analytic sample by primary practice setting and (2) practice settings excluded from this study.
We compared changes from Time 1 to Time 2 by each of the 5 measures at the national and state levels by using bivariate statistics. For the national analysis, we aggregated data from all 50 states and the District of Columbia. Observed trends were compared with other relevant data sources where available. For inclusion in the state-level analysis, we required a minimum of 200 unweighted survey respondents for Time 1 and Time 2 combined. We used SPSS Complex Samples (IBM SPSS Statistics, IBM Corporation)21 for all statistical analysis to account for the stratified survey design of the AAP surveys. An α level of .05 was used for all statistics unless otherwise indicated.
On the basis of an analytic sample of 10 395, Medicaid participation increased nationally from 2011 and 2012 to 2015 and 2016 as measured by 4 of the 5 indicators used in this study (Table 3). Open practice participants (those accepting at least some new Medicaid patients) increased 3.0 percentage points (ppts) to 77.4%. Full participants (those accepting all new Medicaid patients) increased 5.9 ppts to 43.3%, and nonrestrictive participants (those accepting new Medicaid patients at least as often as new privately-insured patients) increased 5.7 ppts to 55.6%. The average PMPPPP increased 6.0 ppts to 31.3%, compared with a 1.5 ppt increase in the proportion of US children enrolled in Medicaid over the study period as reported by MACPAC (Table 1), or a 2.6 ppt increase based on American Community Survey data (Supplemental Table 6). Nonparticipants (those without a single Medicaid patient and not accepting any new Medicaid patients) dropped a marginally significant (P < .10) 2.1 ppts to 14.6%.
Variable changes occurred at the state level (Fig 3). Twenty-seven states, representing 85% of 37 million Medicaid-enrolled US children,8 met our criteria for inclusion in the state trend analysis. Among these, we found gains in participation as measured by 1 or more indicators in 16 states, by 2 or more measures in 11 states, and by all 5 measures in 3 states. Six states (AZ, CT, IL, KY, MA, and OR) showed drops in nonparticipants, 4 (AZ, CT, KY, and OR) showed increases in open practice participants, 7 (AL, AZ, CT, KY, MO, NJ, and WI) showed increases in full participants, 9 (AL, AZ, CO, CT, IL, KY, LO, NJ, and WI) showed increases in nonrestrictive participants, and 14 (AL, AZ, CO, CT, FL, GA, IL, IN, KY, MA, MO, NJ, OR, and WA) showed increases in average PMPPPP. Statistically significant decreases in participation were not found by any indicator for any state (Table 4).
The 2013 to 2014 Medicaid primary care fee increase provides a new opportunity to examine the age-old question of whether raising Medicaid fees increases physician participation.11,19,22 At a total cost of $12 billion, the payment increase that achieved parity with Medicare raised the combined 2013 to 2014 Medicaid budget by 1.3%.23,24 Although published studies reported conflicting evidence on whether physician participation increased after the payment increase, our study adds to a growing body of empirical evidence that documents potential relationships between Medicaid payment increase and pediatrician participation change.
In addition to finding overall increases in self-reported Medicaid participation by office-based primary care pediatricians nationally after Medicaid parity, the multiple participation indicators used in our study demonstrated the various ways in which participation gains were achieved in different states. In partial agreement with state reports of few new Medicaid participants in response to the payment increase,3 we found more participants in only 6 of 27 studied states. However, widespread gains as assessed by the other participation measures accompanied sizable increases in Medicaid panels as measured by the PMPPPP indicator. For example, 4 states (AL, CO, MO, and NJ) that showed increases in either or both of full and nonrestrictive participants demonstrated 7 to 11 ppt increases in PMPPPP despite unchanged rates of nonparticipants. These results echo findings from another study that existing Medicaid participants were able to increase their participation in response to the Medicaid fee increase by adding new capacity for new Medicaid patients.6 In light of such findings, the costs and benefits of recruiting new participants versus incentivizing existing participants to increase their Medicaid capacity merit further investigation and the potential for both because strategies to improve access should be considered.
Importantly, we highlight in our results how nuances in participation measures can change measurement outcomes and potentially support different conclusions about the effects of payment increase on physician participation. For example, had we counted only those who accepted at least some new Medicaid patients, we would have observed that participation remained unaffected in Alabama and New Jersey, despite sizable gains made by these 2 states on most other indicators. In fact, using this single measure, we would have captured participation gains in only 4 states and missed 8 others as identified by the nonparticipant, the full participant, and the nonrestrictive participant indicators. Although a thorough discussion of the merits and weaknesses of the respective measures is beyond the scope of this study, our results suggest that for policy evaluation, the choice of indicators vis-à-vis the policy goal would be at least as important as the measurement outcome.
Our findings should be considered in light of several limitations. First, we use self-reported survey data and cannot independently validate their accuracy. Although an independent public data source, the National Ambulatory Medical Care Survey (NAMCS)25 tracks acceptance of new Medicaid patients for all physicians, and pediatricians comprise approximately a 10th of its sample. Even after the NAMCS augmented its physician sample since 2012, error margins associated with the proportion of pediatricians accepting new Medicaid patients remained too wide to use as a reference (Supplemental Table 6). Thus, we cannot rule out potential biases in terms of over- or underreporting participation based on the AAP surveys. However, the composition of the AAP survey sample did not shift from Time 1 to Time 2, so a major change in the magnitude or direction of any bias remains unlikely. Thus, the net changes measured by the AAP surveys should accurately reflect differences in participation between the 2 times.
Second, response rates for both surveys used in this study are lower than previous AAP member surveys on the same topic.10,12 Many surveyed pediatricians who declined to complete the 2011 to 2012 and 2015 to 2016 surveys indicated that they were working in an employed capacity and did not feel sufficiently knowledgeable about practice operations to answer questions related to Medicaid participation. This is consistent with national data that show a trend toward increasing pediatrician employment by other health care entities.26 Not being able to filter from our initial sample those without the requisite information to complete the surveys depressed response rates.
Third, our study design cannot eliminate potential other influences on participation change occurring contemporaneously with the payment increases. One such consideration is the possibility that a recession-driven shift of health insurance coverage from employer-sponsored plans to Medicaid may have increased the presence of Medicaid patients during our study period, which would have explained our findings independently of physician participation. To assess this theory, we examined employer-based and Medicaid coverage trend data on the basis of American Community Survey data (Supplemental Table 6). From 2008 to 2011, when the economy was struggling to recover from the recession, employer coverage of US children dropped 4.4 ppts to 53.7%, as Medicaid and/or CHIP enrollment surged 7.1 ppts to 33.1%. By 2011, however, employer coverage stabilized and fell by only an additional 0.6% through 2015, whereas the percentage of children insured by Medicaid gradually rose by an average annual rate of 0.7 ppts. These trends suggest that our findings of participation gains after 2011 more likely represented fee increase–induced than recession-driven changes.
The growing prevalence of restrictive networks that accompanied increased Medicaid managed care organization penetration27 during the study period may have also affected participation independently of payment increase, whereas issues with managed care implementation of the fee increase may have limited participation gains in some states more so than in others. Researchers for future studies should examine these complex areas to gain further insight.
Fourth, the AAP survey measured participation 6 to 18 months after the federally funded payment increase expired. Our data may therefore have underestimated participation gains to the extent that participation might have already receded from its peak in some states by Time 2.
Fifth, our findings may not generalize to other Medicaid provider groups because the AAP membership of ∼66 000 accounts for only approximately three-quarters of the total number of self-designated pediatricians as listed in the American Medical Association MasterFile.28 Non-AAP–member pediatricians, nonboard-certified pediatricians, nonpediatrician physicians, and nonphysician providers were not represented in our study.
Our analytic sample further excluded all residents, all pediatric subspecialists, and general pediatricians practicing in safety net and institutional settings. Because these groups are known for high Medicaid volumes, our results underrepresent Medicaid participation by general pediatricians on average and by pediatricians overall. We acknowledge also that although we focused this study on office-based pediatricians who have presumably more autonomy than their colleagues in safety net and institutional settings regarding new patient acceptance policies in their practice, increasing hospital and medical-center ownership of physician practices may be blurring the boundary between the 2 groups.
Finally, we note that physician participation is a proxy measure of patient access. The results of this study should not be interpreted as a direct measure of patient access to care because a host of factors other than physician participation are also influential.
Results of this study provide compelling evidence that office-based primary care pediatricians, who had historically been more resistant to participation than either their subspecialist or safety net and institution-based peers, increased their participation in the Medicaid program nationally from before to after the 2013 to 2014 Medicaid payment increase. That we found participation increases in 4 of 5 indicators nationally and by different measures in the majority of studied states speaks to the robustness of our conclusion. Our findings are even more compelling because one might have expected that widespread implementation problems, delays in increased payment, and the temporary nature of the increases would have blunted an increase in participation. We must caution, however, that our study was not designed to directly examine the relationship between fee increase and participation gains. Whether the magnitude of fee increase predicted the extent of participation improvement or whether observed participation gains may persevere depending on extension of the fee increase funded by some states, for example, are questions that future researchers can investigate.
Physician participation has been and will likely continue to be viewed as a key and integral component for measuring patient access to Medicaid services. In its report to Congress in 2011, MACPAC constructed a framework for examining access to care for Medicaid and CHIP enrollees that identified provider participation as 1 of 3 key components for measuring patient access.18 The collection of robust data and selection of unbiased measures that allow accurate and timely measurement of physician participation in the Medicaid program at national and state levels in response to policy perturbations deserve increased effort. Going forward, continuous monitoring and thoughtful analysis of physician participation can guide policy to achieve the goal of improving timely access to appropriate health care by 37 million children insured by Medicaid.
We thank Dan Walter and Wendy Chill of the AAP (who contributed to survey design and data interpretation) and William Cull and Lynn Olson for their critical reviews of early drafts of the article.
- Accepted October 20, 2017.
- Address correspondence to Suk-fong S. Tang, PhD, Department of Research, American Academy of Pediatrics, 141 Northwest Point Blvd, Elk Grove Village, IL 60009. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Supported by the American Academy of Pediatrics Friends of Children Fund, which provided funding for both the 2011–2012 and 2015–2016 American Academy of Pediatrics Survey of Pediatrician Participation in Medicaid, which collected data for the submitted study.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2017-3241.
- Zuckerman S,
- Goin D
- Medicaid and CHIP Payment and Access Commission
- Tollen L
- Timbie JW,
- Buttorff C,
- Kotzias VI,
- Case SR,
- Mahmud A
- Medicaid and CHIP Payment and Access Commission
- State Health Access Data Assistance Center
- University of Minnesota
- Center for Medicaid and CHIP Services (CMCS). Centers for Medicare & Medicaid Services (CMS). U.S. Department of Health and Human Services (HHS)
- Peterson M
- Yudkowsky BK,
- Cartland JD,
- Flint SS
- Tang SF,
- Yudkowsky BK,
- Davis JC
- Berman S,
- Dolins J,
- Tang SF,
- Yudkowsky B
- Kiely E,
- Robertson L,
- Gore D,
- Farley R
- Hing E,
- Decker SL,
- Jamoom E
- Decker SL
- Decker SL
- Medicaid and CHIP Payment and Access Commission
- Cunningham PJ,
- O’Malley AS
- US Government Accountability Office
- Siller AB,
- Tompkins L
- Medicaid and CHIP Payment Access Commission
- National Center of Health Statistics
- The Physicians Foundation
- Merritt Hawkins
- Giovannelli J,
- Lucia K,
- Corlette S
- American Medical Association
- Copyright © 2018 by the American Academy of Pediatrics