[Skip to Navigation]
Sign In
Figure 1.  Point Estimates and 95% Confidence Intervals for Proportions Responding That They Would Seek an Exemption for the Patient, Stratified by Vignette Scenario (Minor or Major Depression), and Further by State
Point Estimates and 95% Confidence Intervals for Proportions Responding That They Would Seek an Exemption for the Patient, Stratified by Vignette Scenario (Minor or Major Depression), and Further by State

Abbreviations: AR, Arkansas; IN, Indiana; KY, Kentucky; NH, New Hampshire.

Figure 2.  Concordance Between Attitude Regarding Appropriateness of Exemption and Indication of Willingness to Assist, Stratified by Severity Group (Minor vs Major Depression)
Concordance Between Attitude Regarding Appropriateness of Exemption and Indication of Willingness to Assist, Stratified by Severity Group (Minor vs Major Depression)
Table 1.  Descriptive Statistics of Demographic Characteristics, Overall, and Stratified by State
Descriptive Statistics of Demographic Characteristics, Overall, and Stratified by State
Table 2.  Factors Associated With Physician Willingness to Assist Patients With Major Depression (1 = yes; 0 = no), Among Respondents Randomized to the Major Depression Scenarioa
Factors Associated With Physician Willingness to Assist Patients With Major Depression (1 = yes; 0 = no), Among Respondents Randomized to the Major Depression Scenarioa
Table 3.  Descriptive Statistics of Demographic Characteristics, Overall, and Stratified by State
Descriptive Statistics of Demographic Characteristics, Overall, and Stratified by State
1.
Centers for Medicare & Medicaid Services. 2018. Opportunities to Promote Work and Community Engagement among Medicaid Beneficiaries. SMD Letter 18-002, Center for Medicare & Medicaid Services, U.S. Department of Health and Human Services, Washington, DC. Accessed September 3, 2021. www.cms.gov/newsroom/press-releases/cms-announces-new-policy-guidance-states-test-community-engagement-able-bodied-adults
2.
Huberfeld  N.  Can work be required in the Medicaid program?   N Engl J Med. 2018;378(9):788-791. doi:10.1056/NEJMp1800549PubMedGoogle ScholarCrossref
3.
Bagley  N.  Are Medicaid work requirements legal?   JAMA. 2018;319(8):763-764. doi:10.1001/jama.2018.0958PubMedGoogle ScholarCrossref
4.
Rosenbaum  S. Medicaid work requirements: inside the decision overturning Kentucky Health’s approval. Health Affairs Blog 2018. Accessed August 30, 2021. https://www.healthaffairs.org/do/10.1377/hblog20180702.144007/full/
5.
Ayanian  JZ, Tipirneni  R, Goold  SD.  Mitigating the risks of Medicaid work requirements.   N Engl J Med. 2018;379(9):803-805. doi:10.1056/NEJMp1806129PubMedGoogle ScholarCrossref
6.
Wagner  J, Schubel  J.  States’ Experiences Confirm Harmful Effects of Medicaid Work Requirements. Washington, DC: Center on Budget and Policy Priorities;2020.
7.
Chen  L, Sommers  BD.  Work requirements and Medicaid Disenrollment in Arkansas, Kentucky, Louisiana, and Texas, 2018.   Am J Public Health. 2020;110(8):1208-1210. doi:10.2105/AJPH.2020.305697PubMedGoogle ScholarCrossref
8.
Center on Budget and Policy Priorities. Taking Away Medicaid for Not Meeting Work Requirements Harms Low-Wage Workers. 2020. Accessed August 30, 2021. https://www.cbpp.org/research/health/taking-away-medicaid-for-not-meeting-work-requirements-harms-low-wage-workers
9.
Sommers  BD, Goldman  AL, Blendon  RJ, Orav  EJ, Epstein  AM.  Medicaid work requirements–results from the first year in Arkansas.   N Engl J Med. 2019;381(11):1073-1082. doi:10.1056/NEJMsr1901772PubMedGoogle ScholarCrossref
10.
Manivannan  A, Adkins-Hempel  M, Shippee  ND, Vickery  KD.  Experiences with work and participation in public programs by low-income Medicaid enrollees using qualitative interviews.   J Gen Intern Med. 2020;35(10):2983-2989. doi:10.1007/s11606-020-05921-zPubMedGoogle ScholarCrossref
11.
Sommers  BD, Chen  L, Blendon  RJ, Orav  EJ, Epstein  AM.  Medicaid work requirements in Arkansas: two-year impacts on coverage, employment, and affordability of care.   Health Aff (Millwood). 2020;39(9):1522-1530. doi:10.1377/hlthaff.2020.00538PubMedGoogle ScholarCrossref
12.
Venkataramani  AS, Bair  EF, Dixon  E,  et al.  Assessment of Medicaid beneficiaries included in community engagement requirements in Kentucky.   JAMA Netw Open. 2019;2(7):e197209. doi:10.1001/jamanetworkopen.2019.7209PubMedGoogle Scholar
13.
Rosenbaum  S. How the Biden Administration Can Act to Strengthen Medicaid. In: The Commonwealth Fund, ed. To the Point 2020. Accessed August 30, 2021. www.commonwealthfund.org/blog/2020/how-biden-administration-can-act-strengthen-medicaid
14.
Bagley  N.  How to Undo One of Trump’s Worst, Most Despicable Policies.  The Atlantic. 2021. Accessed August 30, 2021. www.theatlantic.com/ideas/archive/2021/01/undoing-trump-damage/617736/
15.
Kaiser Family Foundation. Work Requirement Waivers: Approved and Pending as of June 28, 2021. Published 2021. Accessed July 22, 2021. www.kff.org/medicaid/issue-brief/medicaid-waiver-tracker-approved-and-pending-section-1115-waivers-by-state/#Table2
16.
Wheeler  L. Supreme Court Leaves Fight Over Medicaid Work Rules in Limbo. Bloomberg Law. July 12, 2021.
17.
Huberfeld  N.  The universality of Medicaid at fifty.   Yale J Health Policy Law Ethics. 2015;15(1):67-87.PubMedGoogle Scholar
18.
Goldman  AL, Woolhandler  S, Himmelstein  DU, Bor  DH, McCormick  D.  Analysis of work requirement exemptions and Medicaid spending.   JAMA Intern Med. 2018;178(11):1549-1552. doi:10.1001/jamainternmed.2018.4194PubMedGoogle ScholarCrossref
19.
Solomon  J.  Medicaid Work Requirements Can’t Be Fixed: Unintended Consequences are Inevitable Result. Washington, DC: Center on Budget and Policy Priorities;2019.
20.
Gurley-Calvez  T. KG, Simon  K, Wissoker  D,. Medicaid reform and emergency room visits: evidence from West Virginia’s Medicaid redesign. 2012. Accessed September 1, 2021. www.mercatus.org/publications/government-spending/medicaid-reform-and-emergency-room-visits-evidence-west-virginias
21.
Bonica  A, Rosenthal  H, Blackwood  K, Rothman  DJ.  Ideological sorting of physicians in both geography and the workplace.   J Health Polit Policy Law. 2020;45(6):1023-1057. doi:10.1215/03616878-8641555PubMedGoogle ScholarCrossref
22.
Bonica  A, Rosenthal  H, Rothman  DJ.  The political polarization of physicians in the United States: an analysis of campaign contributions to federal elections, 1991 through 2012.   JAMA Intern Med. 2014;174(8):1308-1317. doi:10.1001/jamainternmed.2014.2105PubMedGoogle ScholarCrossref
23.
Hersh  ED, Goldenberg  MN.  Democratic and Republican physicians provide different care on politicized health issues.   Proc Natl Acad Sci U S A. 2016;113(42):11811-11816. doi:10.1073/pnas.1606609113PubMedGoogle ScholarCrossref
24.
Burgess  DJ, Hardeman  RR, Burke  SE,  et al  Incoming Medical Students' Political Orientation Affects Outcomes Related to Care of Marginalized Groups: Results from the Medical Student CHANGES Study.   Journal of health politics, policy and law. 2019;44(1):113-146. doi:10.1215/03616878-7206755Google ScholarCrossref
25.
Schmidt  H, Hoffman  AK.  The Ethics of Medicaid’s Work Requirements and Other Personal Responsibility Policies.   JAMA. 2018;319(22):2265-2266. doi:10.1001/jama.2018.3384PubMedGoogle ScholarCrossref
26.
Ellsbury  KE, Carline  JD, Wenrich  MD.  Competing professionalism values among community-based family physicians.   Acad Med. 2006;81(10)(suppl):S25-S29. doi:10.1097/01.ACM.0000236562.10528.20PubMedGoogle Scholar
27.
Centers for Medicare & Medicaid Services. Physical and Mental Health Condition Prevalence and Comorbidity among Fee-for-Service Medicare- Medicaid Enrollees, available at: www.cms.gov/Medicare-Medicaid-Coordination/Medicare-and-Medicaid-Coordination/Medicare-Medicaid-Coordination-Office/Downloads/Dual_Condition_Prevalence_Comorbidity_2014.pdf. 2014.
28.
AACP (American Academy of Professional Coders). ICD 10 Resource: Coding for Major Depressive Disorder Availlable at: www.hamiltonhealth.com/wp-content/uploads/2017/08/AAPC_Depressive-Disorder-ICD-10-BH_Ref_Guide.pdf. 2013.
29.
Huber  P. The behavior of maximum likelihood estimates under nonstandard conditions. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability. 1967;1:221-233.
30.
American Association for Public Opinion Research (AAPOR). AAPOR Response Rate Calculator 4.1 availablle at: www.aapor.org/Education-Resources/For-Researchers/Poll-Survey-FAQ/Response-Rates-An-Overview.aspx. 2020.
31.
Zatz  LY, Hersh  E, Gudzune  KA, Thorndike  AN, N Goldenberg  M, Bleich  SN.  Physicians’ political party affiliation and clinical management of obesity.   Clin Obes. 2020;10(5):e12396. doi:10.1111/cob.12396PubMedGoogle Scholar
32.
American Academy of Family Physicians, American Academy of Pediatrics, American Collegeof Obstetricians and Gynecologists, American College of Physicians, American Osteopathic Association, American Psychiatric Association. Section 1115: Demonstration waivers and other proposals to change Medicaid benefits, financing and cost-sharing. Joint position statement. Accessed September 1, 2021. www.aafp.org/dam/AAFP/documents/advocacy/coverage/medicaid/LT-Group6-PrinciplesMedicaidWaivers-120817.pdf
33.
Lee  SY, Dow  WH, Wang  V, VanGeest  JB.  Use of deceptive tactics in physician practices: are there differences between international and US medical graduates?   Health Policy. 2004;67(3):257-264. doi:10.1016/S0168-8510(03)00124-6PubMedGoogle ScholarCrossref
34.
VanGeest  J, Weiner  S, Johnson  T, Cummins  D.  Impact of managed care on physicians’ decisions to manipulate reimbursement rules: an explanatory model.   J Health Serv Res Policy. 2007;12(3):147-152. doi:10.1258/135581907781543102PubMedGoogle ScholarCrossref
35.
VanGeest  JB, Wynia  MK, Cummins  DS, Wilson  IB.  Measuring deception: test-retest reliability of physicians’ self-reported manipulation of reimbursement rules for patients.   Med Care Res Rev. 2002;59(2):184-196. doi:10.1177/1077558702059002004PubMedGoogle ScholarCrossref
36.
Wynia  MK, Cummins  DS, VanGeest  JB, Wilson  IB.  Physician manipulation of reimbursement rules for patients: between a rock and a hard place.   JAMA. 2000;283(14):1858-1865. doi:10.1001/jama.283.14.1858PubMedGoogle ScholarCrossref
37.
Freeman  VG, Rathore  SS, Weinfurt  KP, Schulman  KA, Sulmasy  DP.  Lying for patients: physician deception of third-party payers.   Arch Intern Med. 1999;159(19):2263-2270. doi:10.1001/archinte.159.19.2263PubMedGoogle ScholarCrossref
38.
Rost  K, Smith  R, Matthews  DB, Guise  B.  The deliberate misdiagnosis of major depression in primary care.   Arch Fam Med. 1994;3(4):333-337. doi:10.1001/archfami.3.4.333PubMedGoogle ScholarCrossref
39.
Bogardus  ST  Jr, Geist  DE, Bradley  EH.  Physicians’ interactions with third-party payers: is deception necessary?   Arch Intern Med. 2004;164(17):1841-1844. doi:10.1001/archinte.164.17.1841PubMedGoogle ScholarCrossref
40.
Rushton  JL, Felt  BT, Roberts  MW.  Coding of pediatric behavioral and mental disorders.   Pediatrics. 2002;110(1 Pt 1):e8. doi:10.1542/peds.110.1.e8PubMedGoogle Scholar
41.
Nadler  SF, Malanga  GA, Ciccone  DS.  Positive straight-leg raising in lumbar radiculopathy: is documentation affected by insurance coverage?   Arch Phys Med Rehabil. 2004;85(8):1336-1338. doi:10.1016/j.apmr.2004.02.021PubMedGoogle ScholarCrossref
42.
Werner  RM, Alexander  GC, Fagerlin  A, Ubel  PA.  The “hassle factor”: what motivates physicians to manipulate reimbursement rules?   Arch Intern Med. 2002;162(10):1134-1139. doi:10.1001/archinte.162.10.1134PubMedGoogle ScholarCrossref
Original Investigation
October 1, 2021

Variability in Primary Care Physician Attitudes Toward Medicaid Work Requirement Exemption Requests Made by Patients With Depression

Author Affiliations
  • 1Department of Medical Ethics and Health Policy, Center for Health Incentives and Behavioral Economics, Perelman School of Medicine, University of Pennsylvania, Philadelphia
  • 2Department of Biostatistics, Vanderbilt University Medical Center, Nashville, Tennessee
  • 3Department of Computer and Information Sciences, Fordham University, Bronx, New York
  • 4Perelman School of Medicine, Department of Biostatistics, Epidemiology and Informatics, University of Pennsylvania, Philadelphia
  • 5Perelman School of Medicine, Division of General Internal Medicine, University of Pennsylvania, Philadelphia
  • 6Leonard Davis Institute of Health Economics, University of Pennsylvania, Philadelphia
JAMA Health Forum. 2021;2(10):e212932. doi:10.1001/jamahealthforum.2021.2932
Key Points

Question  To what degree do physicians vary in their willingness to assist patients seeking exemptions from Medicaid work requirements?

Findings  In this mailed survey experiment with 715 responses, we found that 25% of physicians would assist a patient not qualifying under the state’s criteria for exemptions from Medicaid work requirements, and 54% would not, even when a patient qualifies. We found state, administrative effort, political affiliation, and perceived appropriateness to be significantly associated with willingness to assist.

Meaning  It is medically, ethically, and legally imperative that measures aimed at protecting vulnerable patients are not undermined by the burdensomeness of exemption procedures, or by physicians’ political or personal views.

Abstract

Importance  Medicaid work requirements seek to promote health and personal responsibility but can also jeopardize health care access. Physicians have a central function in assisting patients with exemption requests, but it is unclear how their role affects patient welfare, professionalism, and the ethical and legal justification of programs.

Objective  To understand the degree of variability in physician response to assist patients with depression in receiving a Medicaid work requirement exemption.

Design, Setting, and Participants  We conducted a mailed survey experiment among practicing primary care physicians in the first 4 approved states (Arkansas, Kentucky, Indiana, New Hampshire) in July and October of 2019. We report response, cooperation, refusal, and contact rates in line with American Association for Public Opinion Research (AAPOR) standards.

Exposures  In each state, we used an experimental factorial design to randomize recipients to 1 of 4 patient clinical scenarios.

Main Outcomes and Measures  The primary outcome was the indicator of willingness to assist a patient reporting depression with an exemption.

Results  We received 715 responses (overall AAPOR response rate: 21%; cooperation rate: 84%; refusal rate: 4%; contact rate: 25%). Respondents’ mean (SD) age was 54 (12) years; mean (SD) time since graduation, 26 (12) years; 435 (61%) identified as male; 177 as Democrat (25%); 156 as Republican (22%); 197 as Independent/other (28%); and 185 as declined/unknown (26%); the mean (SD) share of Medicaid patients was 29% (21%). We found that 97 of 387 physicians (25%) would offer assistance even when state policy would not support an exemption, and 170 of 315 (54%) would not offer assistance when regulations would require this. Moreover, 49 of 245 respondents (20%) who deemed an exemption appropriate indicated that they would not assist. State, administrative effort, political affiliation, and perceived appropriateness were statistically associated with the odds of assisting with an exemption.

Conclusions and Relevance  In this survey study of primary care physicians, we found substantial variation regarding willingness to assist patients qualifying for a work requirement exemption where none should exist. Insofar as work requirements are implemented again, it is critical to proactively identify measures to ensure that patients qualifying for exemptions are not put at risk due to either the burdensomeness of exemption procedures, or physicians’ political or personal views.

Introduction

Medicaid work and community engagement requirements (henceforth “work requirements”) make health care access conditional on reporting between 80 to 100 hours of qualifying activities per month, such as skills training, job searching, or informal caregiving. In January 2018, the Centers for Medicare & Medicaid Services (CMS) authorized states to test work requirements to promote “better mental, physical, and emotional health” and “to help individuals and families rise out of poverty and attain independence.”1 Critics, however, view the policy as incompatible with Medicaid’s most fundamental objective: to furnish health care access.2-8 Most research has focused on the justification of policies and the ability of beneficiaries to comply with program requirements.2-12 The role of physicians’ behavior merits similar attention: policies in all states assign them an integral role in assisting patients with requests to be exempted, and in doing so, require them to exercise considerable judgment.

Work requirements have been challenged in the courts by Medicaid beneficiaries. After the US Court of Appeals for the District of Columbia Circuit affirmed rulings in favor of the plaintiffs, the Trump administration appealed the ruling. The US Supreme Court had initially agreed to hear the cases of Arkansas and New Hampshire in Spring 2020. However, at the request of the Biden administration, the Court canceled an already scheduled hearing.13,14 As of this writing, work requirements had been approved in 6 states, blocked by the courts in 4 and were halted in 2, whereas 7 applications remained submitted to CMS (eTable 1 in the Supplement). In April 2021, CMS notified Arizona, Arkansas, Indiana, Michigan, New Hampshire, and Wisconsin of its final decision to withdraw work requirement waiver authorities,15 though states may appeal this decision and it is possible that the Supreme Court will still hear the case in the next term.16 Although it is clear that the Biden administration is opposed to work requirements, future administrations could authorize the initiative again; a full assessment of the policy’s effect on different stakeholders hence remains critical and relevant.

All approved work requirement programs recognized that some life circumstances can render compliance unduly burdensome for some groups: full time students, pregnant women, or individuals with severe disabilities, for instance, are routinely exempt, subject to verification by health professionals or other relevant parties.11,12 Moreover, each state established additional medical frailty exemptions12 that may be initiated by patients but, at different points in the process, uniformly require a physician’s attestation. However, procedural arrangements differ across states. Variation exists, for example, in administrative burden and the extent to which a physician’s certification by itself is sufficient to receive an exemption. eTable 1 in the Supplement shows a schematic overview of the processes in the first 4 approved states.

Medical frailty exemptions matter because research has shown that most beneficiaries either already meet work requirements or would qualify for such exemptions.2,9,11,12,17-19 However, because exemptions require administrative effort, physicians might be reluctant to assist even when beneficiaries might be harmed as a consequence. Evaluations of prior related reforms suggested that reluctance to request exemptions led to a 10% overall increase in avoidable emergency department use.20

There is also evidence that political affiliation of physicians varies by geography and specialty,21,22 and that physician behavior is influenced by political views or personal attitudes.23,24 Physician decisions could increase risks of harm to beneficiaries, or, conversely, lead to beneficiaries being exempt even if, on regulatory intent, they should not be exempt.25 Acting in nonconcordant ways may create tensions for physicians as they navigate competing personal, professional, and organizational values.26 Further, nonconcordant behavior risks compromising program evaluations, as possible benefits—or, more problematically, harms—may be underestimated. Understanding possible variability therefore matters for patient welfare, physician professionalism, evidence about the effectiveness of programs, as well as ethically and legally.

Methods
Design

We conducted a survey experiment among practicing primary care physicians (PCPs) to understand (1) the degree of uniformity with which PCPs respond in implementing medical frailty exemptions, and (2) insofar as there might be variations, what can explain them. We mailed a 1-page cover letter, survey instrument (4 pages), and brief background information on each state’s work requirements provisions and exemption processes, using reproductions of the state’s information for the public and clinicians (ranging by state from 4-12 pages). Where email addresses were available, recipients also received an electronic invite to view the same information about the policy features online, and to complete the survey on a computer, tablet, or phone. The initial survey was sent July 15, 2019; after 3 reminders, all remaining PCPs with unconfirmed dispositions were contacted again October 4, 2019. We focused on depression as, compared to the general population, prevalence is markedly higher in low-income Americans, with around 30% having been diagnosed with it.27 Prior research also suggests that views on depression, although often associated with stigma, are not associated directly with political partisanship.23 This study was approved by the University of Pennsylvania institutional review board (protocol #832562). Survey participants provided implied consent by returning the survey by mail or completing it online (data use arrangements were detailed in all versions of the instrument).

Sample

We obtained a sample of mail and email addresses of practicing PCPs accepting Medicaid patients in the first 4 approved states (Arkansas [AR], Kentucky [KY], Indiana [IN], New Hampshire [NH]) from SK&A/IQVIA, comprising 5561 individuals. In light of study resources and power calculations, we fielded to 100% of PCPs in AR and NH but selected a random subset of 80% of PCPs in the 2 larger states (IN, KY), yielding a total of 4160 PCPs (AR = 923, IN = 1332, KY = 1331, NH = 574).

Instrument and Experimental Design

Using a 2×2 factorial experimental design, we randomized recipients in each state to respond to a patient clinical scenario. The clinical scenario varied in the disease severity (minor vs major depression) and duration of the physician-patient relationship (new patient vs having been seen for 2 years). On both International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) and Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition), diagnosing a major depressive disorder is appropriate if 5 or more of 9 overall symptoms are present during the same 2-week time period that represents changes in functioning, and if at least 1 symptom is either a depressed mood or loss of interest.28 The minor depression vignette included 3 of 9 symptoms, not triggering an exemption in any of the 4 states. The major depression vignette included 7 of 9 symptoms. In all 4 states, on regulatory intent, this condition would either be sufficient by itself to trigger an exemption review or play a major role. The vignette also stated that the beneficiary told the PCP that he had heard he could be exempted because of his depression, that a friend gave him the state’s information sheet (included in the survey letter) and that, should he receive no exemption, he would lose Medicaid coverage in about 3 weeks, due to not meeting work requirements (eMethods in the Supplement). Physician participants were asked whether they would assist the patient with an exemption. Respondents were also asked to rate the appropriateness of the exemption and the administrative effort required to provide it. Separately, we elicited baseline knowledge about the exemption process, PCP perception of the likelihood that they would encounter patients who lost coverage due to the policy, and overall approval of the policy. Two free-text fields enabled PCPs to share additional thoughts. We also captured 9 personal, professional, and practice characteristics (Table 1).

Outcomes

The primary outcome was the indicator of willingness to assist a patient seeking an exemption due to medical frailty. Secondary outcomes included an ordinal measure of attitude toward appropriateness of exemption.

Statistical Analysis

All analyses were conducted in R statistical software (version 4.0.3, R Foundation). Descriptive statistics were determined both overall and stratified by state for age, gender, race and ethnicity, political affiliation, time since graduation, medical specialty, proportion of Medicaid patients, and number of physicians in the respondent's practice. Analytically, continuous variables were not categorized; however, to report statistics uniformly as absolute and relative frequencies, we treated continuous variables categorically. To address missing data in all regression-based analyses, we used multiple imputation and in each model, a corresponding robust variance estimator to accommodate possible model misspecification.29 Statistical significance was determined to be achieved at the nominal α = .05 level (2-sided).

To investigate factors predicting whether a PCP would seek to assist with an exemption, we fit a logistic regression model with binary vignette response as the outcome using the following predictors: age, gender, race/ethnicity, state, political affiliation, time since graduation, medical specialty, proportion of Medicaid patients, number of physicians in practice, duration scenario, and attitude regarding administrative burden. To determine predictors of perceived appropriateness of exemption, we used an analogous proportional odds model (ie, using the same predictors as stated) categorizing the outcome in ascending order of appropriateness. We further conducted complete-case analyses for regression models as a sensitivity analysis.

Results
Study Population

A total of 715 complete responses were received, yielding, by American Association for Public Opinion Research standards,30 an average overall response rate (RR) of 20.9% (RR1, 20.6%; RR2, 21.2%; RR3, 20.6%; RR4, 21.2%), average cooperation rates (CRs) of 83.8% (CR1, 82.6%; CR2, 84.9%; CR3, 82.6%; CR4, 84.9%) refusal rate of 3.8% (RefR, 1-3 all identical) and a contact rate of 24.9% (ContR, 1-3 all identical). Respondents’ mean (SD) age was 54 (12) years; mean (SD) time since graduations from medical school was 26 (12) years; 435 of 715 (61%) identified as male, 177 as Democrat (25%), 156 as Republican (22%), 197 as Independent or other (28%), and 185 respondents declined or were unknown (26%). The mean (SD) share of Medicaid patients was 29% (21%). Descriptive statistics on demographic factors are shown in Table 1. Nonresponse bias analyses were conducted on variables for which data on responders and the pool of contacted individuals were available, and demonstrated only modest differences: The overall standardized mean difference in age and time since graduation was less than 0.4 years; for categorical variables (gender, specialty, physicians in practice, Area Deprivation Index of practice location), the difference in proportions between responders and the different types of nonresponders never exceeded 4.4% (eResults and eTables 2-4 in the Supplement).

Physician Willingness to Assist With a Medical Frailty Exemption Request

Across the 4 states, counter to regulatory intent, 97 of 387 (25.1%) of PCPs randomized to the minor depression scenario would assist with a medical frailty exemption. Among PCPs randomized to the severe depression, 145 of 315 (46.0%) indicated that they would assist the patient seek an exemption, indicating that most are not complying with regulatory intent. We observed little variation across the 4 states in the minor depression scenario. However, there was substantial variation in the major depression scenario, ranging from 22 of 63 (34.9%) PCPs indicating that they would assist (AR) to 27 of 39 (69.2%; NH); point estimates and corresponding 95% confidence intervals are depicted in Figure 1 across states for each scenario. We did not find sufficient evidence of an association between PCP willingness to assist with an exemption and described length of the clinical relationship. Adjusted for state, those randomized to the longer duration scenario had an estimated 4.2% higher odds of indicating willingness to assist with an exemption as compared with those randomized to the shorter duration scenario (95% CI, −24.0% to 42.9%; P = .80).

Factors Associated With PCP Willingness to Assist When Indicated

Regarding factors associated with assisting with an exemption, we focus on respondents randomized to the major depression scenario, due to the higher stakes for beneficiaries. We hypothesized that several factors may be associated with PCP’s decisions–respondent views related to administrative burden, practice characteristics such as the practice size (larger practices may have access to more administrative or social work support for administrative requests), Medicaid share of patients in the practice, state (given that states’ procedures vary), and political affiliation.

A multivariable model provided strong evidence of an association between state (omnibus P value: P = .01), political affiliation (omnibus P value: P = .002), and perceived administrative effort (omnibus P value: P < .001), and odds of assisting with an exemption. Specifically, NH respondents had the highest odds of indicating willingness to assist. Republicans had an estimated 74.5% lower odds of indicating willingness to assist, compared with Democrats (95% CI, 42.7%-88.7%; P = .001), and Independents/others had an estimated 61.2% lower odds (95% CI, 18.2%-81.6%; P = .01). Those indicating administrative effort as being more appropriate had 313% higher odds of indicating willingness to assist, compared with those viewing administrative effort as highly inappropriate (95% CI, 59.9%-514%; P < .001). Table 2 reports the estimated adjusted odds ratios and their respective confidence intervals and P values for all variables included in this model. Our complete-case sensitivity analysis involved 266 complete cases (out of 321 randomized to the major depression scenario); no conclusions from this sensitivity analysis were found to differ from the main results.

As a descriptive exploratory analysis, Figure 2 illustrates the degree to which responses regarding appropriateness of exemption are concordant with responses regarding willingness to assist. A total of 49 of 245 (20.0%) of respondents indicating that an exemption would be appropriate also indicated that they would not assist with an exemption. Of further note in this subset, 34 of 49 (69.4%) indicated that they believed the administrative effort to be either inappropriate or completely inappropriate, compared with 382 (54.3%) of the overall sample.

Overall PCP attitudes regarding appropriateness of an exemption in the major depression scenario varied by state (omnibus P value: P = .03). Respondents in KY were least likely to find the scenario appropriate for exemption, whereas those in NH were most likely. Republicans had a 79.7% lower odds of reporting a higher degree of exemption appropriateness compared with Democrats (95% CI, 60.8%-89.5%; P < .001); Independents were estimated to have a 65.2% lower adjusted odds compared with Democrats (95% CI, 30.5%-82.6%; P = .003). Further, those finding the administrative effort to be appropriate were estimated to have a 115% higher adjusted odds of reporting a higher degree of exemption appropriateness compared with those finding the administrative effort to be inappropriate (95% CI, 23.6%-273%; P = .007). Further results from this model are shown in eTable 5 in the Supplement.

Physicians’ Perceptions of Work Requirement Policies

Respondents were asked to indicate how informed they were about the policy, whether the administrative effort was appropriate, and whether they approved of the policy. The results are presented in Table 3. With little variation across states, 383 of 708 respondents reported feeling uninformed about their role in exemption processes (54.1%), compared with 139 who reported feeling informed (19.6%). Similarly, 382 of 704 (54.2%) felt that the administrative effort was inappropriate, again, with little variation across states. However, 419 of 715 respondents approved of the work requirement policy (58.6%), with particularly high approval in IN (172/260 [64.2%]), and particularly low approval in NH (41/105 [38.7%]).

Discussion

In this survey study of PCPs, we obtained responses from more than 700 PCPs practicing in the first 4 states approved to test Medicaid work requirements. We found that 25% would offer assistance even when the state’s medically frail criteria would not support an exemption, and 54% would not offer assistance when a patient does quality for assistance. We found that 20% of respondents who deem an exemption appropriate indicated that they would not assist a patient with requesting one. We further noted substantial variation by state, and that political affiliation, administrative effort, and perceived appropriateness were statistically associated with the odds of assisting with an exemption and can explain heterogeneity in physician decisions where none should exist.

In line with earlier research, we find that physicians are unwilling to simply implement in mechanistic ways rules that conflict with their personal and professional preferences, and that partisan bias can lead to unwarranted variation in patient care.23,31 From a patient welfare perspective, our findings have both upsides and downsides.

On the one hand, a quarter of PCPs signal that they would assist with an exemption even if regulations would not permit this. This aligns with position statements of major medical associations that expressed major concern about the likelihood that the work requirements jeopardize health care access32 and prior attitudinal and empirical research suggesting that between 10% to 60% of physicians will find workarounds to ensure that patients receive needed care, even when they risk disciplinary sanctions (for example, by exaggerating the severity of conditions; changing billing diagnoses; and/or reporting signs or symptoms patients did not have, absorbing cost of treatment, actively referring patients to no-fee safety-net physicians, or violating duty hour restrictions).33-42

On the other hand—and far more concerning—half of PCPs indicate that they would not assist a patient even when regulations suggest that they should do so, to mitigate the risk of harming patients by revoking health care access.

In terms of implications for ongoing legal challenges that center on whether work requirements are incompatible with Medicaid’s objective of furnishing health care access, our study highlights that there is a need to address the fact that even when work requirement’s procedures include rules seeking to protect beneficiaries, variability in PCP’s behavior can directly harm beneficiaries. Our findings suggest that the administrative burden of exemption procedures is regarded as inappropriately high by many PCPs, and that it may cause a PCP to not seek an exemption even if it is warranted. Insofar as policies such as work requirements should be deemed acceptable by the Supreme Court or are otherwise implemented again, it is critical that CMS proactively identify measures to ensure that patients qualifying for exemptions are not put at risk due to either the burdensomeness of exemption procedures, or physicians’ political or other views.

Strengths and Limitations

In absolute terms, our response rate was low. However, we have sufficient statistical power to support our central conclusions and the experimental design provides strong internal validity in a balanced sample, including political affiliation. We reported stated, not revealed preferences, and social desirability bias can convey a rosier picture, moreover, nonresponse bias analyses revealed only modest differences. Although generally a limitation, in our case this only strengthens concerns about the actual effect of nonconcordant behavior because respondents could easily have made statements in line with regulatory intent. The condition used in the vignette was depression: results may differ for other conditions. Yet, given the noted high prevalence among the Medicaid population,36 the focus is certainly meaningful. Frailty determinations are only 1 way in which PCPs are involved in the process of deciding who is expected to meet work requirements (others are disability status determinations, and related issues arise routinely in sick-leave notes and other decisions). Medical frailty determinations therefore are not singularly unique, but still matter centrally, given the large number of eligible beneficiaries.12

Conclusions

Work requirements established new obligations for both beneficiaries and physicians. Although concerns about the demandingness of exemption procedures are entirely understandable, the willingness to assist beneficiaries qualifying for medical frailty exemptions is also determined by political views, and the perceived appropriateness of exemptions. Work requirements are not the only case where personal preferences can make a difference between vulnerable populations retaining health care access or not, but the effect of these preferences needs to be considered more fully in program design and evaluations, as well as in the ongoing review of the legal justification for the policy.

Back to top
Article Information

Accepted for Publication: August 8, 2021.

Published: October 1, 2021. doi:10.1001/jamahealthforum.2021.2932

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2021 Schmidt H et al. JAMA Health Forum.

Corresponding Author: Harald Schmidt, PhD, Medical Ethics and Health Policy, Perelman School of Medicine, University of Pennsylvania, 3401 Market St, Ste 320, Philadelphia, PA 19104 (schmidth@mail.med.upenn.edu).

Author Contributions: Drs Schmidt and Spieker had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Schmidt and Spieker were co-first authors. Drs Szymczak and Grande were co-senior authors.

Concept and design: Schmidt, Spieker, Szymczak, Grande.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Schmidt, Spieker, Luo.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Spieker, Luo.

Obtained funding: Schmidt, Spieker, Szymczak.

Administrative, technical, or material support: Schmidt, Luo, Szymczak, Grande.

Supervision: Schmidt, Spieker, Szymczak.

Conflict of Interest Disclosures: Drs Schmidt, Spieker, Szymczak, Grande reported grants from the Robert Wood Johnson Foundation’s Research in Transforming Health and Health Care Systems program (Grant # 76037) during the conduct of the study. No other disclosures were reported.

Funding/Support: This study was funded by the Robert Wood Johnson Foundation’s Research in Transforming Health and Health Care Systems program (Grant # 76037).

Role of the Funder/Sponsor: The Robert Wood Johnson Foundation had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional ContributionS: We thank Loretta Alexander, MHSA, Arkansas Advocates for Children and Families; Amy Ashbridge, MBA, University of Pennsylvania; Sarah Bart, MBE, University of Pennsylvania; Alexandra Belfi, University of Pennsylvania; Kat Berkowitz, University of Pennsylvania; Susanna Birnbaum, University of Pennsylvania; Lauren Counterman, University of Pennsylvania; Myles Deal, MBE, University of Pennsylvania; Kevin De Liban, JD, Legal Aid of Arkansas--West Memphis; Roma Dhamanaskar, MBE, University of Pennsylvania; Claire Dinh, MBE, University of Pennsylvania; Erica Dixon, PhD, University of Pennsylvania; Raksha Dondapati, BA, University of Pennsylvania; Alec Hilton, BA, Washington University in St. Louis; Nancy Kennedy Smith, University of Pennsylvania; Victoria Lee, University of Pennsylvania; Ashwini Nagappan, MBE, University of Pennsylvania; Mel Nasielski, Kennedy Printing Company, Inc, Philadelphia; AJ Roholt, University of Pennsylvania; Theodore Satterthwaite, MD, University of Pennsylvania; Stephanie Seghers, University of Pennsylvania; Ron Shapiro, Campus Copy, Philadelphia; Meghana Sharma, MPH, University of Pennsylvania; Judy Shea, PhD, University of Pennsylvania; Atheendar Venkataramani, PhD, University of Pennsylvania; and in particular the primary care physicians who kindly shared their views.

References
1.
Centers for Medicare & Medicaid Services. 2018. Opportunities to Promote Work and Community Engagement among Medicaid Beneficiaries. SMD Letter 18-002, Center for Medicare & Medicaid Services, U.S. Department of Health and Human Services, Washington, DC. Accessed September 3, 2021. www.cms.gov/newsroom/press-releases/cms-announces-new-policy-guidance-states-test-community-engagement-able-bodied-adults
2.
Huberfeld  N.  Can work be required in the Medicaid program?   N Engl J Med. 2018;378(9):788-791. doi:10.1056/NEJMp1800549PubMedGoogle ScholarCrossref
3.
Bagley  N.  Are Medicaid work requirements legal?   JAMA. 2018;319(8):763-764. doi:10.1001/jama.2018.0958PubMedGoogle ScholarCrossref
4.
Rosenbaum  S. Medicaid work requirements: inside the decision overturning Kentucky Health’s approval. Health Affairs Blog 2018. Accessed August 30, 2021. https://www.healthaffairs.org/do/10.1377/hblog20180702.144007/full/
5.
Ayanian  JZ, Tipirneni  R, Goold  SD.  Mitigating the risks of Medicaid work requirements.   N Engl J Med. 2018;379(9):803-805. doi:10.1056/NEJMp1806129PubMedGoogle ScholarCrossref
6.
Wagner  J, Schubel  J.  States’ Experiences Confirm Harmful Effects of Medicaid Work Requirements. Washington, DC: Center on Budget and Policy Priorities;2020.
7.
Chen  L, Sommers  BD.  Work requirements and Medicaid Disenrollment in Arkansas, Kentucky, Louisiana, and Texas, 2018.   Am J Public Health. 2020;110(8):1208-1210. doi:10.2105/AJPH.2020.305697PubMedGoogle ScholarCrossref
8.
Center on Budget and Policy Priorities. Taking Away Medicaid for Not Meeting Work Requirements Harms Low-Wage Workers. 2020. Accessed August 30, 2021. https://www.cbpp.org/research/health/taking-away-medicaid-for-not-meeting-work-requirements-harms-low-wage-workers
9.
Sommers  BD, Goldman  AL, Blendon  RJ, Orav  EJ, Epstein  AM.  Medicaid work requirements–results from the first year in Arkansas.   N Engl J Med. 2019;381(11):1073-1082. doi:10.1056/NEJMsr1901772PubMedGoogle ScholarCrossref
10.
Manivannan  A, Adkins-Hempel  M, Shippee  ND, Vickery  KD.  Experiences with work and participation in public programs by low-income Medicaid enrollees using qualitative interviews.   J Gen Intern Med. 2020;35(10):2983-2989. doi:10.1007/s11606-020-05921-zPubMedGoogle ScholarCrossref
11.
Sommers  BD, Chen  L, Blendon  RJ, Orav  EJ, Epstein  AM.  Medicaid work requirements in Arkansas: two-year impacts on coverage, employment, and affordability of care.   Health Aff (Millwood). 2020;39(9):1522-1530. doi:10.1377/hlthaff.2020.00538PubMedGoogle ScholarCrossref
12.
Venkataramani  AS, Bair  EF, Dixon  E,  et al.  Assessment of Medicaid beneficiaries included in community engagement requirements in Kentucky.   JAMA Netw Open. 2019;2(7):e197209. doi:10.1001/jamanetworkopen.2019.7209PubMedGoogle Scholar
13.
Rosenbaum  S. How the Biden Administration Can Act to Strengthen Medicaid. In: The Commonwealth Fund, ed. To the Point 2020. Accessed August 30, 2021. www.commonwealthfund.org/blog/2020/how-biden-administration-can-act-strengthen-medicaid
14.
Bagley  N.  How to Undo One of Trump’s Worst, Most Despicable Policies.  The Atlantic. 2021. Accessed August 30, 2021. www.theatlantic.com/ideas/archive/2021/01/undoing-trump-damage/617736/
15.
Kaiser Family Foundation. Work Requirement Waivers: Approved and Pending as of June 28, 2021. Published 2021. Accessed July 22, 2021. www.kff.org/medicaid/issue-brief/medicaid-waiver-tracker-approved-and-pending-section-1115-waivers-by-state/#Table2
16.
Wheeler  L. Supreme Court Leaves Fight Over Medicaid Work Rules in Limbo. Bloomberg Law. July 12, 2021.
17.
Huberfeld  N.  The universality of Medicaid at fifty.   Yale J Health Policy Law Ethics. 2015;15(1):67-87.PubMedGoogle Scholar
18.
Goldman  AL, Woolhandler  S, Himmelstein  DU, Bor  DH, McCormick  D.  Analysis of work requirement exemptions and Medicaid spending.   JAMA Intern Med. 2018;178(11):1549-1552. doi:10.1001/jamainternmed.2018.4194PubMedGoogle ScholarCrossref
19.
Solomon  J.  Medicaid Work Requirements Can’t Be Fixed: Unintended Consequences are Inevitable Result. Washington, DC: Center on Budget and Policy Priorities;2019.
20.
Gurley-Calvez  T. KG, Simon  K, Wissoker  D,. Medicaid reform and emergency room visits: evidence from West Virginia’s Medicaid redesign. 2012. Accessed September 1, 2021. www.mercatus.org/publications/government-spending/medicaid-reform-and-emergency-room-visits-evidence-west-virginias
21.
Bonica  A, Rosenthal  H, Blackwood  K, Rothman  DJ.  Ideological sorting of physicians in both geography and the workplace.   J Health Polit Policy Law. 2020;45(6):1023-1057. doi:10.1215/03616878-8641555PubMedGoogle ScholarCrossref
22.
Bonica  A, Rosenthal  H, Rothman  DJ.  The political polarization of physicians in the United States: an analysis of campaign contributions to federal elections, 1991 through 2012.   JAMA Intern Med. 2014;174(8):1308-1317. doi:10.1001/jamainternmed.2014.2105PubMedGoogle ScholarCrossref
23.
Hersh  ED, Goldenberg  MN.  Democratic and Republican physicians provide different care on politicized health issues.   Proc Natl Acad Sci U S A. 2016;113(42):11811-11816. doi:10.1073/pnas.1606609113PubMedGoogle ScholarCrossref
24.
Burgess  DJ, Hardeman  RR, Burke  SE,  et al  Incoming Medical Students' Political Orientation Affects Outcomes Related to Care of Marginalized Groups: Results from the Medical Student CHANGES Study.   Journal of health politics, policy and law. 2019;44(1):113-146. doi:10.1215/03616878-7206755Google ScholarCrossref
25.
Schmidt  H, Hoffman  AK.  The Ethics of Medicaid’s Work Requirements and Other Personal Responsibility Policies.   JAMA. 2018;319(22):2265-2266. doi:10.1001/jama.2018.3384PubMedGoogle ScholarCrossref
26.
Ellsbury  KE, Carline  JD, Wenrich  MD.  Competing professionalism values among community-based family physicians.   Acad Med. 2006;81(10)(suppl):S25-S29. doi:10.1097/01.ACM.0000236562.10528.20PubMedGoogle Scholar
27.
Centers for Medicare & Medicaid Services. Physical and Mental Health Condition Prevalence and Comorbidity among Fee-for-Service Medicare- Medicaid Enrollees, available at: www.cms.gov/Medicare-Medicaid-Coordination/Medicare-and-Medicaid-Coordination/Medicare-Medicaid-Coordination-Office/Downloads/Dual_Condition_Prevalence_Comorbidity_2014.pdf. 2014.
28.
AACP (American Academy of Professional Coders). ICD 10 Resource: Coding for Major Depressive Disorder Availlable at: www.hamiltonhealth.com/wp-content/uploads/2017/08/AAPC_Depressive-Disorder-ICD-10-BH_Ref_Guide.pdf. 2013.
29.
Huber  P. The behavior of maximum likelihood estimates under nonstandard conditions. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability. 1967;1:221-233.
30.
American Association for Public Opinion Research (AAPOR). AAPOR Response Rate Calculator 4.1 availablle at: www.aapor.org/Education-Resources/For-Researchers/Poll-Survey-FAQ/Response-Rates-An-Overview.aspx. 2020.
31.
Zatz  LY, Hersh  E, Gudzune  KA, Thorndike  AN, N Goldenberg  M, Bleich  SN.  Physicians’ political party affiliation and clinical management of obesity.   Clin Obes. 2020;10(5):e12396. doi:10.1111/cob.12396PubMedGoogle Scholar
32.
American Academy of Family Physicians, American Academy of Pediatrics, American Collegeof Obstetricians and Gynecologists, American College of Physicians, American Osteopathic Association, American Psychiatric Association. Section 1115: Demonstration waivers and other proposals to change Medicaid benefits, financing and cost-sharing. Joint position statement. Accessed September 1, 2021. www.aafp.org/dam/AAFP/documents/advocacy/coverage/medicaid/LT-Group6-PrinciplesMedicaidWaivers-120817.pdf
33.
Lee  SY, Dow  WH, Wang  V, VanGeest  JB.  Use of deceptive tactics in physician practices: are there differences between international and US medical graduates?   Health Policy. 2004;67(3):257-264. doi:10.1016/S0168-8510(03)00124-6PubMedGoogle ScholarCrossref
34.
VanGeest  J, Weiner  S, Johnson  T, Cummins  D.  Impact of managed care on physicians’ decisions to manipulate reimbursement rules: an explanatory model.   J Health Serv Res Policy. 2007;12(3):147-152. doi:10.1258/135581907781543102PubMedGoogle ScholarCrossref
35.
VanGeest  JB, Wynia  MK, Cummins  DS, Wilson  IB.  Measuring deception: test-retest reliability of physicians’ self-reported manipulation of reimbursement rules for patients.   Med Care Res Rev. 2002;59(2):184-196. doi:10.1177/1077558702059002004PubMedGoogle ScholarCrossref
36.
Wynia  MK, Cummins  DS, VanGeest  JB, Wilson  IB.  Physician manipulation of reimbursement rules for patients: between a rock and a hard place.   JAMA. 2000;283(14):1858-1865. doi:10.1001/jama.283.14.1858PubMedGoogle ScholarCrossref
37.
Freeman  VG, Rathore  SS, Weinfurt  KP, Schulman  KA, Sulmasy  DP.  Lying for patients: physician deception of third-party payers.   Arch Intern Med. 1999;159(19):2263-2270. doi:10.1001/archinte.159.19.2263PubMedGoogle ScholarCrossref
38.
Rost  K, Smith  R, Matthews  DB, Guise  B.  The deliberate misdiagnosis of major depression in primary care.   Arch Fam Med. 1994;3(4):333-337. doi:10.1001/archfami.3.4.333PubMedGoogle ScholarCrossref
39.
Bogardus  ST  Jr, Geist  DE, Bradley  EH.  Physicians’ interactions with third-party payers: is deception necessary?   Arch Intern Med. 2004;164(17):1841-1844. doi:10.1001/archinte.164.17.1841PubMedGoogle ScholarCrossref
40.
Rushton  JL, Felt  BT, Roberts  MW.  Coding of pediatric behavioral and mental disorders.   Pediatrics. 2002;110(1 Pt 1):e8. doi:10.1542/peds.110.1.e8PubMedGoogle Scholar
41.
Nadler  SF, Malanga  GA, Ciccone  DS.  Positive straight-leg raising in lumbar radiculopathy: is documentation affected by insurance coverage?   Arch Phys Med Rehabil. 2004;85(8):1336-1338. doi:10.1016/j.apmr.2004.02.021PubMedGoogle ScholarCrossref
42.
Werner  RM, Alexander  GC, Fagerlin  A, Ubel  PA.  The “hassle factor”: what motivates physicians to manipulate reimbursement rules?   Arch Intern Med. 2002;162(10):1134-1139. doi:10.1001/archinte.162.10.1134PubMedGoogle ScholarCrossref
×