Eric T. Roberts

Objective: To assess existing algorithms for measuring Home‐ and Community‐Based Services (HCBS) in Medicaid claims, and to develop and evaluate a unified, modified algorithm. Study Setting and Design: Using Transformed Medicaid Statistical Information System Analytic Files (TAF), we compared two existing algorithms for identifying and classifying 1915(c) waiver and state plan HCBS. Informed by that analysis, we developed a third, unified and modified algorithm. Using each algorithm, we constructed annual state‐level estimates of HCBS use per 100,000 residents, which we compared from 2021 to 2023. Data Sources and Analytic Sample: National 20% sample of TAF data. The sample included continuously enrolled, full‐benefit Medicaid beneficiaries age 19 or older in 2021 for primary analyses, and 2022–2023 to compare counts over time. Principal Findings: The existing algorithms showed a high level of concordance (observed agreement, P _o = 93% and kappa = 0.77). Disagreement resulted from missing or inconclusive program type codes, differences in service‐related codes used to classify HCBS, and differences in classifying claims for capitated payments. Our algorithm addressed some of this disagreement with the novel use of waiver identification numbers and modification of some TAF data elements and codes. HCBS user counts were mostly consistent across algorithms, with our algorithm identifying the broadest population of users. However, some states exhibited inconsistent counts depending on the algorithm, implausibly high counts (Pennsylvania and Illinois), or implausibly low counts (Mississippi). In several states, HCBS user counts varied substantially from 2021 to 2023 (> 50% in absolute value), suggesting changes in data quality or reporting in those states across years. Conclusions: Our algorithm resolves some discordances between two existing algorithms for measuring HCBS use in TAF. However, in certain states and years, inconsistent or implausible HCBS user counts suggest variations in data reliability and coding patterns that could limit accurate measurement of HCBS, regardless of the algorithm used.

Objective: To investigate whether eligibility for Veterans Health Administration (VA)‐purchased community care, which expanded Veterans' access to care outside VA, was associated with increased polypharmacy or potentially inappropriate medication use among older adult Veterans. Study Setting and Design: Regression discontinuity design, leveraging the distance threshold for community care eligibility (residing > 40 miles from the nearest VA facility with ≥ 1 or more full‐time primary care physician), to examine the effects of community care eligibility on polypharmacy and potentially inappropriate medication use among Veterans aged ≥ 65 years. Data Sources and Analytic Sample: VA pharmacy data for all prescriptions filled at VA facilities, VA Program Integrity Tool files for prescriptions paid by VA and filled in community pharmacies, and Medicare Part D data. Analyses included annual cross‐sectional samples of Veterans 36–39 miles or 41–44 miles from their nearest VA facility during FY 2016–2019. Principal Findings: The sample included 399,250 Veteran‐year observations, of which 226,157 (56.6%) were 36–39 miles and 173,093 (43.4%) were 41–44 miles from the nearest eligible VA facility. Overall, we observed no discontinuities across the 40‐mile threshold in the number of unique medications filled annually (−0.06 medications; 95% confidence interval [CI], −0.15 to 0.03). There were no discontinuities in proportions of Veterans filling ≥ 5 unique medications (−0.25 percentage points [pp]; 95% CI, −0.84 to 0.34), ≥ 10 medications (−0.55 pp.; 95% CI, −1.24 to 0.14), ≥ 1 medication on the Beers list (−0.02 pp.; 95% CI, −0.63 to 0.59), or ≥ 1 high‐risk drug–drug interaction (−0.05 pp.; 95% CI, −0.17 to 0.07). Among Veterans with mental health conditions, exceeding the 40‐mile threshold was associated with a higher likelihood of filling ≥ 10 unique medications annually (2.06 pp.; 95% CI, 0.42 to 3.70). We did not observe clinically or statistically significant discontinuities in other subgroups. Conclusions: Overall, eligibility for VA‐purchased community care was not associated with increased polypharmacy or potentially inappropriate medication use among older adult Veterans.

PURPOSE: People experiencing housing insecurity (PEHI) may face barriers to preventive cancer screening, but the extent to which PEHI receive screening at the population level is understudied. Therefore, this study examined disparities in utilization of breast cancer screening associated with housing insecurity (HI) among women beneficiaries enrolled in a large Medicaid program. METHODS: We assessed receipt of biennial mammography among women Medicaid beneficiaries, age 50-64 years, in administrative Medicaid data linked to encounter-level records on housing-related services in Pennsylvania (PA; Homelessness Management Information System [HMIS]) from 2011 to 2019. We characterized disparities in screening between beneficiaries with an encounter for housing-related services in HMIS (indicating experience or history of HI) compared with beneficiaries without any such encounters. RESULTS: We included 73,456 women Medicaid beneficiaries in PA, including 1,792 HMIS service recipients and 71,664 comparison beneficiaries. The mammography rate was 44.8% among HMIS service recipients and 50.1% among comparison beneficiaries, representing a disparity of 5.3 percentage-points (pp; P < .001). This disparity was pronounced between HMIS service recipients with chronic experience of homelessness and comparison beneficiaries (–9.4 pp; P = .036). In stratified analyses, beneficiaries without any primary care visits had the lowest rate of mammography, although realized access to primary care did not eliminate the disparity in screening associated with HI. CONCLUSION: HI was associated with a significant disparity in mammography among women Medicaid beneficiaries, underscoring the need for efforts to improve preventive care among housing-insecure populations.

Medicare and Medicaid are separate programs that together cover 13 million low-income older adults and people with disabilities, known as dual-eligible individuals. Concern about a lack of coordination across Medicare and Medicaid has prompted the development of Integrated Care Programs (ICPs). Although the primary goal of ICPs is to coordinate financing and care across Medicare and Medicaid, ICPs may also influence whether low-income individuals obtain or keep Medicaid. We evaluated whether the rollout of Medicare-Medicaid Plans (MMPs)—one of the largest ICPs—was associated with changes in Medicaid take-up and retention among Medicare beneficiaries residing in high-poverty zip codes. Using a stacked difference-in-differences design and variation in MMP rollouts across nine states, we found no evidence that MMPs increased monthly or continuous Medicaid enrollment in this population. These findings highlight the need for focused policies to address Medicaid enrollment gaps among low-income Medicare beneficiaries, which could complement broader integration efforts.

Objective: To examine whether eligibility for Veterans Health Administration (VA) community care, which expanded Veterans' access to VA‐funded care outside VA, increased the likelihood of Veterans concurrently filling prescriptions for opioids and central nervous system (CNS)‐active medications. Study Setting and Design: We used a regression discontinuity design to analyze Veterans across a distance threshold for community care eligibility in the Veterans Choice Program, under which Veterans residing > 40 miles from the closest VA medical facility staffed by ≥ 1 full‐time primary care physician qualified for community care. We used local linear regression to test whether exceeding this 40‐mile threshold was associated with discontinuities in the probability of receiving overlapping supplies of opioids and another CNS medication (benzodiazepine, muscle relaxant, antiepileptic, or sleep aid) for ≥ 30 days per year. Data Sources and Analytic Sample: We used VA pharmacy data for prescriptions filled at VA facilities, VA Program Integrity Tool files for prescriptions paid by VA and filled in community pharmacies, and Medicare and Medicaid data for prescriptions covered by those programs. Our analysis included annual cross‐sectional samples of Veterans who filled ≥ 1 opioid prescription through VA, community care, Medicare, or Medicaid and lived 36–39 or 41–44 miles from the nearest VA facility during federal FYs 2016–2019 ( n = 180,903 Veteran‐year observations). Principal Findings: Among Veterans who filled an opioid prescription, 34.1% concurrently received another CNS medication for ≥ 30 days. Exceeding the threshold for community care eligibility was associated with a 1.14 percentage point (pp) increase (95% CI: 0.08, 2.20) in the probability of concurrently receiving an opioid and another CNS drug during 2016–2019. Discontinuities in overlap were larger among Veterans with a serious mental illness (2.7 pp.; 95% CI: 0.6, 4.9) during 2016–2019. During 2018–2019, discontinuities were larger in the overall sample (1.6 pp.; 0.0, 3.1) and among non‐Hispanic Black Veterans (5.4 pp.; 95% CI: 0.5, 10.4). Conclusions: Overall, VA community care eligibility was associated with a small increase in medication overlap involving opioids and other CNS‐active medications. Increases in overlap were larger in certain Veteran subgroups and later study years, underscoring a need for continued monitoring of higher‐risk co‐prescribing in VA community care.

Objective: This study aimed to measure the changes in rural hospital bypass for 11 common elective surgeries following the implementation of the Pennsylvania Rural Health Model (PARHM), a global budget payment model.Study Setting and Design: We leveraged a natural experiment arising from the phase‐in of PHARM in Pennsylvania. We conducted a comparative interrupted time series analysis to assess changes in rural hospital bypass, comparing trends in rural hospital bypass among patients in hospital service areas (HSAs) with PARHM‐participating hospitals to patients in control HSAs with hospitals eligible for but not participating in PARHM. Analyses accounted for staggered entry into PARHM and examined outcomes up to 4 years post‐entry.Data Sources and Analytic Sample: We used Pennsylvania all‐payer visit‐level inpatient discharge data (2016–2022) to measure rural hospital bypass, encompassing 175,138 surgeries.Principal Findings: The average bypass rate for elective surgeries was 59.9%, with an increasing trend observed during the study period. Overall, differential changes in bypass rates between PARHM‐participating and control HSAs were not statistically significant, from a low of 0.53 percentage points (−8.17–9.22) among Cohort 2 HSAs and a high of 5.96 percentage points (−4.63–16.55) among Cohort 1 HSAs. However, among critical access hospitals, PARHM participation was associated with a significant relative increase in levels and trends in bypass rates compared to controls, from a low of 9.12 percentage points (2.45–15.79) among Cohort 1 HSAs and a high of 29.70 percentage points (12.54–46.86) among Cohort 2 HSAs. These relative increases were largely due to a stable rate in PARHM‐participating HSAs and a marked decrease in control HSAs.Conclusions: This study fills a gap in the relationship between global budgets and hospital bypass. Although PARHM did not broadly alter rural bypass rates overall, the differential increase in bypass among HSAs with CAHs participating in PARHM suggests meaningful effect heterogeneity, warranting further research and analysis of impacts on patient outcomes.

Objective: To examine changes in late‐ versus early‐stage diagnosis of cancer associated with the introduction of mandatory Medicaid managed care (MMC) in Pennsylvania.Data Sources and Study Setting: We analyzed data from the Pennsylvania cancer registry (2010–2018) for adult Medicaid beneficiaries aged 21–64 newly diagnosed with a solid tumor. To ascertain Medicaid and managed care status around diagnosis, we linked the cancer registry to statewide hospital‐based facility records collected by an independent state agency (Pennsylvania Health Care Cost Containment Council).Study Design: We leveraged a natural experiment arising from county‐level variation in mandatory MMC in Pennsylvania. Using a stacked difference‐in‐differences design, we compared changes in the probability of late‐stage cancer diagnosis among those residing in counties that newly transitioned to mandatory managed care to contemporaneous changes among those in counties with mature MMC programs.Data Collection/Extraction Methods: N/A.Principal Findings: Mandatory MMC was associated with a reduced probability of late‐stage cancer diagnosis (−3.9 percentage points; 95% CI: −7.2, −0.5; p = 0.02), particularly for screening‐amenable cancers (−5.5 percentage points; 95% CI: −10.4, −0.6; p = 0.03). We found no significant changes in late‐stage diagnosis among non‐screening amenable cancers.Conclusions: In Pennsylvania, the implementation of mandatory MMC for adult Medicaid beneficiaries was associated with earlier stage of diagnosis among newly diagnosed cancer patients with Medicaid, especially those diagnosed with screening‐amenable cancers. Considering that over half of the sample was diagnosed with late‐stage cancer even after the transition to mandatory MMC, Medicaid programs and managed care organizations should continue to carefully monitor receipt of cancer screening and design strategies to reduce barriers to guideline‐concordant screening or diagnostic procedures.

The Pennsylvania Rural Health Model (PARHM) is a novel alternative payment model for rural hospitals that aims to test whether hospital-based global budgets, coupled with delivery transformation plans, improve the quality of health care and health outcomes in rural communities. Eighteen hospitals joined PARHM in 3 cohorts between 2019 and 2021. This study assessed PARHM’s impact on changes in potentially avoidable utilization (PAU)—a measure of admission rates policymakers explicitly targeted for improvement in PARHM. Using a difference-in-differences analysis and all-payer hospital discharge data for Pennsylvania hospitals from 2016 through 2022, we found no significant overall reduction in community-level PAU rates up to 4 years post–PARHM implementation, relative to changes in rural Pennsylvania communities whose hospitals did not join PARHM. However, heterogeneous treatment effects were observed across cohorts that joined PARHM in different years, and between critical access vs prospective payment system hospitals. These findings offer insight into how alternative payment models in rural health care settings may have heterogeneous impacts based on contextual factors and highlight the importance of accounting for these factors in proposed expansions of alternative payment models for rural health systems.

Although statins are a class I recommendation for prevention of atherosclerotic cardiovascular disease and its complications, their use is suboptimal. Differential underuse may mediate disparities in cardiovascular health for systematically marginalized persons. To estimate disparities in statin use by race-ethnicity-gender and to determine whether these potential disparities are explained by medical appropriateness of therapy and structural factors. Cross-sectional analysis. National Health and Nutrition Examination Survey from 2015 to 2020. Persons eligible for statin therapy based on 2013 and 2018 American College of Cardiology/American Heart Association blood cholesterol guidelines. The independent variable was race-ethnicity-gender. The outcome of interest was use of a statin. Using the Institute of Medicine framework for examining unequal treatment, we calculated adjusted prevalence ratios (aPRs) to estimate disparities in statin use adjusted for age, disease severity, access to health care, and socioeconomic status relative to non-Hispanic White men. For primary prevention, we identified a lower prevalence of statin use that was not explained by measurable differences in disease severity or structural factors among non-Hispanic Black men (aPR, 0.73 [95% CI, 0.59 to 0.88]) and non-Mexican Hispanic women (aPR, 0.74 [CI, 0.53 to 0.95]). For secondary prevention, we identified a lower prevalence of statin use that was not explained by measurable differences in disease severity or structural factors for non-Hispanic Black men (aPR, 0.81 [CI, 0.64 to 0.97]), other/multiracial men (aPR, 0.58 [CI, 0.20 to 0.97]), Mexican American women (aPR, 0.36 [CI, 0.10 to 0.61]), non-Mexican Hispanic women (aPR, 0.57 [CI, 0.33 to 0.82), non-Hispanic White women (aPR, 0.69 [CI, 0.56 to 0.83]), and non-Hispanic Black women (aPR, 0.75 [CI, 0.57 to 0.92]). Cross-sectional data; lack of geographic, language, or statin-dose data. Statin use disparities for several race-ethnicity-gender groups are not explained by measurable differences in medical appropriateness of therapy, access to health care, and socioeconomic status. These residual disparities may be partially mediated by unobserved processes that contribute to health inequity, including bias, stereotyping, and mistrust. National Institutes of Health.