key: cord-0074541-6srg253v authors: Bhanja, Aditi; Lee, Dennis; Gordon, Sarah H.; Allen, Heidi; Sommers, Benjamin D. title: Comparison of Income Eligibility for Medicaid vs Marketplace Coverage for Insurance Enrollment Among Low-Income US Adults date: 2021-06-14 journal: JAMA Health Forum DOI: 10.1001/jamahealthforum.2021.0771 sha: 32d4941334ccd4e4959322eb7626992211ce8b3c doc_id: 74541 cord_uid: 6srg253v IMPORTANCE: The Affordable Care Act created 2 new coverage options for uninsured adults: Medicaid expansion, which in most states provides comprehensive coverage without premiums and deductibles; and private marketplace coverage, which requires a premium contribution and cost-sharing, though with generous federal subsidies at lower incomes. How enrollment rates compare in the marketplace vs Medicaid is an important policy question as states continue to weigh alternative coverage options such as Medicaid buy-in programs, enrolling Medicaid-eligible populations into marketplace plans, or creating a public option. OBJECTIVE: To assess the association between income eligibility for Medicaid vs marketplace coverage and insurance enrollment among low-income adults in Colorado. DESIGN, SETTING, AND PARTICIPANTS: Using 2014 and 2015 all-payer claims data from Colorado and detailed income eligibility information, we used a regression discontinuity design to assess the difference in Medicaid and marketplace enrollment just below and just above 138% of the federal poverty level (FPL), the eligibility threshold between the 2 programs. The sample included nonpregnant adults aged 19 to 64 years with incomes between 75% to 400% FPL. We stratified our analysis by age, sex, chronic condition status, and urban vs rural residence. Analysis was conducted from January to October 2020. MAIN OUTCOME AND MEASURES: The main outcome was total enrollment in either Medicaid or marketplace coverage during marketplace’s Open Enrollment period. Income-based health insurance eligibility was assessed as a percentage of FPL at the time of initial application for coverage. RESULTS: The primary analytical sample included 32 091 enrollees in 2014 and 55 451 in 2015, with incomes ranging from 120% to 156% FPL. Most enrollees were women (59.26% in 2014, 59.20% in 2015), resided in urban areas (70.36% in 2014, 73.08% in 2015), and had no chronic conditions (74.66% in 2014, 76.11% in 2015). For age, in 2014 and 2015, respectively, 13.22% and 13.93% were aged 19 to 25 years, 27.85% and 28.54% were aged 26 to 34 years, 23.58% and 24.34% were aged 35 to 44 years, 18.35% and 17.75% were aged 45 to 54 years, and 17.00% and 15.44% were aged 55 to 64 years. Marketplace enrollment was 81.3% (95% CI, −86.0% to −75.0%) lower than Medicaid enrollment in 2014 and 88.6% (95% CI, −90.8% to −86.0%) lower in 2015 among those close to the 138% FPL eligibility threshold. The drop-off in marketplace enrollment was largest among younger adults, aged 26 to 34 and 35 to 44 years: relative drop off −88.7% (95% CI, −93.3% to −80.8%) and −87.8% (95% CI, −90.8% to −83.9%) in 2014, and relative drop off –91.9% (95% CI, −94.5% to −87.9%) and −93.0% (95% CI, −94.5% to −91.1%) in 2015, respectively. CONCLUSIONS AND RELEVANCE: In this cross-sectional study using a regression-discontinuity analysis, meaningful gaps in insurance enrollment may have existed for those with incomes just above the eligibility threshold for Medicaid expansion, especially among younger adults. Policies expanding Medicaid income eligibility or zero-dollar premium marketplace plans are likely to be more effective at inducing enrollment than subsidized private plans with premium requirements. Over the study period, 11 health insurance plans operated in the Colorado Marketplace, excluding dental insurance plans. The CO APCD includes six of these, while BEST Life and Health Insurance Company left the Marketplace in 2014 and ceased submitting claims and Colorado HealthOP left in 2016 but does not appear in the data. We assessed the share of Marketplace enrollees captured by our dataset, by comparing the sample size in our data files to the publicly-reported totals of individuals in the Colorado Marketplace. Our 2014 sample contained 62,336 Marketplace individuals with income data and FPL < 400, compared to approximately 76,110 individuals who enrolled in C4CHO and received subsidies in 2014 (59% of the 129,000 reported by C4HCO's official statistics. 3 For 2015, the comparable numbers were 64,165 Marketplace individuals with income data and FPL < 400 and 76,700 who enrolled in C4CHO and received subsidies in 2015. 4 For our analyses, we excluded individuals who qualified for Medicaid via disability 6 or pregnancy-related eligibility due to differences in income eligibility criteria. We also excluded marketplace individuals who had incomes below 138% of FPL because this group consists primarily of individuals who are ineligible for Medicaid by immigration status. 7 For our outcome, enrollment during the open enrollment period, , where i is any subgroup of enrollment (i.e. year, age, sex, etc.), we used the following regression discontinuity equations. The treatment effect of Marketplace eligibility vs. Medicaid eligibility is captured by the coefficient, , expressed as an incident rate ratio and calculated as percent change in enrollment below and above our threshold. was defined by the total persons enrolled in coverage by year and by age, sex, chronic condition status or urban-rural status, in our data. D is a binary indicator for whether a person's income was above the Medicaid threshold of 138% of FPL or not. The Beta coefficients control for the linear (or quadratic and cubic, depending on the model) relationship between income and enrollment. All equations used a generalized linear model with a negative binomial distribution and log link. ℎ was set using the Stata SE 14 MSE−optimal bandwidth selector for the RD treatment effect estimator e(h_mserd), defined by the command rdbwselect using the 2014 enrollment outcome, (Bandwidth = 18). 8, 9 Sensitivity Analyses To confirm our findings with methods that are more typical of regression discontinuity designs, we replicated our models using a linear regression on enrollment in each bin of income (a percentage-point of FPL). We presented the Comparison of Utilization, Costs, and Quality of Medicaid vs Subsidized Private Health Insurance for Low-Income Adults A modification of the Elixhauser comorbidity measures into a point system for hospital death using administrative data By the Numbers: The First Open Enrollment of Connect For Health Colorado By the Numbers: Colorado's Second Open Enrollment Connect for Health Colorado. Marketplace Dashboard Coverage for lawfully present immigrants Optimal bandwidth choice for the regression discontinuity estimator. The Review of economic studies Robust nonparametric confidence intervals for regressiondiscontinuity designs 79 82 85 88 91 94 97 100 103 106 109 112 115 118 121 124 127 130 133 136 139 142 145 148 151 154 157 160 163 166 169 172 175 178 181 184 187 190 193 196 199 Population Income by Federal Poverty Level