The Affordable Care Act (ACA) was designed to increase access to health insurance by: 1) requiring states to expand Medicaid eligibility to people with incomes less than 138 percent of the Federal Poverty Level (FPL) ($19,530 for a family of three in 2013), with the cost of expanded eligibility mostly paid by the federal government; 2) establishing online insurance “exchanges” with regulated benefit structures where people can comparison shop for insurance plans; and 3) requiring most uninsured people with incomes above 138 percent FPL to purchase insurance or face financial penalties, while providing premium subsidies for those up to 400 percent of FPL.
Recent studies suggest that Medicaid expansion will result in health and financial gains. Older studies also found salutary health effects of expanded or improved insurance coverage, particularly for lower income adults. These studies also document an increase in utilization of most health care services. Most recently, the Oregon Health Insurance Experiment (OHIE) found a striking increase in emergency department use as well as other outpatient care.
The Supreme Court ruled in June 2012 that states may opt out of Medicaid expansion, and as of November 2013, 25 states have done so. These opt-out decisions will leave millions uninsured who would have otherwise been covered by Medicaid, but the health and financial impacts have not been quantified.
In this post, we estimate the number and demographic characteristics of people likely to remain uninsured as a result of states’ opting out of Medicaid expansion. Applying these figures to estimates of the effects of insurance expansion from prior studies, we calculate the likely health and financial impacts of states’ opt-out decisions.
The Consequences of Opting Out
The Supreme Court’s decision to allow states to opt out of Medicaid expansion will have adverse health and financial consequences. Based on recent data from the Oregon Health Insurance Experiment, we predict that many low-income women will forego recommended breast and cervical cancer screening; diabetics will forego medications, and all low-income adults will face a greater likelihood of depression, catastrophic medical expenses, and death. Disparities in access to care based on state of residence will increase. Because the federal government will pay 100 percent of increased costs associated with Medicaid expansion for the first three years (and 90 percent thereafter), opt-out states are also turning down billions of dollars of potential revenue, which might strengthen their local economy.
The ACA’s tax subsidy for insurance purchase on the Exchanges is only available to persons with incomes above 100 percent of FPL. People below this threshold in opt-out states (the so-called low-income “coverage gap”) will see no benefit as the law goes into effect. They may even see harm because the ACA cuts disproportionate share (DSH) funding to safety net hospitals, reducing the resources available to care for the remaining uninsured.
Despite the widely held belief that almost all Americans will be insured under the ACA, more than 32 million people will remain uninsured after the law goes into effect. Even in states that opt in to Medicaid expansion, millions will remain without coverage.
Low-income adults in states that have opted out of Medicaid expansion will forego gains in access to care, financial well-being, physical and mental health, and longevity that would be expected with expanded Medicaid coverage.
Examining the numbers. The number of uninsured people in states opting in and opting out of Medicaid expansion is displayed in Exhibit 1. Nationwide, 47,950,687 people were uninsured in 2012; the number of uninsured is expected to decrease by about 16 million after implementation of the ACA, leaving 32,202,633 uninsured. Nearly 8 million of these remaining uninsured would have gotten coverage had their state opted in. States opting in to Medicaid expansion will experience a decrease of 48.9 percent in their uninsured population versus an 18.1 percent decrease in opt-out states.
Exhibit 1: Uninsured Population by State, Pre- and Post-ACA
Predicted national-level consequences of states opting out of Medicaid expansion are displayed in Exhibit 2. We estimate the number of deaths attributable to the lack of Medicaid expansion in opt-out states at between 7,115 and 17,104. Medicaid expansion in opt-out states would have resulted in 712,037 fewer persons screening positive for depression and 240,700 fewer individuals suffering catastrophic medical expenditures. Medicaid expansion in these states would have resulted in 422,553 more diabetics receiving medication for their illness, 195,492 more mammograms among women age 50-64 years and 443,677 more pap smears among women age 21-64. Expansion would have resulted in an additional 658,888 women in need of mammograms gaining insurance, as well as 3.1 million women who should receive regular pap smears.
Exhibit 2: Effects of Medicaid Expansion on Health and Financial Outcomes, and National Estimates of Adverse Outcomes Avoided or Appropriate Screening/ Treatment Provided If Current Opt-out States Accepted Medicaid Expansion (for an enlarged view, click on the chart below)
State-level estimates for post-ACA effects of opting out of Medicaid expansion are displayed in Exhibit 3. In Texas, the largest state opting out of Medicaid expansion, 2,013,025 people who would otherwise have been insured will remain uninsured due to the opt-out decision. We estimate that Medicaid expansion in that state would have resulted in 184,192 fewer depression diagnoses, 62,610 fewer individuals suffering catastrophic medical expenditures, and between 1,840 and 3,035 fewer deaths.
Exhibit 3: State Estimates of Adverse Outcomes Avoided or Appropriate Screening /Treatment Provided If Current Opt-out States Accepted Medicaid Expansion (for an enlarged view, click on the chart below)
We categorized states as opting in or opting out of Medicaid expansion using the Kaiser Family Foundation’s “Status of State Action on the Medicaid Expansion Decision,” which was updated on November 22, 2013. We used the Census Bureau’s 2013 Current Population Survey, a nationally representative survey of the non-institutionalized US population, to determine the number of uninsured people in each state before implementation of the ACA. We then projected the number of uninsured people in each state after implementation of the ACA depending on whether the state is opting in or opting out of Medicaid expansion. Based on previously published estimates of take-up rates and estimates from the Congressional Budget Office, we assumed that in states opting out, 90 percent of currently uninsured people with incomes below 138 percent of FPL will remain uninsured, as will 75 percent of uninsured people with incomes above 138 percent FPL. In states opting in, we assume that 40 percent of currently uninsured people with incomes below 138 percent FPL will remain uninsured, as will 60 percent of uninsured people with incomes above 138 percent FPL. These estimates incorporate the assumption that enrollment of people with incomes above 138 percent FPL through the exchanges will be higher in states that opt to expand Medicaid.
We used data from three sources to estimate the effects of Medicaid expansion: The Oregon Health Insurance Experiment; and two widely cited estimates of the impact of coverage expansion on mortality. The OHIE is a randomized study that examined the effects of expanding public health insurance for low-income (less than 100 percent FPL) adults on health, financial strain, health care use, and self-reported well-being. It found that after an average of 17 months of exposure to Medicaid coverage, improvements occurred in rates of depression (based on the eight-question version of the Patient Health Questionnaire (PHQ-8)), and catastrophic medical expenditures. In addition, the OHIE found that acquisition of coverage led to increased utilization of most types of health care, including several types of care that has been linked to improved outcomes such as diabetics receiving medication to treat their diabetes and clinically indicated mammograms and cervical pap smears (in the past 12 months). An estimate of the number needed to insure was calculated by dividing the number of newly insured persons by the number of outcomes achieved.
To estimate the effect of Medicaid expansion on catastrophic medical expenditures (i.e. medical expenditures greater than 30 percent of annual income), we used the observed effect size from OHIE for adults up to 100 percent FPL. In order to extrapolate this financial impact finding from the OHIE to near-poor and middle income persons, we assumed that the effect size of Medicaid expansion among adults between 100 percent and 138 percent FPL would be only half as large, and among adults between 138 percent and 400 percent FPL, only one quarter as large as the effect size observed in the OHIE. To estimate the number of women eligible for cervical cancer screening and mammography, we used the age ranges for screening suggested by national consensus guidelines (21 to 64 years for pap smears and 50 to 64 years for mammograms), and applied the increase in pap smear and mammogram rates observed in the OHIE.
We estimated the range of likely mortality effects of Medicaid expansion. For our high estimate, we used the recent study by Sommers and colleagues that compared trends in mortality rates in states with Medicaid expansions (New York, Maine, and Arizona) to trends in states without such expansions. The Medicaid expansions were associated with a 6.1 percent decrease in mortality, or 19.6 deaths per 100,000 non-elderly adults. We conservatively used this population-based estimate, rather than their number-needed-to-insure figure of 176, because, as Sommer et al. pointed out, the latter figure reflects the fact that in their study, Medicaid preferentially enrolled sicker than average adults. For our low estimate, we used a study based on mortality follow-up of participants in the National Health and Nutrition Examination Study, which found a 40 percent increase in death rates among the uninsured, an effect size approximately 42 percent that found by Sommers.
Several caveats apply to our findings. Our figures, which use the number of uninsured in 2012 as the baseline, differ slightly from Congressional Budget Office figures based on projections of the numbers who would have been uninsured in several future years had the ACA not been passed. We could not take into account several factors that might influence the impact of Medicaid expansion. For instance, both the OHIE and Sommers estimates are based on Medicaid expansions that paid doctors pre-ACA reimbursement rates. Since the ACA will provide a two-year increase in Medicaid rates for primary care services, it is possible that access to care will improve more than was observed in those studies if more providers start accepting Medicaid. In addition, Oregon’s health costs (and presumably its rates of catastrophic medical expenditures) are slightly lower than national average.
The patients studied in the OHIE were slightly older than the uninsured poor in opt-out states, and more often female. While we were able to adjust for these demographic differences in estimating cancer screening rates, it was not possible to do so for other effects. Similarly, we did not attempt adjustment for regional differences in depression prevalence, in the uninsured population, although such differences are probably small. If anything, the adjusted prevalence of major depression in Oregon appears slightly below the national average. An older sample population in the OHIE may have resulted in greater improvements in health and screening following Medicaid expansion, leading to a slight overestimate of effects in states with a younger uninsured population, whereas the female predominance in the OHIE may have resulted in a slight underestimate of effects in other states because males are more likely to have diabetes and other chronic conditions. In the OHIE, a relatively small number of persons were covered by the Medicaid expansion. The broader expansion under the ACA may put greater strain on the limited capacity of providers who accept Medicaid patients, curtailing utilization. Finally, participants in the OHIE had been uninsured for at least six months, and were concentrated in the Portland area. Impacts elsewhere might differ.
We used data from the Sommers and Wilper studies to calculate mortality impacts because the OHIE was underpowered to detect changes in death rates. Although small improvements in hypertension prevalence (-1.3 percent) and Framingham risk score (-0.2 points) were observed in the OHIE, these did not achieve statistical significance.