br Patients that live in nonexpanded states are at risk
Patients that live in nonexpanded states are at risk for falling into a ‘‘coverage gap’’ where their access to Medicaid and marketplace subsidies is simultaneously restricted. These patients have income above the current Medicaid eligibility, but below the lower limit for market-place premium tax credits. Those that fall into this coverage gap are likely to face barriers to needed health services, and if they cck-8 do require care, they are more likely to endure dele-terious financial consequences.
Patients diagnosed with cancer are at increased risk of financial toxicity and those without insurance are most vulnerable (4e7). The purpose of this study was to deter-mine the effect of Medicaid expansion on the insurance sta-tus at diagnosis of brachytherapy patients. This study compared insurance status before and after Medicaid expansion under the ACA on a state level and compared expanded states with nonexpanded states.
Materials and methods
This is a retrospective cohort study that utilized the Sur-veillance, Epidemiology, and End Results (SEER) data-base. The SEER program provides statistical software, SEER*Stat 8.3.4, which includes population-based cancer registries in 14 states and covers 28% of the U.S. popula-tion. The Duke University Medical Center Institutional Re-view Board provided a waiver for this study, given that the SEER program is publicly available and contains deidenti-fied data.
Patients aged 19e64 years with newly diagnosed breast, cervical, uterine, or prostate cancer from January 2011 through December 2014 with known insurance status who were treated with brachytherapy were included in the analysis. Those 18 years and under were excluded due to other programs providing medical insurance to minors in the United States; those aged 65 years and over were excluded due to Medicare eligibility. Brachytherapy utili-zation was identified within SEER as those coded as receiving ‘‘radioactive implants’’ or ‘‘Combination of beam with implants or isotopes.’’ Of note, owing to known ascertainment issues within the SEER registries, while the cohorts that are recorded as having received brachytherapy are highly likely to be correctly coded, there is no corre-sponding ‘‘No Brachytherapy’’ group that can be
established with certainty in the data set (8). Insurance sta-tus at diagnosis was divided into insured, any Medicaid, and uninsured. Insured patients included only those with non-Medicaid insurance options, including private, Medi-care, or military/VA insurance, based on coding for pri-mary payer at diagnosis. Patients were categorized as white, black, or other (including all Asian races, indige-nous populations and patients with ‘‘other’’ race coding). Ethnicity is coded in SEER as either ‘‘Hispanic’’ or ‘‘non-Hispanic.’’
Expanded versus nonexpanded states
For the purposes of this study, of the 14 states included in the SEER database, all states that fully expanded Medicaid in 2014 are designated expanded states (CA, CT, HI, KY, MI, NJ, NM, and WA). All states that did not fully expand Medicaid are designated as nonexpanded states (AK, GA, IA, LA, and UT). Iowa had an alternative expansion including funding for private insurance and a modest pre-mium for qualified Medicaid enrollees at 100e133% of the poverty level (9). For this analysis, Iowa was considered nonexpanded, as they did not fully expand Medicaid as rec-ommended by the ACA until 2015. Figure 1 demonstrates the expansion status of all states and highlights the states included in the SEER database.
The two primary hypotheses of this study were that rates of uninsurance (lack of either ‘‘insured’’ or ‘‘Medicaid’’ status in SEER) for patients receiving brachytherapy were reduced in states that expanded Medicaid, and that the rate of reduction in these states was greater than in nonex-panded states. Two-tailed Fisher’s exact test was used to test for associations of insurance status with expanded versus nonexpanded states with regard to residency, age, race, ethnicity, diagnosis, marital status, and percent of poverty in the patient’s county of residence. A ‘‘difference of differences’’ analysis was then performed by propa-gating the standard deviation estimate of the respective binomial distributions to test if insurance rates changed more in expanded versus nonexpanded states in the year 2014, when Medicaid was selectively expanded. For multi-variate analysis, a binomial logistic regression was per-formed using these same factors with insurance status dichotomized to uninsured versus any insurance. Signifi-cance was assumed if p !0.05. All statistical analyses were performed with SPSS version 21 (IBM, Armonk, NY).