We analyzed data from the 2008–2016 National Survey of Drug Use and Health,7 a nationally representative cross-sectional survey that measures the prevalence of mental health and substance use in the USA.7 Our study population was restricted to non-elderly adults aged 18–64. Individuals on probation were identified as such if they reported being on probation in the past year.
Our primary outcome variable was insurance status. An individual was considered to be insured if they reported any private insurance, Medicare, Medicaid/Children’s Health Insurance Program, military-related insurance, or other health insurance. Individuals with any private insurance were labeled private, those with Medicaid but not private insurance were labeled Medicaid, and those with any other insurance were labeled other.
We estimated insurance coverage during each year of our study for individuals with and without probation using logistic regression and predictive margins. We used logistic regression models and predictive margins, controlling for age, sex, education level, race/ethnicity, income, marital status, and employment status to obtain a difference-in-differences (DID) estimate of the ACA’s impact among individuals with probation in the past year, relative to the general population, from 2008–2013 to 2015–2016. We excluded data from 2014 as a transition period due to implementation of the marketplaces and Medicaid expansion that year.
Unadjusted frequencies were used to describe health insurance categories for individuals on probation from 2013 to 2016. Significance testing was conducted using logistic regression.
Analyses were conducted with Stata/MP, version 15.1 (Stata Corp), and accounted for the NSDUH’s complex survey design.