A Longitudinal Examination of Behavioral Health Service Need and Treatment Utilization Among Poor Women in the U.S

Background/Purpose: Over the last decade, the poverty rate among women has increased.  At the same time, generosity of welfare benefits and availability of living wage jobs have declined.  These trends are of concern given evidence that low socioeconomic status is positively associated with psychological distress.  Moreover, if appropriate interventions and treatments are not utilized, longer-term negative psychiatric and economic sequelae are likely.  This research examines trends in the prevalence of behavioral health problems (BHP) and rates of treatment utilization (TU) among poor women in the U.S.  We hypothesize that the rate of behavioral health problems has increased over time, as has treatment utilization among those with a problem.   We also investigate the potentially modifiable correlates of behavioral health problems, as well as utilization, in order to offer policy recommendations that could improve the well-being of poor women.

Methods: Trend analyses were conducted using 10 years of pooled cross-sectional data (2004-2013) from the nationally representative National Survey on Drug Use and Health. Analyses were limited to women aged 18-64 with household incomes < 200% of the poverty threshold (N=84,111).  Dichotomous dependent variables were created from self-report data: BHP is based on responses to questions about past year symptoms of depression, and alcohol or illicit substance abuse or dependence; TU is based on responses to questions about past year use of any professional services (inpatient, residential, outpatient). Rates were compared across all study years with a chi-square test for linear trend.  Logistic regression predicted BHP among all low-income women, and TU among low-income women with a BHP. Other predictors included welfare receipt (e.g., TANF, SNAP), employment status (employed, unemployed, not in labor force), and health insurance (only for TU; public, private, none).  Demographic controls were age, race-ethnicity, number of children, educational attainment, and separate dummy variables for survey year.  Interactions between year dummies and all variables except demographics were added in a separate step to investigate temporal changes in the effect of policy-related variables. To estimate statistical significance, conservative 99% confidence intervals were generated because the number of observations was so large.  All analyses were weighted and adjusted for sample design using Stata.

Results: Neither hypothesis was supported.  Each year, approximately one-quarter of the women reported a BHP, and slightly less than one fifth of women with a BHP received services.  For both outcomes, multivariate results were similar.  The odds of any BHP were significantly greater for women who were unemployed [OR 1.21] and not in the labor force [OR 1.20] compared to employed women. Welfare benefits increased the odds of BHP and TU [OR 1.42, 1.34, respectively].  Interactions between year dummies, employment and welfare did not improve model fit for either BHP or TU.

Conclusions/Implications:  In the U.S., rates of TU among economically marginalized women who have BHP are stagnant.  Initiatives to better link welfare and behavioral health systems may be needed, as is research that tracks how well the reforms of the Affordable Care Act increase access to behavioral health services for poor women.