Trends in mental health impairment, service use, and expenditures by household income and parental education

Monday, June 13, 2016: 3:00 PM
F55 (Huntsman Hall)

Author(s): Ana M. Progovac; Benjamin Le Cook

Discussant: Nathan Tefft

Childhood use of mental health services has increased over the last 20 years, with the greatest relative increase occurring among youths with the most severe impairment in mental health functioning. This trend is observed despite the fact that the percentage of children with more severe impairment has declined over this time frame. How these trends may differ based on childhood exposures to lower-SES environments (such as household income under federal poverty level or relative low-educational attainment by parents) has not been characterized.

Using data for children ages 6-17 from the Medical Expenditure Panel Survey (MEPS) years 1997-2012, we examined (1) trends in severe childhood impairment, (2) mental health service use and (3) total mental health spending. We stratified these trends based on poverty status, and the educational attainment of youths’ parents. Childhood impairment was assessed in the MEPS using the Columbia Impairment Scale (CIS, range 0-52 where 16 or higher is considered more severe impairment). Poverty status was determined by household income less than or equal to the Federal Poverty Level (FPL). Parents’ education was considered as a binary variable determining whether any of the child’s parents completed college.

We used Logit regression models to assess severe impairment, and 2-part GLM models (log link, gamma distribution) to assess any mental health expenditures and total mental health expenditures (dependent on any expenditures). These models were adjusted for year, age, sex, race/ethnicity (White, Black, Hispanic/Latino), region (NE, MW, S, W), parents’ employment status, and Medicaid Status.

In unadjusted models, trends for significant impairment over time (though declining) were consistently higher for children in poverty than their peers, a gap which appeared to widen. In fully adjusted models, children who had at least one parent completing college had 23% lower odds of severe impairment, yet 25% higher odds of any mental health spending compared to children with no parent completing college. When we adjusted for severity of mental health need (based on continuous CIS), there was an even greater disparity: those with a parent who had completed college had a 40% higher odds of any mental health spending than a youth with no parents completing college. Similar results were found when assessing trends by family income level. For children living in poverty, the odds of severe impairment were 17% higher, yet the odds of any mental health spending were 13% lower (and 24% lower when adjusting for severity of impairment). 

Children who live in poverty and have less educated parents are at an especially high risk of high impairment and no mental health treatment. Our next phase of analysis will assess the role of parents’ employment status, noting that the effects of maternal and paternal unemployment differed in initial analyses, and to study the specific impact of the Great Recession (Dec 2007-June 2009, with peak unemployment in October 2009) on subsequent disparities in impairment and mental health spending.