Symptoms of depression among urban, public middle and high school students: A Classification and Regression Tree analysis

Depressive symptoms in early adolescence are associated with subsequent onset of health risk behaviors and negative school outcomes, including dropout. We used classification and regression tree (CART) analysis to identify cross-sectional relationships between depressive symptoms and demographic, family, neighborhood, school, and health variables, and risk behavior. 3,654 6th-12th grade students (81% Latino) completed an in-class survey in 2009. We used a CART analysis to explore the data structure and visually represent decision rules for predicting depressive symptoms (yes/no), based on a 4-item index derived from the CES-D. Variables related to depression were included as covariates in a multivariable logistic regression analysis. 21% of respondents scored four or higher, indicating symptoms of depression. Family bond (“FB” or perceived caring from parents) was the most important correlate of depression, with gender second. Being teased or bullied at school, perceived health, and neighborhood satisfaction also emerged as important correlates. Among girls who reported low FB, fair or poor health, and being bullied, 74% reported depressive symptoms; as did 62% of males with low FB who reported bullying. Ethnicity, age, onset of sexual behavior, and parental communication and monitoring were not found to be associated with depressive symptoms. CART analysis identified several subgroups among whom more than half had symptoms of depression. Some identified factors (e.g., school bullying) may be more amenable to intervention than others (e.g., FB). Adolescents should be screened for mental health issues in school and referred to services to improve a wide range of health outcomes.