Predictors of Depression in a Low-Income, Uninsured, Urban Primary Care Clinic Popluation

Background and Purpose: Major depression annually affects 14.8 million American adults (6.7% of the U.S. adult population) (NIMH, 2008). Another 1.5% (3 million) have symptoms of dysthymic disorder. Depression is also associated with chronic illnesses (Long, 1996) and reduced work attendance and performance (Stewart, Ricci, Hahn, & Morganstein, 2003), in addition to affecting life quality, well-being, and coping. Most epidemiologic studies showing a high prevalence of depression and low identification and treatment in primary care settings have been conducted in private offices. The purpose of this study was to establish the prevalence of depression in a low-income, urban, uninsured primary care clinic population, identify rates of co-morbid conditions, and identify supports and barriers to effective treatment.

Methods: A cross-sectional design was used in the study of 1,793 patients enrolling between April 7, 2007 and April 29, 2008 for free or reduced fee primary care at three clinics in a major Midwest metropolitan community. Eligibility was based on: 18-64 years-of-age, no health insurance or eligibility, household income below 250 per cent of federal poverty guidelines, and living in the community served or in two designated adjacent communities. Sample demographics included: marital status: 69.5% single, 10.8% married or common-law, 17.1% divorced/separated, 2.6% widowed; race: 88.6% African American, 7.7% White; 3.7% other races/ethnicities; gender: 51.5% female; 71.9% unemployed; mean age: 42.88 years; mean monthly household income: $479; mean monthly per person household income: $242. Data were collected from patient reports on demographic and health risk appraisal questionnaires that included the Patient Health Questionnaire (PHQ)-2 (measuring depressed mood and anhedonia). Descriptive analyses (frequencies and percentages), chi square tests, and independent samples t-tests were conducted. Prior to the multivariate logistic regression analysis, potential independent variables containing 20% missing data or whose bivariate relations with the dependent variable (positive screen for depression) exceeded p = .25 or that were strongly collinear with another, conceptually related variable were dropped from the multivariate analysis. Remaining variables were entered hierarchically. All 2- and 3-way interactions among independent variables that made a significant improvement to model fit were retained.

Results: Twenty-six percent screened positive for depression. A positive depression screen was predicted by being treated at two of the three participating clinics, being denied Medicaid, having missed two or more medical appointments in the past 6 months, and being a cigarette smoker, while married patients were less likely to screen positive. Living alone increased the odds of screening positive among married patients and those who were single or widowed but not those separated or divorced. Employment was negatively related to depression for individuals who lived alone but unrelated to depression for those who lived with others.

Conclusions and Implications: Findings imply a need to (1) offer depression screening to low-income, uninsured primary care patients; (2) offer services that address depression concurrently with physical health care; and (3) include attention to the effects of systemic barriers to care, social support availability, and co-morbidity (nicotine dependence). Potential supports and barriers to such policies and services will be discussed.