Methods: Data for this study were collected from a Hospital Emergency Department in the South Bronx from August 2001 to April 2003. A convenience sample of women (N=241) were recruited during randomly selected 6 h time blocks. Eligibility criteria for this study included being a female aged 18 years and older and being admitted to the ED during a time block selected for inclusion in the study. Bivariate associations between homelessness and risk factors (depression, intimate partner violence, and prior incarceration) were analyzed. A multivariate logistic regression model with homelessness as the dependent variable was tested for depressive symptoms, prior incarceration, and intimate partner violence, controlling for personal demographic characteristics, social and economic marginalization, and personal, family and relationship instability. Odds ratios (OR) with their 95% confidence intervals were calculated using SAS 9.3 statistical software.
Results: Fifty-three percent (n=121) of women reported having been homeless at least once in their lifetimes. Bivariate analyses indicated that higher rates of depressive symptoms and a history of prior incarceration were individually associated with homelessness at the .05 level in this sample of low-income women. Intimate partner violence was not significantly associated with homelessness in bivariate analyses. In the multivariate model, depression was significantly associated with homelessness (OR = 1.6; p=0.0087) while prior incarceration was associated with an even higher odds of homelessness (OR = 2.9; p=0.0051). Intimate partner violence was also not associated with homelessness in the multivariate model.
Conclusions and Implications: Findings underscore the need for homeless prevention efforts that target sub-groups of poor women, especially those with a history of depression and/or incarceration. Although IPV was not significantly associated with homelessness, future research should consider other forms of violence (i.e. sexual violence) and potential interaction effects between depression and violence. The study also emphasizes the need to develop evidenced-based interventions that can be utilized in and tailored for the various settings (i.e. shelters, emergency departments, criminal justice), inner-city and otherwise, in which homeless women present. Future research, particularly using longitudinal data, that explores possible synergistic impact of these co-morbid risk factors on homelessness outcomes will be important to further the initial findings from this study.