Prevalence of Mental Health and Substance Use Disorders Among Homeless and Low-Income Housed Mothers

Sampling and data collection procedures have previously been described (11, 12). We employed an unmatched case-control design. We enrolled homeless single mothers (N=220) from all nine emergency shelters in Worcester, Mass., and a comparison group of single mothers (N=216) receiving public assistance (Aid to Families With Dependent Children) who had never been homeless. The age, income, and number of children of our combined group were similar to the data for low-income women in other U.S. cities. However, reflecting Worcester’s population, our group had more Puerto Rican and fewer African American women than are found in many other midsize cities.

We obtained written informed consent from all mothers in our group. We collected data over three to four sessions (6–7 hours). The SCID-NP (10) assessed mood, anxiety, somatization, and substance use disorders according to DSM-III-R criteria. We also asked questions from the SCID-NP that screen for psychosis. When the results were positive, the patient version of the SCID determined the specific disorder. The interviewers, extensively trained in the administration of the SCID-NP, interviewed both subgroups.

We compared, to the extent possible, the prevalence rates of disorders in our sample with comparable DSM-III-R rates for all women in the National Comorbidity Survey (9). The National Comorbidity Survey interviewed a stratified, multistage sample of 8,098 individuals aged 15–54 years but excluded homeless persons. The National Comorbidity Survey used a modified version of the Composite International Diagnostic Interview, which is designed to be used by trained interviewers who are not clinicians. Overall reported prevalence rates from the National Comorbidity Survey were weighted to adjust for household size, nonresponse, and poststratification. Various diagnoses (e.g., certain substance use disorders, posttraumatic stress disorder [PTSD]) made in the National Comorbidity Survey were not available for secondary analyses, which limited our ability to make comparisons for these particular disorders as well as for summary diagnoses that spanned diagnostic categories. We limited the National Comorbidity Survey subjects to women aged 15–40 years and achieved approximate comparability between the two study groups at each quartile of their respective age distributions. We analyzed data by using t tests for continuous variables and chi-square tests for discrete variables. Statistical significance was tested with alpha set at 0.05 (two-tailed).

This is the first study to date that documents the prevalence (lifetime and 1 month) of psychiatric and substance use disorders among homeless and low-income housed mothers and compares it to the prevalence of such disorders among women in the general population. More than two-thirds of the mothers in our study had a lifetime diagnosis of at least one axis I disorder, and almost half had two or more lifetime disorders. Contrary to reports citing mental disorders as a cause of homelessness, DSM-III-R diagnoses were not associated with current housing status among this group of low-income mothers with dependent children.

Unlike solitary homeless women, homeless mothers do not suffer disproportionately from schizophrenia (7). However, lifetime rates of PTSD, major depression, and substance use disorders are overrepresented when compared with rates for the National Comorbidity Survey. With regard to lifetime history of mental disorder, approximately two-thirds of those with diagnoses had two or more lifetime disorders. The rate of two or more current disorders was far lower, with substance abuse most likely to be one of these current (and lifetime) conditions.

Violence has been recognized as a subtext of family homelessness (11, 13). A high percentage (83%) of low-income women in Worcester had been physically or sexually assaulted during their life span (10). Thus, it is not surprising that PTSD, substance use disorders, and major depression were disproportionately represented with lifetime rates of PTSD three times greater than in the general female population.

Various limitations must be considered. This study was conducted in one locale and can best be generalized to midsize American cities. We did not determine axis II diagnoses. Because of resource limitations, we did not assess interrater reliability among interviewers. Recall bias also is an issue in conducting lifetime assessments, but is almost always in the direction of underreporting. Current substance use may have been underdiagnosed since Worcester’s shelters do not accept women who are active users. While our study and the National Comorbidity Survey used identical diagnostic criteria, different assessment instruments were used in the two studies; the difference in instruments made comparison of results slightly more problematic than if the same diagnostic device had been used. Finally, not all National Comorbidity Survey variables were available to us for secondary analyses, limiting our ability to make comparisons for certain diagnoses and across diagnostic conditions.

Our findings support Belle’s argument that psychiatric disorders are more common among women in lower socioeconomic groups, largely because of the multiple stressors associated with poverty (14). Pervasive violence in the context of poverty may account for many of the emotional disorders in our group, particularly the disproportionately high rates of PTSD. Only by acknowledging the bleak reality of poor women’s lives, especially the high rates of traumatization, can we improve their mental health and well-being.

Received Dec. 16, 1996; revisions received Nov. 10, 1997; and Feb. 17, 1998; accepted April 24, 1998. From the Better Homes Fund, the Department of Psychiatry, Harvard Medical School; and the Department of Epidemiology, Harvard School of Public Health. Address reprint requests to Dr. Bassuk, The Better Homes Fund, 181 Wells Ave., Newton Centre, MA 02159. Supported by NIMH grants MH-47312 and MH-51479 and Maternal and Child Health Bureau grant MCJ250809. The National Comorbidity Survey was funded by grants from NIMH (MH/DA-46376 and MH-49098), the National Institute on Drug Addiction (supplement to MH/DA-46376), and the W.T. Grant Foundation (90135190). (Information on the availability of National Comorbidity Survey data to outside researchers for secondary analyses can be found on the Internet [http://www.hcp.med.harvard.edu/ncs/].)The authors acknowledge the contributions of the following persons in the conduct of this research: Carolyn Banach, M.A., Jocelyne Bauduy, M.A., Margaret Brooks, B.A., Angela Browne, Ph.D., Linda Cohan, M.S.W., Mardia Coleman, Ann Dalianis, M.A., Veronica Guerrero-Macia, M.A., Barbra Page, M.A., Nancy Popp, Ed.D., Laurie Ross, M.A., Amy Salomon, Ph.D., Julia Vera, Linda Wein­reb, M.D., Constance Wood, M.S.W., and Dorothy St. Cyr, B.A.