Mental Health Difficulties of Incarcerated Women: Using Latent Class Analysis to Promote Tailored Interventions

Background and Purpose

Research has consistently shown an elevated prevalence of co-occurring mental health difficulties among justice-involved women (Salina et al., 2007, 2011), a population with which social workers regularly interface (Bewley & Morgan, 2011). However, no studies have investigated specific patterns of these co-occurring mental health difficulties. A better understanding of diagnostic patterns among incarcerated women could help clarify incarcerated women’s service needs, tailor treatment approaches, and inform programmatic priorities.  

Methods

To identify patterns of mental health difficulties among incarcerated women, latent class analysis (LCA) was conducted using secondary data from the Bureau of Justice Statistics Survey of Inmates in State Correctional Facilities. The final sample comprised 2553 women from state correctional facilities throughout the United States. Eight variables designating diagnoses of mental health disorders (e.g. depression, anxiety, substance abuse, personality disorders) served as the indicator variables for the LCA; history of attempted suicide was also included as an indicator variable. Analyses were conducted in Mplus 7.1, and model fit was assessed using multiple fit indices (i.e. AIC, BIC, MLRT, BLRT) and the substantive criteria of class proportions and conditional response probabilities. The number of classes was systematically increased until no longer warranted. Class validation procedures included a series of multinomial logistic regression analyses.

Results

Results of the LCA indicated that the 4-class solution was the best-fitting model. The classes were labeled as follows: 1) serious mental illness, 2) mood and drug use disorders, 3) substance use only, and 4) low mental health difficulties. Comprising 8.7% of the sample, the group labeled as the “serious mental illness” subgroup included women with the highest risk of endorsing almost every indicator of mental health difficulties. Another subgroup—labeled the “mood and drug use disorders” subgroup—contained almost one third of the sample; women in this group had high probabilities of endorsing diagnoses of depression, bipolar disorder, and drug use disorder. The “substance use only” subgroup comprised 11.7% of the sample and was characterized by elevated probabilities for endorsement of both alcohol use disorder and drug use disorder. Accounting for almost half of the sample, the final subgroup comprised women who had the lowest probabilities of endorsing various mental health difficulties. Results of the class validation procedures supported the empirical and clinical value of the classes.

Conclusions and Implications

Our results indicate that there is heterogeneity of mental health difficulties among incarcerated women. These findings offer a compelling framework around which correctional facilities might calibrate the provision of mental health services to more efficiently allocate limited resources. Should future research efforts confirm that comparable subgroups exist in other samples of incarcerated women, social workers engaged with that population might begin targeting dual diagnosis treatment toward the one third of incarcerated women that seems especially likely to present with “co-occurring mood and drug use” disorders, for example. Future research should determine the extent to which other psychosocial needs differ between these subgroups of women to identify additional avenues for holistic, tailored intervention approaches.