Abstract: Predictors of Treatment Retention Among Persons with Serious Mental Illness Enrolled in an Integrated Primary and Behavioral Healthcare Program (Society for Social Work and Research 22nd Annual Conference - Achieving Equal Opportunity, Equity, and Justice)

628P Predictors of Treatment Retention Among Persons with Serious Mental Illness Enrolled in an Integrated Primary and Behavioral Healthcare Program

Schedule:
Sunday, January 14, 2018
Marquis BR Salon 6 (ML 2) (Marriott Marquis Washington DC)
* noted as presenting author
Catherine Lemieux, PhD, Professor, Louisiana State University at Baton Rouge, Baton Rouge, LA
Amber Hebert, MSW, Doctoral Student, Louisiana State University at Baton Rouge, Baton Rouge, LA
Chrisann Newransky, PhD, Assistant Professor, Adelphi University, Garden City, NY
Hebah Khalifa, MSW, Doctoral Student, Adelphi University, New York, NY
Katherine Thomas, BS, MSW Student, Louisiana State University at Baton Rouge, Baton Rouge, LA
Purpose: Persons with serious mental illness (SMI) are disproportionately affected by largely preventable chronic cardiometabolic conditions. Community mental health (CMH) centers have aimed to reduce longstanding health disparities by providing whole-person care through a variety of integrated, primary and behavioral healthcare (PBHC) arrangements. For persons with SMI, retention in care is associated with improved overall health and psychosocial outcomes. In CMH settings, treatment retention has been associated with demographic, clinical, and psychosocial characteristics. However, the treatment experiences of persons in integrated health programs remain largely unexamined.  This study examines predictors of retention among clients with SMI receiving integrated PBHC services.

Methods: Utilizing a health data set, this study employed multivariate binary logistic regression (LR) analysis to examine predictors of retention at 6 months (outcome of interest), using a sample of clients with SMI (N=450) enrolled in an integrated PBHC program. A review of relevant research yielded several categories of predictors, and bivariate statistics were computed to identify correlates of retention for inclusion in the LR model. The following potential predictors were assessed. Sociodemographic characteristics included age, race (1=African American, 0=White), gender (0=male, 1=female), employment status (0=unemployed, 1=employed), educational attainment (0=<12, 1≥12), and access to transportation, having a primary healthcare provider (PHP, 0=no, 1=yes). Health-related characteristics included type of mental disorder (0=thought, 1=mood) and number of medications, cardiometabolic conditions, and at-risk health indicators. Health-risk behaviors included tobacco, alcohol, and substance use (0=no, 1=yes). Overall health was assessed with one general self-rated health question, “How would you rate your overall health right now?” (1=poor, 5=excellent). Psychosocial characteristics included measures of psychological distress (K6), daily functioning (8-item Perception of Functioning), presence of PTSD-like symptoms (4-item PTSD Symptom), and social support (4-item Perception of Social Connectedness).  Retention at 6 months, was dichotomized (0=no, 1=yes). The following predictors were significant and therefore included in the final LR model: number of medications, overall health, access to transportation, having a PHP, type of disorder, and use of illicit drugs.

Results: The sample was primarily female (n=255; 56.0%) and African American (n=322; 72.4%). Less than half was retained in treatment at 6 months (n=195, 43.3%).  The binary LR indicated that the overall model correctly classified 65.0% of cases and it was statistically reliable in distinguishing between clients who were and were not retained in treatment at 6 months (-2 Log Likelihood=340.940;χ2(6)=35.36, p<.000). Relevant predictors accounted for approximately 16% of the variance in retention (Nagelkerke R2=.16, p <.000). Among predictors, type of disorder (Wald statistic=6.86, p<.0001) and number of medications (Wald statistic=13.31, p<.000) significantly predicted six-month retention. Clients with thought disorders were twice as likely as those with mood disorders to be retained (OR=2.03). As the number of medications increased by one, the odds of remaining in care increased by 1.21.

Implications: Efforts are needed to improve retention overall and in those with mood disorders. Clients with thought disorders may remain in care to obtain medications that lessen symptomology. Future multivariate research should include other relevant correlates (e.g., treatment satisfaction) that may predict retention in PBHC programs.