Abstract: Predictors for Turnover Intention and Actual Turnover Among Community Mental Health Providers: Exploration of Provider Characteristics and Job Factors (Society for Social Work and Research 24th Annual Conference - Reducing Racial and Economic Inequality)

Predictors for Turnover Intention and Actual Turnover Among Community Mental Health Providers: Exploration of Provider Characteristics and Job Factors

Friday, January 17, 2020
Marquis BR Salon 16, ML 2 (Marriott Marquis Washington DC)
* noted as presenting author
Sadaaki Fukui, PhD, Associate Professor, Indiana University, IN
Michelle Salyers, PhD, Professor, Indiana University - Purdue University, Indianapolis, Indianapolis, IN
Background and Purpose: The turnover of mental health providers is a significant problem (30-60% annually), compounded by enduring and severe provider shortages. Turnover can negatively affect providers and agencies, and most critically  ̶  the quality of care clients receive. Identifying factors that contribute to turnover is critical, yet research is scarce. Commonly, turnover intention (i.e., thoughts about leaving a job) is measured as a proxy outcome in research when actual turnover data are not available. This study aims to explore predictors for both turnover intention and actual turnover, which have critical implications for developing turnover prediction and intervention models to prevent excessive turnover.    

Methods: We used data originally collected for a burnout-reduction intervention (Salyers et al., 2019) from 195 direct care providers at two community mental health centers. They were predominantly female (80%) and white (85%). The average age was 40 (±12) years old, and they worked an average of 3.3 (±5) years at the agencies. We excluded involuntary termination cases, thus the total sample size was 186. Forty-eight participants (26%) voluntarily left the agencies within 12 months after the baseline. We used correlations, chi-square tests, and independent samples t-tests to identify predictors for turnover intention and actual turnover. Provider characteristics included gender, race, age, exempt status, educational degree, marital status, having children under age of 5, and work years in the agency and field. Job factors included overtime work (estimated based on scheduled and actual work hours), burnout, work-life conflict, and job satisfaction. We obtained these measures and turnover intention from the participants at baseline, then collected actual turnover data from their agencies 12 months later.

Results: The study found that those who had higher turnover intentions were more likely to experience higher work-life conflict and burnout, along with lower job satisfaction (p<.001). Females tended to report higher turnover intentions, with marginal significance (p=.07). On the other hand, those who voluntarily left the agencies after 12 months were more likely to have children under age of 5 (p=.01), be younger (p=.03), work fewer years in the field (p=.04), and hold a bachelor’s degree or less (p=.05). Those who worked overtime tended to remain in the agencies (p=.08). Turnover intention and actual turnover were significantly correlated (p<.001). Exempt status, race, marital status, and work years in the agency were not significant predictors for turnover intention or actual turnover.

Conclusions and Implications: Turnover intentions and actual turnover are highly correlated outcomes, yet the predicting factors are different. Specifically, turnover intentions may be more influenced by job factors (e.g., work-life conflict, burnout, job satisfaction), while provider characteristics may be more influential on actual turnover (possibly related to domestic duties/expectations with young children). Further, those who are younger, at earlier career stages, and with minimal degrees, may be more open to obtaining outside opportunities. To draw more rigorous conclusions, building prediction models by controlling for variable associations is necessary. Given both turnover intention and actual turnover are important outcomes (Fukui et al, 2019), further examination is needed.