Predicting mental health provider response to BREATHE, a burnout intervention program

Indiana University-Purdue University Indianapolis (IUPUI)Within the mental health field, provider burnout is widespread and associated with far-reaching negative outcomes for providers, consumers, and organizations. Over the past four decades, various burnout interventions have been tested and found to be minimally effective, leading several researchers to suggest an increased focus on targeted recruitment (i.e., targeting providers who are most likely to benefit from a particular burnout intervention approach) and/or modifications to the interventions (e.g., format and content). Accordingly, the present study examined several person-related and intervention-related variables that were hypothesized to be predictive of response to BREATHE, a burnout intervention for mental health providers. Data from four prior studies that assessed the effectiveness of the BREATHE intervention were amalgamated. For the primary analyses, hierarchical linear regression was used to determine whether the person-related and/or intervention-related variables were predictive of treatment response. Additionally, the BREATHE studies were examined to determine whether the intervention became less effective at reducing burnout with each subsequent iteration. With respect to person-related predictors of response to the BREATHE intervention, age and turnover intentions were significant. Specifically, younger participants had higher post-intervention levels of depersonalization than older participants (β = -.13, p = .023), and higher baseline intentions to turnover were associated with greater post-intervention levels of emotional exhaustion (β = .11 p = .041) and depersonalization (β = .12, p = .023). In terms of intervention-related predictors of treatment response, participants who received the BREATHE intervention in a multi-session format had higher post-treatment levels of emotional exhaustion than those who received the BREATHE intervention in a single session format (β = .13, p = .015). Notably, across these primary analyses, baseline levels of burnout (i.e., emotional exhaustion, depersonalization, and personal accomplishment) were consistently the strongest predictors of post-intervention levels of burnout. Lastly, the data suggests that the BREATHE intervention became less effective with subsequent iterations. For example, earlier BREATHE studies had larger effect sizes than more recent studies. Additionally, there was a significant difference between the studies with respect to the change in emotional exhaustion (F(3, 230) = 4.86, p = .001, η2 = .06), such that participants in the first BREATHE study had a significantly larger reduction in emotional exhaustion than participants in the three subsequent studies. The present study was the first to examine potential predictors of response to the BREATHE intervention. Although the hypotheses were not supported, the implications of these findings are discussed and suggestions for future research directions are provided