Mental Health Navigation for Depression in a Community Cancer Clinic: Predictors of Navigation Engagement and Response

Guideline-driven detection and treatment of depression is a priority; however, little evidence exists to show that these promising guidelines are effective. To address this problem, the UT Southwestern Moncrief Cancer Institute implemented universal depression screening plus mental health navigation (i.e., brief mental health assessment and treatment recommendations via phone with ongoing, periodic calls for symptom monitoring and recommendations). This project evaluated a set of a priori defined sociodemographic, depression, navigation, and treatment variables using the Least Absolute Selection and Shrinkage Operator to create predictive models of characteristics hypothesized to be associated with navigation engagement, depression remission, and ≥50% symptom reduction. Retrospective, longitudinal medical record data were compiled. Of 991 adult patients screening positive for depression (PHQ-9≥5), 21% completed a navigation call. Patients (N=207) were mostly middle-aged (50±10.98), female (88%), racially diverse (39% Hispanic white, 26% black), and English speakers (76%). Most were unemployed (57%), uninsured (48%), and were cancer survivors (60%). Most patients (n=125; 60%) completed 2+ navigation calls. Being employed was associated with a lower likelihood of engaging in multiple mental health navigation calls (odds ratio=0.71). Few patients (n=24; 19%) reached depression remission (PHQ-9<5), while many (n=51; 41%) achieved ≥50% symptom reduction. Sociodemographic characteristics were influential in predicting remission and symptom reduction (odds ratios=0.65-1.57). As hypothesized, suicidal ideation reduced the odds of remission by 35%. Contrary to hypotheses, the odds of achieving remission or reduction in symptoms were reduced by 11% and 7%, respectively, as navigation calls exceeded 2 calls. Furthermore, clinical characteristics like depression treatment utilization and baseline depression severity were not predictive of engagement in navigation, depression remission, or symptom reduction. Predictors for remission and symptom reduction differed, indicating a need to examine these outcomes separately. These data suggest the need to re-evaluate guidelines for these distinct goals using real-world data and robust statistical techniques.