Addressing perinatal mood and anxiety disorders in obstetric settings: results of a cluster randomized controlled trial of two approaches

Background

Mood and anxiety disorders affect one in 5 perinatal individuals and are undertreated. While professional organizations and policy makers recommend that obstetric practices screen for, assess and treat mood and anxiety disorders, multi-level barriers to doing so exist. To help obstetric practices implement the recommended standard of care, we developed implementation assistance, an approach to guide practices on how to integrate screening, assessment, and treatment of mood and anxiety disorders into the obstetric practice workflow. To teach obstetric care clinicians how to treat perinatal mood and anxiety disorders, we also developed an e-learning course and toolkit.

Objective

Evaluate the extent to which 1) implementation assistance + e-learning/toolkit, and 2) e-learning/toolkit alone improved the rates and quality of care for perinatal mood and anxiety disorders in obstetric practices, as compared to usual care.

Study Design

We conducted a cluster randomized controlled trial involving 13 obstetric practices across the United States (US). Using 2:2:1 randomization, 13 obstetric practices were assigned to 1) implementation assistance + e-learning/toolkit (n=5), 2) e-learning/toolkit alone (n=5), or 3) usual care (n=3). We measured obstetric care clinicians’ quality of care for perinatal mood and anxiety disorders (as measured by medical record documentation of screening, assessment, treatment initiation, and monitoring) documented in patient charts (n=1040). Effectiveness was assessed using multilevel generalized linear mixed models, accounting for clustering of repeated measurements (n=2, i.e., pre and post) within obstetric care clinicians’ patient charts (n=40) nested within practices (n=13). Intention-to-treat and per-protocol analyses were conducted.

Results

At baseline, no significant differences were observed among the 3 groups regarding documented mental health screening. Chart abstraction at 8 months post-training revealed a significant increase in recommended bipolar disorder screening only among the practices that received the implementation plus e-learning/toolkit (from 0.0% to 30.0%; p=.017). Practices receiving the e-learning/toolkit alone or usual care continued to not screen for bipolar disorder. Documented screening for anxiety also increased in the implementation + e-learning/toolkit group (from 0.5% to 40.2%), however, it did not reach statistical significance when compared to the other groups (P=.09). A significant increase in documented post-traumatic stress disorder (PTSD) screening was observed among practices receiving the implementation plus e-learning/toolkit (0.0% to 30.0%; P=.018). The quality-of-care score in the implementation + e-learning toolkit group increased from 20.5 at baseline to 42.8 at follow-up and was significantly different from both the e-learning/toolkit alone group (P=.02) and the usual care group (P=.03). At 8 months post-training, the implementation + e-learning/toolkit group had higher mean provider readiness scores than the other 2 groups for documentation of screening, assessment, and monitoring. However, documentation of treatment was the only component that reached statistical significance (P=.025).

Conclusion

Among the practices that followed the implementation protocols, implementation assistance + e-learning/toolkit was effective in improving rates of screening for bipolar disorder, anxiety, and PTSD. However, 3 of the 5 practices did not follow the implementation protocols, suggesting that the intensity of the implementation needs to be tailored based on practice readiness for implementation.