Abstract
BACKGROUND: Concerns about randomized controlled trial (RCT) generalizability typically center on characteristics of RCT patient participants. Possibly there are RCT site characteristics that distinguish RCT outcomes from those that can be expected in non-RCT settings. OBJECTIVES: To examine whether site propensity toward RCT enrollment is associated with recovery outcomes for patients and whether the association differs between patients who participate in a RCT compared with those who remain in an observational (OBS) treatment environment. DATA: Study participants with acute bipolar depression from The Systematic Treatment Enhancing Program for Bipolar Disorder acute depression pharmacotherapy RCT (N=337) and OBS treatment arm (N=1581). METHODS: A longitudinal OBS study comparing the likelihood of recovery in the RCT to the OBS arm, allowing effect modification by site high RCT enrollment propensity (defined as >the median) and other predictors over a 6-month follow-up period. RESULTS: Non-RCT participants who received care in sites with high RCT enrollment propensity had a higher probability of recovering from their bipolar-depression episode compared with participants from low propensity sites [odds ratio (95% confidence interval)=2.13 (1.28-3.55)]. RCT enrollment propensity was not associated with recovery outcomes for RCT participants [1.03 (0.35-3.03)]. CONCLUSIONS: Sites with high propensity to enroll patients in RCTs appear to have unobserved characteristics, which play a significant role in outcomes for non-RCT patients. For RCT participants in low-enrollment sites, possibly RCT protocols, which proscribe care delivery and monitoring, attenuate this effect. These results have implications for future research to improve outcomes in nonresearch care settings.