Guiding Good Choices for Health (GGC4H): Early Lessons from a Pragmatic Trial in Three Large Healthcare Systems

Introduction. Pediatric primary care has the potential to significantly improve adolescent behavioral health through broad implementation of effective prevention programs. Pragmatic approaches that align programs with the needs of busy, real-world clinics and healthcare systems (HCS) can promote adoption, but they also need to achieve high rates of enrollment and generate strong evidence of effectiveness to garner sustained HCS support. GGC4H is a pragmatic, cluster-randomized controlled trial evaluating the feasibility and effectiveness of implementing Guiding Good Choices (GGC), an evidence-based prevention program for parents of adolescents, in three large, integrated HCS. It aims to enroll 4,608 parents under the auspices of the NIH Healthcare Systems Research Collaboratory, whose goal is to strengthen national capacity to implement large-scale pragmatic trials conducted in collaboration with HCS. This paper reports on challenging implementation and design issues that needed to be resolved in the pilot year to achieve a successful large-scale trial: A referral approach that could be sustained by pediatricians and lead to high parent enrollment rates, and an analysis plan that addressed two problems: gaps in electronic health record (EHR) data and a pragmatic implementation plan that led to a partially cross-nested design and problems with statistical inference.

Methods. We used multiple data sources from the three HCS in descriptive analyses: (a) ~371 parent contact records documenting enrollment rates from two strategies (pediatrician’s in-person referral to GGC v. letter/email referral), (b) cross-site comparisons of EHR data about adolescent substance use and mental health, and (c) team meeting notes weighing statistical and pragmatic implications of different implementation strategies.

Results: Early results indicated that pediatricians referred half of eligible parents to GGC. About one quarter enrolled once reached by telephone, exceeding rates found in community trials, particularly following in-person referral. Intractable gaps in EHR data about substance use and mental health led us to create a behavioral health survey to evaluate GGC effectiveness. We developed an innovative, hierarchical statistical model of our partially cross-nested design to avoid bias, increased Type I error, and problems with statistical inference.

Conclusions: HCS-based pragmatic trials that generate strong scientific evidence are feasible and can support prevention program uptake and sustainability. Preliminary data from GGC4H suggest that parent enrollment was enhanced by pediatrician referral, but pediatricians may need support to refer consistently. Expanding EHR data on key adolescent prevention outcomes would facilitate pragmatic evaluation.