Implementing adolescent SBIRT in an urban federally qualified health center: generalist vs. specialist service delivery models

摘要

BackgroundLittle is known about how best to implement SBIRT servicesin pediatric health care settings or who, optimally,should provide brief interventions when on-site behavioralhealth is available. The objective of this presentation is topresent results from a cluster randomized trial examiningimplementation of adolescent SBIRT services for substanceuse within a US federally qualified healthcare system.Two different implementation models for conductingbrief interventions (BIs) were compared using randomizationat the clinic level to either: the Generalist Model (BIprovided by primary care provider) or the SpecialistModel (BI provided by behavioral health specialist).Material and methodsMultilevel logistic regression modeling was used to examinedifferences by Condition in rates of successful deliveryand documentation of the following services: (a) screening(of all adolescent patients ages 12-17), (b) brief advice (forpatients reporting alcohol or drug use but scoring ≥2 onthe CRAFFT), and (c) brief intervention (patients scoring<2 on CRAFFT, delivered using either the Specialist orGeneralist models). Due to the organization transitioningto a new electronic medical record (EMR) in month 6 ofthe study, data on BA and BI are currently limited toextractions from the new EMR.ResultsMultilevel logistic regression analyses taking intoaccount the cluster-randomized design showed no significantdifferences between Generalist and Specialistconditions in rates of screening (OR=1.27; p=.55), withsignificant volatility over time (<.001) and variation bysites. In the post-EMR transition, Generalist sites werenot significantly more likely to deliver appropriate BA(OR=1.34; p=.70) or BI (OR=1.53; p=.36) than Specialistsites. Site-level intraclass correlations were higher thananticipated. Future analyses will examine practices forthe full implementation period and subsequent to theremoval of implementation support resources.ConclusionsBoth service delivery models showed promise for deliveringBIs but the high rates of variability within sitesdemonstrate a need for further examination.