Evidence synthesis methods to inform decisions about complex behaviour change interventions in public health : brief alcohol interventions as a case study

Abstract

Background Decisions about health interventions should be based on best available evidence. Evidence synthesis comprises a suite of methods for robustly collating and integrating existing research to inform these decisions. I explored the use and limitations of established evidence synthesis methods for informing policy decisions about the use of complex interventions, using brief alcohol interventions as a case study. Hazardous and harmful alcohol consumption costs the NHS £3.5 billion annually, and brief alcohol interventions aim to reduce it and prevent alcohol-related harm in people who are neither aware of the risks nor seeking treatment. Practitioner delivered interventions involve in-person conversations that provide feedback and options for reducing consumption; digitally delivered interventions have become more common as personal devices have become more sophisticated and convenient. Methods I followed established systematic review and pairwise and network meta-analysis methods to estimate the effectiveness of brief alcohol interventions, and used subgroup and meta-regression methods to explore heterogeneity. I critically appraised the methods and limitations of metaanalysis of randomised controlled trials (RCTs) for evaluating such complex interventions, and explored elements of complexity missed by these analyses. Results Both practitioner and digitally delivered interventions were effective compared to no or minimal treatment control groups for reducing weekly but not episodic drinking. Practitioner delivered interventions reduced consumption more than digitally delivered interventions for the first six months post-intervention. However, the meta-analyses of RCTs could neither incorporate all relevant data nor fully explore contextual factors for these complex interventions. Conclusions Despite a wealth of systematic reviews and RCTs in this field, conventional evidence synthesis methods were unable to fully evaluate these complex interventions. Further RCTs alone are unlikely to enhance this evidence base; rather, observational and real-world data should be utilised to enrich our understanding of how best to use complex interventions.