Behaviour Change Update: Stage 3. BCT analysis of behaviour change interventions reported in studies of cost effectiveness

Using a search strategy of studies with economic analyses by Bazian, 251 interventions targeting smoking cessation, diet, physical activity, sexual health, alcohol and multiple health-related behaviours were identified. Of these, 102 provided cost-utility (CUA) estimates and 85.3% were considered to be cost-effective based on a conservative NICE threshold. Overall, smoking cessation interventions provided lower CUA values and were more likely to be cost-effective than interventions for multiple health-related behaviours. Across all interventions, those targeting the general population had lower CUA results and were more likely to be costeffective than those aimed at vulnerable populations. In addition, interventions featuring behaviour change techniques (BCTs) related to “Reduce negative emotions” had higher CUA values and those featuring the BCT “Monitoring outcome(s) of behaviour by others without feedback” were less likely to be cost-effective. When looking at health-related behaviours separately, diet interventions that provided medication only had higher CUA outcomes than other types of diet interventions. Moreover, diet interventions including BCTs related to “Comparison of outcomes” and interventions for multiple health-related behaviours that used electronic supporting material were less likely to be cost-effective than interventions that did not. Cost-effective interventions included in this report had CUA estimates broadly similar to interventions already appraised as cost-effective by NICE (reported in Stage 1). However, a higher proportion of interventions in this report focused on multiple health-related behaviours, were set in primary care, aimed at vulnerable populations, involved training and included BCTs pertaining to practical and social support and to discussing body changes. By contrast, fewer interventions than in Stage 1 focused on alcohol, were set in the work place, delivered at population level, used self-help material or incentives, and featured ‘choice architecture’ (CA). In general, interventions in this report served fewer functions, covered fewer BCT clusters and included fewer individual BCTs compared with Stage 1 interventions. Based on the present analysis, there is no consistent and little association between the presence of an individual BCT or BCT cluster and an intervention being considered cost-effective. These findings need to be interpreted cautiously given 1) different search strategies for this and the Stage 1 report, 2) reliance on incomplete information in published papers, 3) heterogeneity in economic analyses, 4) lack of consensus for a definition of CA and 5) bias in reporting of study findings.