Commentary on Tanner‐Smith et al.: Complexity matters—why we need to move beyond ‘what works’ when evaluating substance use interventions

Evidence of the effectiveness of brief interventions for substance use provides limited insight into what works, for whom, and in which context. Future evaluations should embrace more varied and innovative study designs to provide a richer understanding of how content, context and relational dimensions impact upon brief intervention outcomes.

Tanner-Smith et al. [1] have conducted a uniquely comprehensive review of the effectiveness of brief interventions (BI) for substance use delivered in general medical settings. While alcohol-targeted BI were associated with small but significant effects, they found limited evidence that drug-focused BI resulted in similarly positive outcomes for other substance use. As the authors highlight, studies of BI for drug use other than alcohol remain sparse, making it challenging to draw firm conclusions about implications for practice. Nevertheless, their findings prompt us to ask important questions about how the content, context and relational dimensions of BI might ultimately shape outcomes.

There have been previous efforts to open-up the ‘black box’ of BI to understand their active ingredients. One review of the mechanisms of action of alcohol BI suggested that interventions that include a personalized feedback component and change of planned exercises were most effective at reducing alcohol consumption [2]. Another examined the effectiveness of specific behaviour change techniques in alcohol BI and found promoting self-monitoring of alcohol consumption to be associated with positive results [3]. Others have explored the impact of setting and provider on outcomes, with evidence suggesting that alcohol BI are most effective when delivered in primary care by a generalist health-care provider, such as a general practitioner (GP) or nurse [4]. Indeed, one key BI ‘selling point’ has been their relative simplicity; 5 minutes of straightforward advice to reduce your drinking seems to offer comparable performance to long, complex interventions that require providers to have significant time, training and support [5].

However, as this review reminds us, that appealing sales package of simplicity masks a much more complicated reality of practice where an inter-related bundle of factors can shape how, where and when BI is delivered and whether it will be effective because, in short, brief interventions are complex interventions. They are complex because the interventions themselves comprise multiple behaviour change techniques and target what is itself a complicated and highly socially situated practice (substance use) [6, 7]. They are complex because the practitioners who implement these interventions bring varied skills, experiences, values and expectations to the table, with different levels of time, resources and professional agency [8, 9]. They are also complex because the contexts in which these interventions are delivered (primary care practices, emergency rooms, outpatient settings, the wider health and social care system, etc.) are themselves highly varied, constantly transforming and often unpredictable [10, 11].

Which brings us to the most important question: how might we better measure, evaluate and understand the impact of these complex effect moderators on the performance of BI for substance use? One (relatively) straightforward improvement that could support more robust and nuanced assessment of new evidence of BI effectiveness for alcohol and other drug use would be consistent reporting of relevant outcome data in trials [12]; in the alcohol BI field alone, one review identified more than 400 articles reporting 2641 outcomes that were measured in 1560 unique ways [13]. Using the consensus-based ORBITAL (Outcome Reporting in Brief Intervention Trials: Alcohol) core outcome set (COS) going forward could help to improve the quality of evidence synthesis and, ultimately, enhance the relevance of trial results to decision-makers [14]. In turn, there is scope to extend and adapt this COS to cover additional dimensions of BI impact (e.g. cost), and make it more appropriate to assess delivery in novel settings (e.g. criminal justice) or with different types of substances.

However, whether or not BI ‘work’ to reduce the consumption of alcohol or other substances is only part of the knowledge gap. As the new UK Medical Research Council framework for developing and evaluating complex interventions stresses, there is a need to ask a broader range of questions if our research is to lead to interventions that deliver meaningful and desirable outcomes for patients and are acceptable and feasible to implement in real-world practice [15]. Key questions that future drug use BI studies need to ask include: how the intervention reacts in different contexts; how can we consider and account for the perspectives of different types of stakeholders; and how do we expect our intervention to work and under what conditions? To generate evidence that might help to answer these questions, we also need research that embraces more varied and innovative study designs, including qualitative, theory-driven and mixed-methods approaches. As well as having implications for researchers, those responsible for commissioning, publishing and implementing such evidence must also be ready to embrace and reward greater novelty and diversity in BI research.

Amy O'Donnell is a National Institute for Health Research Advanced Fellow. She has no financial or other relevant links to companies with an interest in the topic of this article.

留言 (0)

沒有登入
gif