Commentary on Zur &amp; Zaric and Shepard<i>et al</i>. (2016): Cost‐effectiveness of SBI for alcohol—where are we and where do we want to go?

The analyses by Shepard et al. 1 and Zur & Zaric 2 on the cost-effectiveness of alcohol screening and brief intervention (SBI) add to a relatively limited evidence base and suggest that SBI is a good use of resources. The studies contribute to the field by considering the impacts across a nation's population, incorporating significant others in the intervention and using a preferred common health outcome, and they prompt several directions for future research, such as accounting for several domains of alcohol-related consequences and including referral to specialist treatment as part of SBI. Screening and brief intervention (SBI) for alcohol in a primary care setting has been shown to be clinically effective at identifying people who consume alcohol above recommended levels and reducing their alcohol consumption 3, 4. SBI has a particularly strong effect among patients who are not seeking treatment 3. In the United States, Europe, Canada and other regions, SBI is a recommended strategy to prevent and reduce excessive alcohol consumption 5, and policymakers world-wide are advocating for its widespread adoption. Economic evaluations provide a valuable framework to allocate scarce health-care resources to the most effective treatment and prevention strategies. Although there is substantial literature on the costs and benefits of alcohol dependence treatment, until recently relatively few economic studies have focused on alcohol SBI in medical settings 6. Cost-effectiveness analysis (CEA) quantifies the trade-off between costs and outcomes and uses non-monetary measures of outcomes, such as drinking and quality of life, to help determine which study condition represents better value for the money. Shepard et al. 1 and Zur & Zaric 2 contribute to the body of knowledge on the cost-effectiveness of alcohol SBI. Both studies find that SBI for alcohol is cost-effective and yet they use different populations, medical settings and analytical methods. Shepard et al. conduct a CEA using data from a randomized trial on male patients who drink to hazardous levels and who are located in a large urban emergency department with a high-level trauma unit in the United States. The intervention is built on a motivational interviewing approach and innovatively includes the patient's significant other (SO) in the motivational interview. Zur & Zaric develop a microsimulation model of alcohol consumption in Canada and estimate the cost-effectiveness of implementing universal alcohol SBI in primary care. One potential future direction of research that builds upon these studies is to broaden the analytical perspective, which determines what types of costs and outcomes are relevant. Both studies' main analyses adopt a health systems perspective. Because alcohol misuse has broader consequences on other outcomes—such as employment, productivity, crime, injury and property damage from crashes and accidents and the quality of life of family members and relatives—future work that builds upon these studies should consider a broader analytical perspective that includes such outcomes. The two studies help to advance the field in how costs and effectiveness are assessed and in the methods used. With regard to cost, Shepard et al. 1 collect cost data in two ways: providers self-report the time they spend in clinical-related activities and simulate each activity separately while an observer records the duration of the research and clinical activities. Self-report of short duration activities may be unreliable because respondents would probably round up or down. For example, a clinician may state reasonably that a screen takes ‘1 or 2 minutes’. For a single screen, this response probably covers the range observed, but the ceiling is twice that of the floor. Future research could extend the simulated approach that the authors use to a full time-and-motion study, where observers follow providers to track the time spent on each activity 7. Although this alternative avoids the disadvantages of self-reported data and separates research from clinical-related activities accurately, it is resource-intensive. Nevertheless, Shephard et al. confirm findings elsewhere that non-clinical activities consume a large proportion of clinicians' time 8. If confirmed in other studies, this finding alone should be very valuable in guiding research and clinical practice. With regard to assessing the effectiveness of SBI, both studies adhere to current best practice in CEA by using quality-adjusted life years (QALYs) as the primary end-point 9. QALYs represent a common health output measure that captures changes in morbidity and mortality and allow the economic value of alcohol treatments to be compared to other medical services. Although regulatory agencies world-wide are increasingly requiring the use of QALYs in CEAs 10, alcohol CEAs seldom use QALYs as the primary end-point 11, 12. An important step in the analysis is converting alcohol states into QALYs by applying preference weights 9, and the method used by Zur & Zaric to compute QALYs is a particularly strong contribution. The authors use a nationally representative survey that includes questions to compute the Health Utilities Index (HUI), a generic preference-based system (utility is an economic concept that refers to a specific type of preference 10). Using a similar approach in other observational and experimental studies requires the data collection protocol to include an instrument that can generate preference-based scores to construct QALYs. Without such foresight, studies most probably rely upon less ideal alternatives, such as those used by Shepard et al. In that study, instead of obtaining preference weights, the authors assume a value for each patient for the QALY gained from averting alcohol problems. Moreover, the estimate used was based on a study that computed QALYs that accounted for one specific consequence of alcohol—the reduced mortality from teen drinking and driving—without accounting for the impact of alcohol-related morbidity 13. A key recommendation for future research is to include an instrument routinely in data collection that would allow QALY construction, such as the HUI, EuroQol-5 Dimension (EQ-5D) or Short Form-6 (SF-6D) 9. Shepard et al. 1 raise the intriguing idea of further targeting the intervention to those periods in the emergency department when people who drink to hazardous levels comprise the largest share of admissions. Future research may wish to determine how costs and effectiveness are affected by changes in patient characteristics. Designing and implementing a randomized controlled trial with the required numerous study conditions may be beyond the resources of most studies. Thus, to address this next step in the research agenda, researchers may gain the greatest benefit from using secondary data analysis of trial or observational data in a modeling framework. With regard to the CEA methodology, although decision modeling is used widely in estimating long-term outcomes of chronic diseases, such as diabetes, cancer and heart disease, few published CEAs use decision modeling for substance use 11, 14. Modeling allows researchers to extrapolate cost and effectiveness temporally beyond the time span of the data observed and to assess the implications of varying parameters; it is a particularly powerful approach for understanding long-term alcohol-related consequences, analyzing rare events such as severe injuries and exploring multiple hypothetical clinical and policy scenarios. The microsimulation model of Zur & Zaric 2 is an important contribution to the field. In particular, by applying stringent calibration and validation methods, the authors address a key concern of all modeling, which depends upon the assumptions made. Compared with a base-case of no SBI at baseline, using the full Alcohol Use Disorders Identification Test (AUDIT) at a threshold of 8 provides an average cost-effectiveness ratio of US$8729 per QALY. The Zur & Zaric model prompts three directions for future research. First, if in fact SBI is delivered in Canada to some measurable degree, then a different base case and incremental (rather than average) analysis may be warranted. However, estimates of the penetration of SBI are difficult to come by, as encountered recently in studies of alcohol SBI in primary care in the United Kingdom and Italy 15, 16. Quantifying the extent to which SBI is currently being implemented (i.e. the status quo of SBI) is an important next step in SBI research. Secondly, a modeling approach is well suited to simulate the impact of incorporating referral to treatment. The extent to which referral to specialist services is offered and the extent to which patients then receive the referred services is largely unknown 17, 18. Thirdly, it is also important to understand the circumstances under which SBI is sustainable 19. Implementing SBI in practice introduces economic considerations beyond those considered in CEA, such as how services are funded, patient flow and practitioner availability. None.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.