Defining exposure for estimating the global burden of alcohol consumption: plausibility testing of WHO methodology using ASEAN countries as a case study

BACKGROUND: Comparative risk assessments (CRAs) provide important information for shaping alcohol control policies. Underlying their CRAs, the WHO uses a standardised methodology to assess and detail the levels of alcohol use for all countries and for various regions. This publication uses a case study approach on the member states of the Association of South East Asian Nations (ASEAN) to examine potential biases resulting from the methodology employed by the WHO in calculating exposure values for their CRAs. METHODS: Researchers from each of the 10 ASEAN member states identified large population surveys to improve upon the data collected by the WHO monitoring systems to estimate exposure between 2000 and 2022. From these surveys and aggregate data, key indicators were created for each Member State using WHO standardised methodology. Steps were defined to test for implausible values, particularly for the indicator for average level of alcohol consumption among drinkers. Sensitivity analyses were undertaken to identify possible causes of these values. Finally, we compared the results of the implausibility checks with two other regions, the European Union (EU) and the East African Community (EAC), based on data collected by the WHO. RESULTS: The indicator for average volume of alcohol consumption among drinkers showed implausibly high values for three ASEAN countries, Lao PDR, Thailand and Viet Nam. Further simulations based on assumptions regarding the prevalence of people with heavy or very heavy drinking levels further corroborated a likely bias. An examination of the constituents of the indicator revealed that the bias for Thailand could be due to responses received to survey questions regarding alcohol abstention, in which a high number of respondents claimed no consumption of alcohol over the past year. For the Lao PDR and Viet Nam, the same problem with survey respondent self-reports on alcohol consumption may exist, but we cannot exclude the possibility that answers to the survey question regarding unrecorded alcohol may also have contributed. Investigations of two further regions of the world also showed some implausible values, albeit to a smaller degree for the EU. CONCLUSIONS: Plausibility testing of key monitoring indicators is important and yields important information for improving future monitoring efforts.