How should we set consumption thresholds for low risk drinking guidelines? Achieving objectivity and transparency using evidence, expert judgement and pragmatism

Most high-income nations issue guidelines on low-risk drinking to inform individuals' decisions about alcohol consumption. However, leading scientists have criticized the processes for setting the consumption thresholds within these guidelines for a lack of objectivity and transparency. This paper examines how guideline developers should respond to such criticisms and focuses particularly on the balance between epidemiological evidence, expert judgement and pragmatic considerations. Although concerned primarily with alcohol, our discussion is also relevant to those developing guidelines for other health-related behaviours. We make eight recommendations across three areas. First, recommendations on the use of epidemiological evidence: (1) guideline developers should assess whether the available epidemiological evidence is communicated most appropriately as population-level messages (e.g. suggesting reduced drinking benefits populations rather than individuals); (2) research funders should prioritize commissioning studies on the acceptability of different alcohol-related risks (e.g. mortality, morbidity, harms to others) to the public and other stakeholders; and (3) guideline developers should request and consider statistical analyses of epidemiological uncertainty. Secondly, recommendations to improve objectivity and transparency when translating epidemiological evidence into guidelines: (4) guideline developers should specify and publish their analytical framework to promote clear, consistent and coherent judgements; and (5) guideline developers' decision-making should be supported by numerical and visual techniques which also increase the transparency of judgements to stakeholders. Thirdly, recommendations relating to the diverse use of guidelines: (6) guideline developers and their commissioners should give meaningful attention to how guidelines are used in settings such as advocacy, health promotion, clinical practice and wider health debates, as well as in risk communication; (7) guideline developers should make evidence-based judgements that balance epidemiological and pragmatic concerns to maximize the communicability, credibility and general effectiveness of guidelines; and (8) as with scientific judgements, pragmatic judgements should be reported transparently.