The temporal relationship between per capita alcohol consumption and harm: a systematic review of time lag specifications in aggregate time series analyses.

BACKGROUND Changes in per capita alcohol consumption are temporally linked to changes in rates of alcohol-related harm. Methodological approaches for analysing this relationship have been suggested, however, the problem of time lags is not well-addressed. This study provides a review of time lag specifications, looking at (a) time to first effect on harm, (b) time to full effect and (c) the functional form of the effect accumulation from first to full effect to inform modelling of the relationship between changes in aggregate alcohol consumption and changes in rates of harm. METHODS Bibliographic databases were searched and citation and reference checking was used to identify studies. Included studies were time series analyses of the relationship between aggregated population alcohol consumption and rates of alcohol-related harms where time lag specifications had been derived or tested. RESULTS 36 studies were included with liver cirrhosis, heart disease and suicide dominating the evidence base. For a large number of alcohol-related harms, no literature was identified. There was strong evidence of an immediate first effect following a change in consumption for most harms. Recommended lag specifications are proposed for a set of alcohol-attributable harms. CONCLUSIONS Research on time lag specifications is under-developed for most harms although we provide suggested specifications based on the findings of the review. Greater methodological attention needs to be given to the rationale for choosing or applying lag specifications and the inherent complexity of the time lag process. More consistent and transparent reporting of methodological decisions would aide progress in the field.