Predictors of transitions across stages of alcohol use and alcohol-use disorders in South Africa.
AIMS To determine the relative risk (RR) of non-fatal injury associated with alcohol consumption in a series of emergency departments (EDs), possible effect modifiers and the impact of contextual variables on differences across sites. DESIGN The case-crossover method was used to obtain RR estimates of the effect of alcohol on non-fatal injuries. Meta-analysis was used to evaluate the consistency and magnitude of RR across sites, and the extent to which contextual variables explain differences in effect sizes. PARTICIPANTS Probability samples of 11,536 injured patients attending 28 EDs studies in 16 countries (1984-2002). The majority of the sample was male (65%) and > 30 years old (53%). MEASUREMENTS Exposed cases where those that consumed alcohol 6 hours prior to the injury. Usual alcohol consumption served as the control period. FINDINGS Drinking within 6 hours prior to the injury was reported by 21% of the sample. The estimated (random) pooled relative risk for patients who reported alcohol use within 6 hours prior to injury was 5.69 (95% confidence interval = 4.04-8.00), ranging from 1.05 in Canada to 35.00 in South Africa. Effect size was not homogeneous across studies, as societies with riskier consumption patterns had a higher relative risk for injury. Heavier drinkers also showed lower RR. CONCLUSIONS Acute alcohol was a risk factor for non-fatal injuries in most sites. Policy measures addressed to the general population are recommended, especially in societies with riskier consumption patterns.