Meta-analysis of the association between alcohol consumption and abdominal aortic aneurysm

Background: Alcohol is a possible risk factor for Abdominal Aortic Aneurysm (AAA) but evidence from individual studies is weak and inconsistent. Existing narrative reviews suggest the possibility of nonlinear associations. We therefore aimed to formally quantify the association using a systematic literature review followed by dose-response meta-analysis of prospective studies. Methods: MEDLINE, Embase and Web of Science were systematically searched to January 2017 for relevant prospective studies of alcohol consumption and AAA risk. Summary estimates of highest versus lowest levels of consumption, linear and nonlinear dose-response curves were quantified using random-effects models. Results: Eleven relevant cohorts were identified presenting results from 3580 cases amongst 473092 participants. Data were extracted from 10 cohorts for meta-analyses of high versus low levels of alcohol consumption (risk ratio = 0.93, 95% CI 0.78-1.11, p=0.4, I2=47%). The linear dose-response RR could be derived from 11 cohorts (RR=1.00 per 8g alcohol/day, 95% CI 0.97 to 1.04, p=0.9, I2=73%). Nonlinear dose-response results showed a tick-shaped curve with lower risks up to 2 units/day but increasing risks beyond that (p=0.05). The increase in risk beyond 2 units/day is stronger in men than in women. Conclusions: Whilst the linear dose-response revealed little evidence of an association between alcohol consumption and AAA risk, a tick-shaped trend in the association was observed. This nonlinear dose-response revealed reduced risks for alcohol consumption below 2 units/day masking increasing risks beyond 2 units/day. Randomised controlled trials are unlikely, so this study may present the strongest level of evidence available.