Correlation between alcohol consumption and myocardial infarction: Dose-response meta-analysis of 18 cohort studies

It was demonstrated that alcohol has double effect on myocardial infarction (MI). The results of the studies that investigated the correlation between alcohol drinking and MI are varied. The aim of the present meta-analysis is to combine current prospective cohort studies of alcohol consumption and MI and to analyse the potential dose-response relationship. A complete search on Medline and Embase databases from database creation to May, 2016 was performed. We included in the meta-analysis only prospective cohort studies in English: these containing drinking doses divided into three or more levels and studies that contain relative risk/hazard ratio (RR/HR) and 95% confidence interval (CI) of MI onset risk, or adequate data for calculation of these indexes. RR of the alcohol consumption group, low-dose alcohol consumptiongroup, moderate alcohol consumption group and high-dose alcohol consumption group were 0.770 (95% CI 0.681 - 0.871), 0.861 (95% CI 0.786 - 0.942), 0.721 (95% CI 0.609 - 0.853) and 0.620 (95% CI 0.484 - 0.794) respectively. According to districts, follow-up periods and sample sizes, subgroups were divided and in all subgroups it was seen that alcohol consumption could reduce the onset risk of MI. When the drinking dose was lower than 96 g/day, dose meta-analysis showed that the linear test result p < 0.001; thus the drinking dose and MI onset risk were in non-linear negative correlation. With the increase of drinking dose, the onset risk of MI declined. Sensitivity analysis indicated stable study results; no publication bias showed in begg’s test and egger’s test. In conclusion, there is a negative correlation between drinking doses and MI; with the increase of drinking doses, the incidence rate of MI decreases.