Abstract
BACKGROUND & AIMS: To investigate potential biases that exist in available epidemiological evidence resulting in negative associations or underestimation of cardiovascular (CV) risk associated with alcohol consumption.
METHODS: UK Biobank involved baseline data collection from 22 assessment centres across the United Kingdom. The cohort consisted of 333 259 alcohol consumers and 21 710 never drinkers. Participants were followed up for a median 6.9 years capturing incident fatal and non-fatal CV events, ischemic heart disease and cerebrovascular disease. Alcohol intake was reported as grams/week.
RESULTS: Using never drinkers as reference, alcohol from all drink types combined (hazard ratios ranging between 0.61 and 0.74), beer/cider (0.70-0.80) and spirits combined, and all wines combined (0.66-0.77) associated with a reduced risk for all outcome measures (all CV events, ischaemic heart disease, cerebrovascular disease). In continuous analysis, alcohol captured from all drink types combined (hazard ratio, 1.08, 95% confidence interval, 1.01-1.14), and beer/cider and spirits combined (1.24, 1.17-1.31) associated with an increased risk for overall CV events, however hazard ratios were stronger for beer/cider and spirits (P < 0.0001). Wine associated with a reduced risk for overall CV events (0.92, 0.86-0.98) and ischemic heart disease (0.75, 0.67-0.84). This negative relationship with overall CV events was lost after excluding ischemic heart disease events (1.00, 0.93-1.08), while the positive association of alcohol captured from beer/cider and spirits remained significant (1.30, 1.22-1.40). This positive association with overall CV events was present even when consuming less than 14 units per week.
CONCLUSIONS: Avoiding potential biases prevents underestimation of cardiovascular risk and indicates that consuming up to 14 units per week also associated with increased CV risk in the general population.