42 - LONGITUDINAL PATTERNS OF ALCOHOL USE AND MORTALITY - A 40 YEAR FOLLOW-UP OF THE FINNISH TWIN COHORT

Abstract

Background Although the relationship of alcohol use with mortality has been extensively studied, these epidemiological studies are mainly based on single assessments of alcohol use and do not differentiate between causal and non-causal explanations. With four decades of follow-up in the older Finnish Twin Cohort, we examined the relationship of mortality with patterns of alcohol use in a genetically informative sample of MZ and DZ pairs. Methods Three questionnaire surveys of the Finnish Twin Cohort of like-sex twin pairs born 1930–1957 were conducted in 1975, 1981 & 1990. Alcohol consumption was assessed in an identical fashion in all surveys as the use of beer, wine and spirits. These were converted to the number of standard drinks consumed on average. Based on responses to all three surveys from 11,382 men and women, we first analysed the average consumption, ranging from abstinence over all 15 years (29% of all) to an average of 28 or more drinks per week (2%). Next, 3,656 persons whose consumption level (abstinent, one drink per week, 2–6, 7–13 or 14 or more drinks weekly) was the same across all time points were studied. Total mortality was obtained for response date in 1990 to end of follow-up 12/31/2016 with 1,977 deaths. Individual and within pair analyses were conducted using Cox proportional hazards regression models to derive hazard ratios (HR) for death. Results In analyses of all individuals, the lowest age-sex adjusted mortality was found for those drinking 1 drink per week on average, while consistent abstainers (HR 1.27,95%CI 1.09 to 1.49) and those consuming more had significantly elevated risks. The HR for 14 to 20 drinks weekly was 2.95, 21 to 27 drinks/wk had an HR of 4.42 while consuming 28 or more drinks/wk on average between 1975 and 1990 had an HR of 5.42 (95%CI 4.14to 7.10). For those with stable drinking patterns (n=3,656 twins, 648 deaths) the pattern was similar: those consistently drinking just one drink weekly had the lowest mortality, HR elevated to 1.49 for consistent abstainers; those drinking 2–6 drinks weekly had an HR of 1.43, 7 to 13 drinks an HR of 3.23, and those drinking at least 14 drinks at all three surveys an HR of 4.91. Within-family analyses found a pattern of mortality similar to the individual-based analyses, with an HR of 6.38 (3.14 to 13.0) for those twins consuming, on average, 28 or more drinks weekly in the prior 15 years compared to their co-twins drinking one drink weekly on average, with estimates for intermediate amounts lying between the two extremes. Discussion Our within-pair analyses were consistent with the individual analyses, with increased risks at the high end of use, but with a significant increase in mortality already at levels considered by many to represent safe drinking (i.e. 7 to 13 drinks weekly). Interestingly the results were similar and significant in both MZ and DZ pairs. The increased mortality associated with alcohol use is not due to genetic or other influences shared by co-twins but compatible with a causal effect.