Page 66 - BH-2-3

Brain & Heart Wine intake and 45-year mortality in rural men

2.3. Statistical analysis of all deaths) and cancer (19% of all deaths), while liver

Alcohol intake was classified into six arbitrary classes, partly cirrhosis covered 3% of all deaths.
derived from the inspection of the interquartile distribution. Kaplan–Meier survival curves as a function of the six
Kaplan–Meier survival curves were generated based on these classes of alcohol intake showed a large overlap and were
six classes of alcohol intake. Cox proportional hazard models difficult to read. Therefore, we show a simplified figure
were employed to assess associations with all-cause mortality where only Class 2 and Class 5 are reported, showing a
and mortality from CVD, cancer, and liver cirrhosis, using large difference, with Class 2 having the most favorable
alcohol intake and six confounding covariates as predictors. outcome (p of log-rank Chi-square 0.0006) (Figure 1).
In addition, a multiple regression model was used with the Class 2 was then chosen as a reference in the multivariate
same covariates and age at death as the dependent variable. analysis, although it corresponded to four to seven drinks
Statistical significance was defined as P < 0.05. of wine per day.
Class 2 of wine consumption served as the reference Cox proportional hazard models with alcohol intake
category in multivariate models as it demonstrated the classes, six other covariates, and various mortality end-
lowest risk for all-cause and cardiovascular mortality in points as dependent variables are reported in Table 2,
the preliminary analyses. together with a multiple regression model using the
same covariates and age at death as dependent variables.
Another Cox model was computed on the survivors Mortality in Class 5 for all causes, CVD, and liver cirrhosis
after 20 years of follow-up, considering mortality over the was significantly higher than in Class 2, which was used
subsequent 25 years as the end-point. Baseline alcohol as a reference. This was not the case for mortality from
consumption, changes in alcohol consumption over the cancer, and in this case, not even the difference between
20-year period, and the usual confounding variables were Class 5 and Class 0 was statistically significant, despite
included as covariates. Similarly, a multiple linear regression the presence of a weak, increasing trend from the lowest
model was used to predict age at death. The reduced to the highest class. In the case of all causes and CVD,
denominator for these models was based on 430 men.
Class 0 had an excess of risk versus the reference class,
3. Results but the difference did not reach significance. For these
two endpoints, reference Class 2 carried the lowest risk.
Out of 1284 men examined at entry, only 29 (2.3%) were A definitely increasing mortality was recorded when liver
not drinkers. The six classes of alcohol drinking (together cirrhosis was the endpoint.
with the correspondent wine equivalent) are given in
Table 1, where the overall alcohol intake was very high, Similar findings were obtained analyzing age at death,
with an average of 76 g/day (77.4 for drinkers). Classes 2 with Class 2 showing the highest level, declining for lower
and 3 covered almost half of the sample and had an average and higher alcohol intake classes, and a difference of
alcohol intake of 47 – 95 g/day, corresponding to roughly 3.8 years between Class 2 and Class 5. The relationship of
500 – 1000 mL of wine. alcohol intake in the six classes with age at death is depicted
in Figure 2.
During 45 years, there were 1264 deaths (98.4%), and
the main causes of death were major CVDs (CVD = 44% In a correlation matrix including alcohol intake and
the other covariates fed into the multivariate models,
Table 1. Alcohol and the correspondent wine consumption in
the study population were divided into six arbitrary classes

Classification N % Alcohol, g/day Wine equivalent,
(range) mL/day (range)
Class 0 29 2.3 0 0
Class 1 270 21.0 25 (0.6 – 46) 260 (6 – 483)
Class 2 332 25.9 52 (47 – 70) 543 (500 – 734)
Class 3 270 21.0 82 (71 – 95) 861 (750 – 1000)
Class 4 222 17.3 103 (95 – 134) 1085 (1001 – 1414)
Class 5 162 12.6 176 (142 – 570) 1860 (1500 – 6016)
All 1284 100 76 (0 – 570) 798 (0 – 6016)
Notes: Alcohol among drinkers only: mean=77.4 g/day; wine among Figure 1. Kaplan–Meier survival curves for alcohol intake Class 2 (cl2)
drinkers only: mean=813 mL/day. (average 52 g/day) and 5 (cl5) (average 176 g/day).

Volume 2 Issue 3 (2024) 3 doi: 10.36922/bh.3016