Page 67 - BH-2-3

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

Table 2. Relationships of six classes of alcohol (wine) intake with all‑cause and some specific causes of mortality and with age at death
Models and parameters Hazard rate (95% CI [confidence interval]) for alcohol classes
Cox models for 45‑year mortality Class 0 Class 1 Class 2 Class 3 Class 4 Class 5
All-cause 1.41 1.08 1.00 1.13 1.08 1.34 (*)
(0.97–2.07) (0.92–1.27) reference (0.96–1.33) (0.91–1.29) (1.10–1.62)
Cardiovascular disease 1.74 1.30 1.00 1.12 1.03 1.34 (*)
(1.02–2.96) (1.03–1.65) reference (0.88–1.43) (0.78–1.34) (1.00–1.82)
Cancer 0.75 0.87 1.00 0.84 0.86 1.14
(0.31–1.84) (0.64–1.19) reference (0.62–1.15) (0.62–1.18) (0.81–1.62)
Liver cirrhosis 0.00 0.73 1.00 0.87 3.76 (*) 6.43 (*)
(n.a.) (0.18–3.08) reference (0.19–3.28) (1.29–10.94) (2.23–18.52)
MLR model for age at death
Age at death −2.51 −0.55 1.00 −1.12 −0.99 −2.85 (*)
(−6.23 – 1.21) (−2.13 – 1.03) reference (−2.70 – 0.46) (−2.67 – 0.68) (−4.72 – [−0.98])
Age at death (years)
72.7 74.7 76.2 74.1 74.2 72.4
Notes: Estimates of Cox models and multiple linear regression (MLR) are adjusted for age, BMI (body mass index), physical activity, cigarette smoking,
lnMAI (natural logarithm of the Mediterranean Adequacy Index), and comorbidity index. (*) p<0.05 versus reference following the Cox model or MLR.

Findings related to comorbidity showed that alcohol
intake Class 0 had the highest comorbidity index (0.59),
while Class 1 had the lowest comorbidity index (0.36),
suggesting that men who did not drink had the highest all-
cause mortality, partly explained by an excess of prevalent
diseases.

During the first 20 years of follow-up, alcohol
consumption among survivors decreased from 73 to
44 g/day (i.e., a 29 g/day difference). The role of this time
change in alcohol intake on all-cause mortality was tested
in the Cox model of Table 4, where it appears that it was
associated with significantly lower all-cause mortality
during the next 25 years since the coefficient of change
Figure 2. Age at death in six classes of alcohol intake (c0 [Class 0] was negative and significant. The same conclusion was
through c5 [Class 5]), adjusted in a multiple linear regression model for reached by the multiple linear regression of Table 4,
age, physical activity, cigarette smoking, body mass index, lnMAI, and where the reduction of alcohol intake was associated with
comorbidity index.
a higher age at death during the next 25 years, roughly
corresponding to 1 year for a reduction of 55 g of alcohol
we found relatively high correlations between alcohol intake. In both models, the role of baseline alcohol intake
intake and working physical activity. In fact, average
alcohol intake was 59 g/day in sedentary men, 67 g/day was not significant.
in men with moderate physical activity, and 84 g/day in 4. Discussion
men with vigorous physical activity. In cross-tabulation
of three classes of physical activity with six classes of The specificity and potential interest of this analysis were
alcohol intake, resulting in a total of 18 cells, the highest bound to the following issues: (i) the universal wine intake
age at death of 77.3 years was that with vigorous physical of almost all participants; (ii) the extremely high alcohol
activity plus Class 2 of alcohol intake, while the lowest intake that practically derived only from wine (facilitating
was 69.8 years in sedentary men plus Class 5 of alcohol analysis and conclusions); (iii) the extremely long follow-up
consumption, i.e., a large difference of 7.5 years (Table 3). that reached the quasi extinction of the cohort; and (iv) the
These estimates were derived from the multiple regression fact that among the majority of subjects characterized by a
equation, and therefore, they were adjusted for the other high level of working physical activity, such a high alcohol
covariates. intake was not necessarily dangerous.

Volume 2 Issue 3 (2024) 4 doi: 10.36922/bh.3016