Page 68 - BH-2-3

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

Table 3. Age at death as a function of 3 classes of physical mortality curves for total and CVD remained J-shaped. In
activity combined with six classes of alcohol intake. Data fact, the relationship between alcohol (wine) intake showed
estimated from the multiple linear regression given in Table 2 a J-shaped pattern for both all-cause and CVD mortality,
while it was direct and positive for liver cirrhosis and cancer
Physical activity class Alcohol intake class
0 1 2 3 4 5 (although with some uncertainties in the latter case).
Age at death (years) The curvilinear relationship with age at death was even
1 (sedentary) 70.2 72.1 73.7 71.6 71.7 69.8 more relevant because the piece of analysis on all-cause
2 (moderate) 72.0 74.0 75.5 73.4 73.5 71.7 death could have been biased by the fact that the study
cohort was almost extinct. On the other hand, the age at
3 (vigorous) 73.8 75.8 77.3 75.2 75.4 73.5
death of the reference class was 3.5 years higher versus
abstainers and 3.8 years higher versus men in Class 5.
Table 4. Models estimating all‑cause mortality between All of this was adjusted for the covariates. It is of interest
year 20 and year 45 of follow‑up and age at the death after that the abstainer class carried the highest comorbidity
45 years of follow‑up as a function of alcohol intake at year index across the six classes of alcohol consumption, partly
zero and changes in alcohol intake between year zero and explaining their excess mortality.
year 20 of follow‑up
A peculiar relationship was identified between alcohol
Models and parameters Coefficient t value P value consumption and physical activity because the three classes
Cox model of physical activity had an increasing intake of alcohol,
Alcohol intake: Year 0 −0.0020 −1.34 0.1799 from sedentary to moderate to vigorous physical activities.
Changes in alcohol intake −0.0027 −2.24 0.0250 The consequence of this association was that everything
between year 0 and year 20 else being equal, larger alcohol intake among very active
MLR model men was still associated with a good outcome.
Alcohol intake: year 0 0.0139 1.70 0.0893 Changes in alcohol intake played a role in all-cause
Changes in alcohol intake 0.0177 2.60 0.0098 mortality and age at death starting from the 20-year mark of
between year 0 and year 20 follow-up when another examination allowed for updates
Notes: Estimates are adjusted for entry levels of age, Mediterranean in alcohol consumption. At that point, the entry-level of
adequacy dietary index, physical activity, and cigarette consumption. alcohol intake was no longer significant, while a significant
Denominator=430. For the Cox model, end-point: all-cause mortality decrease over the previous 20 years was associated with
between years 20 and 45 of follow-up. For the multiple linear regression
(MLR) model, the end-point is age at death after 45 years of follow-up. lower all-cause mortality and higher age at death.
Early literature on the issue 1-10 consistently showed a
In fact, 97.7% of subjects were drinkers, and this is J-shaped relationship between alcohol intake, CVD, and
probably a unique case. Moreover, wine consumption in all-cause mortality. More recent meta-analyses and review
the study population was very high, probably among the articles 11-17 produced similar findings but added several
highest ever recorded in other studies, although precise exceptions and doubts. In particular, according to a review
comparisons were difficult since we expressed the intake article, low alcohol consumption may be beneficial to
in g/day of alcohol (and using our specific consumption prevent CHD but not other fatal conditions. In a meta-
11
classes) while consumption is frequently expressed as analysis, alcohol intake of <30 g/day was protective of CHD
drinks per day. Approximately 70% of the population mortality, but episodic heavy drinking carried a risk similar
could be classified as heavy drinkers. The comparison to that of the abstainers. Moreover, the use of abstainers
13
of the various classes of alcohol intake with that of the as reference groups seems to introduce a large bias.
nondrinkers was not promising, and therefore, we used In a large meta-analysis, a curvilinear relationship was
Class 2 as a reference (about 52 g of alcohol/day), but due identified between drinking and mortality risk, but there
to the small numbers involved, the differences versus the were uncertainties in interpretations due to large differences
reference class were not always statistically significant.
in the methods used to measure the exposure. Another
14
We were uncertain whether the few abstainers refrained meta-analysis did not show any benefit for moderate
from alcoholic beverages due to personal preference, drinking for all and some specific mortality causes. In a
15
experienced unpleasant reactions when drinking, or recent meta-analysis, it was found that after adjustment for
had stopped drinking due to illness or medical advice. the role of abstainers and for the quality of the studies, low
However, their small number likely had minimal impact levels of alcohol intake did not provide benefits for all-cause
on our analysis. Even when excluding these abstainers, the mortality. Finally, a meta-analysis involving only studies
16

Volume 2 Issue 3 (2024) 5 doi: 10.36922/bh.3016