Introduction
Adolescence is considered a sensitive period for future health as the brain still undergoes maturation [
1]. Moreover, several behavioural traits including eating and drinking habits are formed during adolescence and are likely to sustain throughout life, which may be especially important in relation to adult risk of obesity [
2].
Alcohol is considered to be a risk factor for obesity due to a high calorie content [
3,
4], and because alcohol inhibits fat oxidation, which may result in accumulation of fat in adipose tissues [
5]. On the other hand, alcohol is also known to have a high thermogenic effect that may result in increased energy expenditure [
6]. Accordingly, results from previous studies examining the relationship between alcohol consumption and subsequent weight development are conflicting [
7‐
15]. The discrepancy in results could partly be attributed to variation in types of alcohol beverage consumed. Studies have found mixed results for wine [
16‐
19], and beer intake [
20‐
22], while spirit intake was more consistently found to be directly related to risk of weight gain [
16,
20,
23]. However, these studies were primarily based on adult populations and cannot be generalized to adolescents.
Wine is reported to be one of the most frequently consumed alcoholic beverages among adolescents [
24,
25]. However, the evidence of a relationship between different types of alcohol during adolescent and body weight is limited. Most studies have examined total alcohol consumption, and either been cross sectional or lasted into young adulthood. Thus, it remains unclear whether adult obesity and weight gain into adulthood may be attributed to the types of alcoholic beverages consumed during adolescence. Of the few longitudinal studies conducted among adolescents, some found a direct association between high alcohol consumption and high self-reported weight gain [
10,
11] while others found that adolescents with a high alcohol intake had a lower risk of becoming obese in young adulthood than adolescents with low intakes [
12]. Most studies, however, did not account for type of alcohol consumed.
Thus, in the present study we examined the association between intake of total alcohol and type of alcoholic beverages (wine, beer, spirit) during adolescence and subsequent weight gain until midlife. We hypothesized that adolescent total alcohol consumption, and in particular beer and spirit consumption would be directly, and wine intake inversely, associated with weight gain into midlife.
Discussion
In this longitudinal analysis with 20 or 22 year follow up, we found that participants with higher wine consumption during adolescence subsequently gained less weight than those with a lower wine consumption. Furthermore, contrary to results from a previous study [
11], in our study smoking status was found to modify the relationship between alcohol consumption and weight development where both total alcohol and wine drinking were associated with less gain in BMI into midlife for those who were non-smokers in adolescence, while no associations were observed for the smokers. The observed difference between smokers and non-smokers could have potentially been driven by differences in SES [
29]. However, the association persisted after adjustment for SES and we found no evidence of an interaction between SES and alcohol intake, suggesting this was not the case in our study. Moreover, there was a considerably lower prevalence of smokers than non-smokers, and hence a lack of power may be responsible for the non-significant results among the smokers.
Results from previous prospective population studies looking at the association between adolescent drinking and subsequent weight gain are few and mixed, generally confined to weight development into young adulthood, only, or of cross-sectional nature [
10‐
12,
15]. Hence, the present results are among the first to examine the long-term consequences of adolescent alcohol consumption and adult weight gain into midlife.
It has been suggested that alcohol consumption has short-term stimulatory effects on appetite and food intake, which may result in progressive weight gain [
6]. Furthermore, it has also been demonstrated that alcohol temporarily inhibits fat oxidation resulting in accumulation of fat in the adipose tissue. These would promote an increased risk of developing obesity on the long run [
30]. On the contrary, alcohol is also known to have a high thermogenic effect, which may result in increased energy expenditure following its intake [
6]. It has also been suggested that some of the energy ingested as alcohol is ‘wasted’, due to the activation of the inefficient hepatic microsomal ethanol-oxidizing system [
31]. These latter, might, in part, explain the observed inverse association in the present study. However, the inverse association observed could be also attributed to residual confounding, e.g. unrecorded differences in lifestyle and nutritional characteristics between those drinking more and less alcohol, wine in particular. Wine intake has repeatedly been associated with better overall nutrition and lifestyle, and moderate wine drinkers have been found to exhibit a better overall health and good quality of life [
32]. On the other hand, others have found similar inverse associations between wine consumption and weight gain [
16,
17,
20], suggesting that phytochemical compounds like resveratrol found in wine may offer additional protection against fat accumulation by inhibiting lipogenesis and de-regulating lipogenic gene expression [
33,
34]. Moreover, resveratrol is known to increase insulin mediated glucose uptake thus, effectively helping to reduce blood glucose levels [
34]. It is thus possible that the combination of antioxidants and ethanol in wine may have been responsible for the apparent benefits against weight gain from adolescence into midlife. This could also possibly explain why a similar association was not observed in relation to beer and spirit.
Strengths of the study include the detailed information on type of beverages and total alcohol consumption as well as the longitudinal nature and long follow up. However, the latter might also be a limitation as over such a long-time span, lifestyle and weight status may have changed several times during follow-up. Also, it can be argued that Danish adolescent’s lifestyle habits including their alcohol intake choices and patterns in 1983/85 were most likely different to today [
35]. However, a potential biologic association between adolescent alcohol intake, which is occurring during a time in life that may leave imprinting of importance for long term weight development, would not be expected to be different. It is also a limitation that we did not have information on drinking patterns (e.g. binge drinking), as data was collected before it was apparent that not only quantity but also pattern may be of importance for health. Moreover, we used self-reported data and hence the study is potentially subjected to reporting bias. Previous studies have noted overestimation of self-reported height and under estimation of weight resulting under-estimated BMI [
36] particularly among the overweight and obese [
37]. The consequence of such a differential reporting bias among the overweight and obese is that significant associations may have been overlooked (as for total alcohol), and that “true” associations were even stronger than those observed (as for wine and weight gain). Although it has been noted that self-reported frequency of alcohol consumption seems adequately reliable and valid for the research purposes [
38,
39] and compares well to peer based observation [
39], information collected via diet surveys [
40] or other questionnaires [
40], both over and under reporting of alcohol in adolescence is still possible. This may have inflated our results and hence we could have seen significant associations that should have been insignificant (like for wine). However, whether this bias in reporting of alcohol was also related to adolescent BMI is unknown, and hence we cannot predict if reporting bias may have influenced our observed associations. Nonetheless, the results should be interpreted with caution as causation cannot be conferred due to the nature of study. Also, the varying size of glasses is unaccounted for in the study, which most likely may have attenuated the results.
Although we adjusted for several potential confounders, we can also not rule out that unmeasured or residual confounding remains. Though we would expect a high alcohol, but potentially not high wine intake, to be associated with poorer dietary and physical activity habits, we cannot exclude that an adolescent healthy lifestyle was associated with adolescent alcohol and wine intakes which may have been responsible for the long-term relationship with weight status into midlife.
Finally, response at follow up was 42% only, with the possibility of selection bias and thus the generalizability of these results is unknown. There was no differences in age between participants and non-participants (
P = 0.25), while participants had slightly lower baseline BMI (20.6 vs. 20.3 kg/m
2), and higher baseline SES (34 vs. 27%) [
41]. Nevertheless, there is little reason to expect that the biological associations we observed in the present study between adolescent drinking and midlife cannot be generalised, as also suggested by the lack of modification or confounding by SES.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.