Gea A, Martinez-Gonzalez MA, Toledo E, Sanchez-Villegas A, Bes-Rastrollo M, Nuñez-Cordoba JM, Sayon-Orea C, Beunza JJ. A longitudinal assessment of alcohol intake and incident depression: the SUN Project. BMC Public Health 2012, 12:954 doi:10.1186/1471-2458-12-954
Background: Longitudinal studies assessing the long-term association between alcohol intake and depression are scarce. The type of beverage may also be important. Therefore we aim to prospectively evaluate the influence of alcohol intake on incident depression in a Mediterranean cohort.
Methods: We assessed 13,619 university graduates (mean age: 38 years, 42% men) participating in a Spanish prospective epidemiological cohort (the SUN Project), initially free of depression. They were recruited between 1999–2008 and biennially followed-up during 2001–2010. At baseline, a 136-item validated food frequency questionnaire was used to assess alcohol intake. Wine was the preferred beverage. Participants were classified as incident cases of depression if they reported a new clinical diagnosis of depression by a physician and/or initiated the use of antidepressant drugs. Cox regression and restricted cubic splines analyses were performed over 82,926 person-years.
Results: Only among women, an U-shaped relationship between total alcohol intake and depression risk was found (P=0.01). Moderate alcohol intake (5–15 g/day) was associated with lower risk (Hazard Ratio: 0.62; 95% Confidence Interval: 0.43-0.89). No association was apparent for higher intakes of alcohol or for any specific type of alcoholic beverage.
Conclusions: Moderate alcohol intake might protect against depression among women. Further confirmatory studies are needed.
A number of studies have related alcoholism or other heavy alcohol use to depression, generally finding an increase in the risk of depression. Few prospective studies have been carried out to evaluate this association among moderate drinkers.
Key features of this paper: This was a large follow-up study with more than 13,000 subjects and with up to 10 years of monitoring. There were reasonable estimates of baseline alcohol intake using a food-frequency questionnaire (FFQ) for alcohol intake over the preceding year; in a validation study, there was high correlation between alcohol intake reported on the FFQ and on four food records.
There were 628 non-drinking males and 2,141 non-drinking females, an adequate number of non-drinkers to use them as the referent group. Most subjects reported <5 or 5-15 g/day, with a small number (mainly males) reporting an average intake of more than 15 g/d. Data on the pattern of drinking were not available.
Incident cases of depression, assessed at the 4th, 6th, 8th, and 10th year of follow up, were determined if the subject reported having been given such a diagnosis by a doctor or if the subject reported habitual use of anti-depressant drugs for four or more years. The reported diagnosis was tested in a sub-sample using the DSM-IV, with high specificity and sensitivity. Excluded were subjects with a history of depression at base-line or those reporting depression within 2 years of the baseline evaluation.
Using cut-points based on spline analyses, subjects were divided into 4 groups according to their daily alcohol intake: 0, < 5 g/day; 5-15 g/day, and > 15 g/day of alcohol. (A typical “drink” is variously defined as between 10 g and 15 g of alcohol; hence the 5-15 g/day group reported that they averaged the equivalent of about ½ to about 1 to 1 ½ typical drinks/day.) Included as co-variates in multi-variable analysis were age, smoking, physical activity, total energy intake, baseline BMI, adherence to a Mediterranean diet plan, marital status, and employment status.
Main findings of the paper: Reported depression was much higher among women than among men. Among women, there were 88 instances of depression, and the estimated hazard ratios for the three drinking groups were 0.97, 0.62, and 0.84 (in comparison with subjects reporting no alcohol). The difference was statistically significant (CI = 0.43-0.89) for the group reporting 5-15 g/day.
Key findings for males were an 18-36% lower risk of depression for the three categories of alcohol consumption, although the number of males reporting depression was very small (n=22) and none of the associations was statistically significant. (However, the hazard ratio for males in the 5-15 g/day category was almost identical to that of females in that drinking category, so a similar decrease in risk for men is also possible.)
While the estimated hazard ratios for subjects consuming more than 15 g/day were not statistically significant, they remained less than 1.0 (the risk of abstainers) for both men (0.68) and women (0.84). However, with so few heavy drinkers in this cohort, the study was not able to estimate the risk of depression for heavy drinkers.
The authors state that there were no differences seen according to the type of beverage consumed; the majority of subjects reported that they consumed red wine.
Potential for reverse causation: Despite the attempts of the authors to control for reverse causation, several Forum members were still unsure of the direction of the association. Stated reviewer Finkel, “Rather than alcohol leading to or protecting from depressive disease, which is certainly near the top causes of disability and misery in our world, I think of depression sometimes leading to the use of alcohol (to medicate the symptoms) or even, in its retardive manifestation, to not drinking.”
Stated reviewer Van Velden: “The question is, does alcohol cause depression, or may moderate alcohol consumption protect against the development of depression. It will be very difficult to answer these questions because moderate alcohol (especially wine) drinkers generally have a very healthy lifestyle, including meaningful exercise, a love for the performing and visual arts, good social skills and an integrated family life. Problem drinkers, on the other hand, may abuse alcohol because they are depressed. There are just too many confounders to draw valid conclusions.”
Reviewer Skovenborg countered: “I tend to agree with Van Velden’s comments, however it should be noted that (1) the confounders seem to have been taken care of pretty well in this study and (2) this is a cohort of moderate red wine drinkers from mostly the same social class eating Mediterranean food, which leaves room for less confounding. The most worrying aspects in my eyes are the wide confidence intervals of most results. A decrease of depression and an increase of depression associated with moderate alcohol consumption are both possible.”
Polyphenols and depression: Reviewer Vauzour adds: “This is indeed a cohort of mainly wine drinkers (and wine contains many polyphenols as well as alcohol). Previous reports demonstrate that as a whole, dietary polyphenols seem to exert positive effects on anxiety and depression, possibly in part via regulation of adult hippocampal neurogenesis. In fact, it has been proposed that depressed individuals have reduced hippocampal volume and there is increasing evidence from animal research that antidepressants might exert a proneurogenic effect in the hippocampus.”
Potential mechanisms; differences between men and women: Vauzour comments: “Potential mechanisms include, but are not limited to, (1) reversal of oxidative damage; (2) antidepressant-like effects potentially involving the central monoaminergic neurotransmitter systems and/or serotonergic and noradrenergic activation; (3) anxiolytic effect through an interaction of polyphenols with GABAA receptors. The only question that remains is why such effects would be more pronounced in women versus men. About twice as many women as men experience depression. Furthermore, typical symptoms we associate with depression such as depressed mood may not be present in depression in men. This can make it more difficult to recognize depression in men.”
Analyses from a prospective study of more than 13,000 adults in Spain related baseline alcohol intake (mainly wine) with the development of depression, with the outcome based on a physician’s diagnosis or on the habitual use of anti-depressant drugs for four or more years. Reported depression was much higher among women than men. Among women, the estimated hazard ratios for groups reporting an average alcohol intake of < 5 g/day of alcohol, 5-15 g/day, and > 15 g/day were 0.97, 0.62, and 0.84 (in comparison with subjects reporting no alcohol). The difference was statistically significant (CI = 0.43-0.89) for the group reporting 5-15 g/day.
Few men reported a diagnosis of depression; among moderately drinking men, the estimated risks of depression were similar to those of women, but none of the differences were statistically different from those of non-drinkers. The investigators found no differences according to the type of alcoholic beverage consumed.
The merits of this study include that the data were collected prospectively, that there were repeated questionnaires seeking data on depression, and that there were a large number of subjects, all of whom were college graduates. There were few men reporting depression, and few heavy drinkers, so results apply only to moderately drinking women. Among these women (those consuming between about ½ drink/day to about 1 or 1 ½ drinks per day) there was evidence of a lower risk of depression during a follow-up period extending up to 10 years.
Forum reviewers thought that this was a well-done analysis, and the use of cut-points for drinking groups according to spline analyses was appropriate. Some reviewers cautioned that moderate drinkers, especially regular wine drinkers, tend to have a large number of other healthy lifestyle habits that could relate to the risk of depression. While many of the recognized potential confounders were included in the Cox regression analysis in this paper, it will be important to determine if future studies support the conclusions of the authors.
Comments on this critique by the International Scientific Forum on Alcohol Research were provided by the following members:
David Vauzour, PhD, Senior Research Associate, Department of Nutrition, Norwich Medical School, University of East Anglia, Norwich, UK
Erik Skovenborg, MD, Scandinavian Medical Alcohol Board, Practitioner, Aarhus, Denmark
Pierre-Louis Teissedre, PhD, Faculty of Oenology – ISVV, University Victor Segalen Bordeaux 2, Bordeaux, France
Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA
Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway
David Van Velden, MD, Dept. of Pathology, Stellenbosch University, Stellenbosch, South Africa
R. Curtis Ellison, MD, Section of Preventive Medicine & Epidemiology, Boston University School of Medicine, Boston, MA, USA