Critique 118: Clustering of alcohol consumption with socioeconomic and biologic risk factors for cancer — 30 July 2013

Touvier M, Druesne-Pecollo N, Kesse-Guyot E, Andreeva VA, Galan P, Hercberg S, Latino-Martel P. Demographic, socioeconomic, disease history, dietary and lifestyle cancer risk factors associated with alcohol consumption.  Int J Cancer 2013;in press

Authors’ Abstract

Data are lacking regarding the association of alcohol consumption with a broad range of other cancer risk factors.

The objectives were: (1) to assess which sociodemographic,  lifestyle and dietary factors were associated with alcohol consumption; (2) to identify profiles of alcohol consumers by beverage type; (3) to estimate the number of cancer risk factors accumulated on the individual level according to alcohol consumption.

Alcohol and dietary intakes were assessed by six 24 h records among 29,566 adults of the NutriNet-Sante cohort.  Factors associated with alcohol consumption: non-drinkers (reference)/<10g/d/≥10g/d) were assessed by polytomic multivariate logistic regression stratified by gender.  Among alcohol consumers, percentages of alcohol brought by each beverage type were compared across sociodemographic and lifestyle characteristics using Kruskal-Wallis rank tests.

Several factors were associated with alcohol consumption ≥10g/d in both genders: older age (Pmen=0.02, Pwomen<0.0001), smoking (Pmen&women<0.0001), higher socioprofessional category (Pmen&women<0.0001), higher income (Pmen=0.003, Pwomen<0.0001), and less healthy dietary intakes.  Profiles of subjects varied across alcoholic beverage types.  Men with history of cardiovascular disease (P=0.0002) or depression (P=0.03) and women with history of cirrhosis (P<0.0001) consumed less alcohol.  In women, personal history of cancer was associated with a lower proportion of moderate alcohol users only (<10g/d, P=0.04).

In both genders, higher alcohol drinkers clustered more cancer risk factors (median=5, apart from alcohol) than non-drinkers (median=4), P<0.0001.  The multiplicity of deleterious lifestyle behaviours combined with alcohol drinking must be taken into account in cancer prevention efforts.  Gender-specific medical advice for people with personal or family history of alcohol-related diseases, including cancer, should be strengthened.

Forum Comments

As described by Forum reviewer Skovenborg: ‘The present study highlighted the clustering of alcohol consumption with other demographic, lifestyle and dietary cancer risk factors, such as excess body weight (in men), tobacco use, and poor compliance with cancer-preventive dietary recommendations.  The study managed to collect a very large number of men and women with moderate alcohol consumption and a sufficient number of non-drinkers as well.  The study is essentially a study of confounding and of clustering of lifestyle habits.  The difficult question is how to separate the drink from the drinker [Day NL. Alcohol and mortality: Separating the drink from the drinker (thesis). Ann Arbor: University Microfilms International, 1978.]  The positive association of alcohol consumption and tobacco use is well known and found in most studies.  And the association of beverage preference and diet is an example of the influence of regional culture and lifestyle habits.”

A number of factors that may diminish the applicability and generalizability of the present findings were noted by reviewer Ellison and other Forum reviewers.  As stated, all subjects were computer-literate volunteers, as data were collected over the internet, so results obviously do not apply to non-computer-literate people.  Further, the study included subjects who already had cancer, who may well have modified their lifestyle factors after diagnosis.  Much space is given to reporting p-values, whereas with this large dataset almost any relation, however trivial, will be found to be “statistically significant.”  The magnitude of effect is of much more interest.

As stated by Skovenborg: “The study is prone to selection bias due to the process of recruiting participants using the internet.  All observational studies are at risk of information bias; however, the use of internet-based questionnaires providing detailed data on the nature of the consumed beverage (83 different items were proposed) and using validated photographs to estimate the portion sizes for each reported food and beverage item seems to be a novel and very clever method of data-collection.”

The use of data from 6 randomly chosen days (both week days and weekend day) of dietary records should provide a reasonable estimate of most nutrient intake.  “Dairy” apparently did not separate milk from butter or cheese; there are differences among these types of dairy products for cardiovascular risk, although the association with cancer risk is unclear.

The main concern of many Forum reviewers was the grouping of all factors into a single, total number for assessing the relation with alcohol intake.  Essentially all associations with alcohol consumption shown in this study are known from previous epidemiologic research.  Some are very weak predictors of cancer, so adding them all up and calling them “cancer risk factors” may not be justified.  In other words, the use of the total number of “cancer risk factors” to relate to alcohol intake,  without attention to the relative importance of each (e.g., smoking is a major factor for cancer risk, while occupation or supplement intake may be less important), is a major problem in interpreting the results.

Relation of present findings to previous research:  Reviewer Skovenborg provided an excellent summary of previous data on this subject.  “In previous studies of beverage preference and diet conducted in the Mediterranean area, results are discordant.  A study carried out in a French population showed that moderate alcohol drinkers or wine drinkers had better dietary habits and healthier behaviours compared to other drinkers or abstainers (Ruidavets JB, et al. Alcohol intake and diet in France, the prominent role of lifestyle. Eur Heart J 2004;25:1153–1162).  However, in a study carried out in an Italian population, where wine was the most common type of alcoholic beverage consumed, wine drinking was not an indicator of a healthier diet (Chatenoud L ,et al. Wine drinking and diet in Italy. Eur J Clin Nutr 2000;54:177–179) with no appreciable differences in either sex for any of the food items considered (fruits, raw vegetables, cooked vegetables, salad and fish) between wine drinkers and other alcoholic beverage drinkers and abstainers.  In a study from Spain, no relevant differences were found in adherence to the Mediterranean food pattern according to alcoholic beverage preference (Alcacera MA, et al. Alcoholic beverage preference and dietary pattern in Spanish university graduates: the SUN cohort study. European Journal of Clinical Nutrition 2008;62:1178–1186).”

As the summary by Skovenborg continues, “In contrast, other investigations conducted outside the Mediterranean area, such as the United States or Northern Europe, have reported that wine drinkers tended to adopt healthier dietary patterns.  In a Danish study, wine drinking compared to other alcoholic beverages was associated with a higher intake of fruit, fish, cooked vegetables, salad and olive oil (Tjønneland A et al., Wine intake and diet in a random sample of 48 763 Danish men and women. Am J Clin Nutr 1999;69:49-54).  Another study from Denmark showed that wine buyers made more purchases of healthy food items than people who routinely bought beer (Johansen D et al. Food buying habits of people who buy wine or beer: cross sectional study. BMJ 2006;332:519–522).  A cross-sectional study from a large American cohort showed that wine drinkers had healthier diets than did those who preferred beer or spirits or had no preference(Barefoot JC et al. Alcoholic beverage preference,
diet, and health habits in the UNC alumni heart study. Am J Clin Nutr 2002;76:466–472).  Similarly, in another study conducted in western New York, wine drinkers tended to have diets more consistent with recommendations to reduce chronic diseases, whereas beer drinkers, liquor drinkers and non-drinkers had patterns less consistent with these guidelines (McCann SE, et al. Alcoholic beverage preference and characteristic of drinkers and nondrinkers in western New York. Nutr Metab Cardiovasc Dis 2003;13, 2–11).

Overall, this paper presents an interesting tabulation of lifestyle factors associated with alcohol consumption.  However, the grouping of all “cancer risk factors” and relating only the total number to alcohol intake makes the relevance of these findings unclear.  As added by Forum member Finkel: “After casually accepting alcohol as leading to cancer, a dangerous generalization, the authors laudably recognize that people are not inbred laboratory rats, that they are beset by nearly infinite confounding or possibly confounding variables.  I find this paper, though interesting in intent, difficult to interpret.”

Forum Summary

It has been shown in most previous studies that moderate drinkers, especially those who generally consume wine, tend to have other “moderate” lifestyle factors.  For example, they tend to be better educated, of higher socio-economic status, and generally are more active and eat a healthier diet than non-drinkers.  An exception is cigarette smoking, which tends to be more common in drinkers than in abstainers.  Epidemiologists studying alcohol and health are always striving to adjust appropriately for other lifestyle factors, attempting to assure that the difference in health outcomes among subjects relates to their drinking, and not to associated lifestyle factors.

In the present study, the investigators related alcohol  consumption to a large number of socio-demographic and lifestyle factors that relate to the risk of cancer.  They report that several factors were associated with alcohol consumption ≥10g/d in both genders: older age, smoking, higher socio-demographic category, higher income, and less healthy dietary intakes.  Other factors were associated with alcohol consumption differently for men and women.  The authors then report that the total number of such factors was higher among consumers of ≥10g/d of alcohol than among abstainers or lighter drinkers.  They conclude: “The multiplicity of deleterious lifestyle behaviours combined with alcohol drinking must be taken into account in cancer prevention efforts.  Gender-specific medical advice for people with personal or family history of alcohol-related diseases, including cancer, should be strengthened.”

While Forum reviewers found the data presented in this paper to be of interest, they did not believe that simply adding up all adverse risk factors provided useful information for the prevention of cancer.  It is well known that smoking is a major risk factor for certain cancers, and in almost all studies alcohol consumers are more likely to be smokers; hence, drinkers should certainly be advised about the dangers of smoking.  On the other hand, by simply summing the number of factors into a total score, the authors included a number of risk factors for which there is less of a scientific basis for their effect on cancer risk (e.g, the intake of supplements).  Such factors should not be given the same weight as smoking as risk factors for cancer.  Further, this study was carried out among computer-literate volunteers, so it may have limited relevance for the general population.

There have been extensive data showing that “healthy” and “unhealthy” lifestyle factors tend to cluster: for example, light-to-moderate drinkers tend to be leaner and eat a healthier diet than non-drinkers or heavy drinkers, as shown in this study.  However, the prevention of chronic diseases relates to a large number of behaviors, and alcohol consumption cannot be considered as an isolated factor.  Current epidemiologic data suggest that the combination of not smoking, being physically active, eating a healthy diet, avoiding obesity, and, unless contraindicated, regularly consuming small amounts of an alcoholic beverage, together make up what can be defined as a “healthy lifestyle.”

*                      *                      *                      *                   *

Comments on this paper were provided by the following members of the International
Scientific Forum on Alcohol Research:

Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA

Erik Skovenborg, MD, Scandinavian Medical Alcohol Board, Practitioner, Aarhus, Denmark

R. Curtis Ellison, MD, Section of Preventive Medicine & Epidemiology, Boston University School of Medicine, Boston, MA, USA

Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway

Gordon Troup, MSc, DSc, School of Physics, Monash University, Victoria, Australia

Pierre-Louis Teissedre, PhD, Faculty of Oenology – ISVV, University Victor Segalen Bordeaux 2, Bordeaux, France

Creina Stockley, PhD, MBA, Clinical Pharmacology, Health and Regulatory Information Manager, Australian Wine Research Institute, Glen Osmond, South Australia, Australia

Dee Blackhurst, PhD, Lipid Laboratory, University of Cape Town Health Sciences Faculty, Cape Town, South Africa

Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy