Critique 083: Higher indices of quality of life are seen among persistent moderate drinkers than among abstainers — 20 June 2012
Kaplan MS, Huguet N, Feeny D, McFarland BH, Caetano R, Bernier J, Giesbrecht N, Oliver L, Ross N. Alcohol Use Patterns and Trajectories of Health-Related Quality of Life in Middle-Aged and Older Adults: A 14-Year Population-Based Study. J. Stud. Alcohol Drugs,2012;73, 581–59l
Objective: A 14-year multiwave panel design was used to examine relationships between longitudinal alcohol-consumption patterns, especially persistent moderate use, and change in health-related quality of life among middle-aged and older adults.
Method: A nationally representative sample of 5,404 community-dwelling Canadians ages 50 and older at baseline (1994/1995) was obtained from the longitudinal National Population Health Survey. Alcohol-consumption patterns were developed based on the quantity and frequency of use in the 12 months before the interview. Health-related quality of life was assessed with the Health Utilities Index Mark 3 (HUI3). Latent growth curve modeling was used to estimate the change in HUI3 for each alcohol pattern after adjusting for covariates measured at baseline.
Results: Most participants showed stable alcohol-consumption patterns over 6 years. Persistent nonusers, persistent former users, those decreasing their consumption levels, and those with unstable patterns (i.e., U shaped and inverted U shaped) had lower HUI3 scores at baseline compared with persistent moderate drinkers. A more rapid decline in HUI3 scores than that observed for persistent moderate users was seen only in those with decreasing consumption (p < .001). In a subgroup identified as consistently healthy before follow-up, longitudinal drinking patterns were associated with initial HUI3 scores but not rates of change.
Conclusions: Persistent moderate drinkers had higher initial levels of health-related quality of life than persistent nonusers, persistent former users, decreasing users, U-shaped users, and inverted U-shaped users. However, rates of decline over time were similar for all groups except those decreasing their consumption, who had a greater decline in their level of health-related quality of life than persistent moderate users.
There are a number of important aspects of the present analysis. The aim of the authors was “to explore the relationship between longitudinal patterns of alcohol use (especially ‘persistent moderate use’) and changes in health-related quality of life (HRQL) among middle-aged and older adults.” In a population-based cohort, the investigators used repeated assessments of alcohol among subjects to take potential changes over time in drinking patterns into
Adjustments were made for most key potential confounders, including baseline age, gender, education, household income, marital status, smoking, potentially life-threatening illnesses, other chronic illness, and body mass index. The authors also assessed level of usual physical activity, and nonspecific psychological distress using the Likert scale.
Detailed information was available on alcohol consumption. Moderate drinkers were defined as those having 1–14 drinks per week with no more than 3 on any day for women and no more than on any day for men. The repeated assessments allowed for the investigators to classify subjects according to changes over time in their drinking patterns, so that “persistent moderate drinkers” could be identified. The investigators also did secondary analyses among subjects who did not report any adverse health conditions (heart disease, cancer, stroke, or diabetes) during the first four years of follow up; these subjects were referred to as “consistently healthy.”
Key results of paper: Key results of the study were that the majority of subjects remained in their respective alcohol categories for over 6 years (as seen in most studies). However, 31.4% of their subjects decreased their intake over the follow-up period. In the model using baseline consumption data, lifetime abstainers and former drinkers had lower quality of life (HUI3) scores at baseline (poorer quality of life) than did moderate drinkers. Lifetime abstainers, former drinkers, and infrequent drinkers experienced a greater decline in HUI3 compared with moderate drinkers.
In a separate model incorporating changes in drinking patterns that was limited to subjects consistently reporting good health, trajectories of the quality of life were similar for all groups. The persistent moderate drinkers had higher HUI3 scores at the start of their follow up, but alcohol patterns only affected baseline HUI3 scores and not the rates of change. As stated by the authors, “The findings suggest that alcohol-consumption patterns are associated with HRQL, but the rate of decline in HRQL is similar for all drinking patterns except for persons who decreased their consumption.”
Specific comments on paper by Forum reviewers: There were some concerns with the paper. For example, in the introduction the authors state: “Although alcohol consumption in moderation may have beneficial effects regarding cardiovascular disease, it is associated with elevated risks for several cancers including those of the oral cavity, pharynx, larynx, esophagus, and liver (quoting papers by Latino-Martel et al and Rehm et al).” In fact, scientific studies generally show an increase in these cancers only among heavier drinkers, and almost exclusively among those who are also heavy smokers.1,2 In addition, Rehm et al3 have previously shown from Canadian population data that so-called “moderate” drinkers who do not binge drink show a net favorable effect on mortality, whereas when one includes binge drinkers in that category, there is a net adverse effect.3
More importantly, we do not know the reasons why some people in this study (or in most studies) declined their level of drinking. The statement by the authors that quality of life was lower for people who decreased their intake cannot be interpreted as meaning necessarily that the decrease in alcohol was the reason that they had poorer quality of life. Forum reviewer Harvey Finkel comments: “As people age, even disregarding medical obstacles, social interactions generally decrease, which leads to both less stimulation to drink and less opportunity to drink.” It is thus important that the reasons that someone stops drinking, or decreases his or her intake, are taken into account. If the change was due to the development of a serious illness, that may be the cause of subsequent adverse health effects and not the decrease in alcohol itself.
Problems with statistical approach used: There are problems in judging longitudinal effects of alcohol when alcohol strongly affects the baseline value (obtained in middle-age or later in this analysis). In the present study, the quality of life measures at baseline were highest among moderate drinkers. There are statistical problems if estimates of the effects of change in alcohol intake adjust for this baseline value. Peto stated that in prospective studies, the correlation between exposures (e.g., drinking pattern) and outcomes (e.g., quality of life), assessments of outcome during follow up are likely to be the same as the outcome at the end of follow up. As an analogy he uses a race between ‘slow’ and ‘fast’ horses; it is likely that the fast horses will be ahead at the mid-point of the race, and will win the race. Environmental effects on quality of life presumably begin early in life, and if one adjusts for the mid-life value (as done at “baseline” in the present study), you may throw away much of the expected effect.
Further, as described by reviewer, Yuqing Zhang: “In the paper, the alcohol drinking pattern in model 2 was defined by information collected from Cycle 1 to Cycle 3; however, the authors used it to predict the baseline HUI3 at Cycle 1. This seems very odd, as they are using data after baseline to determine the baseline characteristics of subjects. Similarly, they used this variable (i.e., alcohol drinking pattern) to predict the quality of life changes change over Cycle 1 to Cycle 8. Instead, a more appropriate way would be to use alcohol drinking pattern at Cycle 3 as a baseline value and follow changes from Cycle 3 to Cycle 8.”
Zhang continues: “While the effect of pattern of alcohol drinking associated with HUI3 change is not statistical significant, the beta coefficients all suggest a slower decline rate in HUI3 (negative sign) among persistent moderate alcohol drinkers. In any case, it is hard to make the claim that there is a cross-sectional association between drinking pattern and HUI3 but not with rate of change in HUI3. Unless higher HUI3 levels observed at ‘baseline’ (not at birth, but in mid-life) among persistent moderate alcohol drinkers are due to other factors (such as genetic factors), their high level of HUI3 at baseline is presumed to relate, at least to some degree, to their previous drinking pattern (assuming all subjects have the same HUI3 levels when they were born).”
Other concerns by Forum reviewers: Forum reviewer Erik Skovenborg states that the question of association or causation is very difficult when the outcome is health-related quality of life (HRQL). “This study offers a nice try, though, and it is particularly interesting that even among the group of subjects in ‘consistently good health’ throughout the study period, moderate drinkers reported better HRQL at baseline than did non-drinkers, infrequent drinkers and heavy drinkers.”
Skovenborg comments further: “Another interesting result concerns heavy drinking: compared to consistent moderate use of alcohol, persistent heavy use was not associated with a significant greater decline in HUI3 scores and the heavy users were actually doing pretty well in this study. One explanation could be an inadequate definition of heavy alcohol use, in that 50% of the heavy users averaged 14 drinks or fewer per week plus an occasional intake of more than 4 drinks. According to the Danish rules for sensible drinking (up to 21 drinks/week for men and 14 drinks/week for women and not more than 5 drinks per occasion) these participants would have been categorized as moderate drinkers. And a recent Danish study concluded that occasional binges embedded in a moderate drinking pattern are compatible with a good health.5
1. Anantharaman D, Marron M, Lagiou P, Samoli E, Ahrens W, Pohlabeln H, et al. Population
attributable risk of tobacco and alcohol for upper aerodigestive tract cancer. Oral Oncology 2011;47:725–731.
2. Szymańska K, Hung RJ, Wűnsch-Filho V, Eluf-Neto J, Curado MP, Koifman S, Matos E, Menezes A, Fernandez L, Daudt AW, Boffetta P, Brennan P. Alcohol and tobacco, and the risk of cancers of the upper aerodigestive tract in Latin America: a case–control study. Cancer Causes Control (2011) 22:1037–1046. DOI 10.1007/s10552-011-9779-7.
3. Rehm J, Patra J, Taylor B. Harm, benefits, and net effects on mortality of moderate drinking of alcohol among adults in Canada in 2000. Ann Epidemiol 2007;17(Suppl):S81-S86.
4. Peto R. The horse-racing effect. Letter to the Editor; The Lancet (August 29, 1981); pages 467-468.
5. Skov-Ettrup LS, Eliasen M, Ekholm O, Grønbaek M, Tolstrup JS. Binge drinking, drinking frequency, and risk of ischaemic heart disease: A population-based cohort study. Scandinavian
Journal of Public Health 2011;39:880–887.
Data from a nationally representative sample of 5,404 community-dwelling Canadians ages 50 and older at baseline (1994/1995) was used to estimate the effects of alcohol drinking patterns on indices of quality of life, at baseline (when subjects were aged ≥50 years) and after a follow-up period. Health-related quality of life was assessed with the Health Utilities Index Mark 3 (HUI3). The authors report that most participants showed stable alcohol-consumption patterns over 6 years. Persistent regular moderate drinkers had the highest indices of quality of life at baseline, but subsequent changes in HUI3 were similar in all groups except those reporting decreasing alcohol consumption. The investigators conclude that persistent moderate drinkers had higher initial levels of health-related quality of life than persistent nonusers and those in other groups. However, rates of decline over time were similar for all groups except those decreasing their consumption, which had a greater decline in their level of health-related quality of life than persistent moderate users.
While Forum reviewers admired the intent of this study, there were concerns about some of the statistical and epidemiologic aspects. The reasons that some people stopped drinking or decreased their intake were not known; while the authors state that subjects decreasing their alcohol consumption had a more rapid decline in quality of life measures, it is not known what percentage of such subjects decreased alcohol intake due to the development of serious diseases. If this were the case, such diseases may have been more important in their subsequent adverse health than the change in their drinking habits.
Further, the “baseline” quality of life measures in this study were obtained when subjects were aged 50 or older; this baseline value of quality of life was higher in moderate drinkers. However, there are statistical problems if adjustments are made for this when quality of life is assessed subsequently and related to drinking pattern. Peto has described this problem as a “horse-racing effect.” He states that in prospective studies, the correlation between exposures (e.g., drilking pattern) and outcomes (e.g., quality of life) assessments during follow up are likely to be the same as the outcome at the end of follow up. As an analogy he uses a race between ‘slow’ and ‘fast’ horses; it is likely that the fast horses will be ahead at the mid-point of the race as well as at the end. Environmental effects on quality of life begin early in life, and if one adjusts for the mid-life value (as done and referred to as “baseline” in the present study), you may end up disregarding much of the effect of subsequent alcohol intake.
Overall, this study shows a positive relation between moderate alcohol intake and quality of life in middle-aged adults. The effects on the subsequent quality of life as one ages of continued alcohol consumption, or of decreasing intake, remain unclear.
* * *
Comments on this paper were provided by the following members of the International Scientific Forum on Alcohol Research:
Erik Skovenborg, MD, Scandinavian Medical Alcohol Board, Practitioner, Aarhus, Denmark
Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA
Andrew L. Waterhouse, PhD, Marvin Sands Professor, Department of Viticulture and Enology, University of California, Davis; Davis, CA, USA
Yuqing Zhang, MD, DSc, Epidemiology, Boston University School of Medicine, Boston, MA, USA
Helena Conibear, Executive Director, Alcohol in Moderation (AIM), Dorset, United Kingdom
R. Curtis Ellison, MD, Section of Preventive Medicine & Epidemiology, Boston University
School of Medicine, Boston, MA, USA