Critique 178: The association of alcohol intake with health-related quality of life — 12 January 2016
Schrieks IC, Wei MY, Rimm EB, Okereke OI, Kawachi I, Hendriks HFJ, Mukamal KJ. Bidirectional associations between alcohol consumption and health-related quality of life amongst young and middle-aged women. J Intern Med 2015, pre-publication. doi:10.1111/joim.12453.
Background. Evidence from cross-sectional studies has suggested a positive association between moderate alcohol consumption and health-related quality of life but prospective data remain scarce.
Objectives. To examine the bidirectional relationships between alcohol consumption and health-related quality of life using a longitudinal study design.
Methods. A total of 92,448 participants of the Nurses’ Health Study II reported their alcohol consumption (in 1991, 1995, 1999 and 2003) and health-related quality of life (in 1993, 1997 and 2001). Using generalized estimating equations, we modelled the physical and mental component summary (PCS and MCS) scores as a function of alcohol consumption 2 years earlier (n = 88,363) and vice versa (n = 84,621).
Results. Greater alcohol consumption was associated with better PCS scores 2 years later in a dose–response manner up to ~1 serving daily [mean difference (b) = 0.67 ± 0.06 PCS units, for moderate versus infrequent drinkers]. After adjustment for previous PCS, a similar but attenuated pattern was observed (b = 0.33 ± 0.07). Moderate alcohol consumption was not related to MCS, whereas moderate-to-heavy alcohol consumption was associated with lower MCS scores (b = -0.34 ± 0.15). Higher PCS scores were associated with greater alcohol consumption 2 years later, also after adjustment for previous alcohol consumption (b = 0.53± 0.05 g day -1). MCS was not associated with alcohol consumption 2 years later.
Conclusion. Amongst young and middle-aged women, moderate alcohol intake was associated with a small improvement in physical health-related quality of life 2 years later and vice versa. Moderate alcohol consumption was not associated with mental health-related quality of life in either direction.
Many studies have shown that moderate drinkers tend to have higher ratings of their quality of life (QOL). The directionality of this association has been difficult to ascertain: does moderate drinking improve someone’s QOL, or do people with higher QOL to begin with tend to drink alcohol moderately? The present study attempts to provide data designed to evaluate the directionality of these associations for both physical aspects of QOL as well as mental aspects.
This analysis has a number of merits. It is from a very large cohort (the Nurses’ Health Study II) with excellent and repeated measurements of both alcohol consumption and indices of QOL. It permits, for the first time, a valid assessment of both physical and mental aspects of quality of life as well as the directionality of such an association: subjects were repeatedly assessed for the effects of alcohol intake on subsequent QOL as well as the effects of a given QOL measure on subsequent alcohol intake. The Nurses’ Health Study II has excellent ascertainment of alcohol intake from repeated measures, and used current state-of-the-art instruments for assessing QOL.
The key findings of this study are a clear positive association between alcohol and subsequent indices of physical QOL indices, with the poorest outcomes in abstainers and former drinkers. In spline analyses, the favorable effect appears to be up to about 2 drinks/day (although the number of heavier drinkers was small, making it difficult to judge effects of truly heavy alcohol intake). Further, there was a clear linear relation between QOL and subsequent alcohol intake, even when previous drinking included in the analysis. Somewhat surprising were the finding of little relation between alcohol and mental aspects of QOL, and even an apparent deleterious effect at the highest category of drinking. Thus, the study shows that physical aspects of QOL are positively associated with alcohol consumption, while mental aspects show no relation.
Specific comments on paper by Forum members: Forum members concluded that this was a well conducted study (although, as noted below, there were some questions raised as to bias in the estimates of effect). As stated by reviewer Finkel, “We are familiar with the previous reliable work of much of this group, and both the study material and its handling appear without fault. However, I still cannot but wonder whether the intrinsic character of the study group (productive professionals) plays a role, and how much of the association between alcohol and QOL is in one direction or the other, with or without causation.” He added: “The lack of effect on mental health is counterintuitive, especially because several published studies of older subjects did demonstrate a relationship.” Stated reviewer Waterhouse: “The study looks very solid and provides some new insight into the effects of alcohol consumption.” Forum member Sluik considered that “this study particularly stands out in the assessment of the bidirectionality of the association.”
Forum member Barrett-Connor noted: I agree that this very large prospective study of young and middle-aged women nurses’ alcohol use and health-related quality of life is well done but, as stated by Finkel, the nature of the study group (productive professionals) plays an important role in their behavior and responses. Shift work, which is almost a universal aspect of the nursing profession, may also affect not only nurses’ alcohol use but also what and when they eat and drink. Differences in diet are surely important but not addressed. Overall, this paper strongly supports that modest alcohol intake in young and mid-life nurses is beneficial for health-related quality of life, not likely to be explained completely by confounding.”
Forum member Mattivi stated: “This is an interesting study with robust conclusions. With the food frequency questionnaires collecting information on the consumption of regular beer, light beer, white wine, red wine and liquor with nine frequency responses and available for as many as 92,448 participants, it would have been worth a try a separate evaluation of the effects of the different beverages. After all, people do not drink ‘alcohol’ and it would have been interesting to see if the specific beverage consumed plays a role.” Reviewer Lanzmann-Petithory agreed: “A nice study, but the authors should have reported relations of PCS and MCS separately for each of their categories of alcoholic beverage: beer, light beer, white wine, red wine, and liquor.”
Reviewer De Gaetano noted: “This is a good paper, but I share the regret of others about the lack of a separate analysis of different alcoholic beverages. In a paper we published a couple of years ago on Moli-sani cross-sectional data, the results suggested that better adherence to a Mediterranean Diet was associated with a better health-related QOL (Bonaccio et al). We found that the association was stronger with mental than with physical health. Dietary total antioxidant and fibre content independently explained the observed relationship. We did not analyse alcohol consumption separately from the total Mediterranean Diet. A working hypothesis might be that alcohol would improve mental QOL if consumed in the context of a Mediterranean Diet.”
Reviewer Skovenborg had a number of comments: “This is an area of research brimming over with possible confounding factors and where it is notoriously difficult to establish causation – and even difficult to establish the direction of the causality. There may be a difference between drinking cultures. For example, in a US population the highest odds of reporting above-average health status was restricted to current moderate drinkers (French & Zavala). In a Spanish population with a Mediterranean drinking culture the higher the consumption of total alcohol, wine and beer, the lower the prevalence of suboptimal health (Guallar-Castillon et al). These Spanish authors concluded: ‘The higher the consumption of total alcohol, wine and beer, the lower the prevalence of suboptimal health. These results differ from those obtained in several Nordic countries, where a J-shaped relation has been observed for total alcohol and wine, and suggest that the relation between alcohol consumption and subjective health may be different in Mediterranean countries.’”
Skovenborg continued: “Among Japanese workers, subjects who consumed 25-36 or 49 g alcohol or more per day had a significantly lower risk of self-rated ill-health compared with lifelong abstainers. Workers with those levels of alcohol consumption were described by the authors as ‘moderate drinkers’ (Sakuri et al). In response to a question about beverage-specific effects, in a Danish cross-sectional survey a light to moderate wine intake was related to good self-perceived health, whereas this was not the case for beer and spirits (Grønbæk et al). In a Swedish interview study only consumption of wine was associated with a decreased odds ratio for poor, self-reported health status as compared with non-users. (Theobald et al). In a British study of the 1958 birth cohort with 33-year follow-up, higher rates of self-rated ill health were observed among non-drinkers and heavy drinkers than among moderate drinkers at age 33. The U-shaped relation persisted in repeated analyses excluding heavy or problem drinkers at age 23 (Power et al).”
Reviewer Thelle thought that “It is odd that mental health is ‘unaffected’ by alcohol. In our own studies we used SF 36 in a cross-sectional analysis of the Oslo Study and found a negative association between mental health and smoking. Even if smoking was adjusted for in the present paper, one might suspect residual confounding, i.e. alcohol ‘improves’ mental health, smoking does not, net result is nil. Of course, this is just speculative.”
A potential statistical problem in the analysis: Despite the generally favorable opinions of Forum members on this paper, reviewers Zhang and Ellison had some concerns about the analysis: The authors stated, “To account for the effect of previous outcomes on current outcome variables, we further adjusted for baseline outcomes (HRQOL in the first analysis and alcohol consumption in the second analysis) in each model. For example, when we used alcohol consumption in 1995 to predict HRQOL scores in 1997, we additionally adjusted for HRQOL scores in 1993.”
Zhang commented: “This approach is more valid if the exposure, here alcohol consumption, is an incident exposure (i.e., the start of alcohol drinking) rather than a prevalent exposure. However, if the exposure is a prevalent one, and alcohol consumption remains roughly the same over the time period of a study, any effect of subsequent alcohol on QOL may be subsumed by the effects shown at baseline. The net effect of adjustment for baseline alcohol may attenuate (or remove) the effect of the follow-up exposure on the outcome variable.
“Similarly, for a subject whose QOL did not change between baseline and follow up, then adjustment for previous QOL when assessing alcohol’s effects on subsequent QOL can provide a null association. This is because the previous (baseline) QOL explains 100% of the variability of the subsequent QOL.”
This phenomenon has been well described by Glymour et al, who stated: “In research on the determinants of change in health status, a crucial analytic decision is whether to adjust for baseline health status . . . when exposures are associated with baseline health status, this bias can arise if change in health status preceded baseline assessment or if the dependent variable measurement is unreliable or unstable. In some cases, change-score analyses without baseline adjustment provide unbiased causal effect estimates when baseline-adjusted estimates are biased.”
In a subsequent discussion on this phenomenon, Glymour wrote: “Scientific interest frequently focuses on how factors that influence disease onset subsequently affect disease progression. In this commentary, the author discusses four sources of bias that arise in such work.” Using a paper relating apoliprotein E*4 to cognitive impairment, Glymour describes “four phenomena that can lead to spurious (noncausal) associations between apolipoprotein E *E4 status and rate of progression of cognitive impairments: beginning observations in the middle of a developing pathologic process, survivor bias, uncertainty in the timing of disease diagnosis, and nonlinear disease progression trajectories.”
The authors of the present paper were aware of this potential problem in that they presented results both unadjusted and adjusted for previous values of the exposure variables. And, as expected, the estimated effects of both alcohol (on QOL) and QOL (on alcohol consumption} were attenuated when adjustments for baseline values were added. However, without knowing the exact analytic procedures used, it is difficult to interpret the results, as the estimated effect from adjustment could appear to be null, or even opposite from the baseline association.
Reviewer Zhang concludes: “From the data presented in the present paper by Schrieks et al, it is not possible to ascertain the degree to which such potential problems as discussed by Glymour were addressed in these analyses. The net result is that the final estimates of effect given for alcohol and QOL may remain biased.”
References from Forum review
Bonaccio M, Di Castelnuovo A, Bonanni A, Costanzo S, De Lucia F, Pounis G, Zito F, Donati MB, de Gaetano G, Iacoviello L; Moli-sani project Investigators. Adherence to a Mediterranean diet is associated with a better health-related quality of life: a possible role of high dietary antioxidant content. BMJ Open 2013;3. pii: e003003. doi: 10.1136/bmjopen-2013-003003.
French MT, Zavala SK. The health benefits of moderate drinking revisited: alcohol use and self-reported health status. Am J Health Promotion 2007;21:484-491.
Glymour MM, Weuve J, Berkman LF, Kawachi I, Robins JM. When Is Baseline Adjustment Useful in Analyses of Change? An Example with Education and Cognitive Change. Am J Epidemiol 2005;162:267–278.
Glymour MM. Invited commentary: when bad genes look good – APOE*E4, cognitive decline, and diagnostic thresholds. Am J Epidemiol 2007;165:1239-1246.
Grønbæk M, Mortensen EL, Mygind K, Andersen AT, Becker U, Gluud C, Sørensen TIA. Beer, wine, spirits and subjective health. J Epidemiol Community Health 1999;53:721-724.
Guallar-Castillon P, Rodriguez-Artalejo F, Diez Ganan LD, Banegas Banegas JR, Lafuente Urdinguio PL, Herruzo Cabrera H. Consumption of alcoholic beverages and subjective health in Spain. J Epidemiol Community Health 2001;55:648-652.
Power C, Rodgers B, Hope S. U-shaped relation for alcohol consumption and health in early adulthood and implications for mortality. Lancet 1998;352:877.
Sakurai Y, Hattori N, Kondo T, Teruya K, Shimada N, Honjo S, Umeda T, Muto T, Takemura Y, Todoroki I, Nakamura K. Association between alcohol intake and subjective health: the Sotetsu Study. Keio J Med 1999;48:147-150.
Theobald H, Johansson S-E, Engfeldt P. Influence of different types of alcoholic beverages on self-reported health status. Alcohol & Alcoholism 2003;38:583-588.
Many studies have shown that moderate alcohol drinkers tend to have higher ratings of their quality of life (QOL) than non-drinkers. The directionality of this association has been difficult to ascertain: does moderate drinking improve someone’s QOL, or do people with higher QOL to begin with tend to drink alcohol moderately? The present study attempts to provide data designed to evaluate the directionality of these associations for both physical aspects of QOL as well as mental aspects.
Forum reviewers considered this to be a very well-done study from the Nurses’ Health Study II, with excellent and repeated measurements of both alcohol consumption and indices of QOL. It deals with the assessment of both physical and mental aspects of quality of life as well as the directionality of such an association: subjects were repeatedly assessed for the effects of alcohol intake on subsequent QOL as well as the effects of a given QOL measure on subsequent alcohol intake.
The key reported findings of this study were a positive association between alcohol and subsequent indices of physical QOL indices, with the poorest outcomes in abstainers and former drinkers; in spline analyses, the favorable effect appears to be up to about 2 drinks/day. Subjects with higher physical QOL also appeared to consume more alcohol in subsequent assessments. Somewhat surprising was the finding of little relation between alcohol and mental aspects of QOL, and even an apparent deleterious effect at the highest category of drinking. Thus, the study reports that physical aspects of QOL are positively associated with alcohol consumption, while mental aspects show little relation.
There was some concern by some Forum members about the methods used by the authors in adjusting, or not adjusting, for previous exposures (both alcohol and indices of QOL). Given that both baseline alcohol intake and QOL measures may tend to remain rather static over the period of the study (this “prevalent” exposures), adjusting for the effects of baseline values could attenuate any effect seen on subsequent outcomes. For example, if alcohol consumption indeed affects QOL and consumption remains roughly the same over the time period of a study, any effect of subsequent alcohol on QOL may be subsumed by the effects shown at baseline. In that situation, the net effect of adjustment for baseline alcohol may attenuate (or remove) the effect of the follow-up exposure on the outcome variable. The authors of this paper were aware of this problem, and attempted to deal with it. However, from the data presented, it is not possible to ascertain the specific analytic procedures used, and the final estimates of effect given for alcohol and QOL may still be biased.
Overall, the Forum considers this to be an important contribution to our understanding of how alcohol consumption may affect quality of life, and how quality of life affects alcohol consumption. As interpreted by the authors, physical aspects of QOL may be positively affected by alcohol consumption, but there is little effect on mental aspects of QOL.
Comments on this critique by the International Scientific Forum on Alcohol Research have been provided by the following members:
Yuqing Zhang, MD, DSc, Clinical Epidemiology, Boston University School of Medicine, Boston, MA, USA
Andrew L. Waterhouse, PhD, Department of Viticulture and Enology, University of California, Davis, USA
David Van Velden, MD, Dept. of Pathology, Stellenbosch University, Stellenbosch, South Africa
Dag S. Thelle, MD, PhD, Senior Professor of Cardiovascular Epidemiology and Prevention, University of Gothenburg, Sweden; Senior Professor of Quantitative Medicine at the University of Oslo, Norway
Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway
Diewertje Sluik, DrPH, Division of Human Nutrition, Wageningen University, NL.
Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark
Fulvio Mattivi, MSc, Head of the Department of Food Quality and Nutrition, Research and Innovation Centre, Fondazione Edmund Mach, in San Michele all’Adige, Italy
Dominique Lanzmann-Petithory,MD, PhD, Nutrition/Cardiology, Praticien Hospitalier Hôpital Emile Roux, Paris, France
Maritha J. Kotze, PhD, Human Genetics, Dept of Pathology, University of Stellenbosch, Tygerberg, South Africa.
Ulrich Keil, MD, PhD, Professor Emeritus, Institute of Epidemiology & Social Medicine, University of Muenster, Germany
Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy
Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA
R. Curtis Ellison, MD, Professor of Medicine & Public Health, Boston University School of Medicine, Boston, MA, USA
Elizabeth Barrett-Connor, MD, Distinguished Professor, Division of Epidemiology, Department of Family Medicine and Public Health and Department of Medicine, University of California, San Diego, La Jolla, CA USA