Abstract
BACKGROUND: Although the association of moderate alcohol consumption with specific disorders, such as cardiovascular disease and cancers, has been well documented, the evidence of the broader impact of alcohol consumption on health-related quality of life is less clear. Our objective was to examine the association of drinking patterns with changes in physical and mental well-being across populations.
METHODS: We conducted a multilevel analysis with multivariate responses in the population-representative FAMILY Cohort in the Hong Kong Special Administrative Region, China, to examine the association between alcohol drinking patterns across 2 waves (2009–2013) (i.e., quitters, initiators, persistent drinkers, persistent former drinkers and lifetime abstainers) and changes in physical and mental well-being (Physical and Mental Component Summary of the 12-Item Short Form Health Survey [SF-12]). Analyses were stratified by sex. We validated findings using a nationally representative cohort in the United States, the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC, 2001–2005).
RESULTS: In the FAMILY Cohort (n = 10 386; median follow-up 2.3 yr), the change in mental well-being was more favourable in female quitters than in lifetime abstainers (β = 1.44, 95% confidence interval [CI] 0.43 to 2.45; mean score change of +2.0 for quitters and +0.02 for lifetime abstainers). This association was validated in the NESARC (n = 31 079; median follow-up 3.1 yr) (β = 0.83, 95% CI 0.08 to 1.58; mean score change of −1.1 for quitters and −1.6 for lifetime abstainers).
INTERPRETATION: The change in mental well-being was more favourable in female quitters, approaching the level of mental well-being of lifetime abstainers within 4 years of quitting in both Chinese and American populations.
Global alcohol consumption is rising and is the leading risk factor for the global burden of disease, and a major preventable risk factor of cardiovascular diseases, gastrointestinal diseases and neuropsychiatric disorders.1–4 However, the risks and benefits of moderate drinking are not clear. Cross-national studies have recommended the low-risk limit for alcohol consumption to be 100 g per week for both men and women,9 which is lower than the safe limits recommended by many countries, including Australia, Canada, Italy, Japan, Spain, the United Kingdom and the United States.10 A recent global study further challenged the idea that there is a safe limit.11,12 Indeed, alcoholic beverages have been classified as a Group 1 carcinogen by the International Agency for Research on Cancer.13 The examination of broader health outcomes related to alcohol consumption, such as health-related quality of life, may therefore be particularly valuable. However, few studies have investigated the impact of moderate drinking on health-related quality of life, and many have often relied on a single time-point measurement of alcohol consumption rather than longitudinal patterns of use.14,15 A randomized controlled trial (RCT) showed that initiating moderate drinking had no effect on quality of life.8 In a longitudinal study, reduced alcohol consumption over time was associated with better mental health–related quality of life.16
Validating associations across cohorts can improve causal inference by exploiting differences in confounding patterns between populations.17–21 Confounding patterns and social norms of alcohol consumption have been shown to differ, in that moderate drinkers in the US often have systematically healthier attributes than abstainers.22 In contrast, alcohol consumption is less normative among the Chinese population, and Chinese moderate drinkers have been shown to have less healthy attributes than abstainers.20,23–26
We sought to examine the longitudinal relation between changes in drinking patterns and changes in physical and mental well-being in the Hong Kong Special Administrative Region, China. We further sought to externally validate our findings in a US population to clarify the relevance of the findings across populations.
Methods
Study design and participants
We selected large population-representative cohorts from Hong Kong (FAMILY Cohort) and the US (National Epidemiologic Survey on Alcohol and Related Conditions [NESARC]) to purposively leverage the differences in social norms. Consistent findings between these 2 distinct populations could provide evidence for the external validity for our associations.19,20 We derived the Hong Kong sample from Waves 1 and 2 of the FAMILY Cohort, which was conducted from 2009 to 2013. The study design has been described in detail elsewhere.27 Participants were recruited based on stratified random sampling from all 18 districts in Hong Kong, with sample sizes proportionate to each of the district populations. For each district, we obtained a random sample of households based on a complete list of living quarters provided by the Census and Statistics Department in Hong Kong. All eligible household members residing in the sampled household were invited to participate. Participants were included if they were aged 18 years and older, were nondrinkers or moderate drinkers (on average ≤ 14 drinks per week [196 g of pure alcohol] for men and ≤ 7 drinks per week [98 g of pure alcohol] for women)28 and completed both Waves 1 and 2. People who reported heavy drinking were excluded because the evidence for adverse impacts of heavy drinking on health-related quality of life is well established.29 The follow-up rate was 67.4% in the FAMILY Cohort.
For external validation, we used data from NESARC (2001–2005), a nationally representative survey of US citizens conducted by the National Institute on Alcohol Abuse and Alcoholism.30,31 Similarly, stratified random sampling was conducted, and adults aged 18 years and older were interviewed face-to-face in household visits. The analytic sample was from the Waves 1 and 2 in 2001 to 2005, with the same inclusion criteria used in the FAMILY Cohort. The follow-up rate was 78.8% in the NESARC Cohort.
Alcohol consumption patterns
In both cohorts, alcohol measurements were available at 2 time points over a 4-year period. Lifetime abstainers were participants who reported that they never had a drink in their lifetime at both waves. Quitters during follow-up were participants who were current drinkers (≥ 1 drink in the last 12 months)32 at Wave 1 and then quit drinking. Initiators during follow-up were nondrinkers (i.e., lifetime abstainers or former drinkers who had ≥ 1 drink in their lifetime but not in the last 12 months)32 who started drinking after Wave 1. Persistent drinkers were participants who were current drinkers at both waves. Persistent former drinkers were participants who were former drinkers (including former heavy drinkers and former moderate drinkers) at Wave 1 who continued to not drink.
Changes in physical and mental well-being
Physical and mental well-being were measured by use of the 12-Item Short Form Health Survey (SF-12) version 2, which comprises the Physical Component Summary (PCS) and Mental Component Summary (MCS).33 The PCS and MCS scores were standardized by use of the mean and standard deviation (SD) of the US general population.33 Both scores range from 0 to 100, with higher scores indicating better health. Outcomes of the current study are the changes in PCS and MCS between the 2 waves (i.e., score at Wave 2 subtracting score at Wave 1). Studies have shown that PCS and MCS scores are valid and reliable indicators of physical and mental health in a variety of population groups, including the Chinese and American general populations.34,35
Statistical analysis
We used multivariate multilevel models with random intercepts to examine the association between drinking patterns and changes in physical and mental well-being (i.e., 2 outcomes) simultaneously (Appendix 1, available at www.cmaj.ca/lookup/suppl/doi:10.1503/cmaj.181583/-/DC1). Models were adjusted for potential confounders, including baseline sociodemographic measures (age, marital status, the highest level of education attained, occupation, monthly household income and ethnicity), body mass index (BMI) and smoking status.16,36,37 Models were then additionally adjusted for other potential confounders, including cardiovascular disease, gastrointestinal disease, liver disease, arthritis and mental disorders, because these could lead to abstaining from drinking and a decline in self-perceived health.16,36,38,39 We used the NESARC cohort to externally validate associations in the FAMILY Cohort. Lifetime abstainers were selected as the reference group, as abstainers constitute the largest proportion in the Chinese general population.24,40–42 We did not include former drinkers in the reference group to avoid biased estimates from the inclusion of “sick quitters,” as former drinkers are more likely to quit drinking because of illness.43–45 All analyses were stratified by sex because of differences in alcohol consumption patterns, sociodemographic factors and alcohol metabolism.15,36,46 Each 1 point difference in PCS and MCS scores could be interpreted as one-tenth of an SD, without the need for additional standardization of the β coefficients.33 We conducted a sensitivity analysis with additional adjustment for physical activity (low, moderate and high, according to International Physical Activity Questionnaire).47,48 Another sensitivity analysis was conducted with adjustment for baseline sociodemographic measures only (without adjustment for BMI, smoking status, self-reported diseases and physical activity). We conducted a complete-case analysis, because the percentage of participants with at least 1 sociodemographic variable missing was only 5%. Additional sensitivity analyses were carried out as detailed in the Supplementary Methods, available in Appendix 2, www.cmaj.ca/lookup/suppl/doi:10.1503/cmaj.181583/-/DC1. All analyses were done using R version 3.4.0 and SAS version 9.4.
Ethics approval
The study was approved by the Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster, and written informed consent was obtained from participants before the surveys.
Results
FAMILY cohort
The sample size was 10 386, after the exclusion of 41 participants with incomplete alcohol use patterns or outcomes. The mean age was 49.3 (SD 17.4) years, and the proportion of men was 44.2%. The total follow-up time was 23 055 person-years, with a median follow-up time of 2.3 years. In the FAMILY Cohort, 63.8% of men were nondrinkers (former drinkers and lifetime abstainers, n = 2931) at Wave 1. Of these, 74.8% remained lifetime abstainers at Wave 2, 20.9% started drinking and 4.3% were persistent former drinkers (Table 1A and 1B). The remaining 36.2% were drinkers (n = 1661) at Wave 1, of which 40.3% quit drinking during the follow-up period.
Most women were current nondrinkers at Wave 1 (n = 5080, 87.7%). Of these, most (89.8%) remained lifetime abstainers at Wave 2, 9.2% started drinking and 1.0% were persistent former drinkers (Table 1A and 1B). Among drinkers (n = 714, 12.3%) at Wave 1, 62.2% quit drinking during the follow-up period.
Men and women who were lifetime abstainers had the highest level of mental well-being at baseline, with mean mental health scores of 54.8 (SD 7.3) in men and 53.6 (SD 8.2) in women (Table 1A and 1B). Persistent drinkers had the highest level of physical well-being at baseline.
Drinking patterns and changes in physical and mental well-being
Female quitters had a greater improvement in mental well-being relative to lifetime abstainers (β = 1.44, 95% confidence interval [CI] 0.43 to 2.45) (Table 2), which indicated that the change in mental well-being in female quitters was 1.44 points higher than that in female abstainers. Male persistent former drinkers had a greater improvement in mental well-being (β = 2.10, 95% CI 0.30 to 3.89) (Table 2), which indicated that the change in mental well-being in male persistent former drinkers was 2.10 points higher than that in male abstainers. Sensitivity analyses with different confounder adjustment showed consistent results (Appendix 2, Supplementary Table 1). Sensitivity analyses with multilevel multiple imputation and the exclusion of binge drinkers showed similar findings (results not shown).
NESARC cohort: external validation
In the NESARC cohort, the sample size was 31 079, after the exclusion of 286 participants with missing outcomes. The mean age was 46.3 (SD 17.5) years, and the proportion of men was 40.6%. The total follow-up time was 94 798 person-years, with a median follow-up time of 3.1 years. Baseline characteristics of the validation data in the NESARC cohort are shown in Appendix 2, Supplementary Table 2. Men and women who were lifetime abstainers had the highest level of mental well-being at baseline, with mean scores of 54.0 (SD 9.7) in men and 52.0 (SD 10.6) in women.
We externally validated associations detected in the FAMILY Cohort using the NESARC. Results were consistent in that female quitters had a more favourable change in mental well-being (β = 0.83, 95% CI 0.08 to 1.58 in model 1; β = 0.83, 95% CI 0.07 to 1.58 in model 2). However, the association between male persistent former drinkers and change in mental well-being was not validated (p > 0.05). Analyses were repeated with inverse probability weighting and poststratification weighting in both cohorts, which yielded similar findings (results not shown). We also reverted the use of both samples by conducting the primary analysis with the NESARC cohort and the external validation with the FAMILY Cohort, which led to the same validated findings. The mean scores of mental well-being among female quitters and lifetime abstainers at Waves 1 and 2 are shown in Appendix 2, Supplementary Table 3. In the FAMILY Cohort, female quitters had a greater increase of the score than lifetime abstainers, with a mean score change of +2.0 for quitters and +0.02 for lifetime abstainers; in the NESARC cohort, female quitters showed a smaller decline in mental well-being, with a mean score change of −1.1 for quitters and −1.6 for lifetime abstainers.
Interpretation
We found that lifetime alcohol abstainers reported the highest level of mental well-being. Women who quit drinking were found to have a greater improvement in mental well-being than lifetime abstainers. This association was found in a Hong Kong cohort as well as in a US cohort. On average, the mental well-being of female quitters approached the level of lifetime abstainers within a 4-year period in both cohorts (median 2.3 years of follow-up in the FAMILY Cohort and 3.1 years in the NESARC cohort). In contrast, initiation and persistent moderate drinking for 4 years were not associated with better mental or physical well-being. These results remained consistent after adjustment for sociodemographic characteristics, BMI, smoking status, self-reported diseases and physical activity.
Interventional studies supporting the positive impact of alcohol cessation on mental well-being have been mostly based on individuals with alcohol dependence;49–51 evidence for the positive effect of brief alcohol interventions for people who drink moderately is emerging.52 Cross-sectional studies have suggested a positive association between moderate drinking and mental well-being. 53,54 However, our findings are consistent with those of a previous smaller longitudinal study that showed alcohol reduction to be associated with better mental well-being and an RCT that showed no effect of moderate drinking on quality of life.8,16 Cross-national studies have also challenged the idea that moderate drinking could have health benefits.9,11,12 We found that quitting alcohol was associated with a more favourable change in mental well-being among women. The explanation for our findings and the underlying mechanism are not clear. It is possible that alcohol-related neurotoxicity reverses following abstinence.55–57 Alcohol cessation may also reduce stressful life events, such as conflict within family, difficulties in employment and legal troubles, resulting in improved mental well-being.58,59 It is also possible that improved mental well-being may result from the psychological benefits of “giving up” per se rather than an effect of alcohol, as most Chinese women in Hong Kong use alcohol fewer than 4 times per month, which may not have a physiologic effect.60
Limitations
Our study is subject to the following limitations. First, prospective cohort studies are susceptible to selection bias and attrition bias. The FAMILY Cohort enrolled complete households in which all adult members agreed to participate, thus potentially selecting better-functioning family units. However, the “healthy volunteer effect” could still have occurred if individuals were sampled instead.27 The application of censoring weights in both cohorts did not appreciably alter results, suggesting that attrition had little influence on our results. Second, as alcohol consumption pattern was self-reported, underreporting and some degree of misclassification are possible, which should be considered when interpreting the findings.44 Current drinkers may claim to be nondrinkers because of social desirability bias.36 However, the Cohen’s w effect size for differences in alcohol consumption between high and low social desirability groups was small in our study.61,62 Further, there has been a shift in the social norms of alcohol consumption and increased societal acceptance of social drinking among Hong Kong Chinese.60,63 Misclassification of former drinkers as lifetime abstainers has been documented in previous studies.64,65 In the FAMILY Cohort, 9.6% of the women who reported never having a drink at Wave 2 had reported drinking at or before Wave 1, which is substantially lower than in other cohorts and surveys in the US and UK.64,65 Third, the relatively short period of follow-up does not allow for capturing the dynamic tracks of drinking patterns and health status across the entire life course.15 However, we believe the finding of this study may inform future prospective studies with longer follow-up. Fourth, whether there is a sex difference in the link between alcohol use patterns and mental well-being remains unclear, and assessment of such a difference would require interaction analyses and larger samples. 66 Fifth, missing data must be considered; however, the application of multilevel multiple imputation yielded similar findings, suggesting that the missing data had little influence on our findings.
Conclusion
Our findings, that lifetime alcohol abstainers report the highest level of mental well-being and quitting alcohol improves mental well-being among women, suggest caution in recommending that moderate drinking could improve health-related quality of life.37,67 Instead, quitting drinking may be associated with a more favourable change in mental well-being, approaching the level of lifetime abstainers. This may be analogous to smoking cessation, which results in the recovery of health outcomes to the level of lifetime nonsmokers.68–70 Global alcohol consumption is expected to continue to increase unless effective strategies are employed.1,71 Further studies are needed to establish clearly the impact of alcohol use on mental and physical well-being before alcohol is recommended as part of a healthy lifestyle.
Acknowledgements
The authors are most grateful for the participants’ support in the FAMILY Cohort and the NESARC cohort. NESARC is funded by the National Institute on Alcohol Abuse and Alcoholism, with supplemental support from the National Institute on Drug Abuse. The authors thank Tiffany Leung for her help with the manuscript formatting.
Footnotes
Competing interests: None declared.
This article has been peer reviewed.
Contributors: Xiaoxin Yao, Michael Ni, Gabriel Leung and Herbert Pang conceived the study and planned the analyses. Xiaoxin Yao, Michael Ni and Herbert Pang carried out the study and drafted the manuscript. Xiaoxin Yao and Herbert Pang analyzed the data. All authors contributed to the interpretation of the data and revision of the paper. All authors gave final approval of the version to be published and agreed to be accountable for all aspects of the work.
Funding: The Hong Kong Jockey Club Charities Trust was the sole funder of the FAMILY Project from 2007 to 2014. The funding source had no role in the design of the study; the collection, analysis or interpretation of the data; the writing of the manuscript; or the decision to submit the manuscript for publication.
Data sharing: The data that support the findings of this study are archived at the School of Public Health, The University of Hong Kong. The FAMILY Cohort is set up as a supported-access resource; an application is required to access the data. Further information is available at https://familycohort.sph.hku.hk/en/data_access. The data for the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) cohort is available through the National Institute on Alcohol Abuse and Alcoholism.
- Accepted June 1, 2019.