Critique 280 – Reducing alcohol use to prevent cancer deaths: estimated effects among U.S. adults
Authors
Esser MB; Sherk A; Liu Y; Henley SJ; Naimi TS
Citation
Am J Prev Med 2024 66(4):725−729. doi.org/10.1016/j.amepre.2023.12.003
Author’s Abstract
Introduction The Dietary Guidelines for Americans, 2020−2025 recommends non-drinking or no more than 2 drinks for men or 1 drink for women in a day. However, even at lower levels, alcohol use increases the risk for certain cancers. This study estimated mean annual alcohol-attributable cancer deaths and the number of cancer deaths that could potentially be prevented if all U.S. adults who drank in excess of the Dietary Guidelines had instead consumed alcohol to correspond with typical consumption of those who drink within the recommended limits.
Methods Among U.S. residents aged ≥20 years, mean annual alcohol-attributable cancer deaths during 2020−2021 that could have been prevented with hypothetical reductions in alcohol use were estimated. Mean daily alcohol consumption prevalence estimates from the 2020−2021 Behavioral Risk Factor Surveillance System, adjusted to per capita alcohol sales to address underreporting of drinking, were applied to relative risks to calculate population-attributable fractions for cancers that can occur from drinking alcohol. Analyses were conducted during February−April 2023.
Results In the U.S., an estimated 20,216 cancer deaths were alcohol-attributable/year during 2020 −2021 (men: 14,562 [72.0%]; women:5,654 [28.0%]). Approximately 16,800 deaths (83% of alcohol- attributable cancer deaths, 2.8% of all cancer deaths) could have been prevented/year if adults who drank alcohol in excess of the Dietary Guidelines had instead reduced their consumption to ≤2 drinks/day for men or ≤1 drink/day for women. Approximately 650 additional deaths could have been prevented annually if men consumed 1 drink/day, instead of 2.
Forum Summary
Esser et al., 2024 investigate the effect of quantities of alcohol consumed and the risk of alcohol-attributable cancers in the U.S.A. Data for a range of cancer types and not simply alcohol-related cancers is used, and potential underreporting in the Behavioral Risk Factor Surveillance System is addressed by adjustment with per capita alcohol sales data. Their hypothetical and broad-brush approach has likely overestimated the risks and calculated reduction in risk from reduced consumption and hence promoted poor alcohol public health policy. The conclusions drawn in this U.S. study should not be translated to other world regions as the relationship between alcohol consumption and cancer risk is likely to vary across world regions.
Forum Comments
Background
Cancer is an important leading cause of death in the world. In the U.S., 3.2 million people died in 2022, of which some 608 thousand of cancer being the second leading cause of death[1]. Fortunately, age-adjusted death rates are decreasing in general leading to some 184 thousand less deaths in 2022. Also, for cancer, the age-adjusted death rate decreased by 2.9% from 2021 to 2022.
Since cancer is such a prevalent and drastic disease, ISFAR has produced various critiques on alcohol consumption and cancer risk over the years. In critique 270 and critique 265, we recently discussed the role of alcohol consumption in the prognosis and survival of breast cancer survivors. We concluded that the positive association between alcohol consumption and breast cancer incidence is well-established. At the same time, there is no positive association between pre-diagnosis and post-diagnosis alcohol consumption and breast cancer recurrence and breast cancer-specific mortality.
Cancer is not one disease but a collection of related diseases, which means that each type of cancer has its distinct and different characteristics, risk factors, causes, and treatments. Cancer is just a name for cells rapidly dividing without regard to the body. Each cancer behaves differently, and even cancers from the same site of origin, like breast cancer, can be radically different depending on their unique molecular profiles.
Alcohol has been registered as a group 1 carcinogen by the WHO International Agency for Research on Cancer (IARC), which means there was sufficient evidence from animal studies and epidemiological associations to conclude that drinking alcoholic beverages is carcinogenic for humans. IARC states that alcohol consumption causes cancers of the oral cavity, pharynx, larynx, oesophagus, colorectum, liver (hepatocellular carcinoma), and female breast. These seven types of cancer are referred to as cancers causally related to alcohol or alcohol-attributable cancers. The evidence that alcohol consumption causes other cancer types is inconclusive (Sarich et al. 2022). Furthermore, there are also inverse or negative associations between alcohol consumption and certain cancers (Bagnardi et al. 2015).
Mechanisms of action have been proposed for cancers for instance for breast cancer by disturbing hormone homeostasis or for oral cancers by exposing tissues to high concentrations of the metabolite acetaldehyde possibly in combination with other carcinogens derived from smoking.
Furthermore, cancer is a multifactorial disease, which means that various factors contribute to the emergence of cancer of which alcohol may be one of several combined. Most cancers are primarily attributable to genetic and environmental factors (60+%)[2] and not to lifestyle factors (Anand et al. 2008). Of the lifestyle factors, the American Association for Cancer Research indicated that alcohol is the third contributing factor after smoking (contributing to 19% of cancers), obesity/overweight (contributing 7%), and alcohol (6%)[3].
The estimated contribution of about 6% is for alcohol consumption in general, disregarding quantities and drinking patterns. The current paper by Esser et al., 2024 is interesting since it investigates the effect of quantities drunk and considers the underreporting of alcohol consumption.
Critique
This paper estimated the effects of reducing alcohol consumption to prevent cancer deaths. The authors from the Centers for Disease Control and Prevention (CDC) have published various other papers on alcohol consumption and cancer prevention. The study consisted of a two-step approach. The first step was to use the national cancer mortality data from the CDC’s health statistics system for specific types of cancer and use these data to calculate population attributable fraction (PAF) to estimate cancer deaths due to alcohol consumption for both genders and at low, medium and high mean daily alcohol consumption levels. The second step was to correct for underreporting of alcohol consumption which is considerable in alcohol-disease epidemiology.
The authors choose to not only include all cancers identified by IARC as causally related to alcohol consumption but also include pancreatic, prostate, and stomach cancer. Of these, the CDC’s Alcohol-related Disease Impact (ARDI) application indicates that these specific cancer deaths only apply to those consuming high levels of alcohol. This may have led to a small overestimation of cancer deaths attributable to alcohol consumption at the lower ranges. Also, the cancers that are negatively associated with alcohol consumption, mainly in light to moderate consumption levels, have not been included in the analysis. They include lung cancer (Fehringer et al., 2017), renal cancer (Xu et al., 2015), non-Hodgkin lymphoma (Tramacere et al., 2012b), Hodgkin lymphoma (Tramacere et al., 2012a) and thyroid cancer (Hong et al., 2017).
Omitting negatively associated cancers and focusing on positively associated cancers only may contribute to an overestimation of cancer deaths attributable to alcohol consumption, specifically in the lower consumption ranges. A similar study evaluating the contribution of low and moderate alcohol consumption to the most common 20 cancer types in Europe (Hendriks & Calame, 2018) including both positively and negatively associated cancer types, suggests that up to 15 grams of alcohol per day, overall cancer risk would not increase. In contrast, the risk increased minimally at consumption levels up to 30 g per day.
In general, there are several limitations to the approach of applying population-attributable fractions in generating these estimates. Firstly, there is a high heterogeneity across studies, and it may be difficult to select the relevant relative risks for the entire U.S. population. Also confounding by other lifestyle factors interacting with alcohol consumption such as smoking, physical activity, and body weight cannot be ruled out. Providing estimates with confidence intervals would have been more informative and allowed statistical analysis. Possibly, the estimated number of cancer deaths attributable to alcohol consumption may not have reached statistical significance in the lower alcohol consumption categories as suggested by Rumgay et al. (2021).
Underreporting plays an important role in the associations between alcohol consumption and disease outcomes, as in cancer death risk. The authors have tried to tackle this problem by increasing the amount of alcohol reported as being consumed, by adjusting the consumption level up to 73% of the per capita alcohol sales (Esser et al., 2022). For the methodology involved, the authors refer to a paper describing six different methods to correct for underreporting. Although it is unclear which of the six methods has been used in this specific 2024 study, the overall impression of the recalculation methods in the 2022 study is that fewer people are in the low drinking category and more people are in the high drinking category. This may reflect the drinking population’s composition better, but the risk ratios used to calculate the PAR were not generated with these hypothetical populations. This may have generated an artificially higher PAR, which may not be representative of the associations studied.
Also, the authors assume that all underreporting would be similar throughout drinking categories as have Livingston et al. (2015) and Stockwell et al. (2014) previously. Other studies have indicated, however, that underreporting may vary between persons reporting to drink moderately (Klatsky et al., 2014). Klatsky et al. (2024) indicated that those moderate drinkers suspected of underreporting had increased risks for cancer, whereas those moderate drinkers not suspected of underreporting had not. The authors concluded that the apparent increased risk of cancer among light to moderate drinkers may be substantially due to underreporting of intake. This was also suggested by Boniface et al. (2014) who concluded that heavy drinking and non-routine drinking patterns may be associated with greater under-reporting of alcohol consumption.
All in all, Esser et al. (2024) suggested that reducing alcohol consumption to levels that are in the light to moderate range will reduce cancer deaths, which corresponds to the notion that alcohol consumption and cancer deaths are mainly associated with high levels of alcohol consumption. This is indicative of a non-linear dose-response relationship, where generally high levels of alcohol consumption are linked to greater cancer risk (Jun et al. 2023). Furthermore, the cancer risk from light alcohol consumption appears to vary across cancer types (Jun et al. 2023). Even when reducing alcohol consumption from two to one glass per day, cancer deaths for men will be reduced by a mere 650 deaths (some 3% of alcohol-attributable cancer deaths), consistent with the suggestion of Cao et al. (2015) that light to moderate drinking is associated with minimally increased risk of overall cancer.
It is recognized, however, that even 3% of cancer deaths can be considered important individually and societally. However, the scenarios developed by Esser et al. (2024) resulted in estimates that may or may not be valid without indication of confidence intervals and with a possibility of over-estimation of cancer death risks. The results and conclusions drawn in this U.S. study should not be translated to other world regions as the relationship between alcohol consumption and cancer risk is likely to vary across world regions (Bagnardi et al. 2015). This variation is due in part to regional variation in gene polymorphisms that relate to the metabolism of alcohol (Li et al. 2009).
Furthermore, Esser et al. (2024) conclude that “implementing evidence-based alcohol policies (e.g., increasing alcohol taxes, regulating alcohol outlet density) to decrease drinking could reduce alcohol-attributable cancers, complementing clinical interventions”, yet there is no available peer-reviewed data to support that these alcohol policies reduce alcohol consumption and importantly reduce the incidence of and death from cancer.
Specific Comments from Forum Members
Forum member Skovenborg comments “this study is interesting because of the steps leading up to its conclusions.”
The first step was a meta-analysis of case-control studies and cohort studies that reported findings expressed as odds ratios, relative risks or hazard ratios for light drinkers (≤ 12.5 g ethanol, ≤ 1 drink) versus non-drinkers. However, Bagnardi et al. (2013) list numerous limitations as follows:
• “The first one is that heterogeneity across studies reporting on esophageal, breast and liver cancer was high (and unexplained) therefore… our pooled estimates should be interpreted with caution”.
• “A second issue is that we could not investigate the role of different drinking patterns in modifying the effect of the total amount of alcohol consumed. In fact, the great majority of studies on the alcohol–cancer association reported information on the total amount of alcohol consumed during a period that includes both drinking and non-drinking days.” As most studies focus on the quantity of alcohol consumed during a period, often a week, the participant with a weekly intake of e.g. 84 g ethanol may either be a regular daily drinker consuming <12.5 g ethanol/day = a light drinker or a weekend drinking consuming e.g. 5 drinks Saturday evening and 1 drink occasionally two times a week is a binge drinker. The obvious conclusion is that the proposed light drinker participant is an assumption made by the authors.
• “A third issue is the possible interaction effect between alcohol consumption and tobacco smoking on the development of cancer. A simple yet effective way to clarify whether alcohol is an independent risk factor for cancer is to stratify the investigation by smoking status, but only a small number of studies reported the effect of light drinking in different smoking strata.”
• “Finally, an under-reporting of alcohol consumption in drinkers may partly explain the association with light alcohol drinking. In fact, alcohol consumption might be systematically underreported by both cases and controls (non-differential underreporting). This would lead to an overestimation of the RR for low doses.” According to a study by Klatsky et al. (2014) (ref) “the apparent increased risk of cancer among light-moderate drinkers may be substantially due to underreporting of intake.”
The most difficult aspect and second step of the study is that Esser et al. (2024) estimate the cancer deaths attributable to alcohol by upgrading mean daily alcohol consumption prevalence estimates from the 2020-2021 Behavioural Risk Factor Surveillance System and adjusting the alcohol consumption to per capita alcohol sales to address underreporting of drinking. Although this is not recognised as a solid strategy or methodology, Esser et al. (2024) themselves reflect that their analysis assumed that the distributions of low-level alcohol consumption mirrored current population distributions and that the relative risks did not change in the two hypothetical scenarios analysed compared to the actual estimates.
Forum member Ellison considers that “It is unfortunate that so many analyses attempting to relate alcohol consumption to health outcomes do so using only data on the number of drinks consumed over a period of time (usually averaged over a week). They fail to adjust for the pattern of drinking, which provides a much better assessment of the subsequent effects on health than just considering the number of drinks. The pattern includes adjusting for whether the alcoholic beverage is consumed with or without food, the regularity of consumption (frequently, even daily) rather than in binges, the type of beverage, etc.
Further, analyses based only on the number of drinks consumed are problematic in adjusting for under-reporting. Attempts to control for this often involve assuming everyone under-reports their intake similarly (by 73% of per capita alcohol sales in this paper), which is a crude and unsatisfactory method. As pointed out by other Forum reviewers, Klatsky et al. (2014) have demonstrated that factors that indicate alcohol misuse can be used to identify true under-reporters, who indeed have health risks higher than the large majority of drinkers in the population.
Given that most of the adverse effects of alcohol on cancer risk are among heavy or binge drinkers who generally consume their beverages without food, analyses are hampered when investigators combine truly moderate drinkers with alcoholics or other abusers of alcohol. Again, as shown by Klatsky et al. (2014), using available data to separate these two groups permits an improved ability to judge the extent to which alcohol consumption relates to the risk of cancer or other diseases or death. Among truly moderate drinkers, the effects of their drinking on the risk of such outcomes are minimal (Klatsky et al. 2014).”
Forum member de Gaetano states that “besides the important concept that ‘cancer’ does not exist, one should remember, that there are many different cancers and that even any single type of cancer may develop in different ways. Indeed, there is no cancer but only persons with that or that disease referred to as ‘cancer’. A second point is related to the individual person rather than to a single disease: one should always consider the global risk of an individual, rather than the single risks of different diseases. Drinking moderately might increase the risk of some cancers. Still, in the same individual, it might reduce the risk of more common or more severe disease, such as cardiovascular disease in a post-menopausal woman.”
Forum member Harding suggests that “the evidence cited by the authors does not justify the conclusions reached on public health policy. Given that most deaths in the population under study are due to either cardiovascular conditions or cancer, and that moderate alcohol consumption is protective for cardiovascular conditions, you would expect those who consume alcohol moderately and regularly to live long enough to die from cancer.”
Concluding comments
This study methodology is problematic, and hence so are the conclusions drawn. First, the authors take CDC cancer mortality figures for 2020-21 for a range of cancer types, and not simply alcohol-related cancers. Then, on the assumption that alcohol is a carcinogen, they take self-reported alcohol consumption across the whole population, without any data on the actual alcohol amount and pattern of consumption of those who died from cancer in this period. On this basis, they make poor public health recommendations and fail to adequately address that light-to-moderate alcohol consumption might reduce the risk of the equally important cause of death worldwide, cardiovascular disease.
References
Anand, P., Kunnumakkara, A.B., Sundaram, C., Harikumar, K.B., Tharakan, S.T., Lai, O.S., Sung, B., & Aggarwal, B.B. Cancer is a preventable disease that requires major lifestyle changes. Pharm Res. 2008 Sep;25(9):2097-116. https://doi.org/10.1007/s11095-008-9661-9. Epub 2008 Jul 15. Erratum in: Pharm Res. 2008 Sep;25(9):2200. Kunnumakara, Ajaikumar B [corrected to Kunnumakkara, Ajaikumar B]. PMID: 18626751; PMCID: PMC2515569.
Bagnardi, V., Rota, M., Botteri, E., Tramacere, I., Islami, F., Fedirko, V., Scotti, L., Jenab, M., Turati, F., Pasquali, E., Pelucchi, C., Bellocco, R., Negri, E., Corrao, G., Rehm, J., Boffetta, P., & La Vecchia, C. Light alcohol drinking and cancer: a meta-analysis. Ann Oncol. 2013 Feb;24(2):301-308. https://doi.org/10.1093/annonc/mds337. Epub 2012 Aug 21. PMID: 22910838.
Bagnardi, V., Rota, M., Botteri, E., Tramacere, I., Islami, F., Fedirko, V. et al. Alcohol consumption and site-specific cancer risk: a comprehensive dose-response meta-analysis. Br. J. Cancer 112, 580–593 (2015).
Boniface, S., Kneale, J. & Shelton, N. Drinking pattern is more strongly associated with under-reporting of alcohol consumption than socio-demographic factors: evidence from a mixed-methods study. BMC Public Health 14, 1297 (2014).
Cao, Y., Willett, W. C., Rimm, E.B., Stampfer, M.J., & Giovannucci, E.L. Light to moderate intake of alcohol, drinking patterns, and risk of cancer: results from two prospective US cohort studies BMJ 2015; 351 :h4238 https://doi.org/10.1136/bmj.h4238
Esser, M. B., Sherk, A., Liu, Y., Henley, S. J., & Naimi, T. S. (2024). Reducing Alcohol Use to Prevent Cancer Deaths: Estimated Effects Among U.S. Adults. American Journal of Preventive Medicine, 66(4), 725–729. https://doi.org/10.1016/j.amepre.2023.12.003
Esser, M. B., Sherk, A., Subbaraman, M. S., Martinez, P., Karriker-Jaffe, K. J., Sacks, J. J., & Naimi, T. S. (2022). Improving Estimates of Alcohol-Attributable Deaths in the United States: Impact of Adjusting for the Underreporting of Alcohol Consumption. Journal of Studies on Alcohol and Drugs, 83(1), 134–144. https://doi.org/10.15288/jsad.2022.83.134
Fehringer, G., Brenner, D. R., Zhang, Z. F., Lee, Y. A., Matsuo, K., Ito, H., Lan, Q., Vineis, P., Johansson, M., Overvad, K., Riboli, E., Trichopoulou, A., Sacerdote, C., Stucker, I., Boffetta, P., Brennan, P., Christiani, D. C., Hong, Y. C., Landi, M. T., … Hung, R. J. (2017). Alcohol and lung cancer risk among never smokers: A pooled analysis from the international lung cancer consortium and the SYNERGY study. Int J Cancer. https://doi.org/10.1002/ijc.30618
Hendriks, H. F. J., & Calame, W. (2018). The contribution of alcohol consumption to overall cancer incidence in the Western World: a meta-analysis. Journal of Nutrition and Health Sciences, 5(3), 311.
Hong, S. H., Myung, S. K., & Kim, H. S. (2017). Alcohol Intake and Risk of Thyroid Cancer: A Meta-Analysis of Observational Studies. Cancer Res Treat, 49(2), 534–547. https://doi.org/10.4143/crt.2016.161
Jun, S., Park, H., Kim, U.J., Choi, E.J., Lee, H.A., Park, B., Lee, S.Y., Jee, S.H., & Park, H. Cancer risk based on alcohol consumption levels: a comprehensive systematic review and meta-analysis. Epidemiol Health. 2023;45:e2023092. doi: 10.4178/epih.e2023092. Epub 2023 Oct 16. PMID: 37905315; PMCID: PMC10867516.
Klatsky, A. L., Udaltsova, N., Li, Y., Baer, D., Nicole Tran, H., & Friedman, G. D. (2014). Moderate alcohol intake and cancer: the role of underreporting. Cancer Causes Control, 25(6), 693–699. https://doi.org/10.1007/s10552-014-0372-8
Li, H., Borinskaya, S., Yoshimura, K., Kal’ina, N., Marusin, A., Stepanov, V. A. et al. Refined geographic distribution of the oriental ALDH2*504Lys (nee 487Lys) variant. Ann. Hum. Genet. 73, 335–345 (2009).
Rumgay, H., Shield, K., Charvat, H., Ferrari, P., Sornpaisarn, B., Obot, I., Islami, F., Lemmens, V.E.P.P., Rehm, J., & Soerjomataram I. Global burden of cancer in 2020 attributable to alcohol consumption: a population-based study. Lancet Oncol. 2021 Aug;22(8):1071-1080. https://doi.org/10.1016/S1470-2045(21)00279-5. PMID: 34270924; PMCID: PMC8324483.
Tramacere, I., Pelucchi, C., Bonifazi, M., Bagnardi, V., Rota, M., Bellocco, R., Scotti, L., Islami, F., Corrao, G., Boffetta, P., La Vecchia, C., & Negri, E. (2012a). A meta-analysis on alcohol drinking and the risk of Hodgkin lymphoma. Eur J Cancer Prev, 21(3), 268–273. https://doi.org/10.1097/CEJ.0b013e328350b11b
Tramacere, I., Pelucchi, C., Bonifazi, M., Bagnardi, V., Rota, M., Bellocco, R., Scotti, L., Islami, F., Corrao, G., Boffetta, P., La Vecchia, C., & Negri, E. (2012b). Alcohol drinking and non-Hodgkin lymphoma risk: a systematic review and a meta-analysis. Ann Oncol, 23(11), 2791–2798. https://doi.org/10.1093/annonc/mds013
Xu, X., Zhu, Y., Zheng, X., & Xie, L. (2015). Does beer, wine or liquor consumption correlate with the risk of renal cell carcinoma? A dose-response meta-analysis of prospective cohort studies. Oncotarget, 6(15), 13347–13358. https://doi.org/10.18632/oncotarget.3749
Comments on this critique by the International Scientific Forum on Alcohol Research were provided by the following members:
Henk Hendriks, PhD, Netherlands
Creina Stockley, PhD, MBA, Independent consultant and Adjunct Senior Lecturer in the School of Agriculture, Food and Wine at the University of Adelaide, Australia
Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark
R. Curtis Ellison, MD, Section of Preventive Medicine/Epidemiology, Boston University School of Medicine, Boston, MA, USA (Chair)
Giovanni de Gaetano, MD, PhD President, IRCCS Istituto Neurologico Mediterraneo NEUROMED
Fulvio Mattivi, MSc, Scientific Advisor, Research and Innovation Centre, Fondazione Edmund Mach, in San Michele all’Adige, Italy
Richard Harding, PhD, Formerly Head of Consumer Choice, Food Standards and Special Projects Division, Food Standards Agency, UK
[1] https://www.cdc.gov/nchs/products/databriefs/db492.htm#section_4
[2] https://www.isglobal.org/en/-/90-of-cancer-cases-are-caused-by-environmental-factors
[3] https://www.aacr.org/blog/2023/02/27/how-do-diet-and-exercise-help-prevent-cancer/