Critique 156: The pattern of alcohol consumption and risk of cirrhosis — 10 February 2015

Askgaard G, Grønbæk M, Kjær MS, Tjønneland A, Tolstrup JS.  Alcohol drinking pattern and risk of alcoholic liver cirrhosis:  A prospective cohort study. Pre-publication.  J Hepatol (2015), http://dx.doi.org/10.1016/j.jhep.2014.12.005.

Authors’ Abstract

Background & Aims:  Alcohol is the main contributing factor of alcoholic cirrhosis, but less is known about the significance of drinking pattern.

Methods:  We investigated the risk of alcoholic cirrhosis among 55,917 participants (aged 50–64 years) in the Danish Cancer, Diet, and Health study (1993–2011). Baseline information on alcohol intake, drinking pattern, and confounders was obtained from a questionnaire. Follow-up information came from national registers. We calculated hazard ratios (HRs) for alcoholic cirrhosis in relation to drinking frequency, lifetime alcohol amount, and beverage type.

Results:  We observed 257 and 85 incident cases of alcoholic cirrhosis among men and women, respectively, none among lifetime abstainers. In men, HR for alcoholic cirrhosis among daily drinkers was 3.65 (95% CI: 2.39; 5.55) compared to drinking 2–4 days/week. Alcohol amount in recent age periods (40–49 and 50–59 years) was associated with an increased risk, whereas the amount in 20–29 and 30–39 years was not. In men drinking 14–28 drinks/week, HR was 7.47 (95% CI: 1.68; 33.12), 3.12 (95% CI: 1.53; 6.39), and 1.69 (95% CI: 0.79; 3.65) in drinkers of little (<1% of weekly amount), some (1–15%), and mostly wine (50–100%), compared to drinking <14 drinks/week. In general, results were similar for women.

Conclusions:  In men, daily drinking was associated with an increased risk of alcoholic cirrhosis. Recent alcohol consumption rather than earlier in life was associated with risk of alcoholic cirrhosis. Compared to beer and liquor, wine might be associated with a lower risk of alcoholic cirrhosis.

Forum Comments

Background:  The association between excessive alcohol intake and hepatic cirrhosis has been known for many years, but it has not been determined clearly why only a percentage of alcoholics develop cirrhosis, while the majority do not.  From physiological studies over many decades, it has been suggested that for humans the intake of at least 50 g/day of alcohol (about 4 to 5 typical drinks) over at least 5 years years is required for the development of cirrhosis; some scientists suggest that dietary deficiencies and other environmental factors are also necessary for cirrhosis to develop, and there are surely genetic factors at work.  Often, epidemiologic studies have suggested that consumers of small amounts of alcohol may also develop cirrhosis, but under-reporting of alcohol intake by heavy drinkers has always been considered as a potential confounder of such an association.

Reviewer Skovenborg provided additional background data: “According to the Danish Cancer, Diet, and Health study cohort, in Denmark each year 1 in 2,000 citizens aged 45-64 years is diagnosed with alcoholic cirrhosis; a very serious disease with a 5-year mortality rate from 58% to 85% (Jepsen P et al).  For both sexes, alcohol has been and is closely related to liver cirrhosis rates. The Pearson correlation between in European countries for per capita consumption in 2002 and the liver cirrhosis rates for the time period 2000–2002 was 0.75 for men and 0.80 for women. Thus, for men, alcohol explained 57% of variance of liver rates, and for women 64% (Zatonski et al).  With a third of liver cirrhosis cases attributable to heavy alcohol intake, alcohol intake remains the most common cause of liver cirrhosis in Western Europe.  In Italy 68% of liver cirrhosis cases have been attributed to alcohol consumption (Corrao et al).  The decline in liver cirrhosis mortality in Spain, France and Italy is undoubtedly due to a steady decrease in alcohol consumption, accompanied by an improvement in the quality of wine and restrictions on home-brewed alcoholic beverage making. (Mokdad et al).

Importance of the amount of alcohol consumed and risk of cirrhosis: In a review of the scientific literature, Forum member Skovenborg noted that “The thresholds at which the risk of developing alcohol-related liver injury begins to appear may be about 20 g a day for women and 40 g a day for men, with a sharp increase of risk for men and women at levels of consumption of up to 60 g of alcohol a day, but thereafter the risk levels out.  Men who drank periodically were found to have a significantly lower RR of 0.56 (CI 0.37 – 0.85) compared to a reference group of men who drank daily, adjusted for average number of drinks per day, duration of alcohol misuse, and predominant type of alcohol consumed (Kamper-Jørgensen et al).  In a Danish study of 258 men with an average daily consumption of more than 50 g for more than 1 year who were free from cirrhosis on primary liver biopsy, the heavy-drinking men were followed for 10-13 years during which cirrhosis developed in 38, corresponding to a rate of 2% per year (Sørensen et al).  The likelihood of cirrhosis developing proved to be independent of duration of abuse and of average daily consumption before the primary biopsy.  This result suggests that alcohol abuse leads to cirrhosis by a conditioning effect that enhances the risk of cirrhosis to a set level per unit of time. Patients reporting intermittent abuse did show a lower rate of cirrhosis than those reporting daily intake, but the difference was not significant.  This suggests that the conditioning by the abuse either has a threshold or, if dose-related, has a saturation level that is attained when the daily intake exceeds 50-75 g (Sørensen et al).”

Forum comments on present study:  This is a good cohort (in Denmark) to study long-term health effects, as public health records permit excellent follow up of individuals and ascertainment of disease, although it is somewhat surprising that, in this cohort of Danes (generally considered to be heavy binge drinkers), alcoholic cirrhosis was a rare diagnosis; there were only 342 cases in their cohort of 55,917 subjects, which equals about ½ of 1 percent (0.006195) in this population.  

In this study, the risk of cirrhosis was related to baseline estimates of alcohol and earlier drinking, but no drinking data following the baseline ascertainment were collected.  The finding of a greater relation of more recent drinking to cirrhosis, in comparison with reports of drinking in earlier decades of life, can be interpreted not only as less effect at earlier time periods but could relate to less precise estimates based on poorer memory of drinking in earlier years. 

Under-reporting of alcohol intake could not be evaluated in the present study, making it likely that groups reporting only moderate alcohol intake might include some who were heavier drinkers; this would suggest an increase in cirrhosis at lower levels of alcohol.  This makes it difficult (impossible) to define a clear threshold of effect of alcohol on cirrhosis from observational studies.

Making a diagnosis of alcoholic liver disease:  In the present study, there is special concern by Forum members related to the diagnosis of alcoholic cirrhosis.  A diagnosis of alcoholic cirrhosis generally requires a history of alcoholism or heavy drinking; subjects given that diagnosis would essentially be limited to those who were heavy drinkers.  Thus, it would be no surprise to find a history of heavy drinking when evaluating alcohol as an exposure.  Forum member Barrett-Connor pointed out that the authors’ conclusion states that “daily drinking” was associated with increased cirrhosis, “but they do not state daily drinking of how many drinks, which would be very important.”  She agreed that there is concern that none of the subjects reporting lifetime abstinence had a diagnosis of alcoholic cirrhosis, which is unusual for epidemiologic studies.  (In most studies, due to under-reporting of alcohol by some heavy drinkers or to data being collected at a time when they may be avoiding alcohol, perhaps due to liver dysfunction, there are usually a few cases of cirrhosis among putative “abstainers.”) 

Forum member Skovenborg commented:  “The diagnosis of alcoholic liver disease (ALD) can generally be made based on clinical and laboratory features alone in patients with a history of significant alcohol consumption after other etiologies for chronic liver disease have been ruled out.  However, the diagnosis of ALD can be clinically challenging, as there is no single laboratory or imaging study that can confirm the diagnosis.  Furthermore, patients may be completely asymptomatic, have no clinical signs of early ALD or early cirrhosis and may have normal liver enzymes.  In addition, patients may have co-existing risk factors for non-alcoholic fatty liver disease such as obesity and diabetes, and some may not be entirely forthcoming as to their degree of alcohol consumption.  It is important to rule out other etiologies for the patient’s liver disease before making a definitive diagnosis of ALD, including chronic viral hepatitis, autoimmune hepatitis, hemochromatosis, and drug-related hepatotoxicity.  When the diagnosis is unclear, a liver biopsy may be warranted.  The histological features of ALD can ultimately define the diagnosis according to the typical presence and distribution of hepatic steatosis, inflammation, and Mallory-Denk bodies (Torruellas et al).  There is a considerable variation among specialists in the use of liver biopsy for the diagnosis of alcoholic cirrhosis, which is often based solely on clinical findings; however, when specific diagnostic markers are absent, the grounds for making a diagnosis of alcoholic cirrhosis in a patient who happens to be a heavy drinker are no stronger than those of guilt by association.  As a result some cases of cryptogenic cirrhosis may be wrongly attributed to alcohol (Williams & Davis)”

Forum member Thelle had similar concerns: “In the present study, the end-points are collected from the Danish National Patient Register and the Register of Cause of Death.  It is likely that a diagnosis of alcoholic cirrhosis in the Patient Register is based upon information about the patients’ alcohol habits as well as liver function, whereas the unspecified cirrhosis either lack this information or are true unspecified cases. Thus the diagnostic process may induce a bias.  The authors claim that the validity of this register is good referring to a paper by Lynge et al.  That paper does not assess liver diseases in particular and is a very general assessment of the patient register.  Still, the validity is probably as good as they can get it using public registers.  Only about 10 per cent of all the cases were collected from the Death Register.  Cause of death due to cirrhosis may be under-reported, and especially alcoholic cirrhosis, as this may be a stigmatizing diagnosis in a puritan Nordic country.”

Reviewer Finkel stated: “Going back more than 20 years, the work of Lieber and others had convinced me that, on average, healthy men could drink up to 40 grams of alcohol daily without annoying their livers (e.g., Lieber et al 1965; Lieber & Decarli 1970; Lieber 1995).  If diagnostic criteria and death certificates in Denmark are anything like those in the US, I would discount the accuracy of inscribed diagnoses.”  Reviewer Ellison added: “There are even studies suggesting that moderate alcohol consumption benefits the liver by decreasing inflammation.  For example, Szabo reported that ‘in contrast to the pro-inflammatory activation by chronic excessive alcohol consumption, acute moderate alcohol administration has anti-inflammatory effects,’ and many studies have found that fatty liver disease is less common among moderate drinkers than among abstainers.” 

Focus on frequency of drinking rather than amount:  The focus of the report (and the message being spread widely in the media) relates to the number of days per week when alcohol is consumed, with little attention to the amount of alcohol consumed on each day.  Indeed, the media headlines based on this study have included “Daily Drinking Causes Cirrhosis!”

The use of >0  up to 14 drinks/week as the referent group essentially guarantees that heavier drinking would increase the risk of cirrhosis.  Drinkers in this referent group had, by far, the lowest risk of alcoholic cirrhosis (and this is the generally the accepted range of “sensible drinking” for most guidelines).  Further, there was a much higher risk of of cirrhosis among “current abstainers” than among subjects in this referent group, undoubtedly because the current abstainers included many former heavy drinkers.

In the present study, the fully adjusted risk of alcoholic cirrhosis for men increases from 1.43 (0.84-2.43) for men drinking 5-6 days a week to 3.65 (2.39-5.55) for men drinking 7 days a week.  For women the risk function of drinking days is opposite: the fully adjusted risk of alcoholic cirrhosis for women decreases from 2.30 (1.14-4.67) for women drinking 5-6 days a week to 1.73 (0.85-3.52) for women drinking 7 days a week.  The mean alcohol amounts for daily drinkers were somewhat higher compared to those drinking less frequently in each category of alcohol amount, making it difficult to separate the effects of drinking frequency and alcohol amount.  Indeed, the authors state: “The mean alcohol amounts for daily drinkers were somewhat higher compared to those drinking less frequently in each category of alcohol amount, making it difficult to detangle the effect of drinking frequency and alcohol amount.”

Forum member Ellison added: “Data presented in supplementary tables on a total and larger number of patients with cirrhosis, that included patients in whom their cirrhosis could not be diagnosed as alcoholic cirrhosis, show clearly that for subjects whose total weekly number of drinks was > 0 to 14 (the usual amount of alcohol considered moderate in most international guidelines), there was little increase in risk of total cirrhosis, going from drinking on ≤ 4 days to 5-6 days to 7 days per week (HR for men 1.00, 1.22, and 1.26, respectively; for women, HR were 1.00, 1.63, and 1.22 ; none of these differences were statistically significant).  However, at essentially every level of increase in total number of drinks per week, there was a large increase in the risk of cirrhosis, suggesting that even in patients not meeting criteria for alcoholic cirrhosis, there were probably a large percentage in whom alcohol played a role in their disease.  This also makes us wonder why the focus of the authors was so much on number of days per week without commenting on the amount consumed on each day.”

Reviewer Thelle added: “The authors’ attempt to describe in detail drinking pattern and liver cirrhosis, differentiating between drinking daily versus 4-5, or 6 days a week.  Misclassification between frequency categories is likely to have occurred.  It seems odd that the number of person-years decline from 135,573 for category 2-4/week to 66,512 in 5-6/week, and then up to 103,820 for 7/week.  Whether this really matters is difficult to assess, but it is likely to affect the estimates for the two highest categories.  The authors want to ‘execute evidence-based counselling,’ which of course is praiseworthy, but I have difficulties understanding why the effect of drinking 5-6 days/week should differ much from 7 days/week.” 

Differential effects according to type of beverage:  Data presented in supplementary tables indicate that among wine, beer, and liquor drinkers, there was a large and step-wise increase in risk of cirrhosis with larger total amounts of alcohol.  This also questions why the focus was primarily on drinking days, rather than on the frequency and amount of alcohol per day.  In these same tables, there was a step-wise increase for beer and liquor, but no increase for wine drinkers until their total weekly intake exceeded 35 drinks/week for men or 28 drinks/week for women.  (If wine drinkers indeed had lower risk of cirrhosis, it cannot be determined if it related to different patterns of drinking or to polyphenols or other substances other than alcohol in wine.)  However, the consensus of Forum members is that whether this paper is good evidence for supporting drinking wine instead of other alcoholic beverages remains to be seen.    

Implications of this paper:  Reviewer Skovenborg stated that the concerns discussed above make it difficult to determine the following: (1) Is wine alcohol really less liver-toxic than beer or liquor?  (2) Should we really tell everyone to have one day each week without wine being included with their meal?  (3)  Why is daily drinking dangerous for men but tends to be protective in women regarding liver disease?  I agree with the comments so far, especially the concerns due to limited diagnostic accuracy, the probability of under-reporting of alcohol consumption, and the lack of biological plausibility according to the Bradford-Hill criteria.”

Skovenborg continued: “As the total observation time in the present study was 831,285 person-years, the expected number of incident cases of alcoholic liver cirrhosis would be 416. With a response rate of 35% of the eligible cohort the participants of the Danish Cancer, Diet, and Health Study are probably not a high-risk population and the actual number of alcoholic cirrhosis cases in the study was only 342.

“In general, whenever a certain type of alcoholic drink has been particularly associated with cirrhosis, it has always been the most readily available, inexpensive, or socially acceptable form of alcohol (Williams & David). The explanation why less than 10% of heavy drinkers develop advanced liver disease remains unclear, and both environmental and genetic factors undoubtedly play a role (Day).  The unhealthy diet consumed by beer rather than wine drinkers may explain the increased risk of liver disease associated with beer versus wine drinking (Johansen et al).  However, in The Copenhagen Alcohol Cohort no significant association was observed between predominant type of alcohol consumed and risk of alcoholic cirrhosis mortality (Kamper-Jørgensen M et al).”

Indeed, Reviewer Van Velden indicated that he “Would feel much better if there were liver biopsies done to confirm the diagnosis.”  He added: “Hepatitis B and C have to be taken into consideration as a cause of cirrhosis.  There is considerable individual variation in the dose of alcohol that may cause cirrhosis.  While the association of alcohol and liver cirrhosis is well established, other nutritional factors have to be taken into consideration; heavy drinkers often have a nutrient-deficient diet.”  Reviewer Thelle also mentioned that “It has been hypothesized that coffee is a possible confounding or rather a modifying factor that should have been included in this analysis (Arnesson et al, Ikeda et al, Cardin et al, Danielsson et al).”

References from Forum critique

Arnesen E, Huseby N E, Brenn T, Try K.  The Tromsø Heart Study: distribution of, and determinants for, gamma-glutamyltransferase in a free-living population. Scand J Clin Lab Invest 1986;46:63-70.

Cardin R, Piciocchi M, Martines D, Scribano L, Petracco M, Farinati F. (2013). Effects of coffee consumption in chronic hepatitis C: a randomized controlled trial. Dig Liver Dis 2013;45:499-504.

Corrao G, et al.  Attributable risk for symptomatic liver cirrhosis in Italy.  J Hepatol 1998;28:608-614.

Danielsson J, Kangastupa P, Laatikainen T, Aalto M, Niemelä O.  Dose- and gender-dependent interactions between coffee consumption and serum GGT activity in alcohol consumers. Alcohol Alcohol 2013;48:303-307.

Day CP. Who gets alcoholic liver disease: nature or nurture? J Royal Coll Physicians Lond 2000;34:557-562.

Jepsen P, et al.  Clinical course of alcoholic liver cirrhosis: A Danish Population-Based cohort study. Hepatology 2010;51:1675-1682.

Ikeda M, Maki T, Yin G, Kawate H, Adachi M, Ohnaka K, . . . Kono S.  Relation of coffee consumption and serum liver enzymes in Japanese men and women with reference to effect modification of alcohol use and body mass index. Scand J Clin Lab Invest 2010;70:171-179.

Johansen D, et al.  Food buying habits of people who buy wine or beer: cross sectional study.  BMI 2006;332:519-522.

Kamper-Jørgensen M, et al.  Alcohol and cirrhosis: dose-response or threshold effect?  Journal of Hepatology 2004;41:25-30.

Lieber CS.  Medical disorders of alcoholism.  New England Journal of Medicine 1995;l333:1058-1065.

Lieber CS, Decarli LM.  Hepatic microsomal ethanol oxidizing system: In vitro characteristics and adaptive properties in vivo. Journal of Biological Chemistry 1970;245:2505-2512.

Lieber CS, Jones DP, Decarli LM.  Effects of prolonged ethanol intake: Production of fatty liver despite adequate diets. Journal of Clinical Investigation 1965;44:1009-1021.

Lynge E, Sandegaard JL, Rebolj M.  The Danish National Patient Register.  Scand J Public Health 2011;39:30–33.

Mokdad AA, et al.  Liver cirrhosis mortality in 187 countries between 1980 and 2010: a systematic analysis.  BMC Medicine 2014;12:145.

Sørensen TIA, et al.  Prospective evaluation of alcohol abuse and alcoholic liver injury in men as predictors of development of cirrhosis. Lancet 1984;324:241-244.

Szabo G.  Moderate drinking, inflammation, and liver disease.  Ann Epidemiol 2007;17:S49–S54.

Torruellas C, et al.  Diagnosis of alcoholic liver disease.  World J Gastroenterol 2014;20:11684-11699.

Williams R, Davis M.  Alcoholic liver injury.  Proc Roy Soc Med 1977;70:333-337.

Zatonski WA, et al.  Liver Cirrhosis Mortality in Europe, with Special Attention to Central and Eastern Europe.  Eur Addict Res 2010;16:193–201.

Forum Summary

This paper, from a group of experienced investigators in Denmark using data from a large population-based cohort, attempted to judge how drinking pattern affects the risk of a subject developing alcoholic cirrhosis.  From a cohort study of 55,917 participants (aged 50–64 years), the authors calculated hazard ratios (HRs) for alcoholic cirrhosis in relation to drinking frequency, lifetime alcohol amount, and beverage type.  A total of 342 subjects developed alcoholic cirrhosis.  The authors concluded that, in men, daily drinking was associated with an increased risk of alcoholic cirrhosis as compared with less frequent consumption; they also concluded that wine consumption might be associated with a lower risk of alcoholic cirrhosis that associated with the consumption of beer or liquor.

The major concern of Forum reviewers regarding this paper was the use of a diagnosis of “alcoholic cirrhosis” to judge the effects of alcohol on the disease; this was considered a type of circular reasoning.  A diagnosis of alcoholic cirrhosis generally requires a history of alcoholism or heavy drinking, and subjects given that diagnosis would essentially be limited to those who were heavy drinkers.  Thus, it would be no surprise to find a history of heavy drinking when evaluating alcohol as an exposure. 

There was also concern about the implications of the author regarding frequency of drinking.  The authors’ described in detail the relation of drinking pattern to liver cirrhosis, differentiating, in particular, effects associated with drinking daily versus drinking on 4-5, or 6 days a week.  However, Forum members were concerned that as the overall average amount of alcohol for the daily drinkers exceeded that of those drinking less frequently, it was difficult to know if it was primarily the frequency of drinking (as concluded by the authors) or the total amount of alcohol consumed by subjects. The authors stated that they wanted to “execute evidence-based counselling,” which of course is praiseworthy, but Forum members had difficulty understanding why the effect of drinking 5-6 days/week should differ much from 7 days/week, especially if the usual amount was moderate.  Indeed, the investigators state: “The mean alcohol amounts for daily drinkers were somewhat higher compared to those drinking less frequently in each category of alcohol amount, making it difficult to detangle the effect of drinking frequency and alcohol amount.”  Forum members noted that the authors conclude that “daily drinking” was associated with increased cirrhosis, but they do not state daily drinking of how many drinks, which would be especially important. The study did not provide reliable data that would support alcohol-free days during the week.

The referent group in this study was made up of subjects reporting < 0 to 14 drinks/week, which would consist of light-to-moderate drinkers, those considered to be “sensible drinkers” in most cultures.  Hence, it is no surprise that as subjects reported greater amounts of alcohol in the study, the risk of alcoholic cirrhosis increased.  No one advises people to drink more than sensible limits. Further, there was a much higher risk of of cirrhosis among “current abstainers” than in the referent group, undoubtedly indicating that the current abstainers group included many former heavy drinkers.

Overall, the Forum thought that this paper raises, but does not answer, a number of questions about the relation of alcohol to the risk of cirrhosis: (1) Is wine really less liver-toxic than beer or liquor?  (2) Should we really tell everyone to have one day each week without wine being included with their meal, or on which to avoid any alcohol?  (3)  Why is daily drinking, rather than 5 or 6 days per week dangerous for men but tends to be even slightly protective in women regarding liver disease?  Because of concerns due to problems with bias from the diagnostic approach, the probability of under-reporting of alcohol consumption, and the lack of biological plausibility, we will need further scientific data to answer these important questions.

                                                         *                *                  *

Contributions to this critique by the International Scientific Forum on Alcohol Research have been provided by the following members:

Elizabeth Barrett-Connor, MD, Chief of the Division of Epidemiology, Distinguished Professor in the Departments of Family and Preventive Medicine & Medicine, University of California, San Diego, La Jolla, CA, USA.

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

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

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

Ulrich Keil, MD, Institute of Epidemiology & Social Medicine, University of Muenster, Germany

Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark

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

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

Fulvio Ursini, MD, Dept. of Biological Chemistry, Universityof Padova, Padova, Italy

David Van Velden, MD, Dept. of Pathology, Stellenbosch University, Stellenbosch, South Africa

Yuqing Zhang, MD, DSc, Clinical Epidemiology, Boston University School of Medicine, Boston, MA, USA