Research Article

Heavy daily alcohol intake at the population level predicts the

weight of alcohol in cirrhosis burden worldwide
Eva Stein1,2,y, Monica Cruz-Lemini3,y, Jose Altamirano3,4, Nambi Ndugga1, David Couper5,
Juan G. Abraldes6, Ramon Bataller1,4,7,
1
Division of Gastroenterology and Hepatology, Department of Medicine, University of North Carolina, Chapel Hill, NC, USA; 2University of North
Carolina School of Medicine, Chapel Hill, NC, USA; 3Vall dHebrn Institut de Recerca, Barcelona, Spain; 4Institut dInvestigacions Biomdiques
August Pi i Sunyer (IDIBAPS), Barcelona, Spain; 5Department of Biostatistics, University of North Carolina, Chapel Hill, NC, USA; 6Cirrhosis Care
Clinic, Liver Unit, Division of Gastroenterology, Department of Medicine, CEGIIR, University of Alberta, Edmonton, Canada; 7Division of
Biochemistry, Department of Nutrition, University of North Carolina, Chapel Hill, NC, USA

Background & Aims: Studies assessing alcohol as a population Lay summary: We carried out an analysis of the WHO 2014 Glo-
level risk factor for cirrhosis typically focus on per capita con- bal Status Report on Alcohol and Health, and categorized coun-
sumption. However, clinical studies indicate that daily intake is tries by their level of drinking (heavy or moderate). We found
a strong predictor of alcoholic cirrhosis. We aimed to identify that half of the global cirrhosis cases, and 60% in both North
the determinants of alcohols contribution to the global cirrhosis America and Europe are associated with alcohol intake. We con-
burden and to evaluate the influence of daily drinking on a pop- cluded that on a population level heavy daily drinking signifi-
ulation level. cantly influences the impact of alcohol on the cirrhosis burden.
Methods: We performed a comprehensive analysis of the WHO 2016 European Association for the Study of the Liver. Published
2014 Global Status Report on Alcohol and Health. We categorized by Elsevier B.V. All rights reserved.
countries by heavy or moderate drinking based on daily con-
sumption, using U.S. Department of Agriculture definitions of
heavy drinking. Additional data on cirrhosis cofactors were also
Introduction
obtained. Uni- and multivariate models were fitted to identify
independent predictors of the alcohol-attributable fraction of
Excessive use of alcohol is a main cause of global mortality, con-
cirrhosis.
tributing to deaths through cancers, toxic organ damage, and
Results: The WHO 2014 Report found that half of cirrhosis mor-
accidents and injuries [1]. Detailed data on the burden of
tality worldwide is attributable to alcohol, approximating 60% in
alcohol-associated mortality were released in the 2014 World
North America and Europe. In an integrative multivariate model,
Health Organization (WHO) Global Status Report on Alcohol
the designation of countries by moderate or heavy daily drinking
and Health, which is the most detailed and comprehensive report
had the strongest influence on the weight of alcohol in the cirrho-
of this kind [2]. This publication emphasizes the significant con-
sis burden. The relative contribution from alcohol increased by
tribution alcohol use makes to the global cirrhosis burden, with
11% with a transition from the moderate to heavy classification
approximately half of global cirrhosis deaths attributed to alcohol
(p <0.001). Importantly, drinking patterns such as heavy episodic
use. Attention should be placed on detrimental patterns and
drinking and the type of alcohol did not independently predict
levels of alcohol use that lead to cirrhosis in order to reduce asso-
the alcohol-attributable fraction of cirrhosis.
ciated mortality.
Conclusions: Heavy daily drinking on a population level signifi-
Previous studies have demonstrated a dose-response relation-
cantly influences the weight of alcohol in the cirrhosis burden.
ship between alcohol intake and an individuals risk for develop-
Reducing heavy drinking should be considered as an important
ing cirrhosis [36]. There is also substantial support for a
target for public health monitoring and policies.
threshold effect, wherein the risk for developing cirrhosis is con-
stant over time if a threshold of intake is surpassed [79]. For a
Keywords: Alcohol; Liver cirrhosis; Cirrhosis mortality; Global cirrhosis; Drinking given level of total consumption, the pattern of drinking may
patterns; Daily drinking. confer higher risk for cirrhosis; for example, drinking on a daily
Received 8 September 2015; received in revised form 27 May 2016; accepted 14 June basis has been shown to be associated with higher risk [6,7,9].
2016; available online 5 July 2016 Some studies have investigated the potential influences of factors
Corresponding author. Address: 7340A Medical Biomolecular Research Build-
including binge drinking [10,11], duration of consumption over
ing, 111 Mason Farm Rd, University of North Carolina, Chapel Hill, NC 27599-
7032, USA. Tel.: +1 919 966 4812. lifetime [7,9], and alcohol type [6,12,13]. Cofactors that have been
E-mail address: [email protected] (R. Bataller). shown to increase risk of alcoholic cirrhosis include cigarette
y
These authors contributed equally as joint first authors. smoking [14,15], excess weight [16], and viral hepatitis [17].
Abbreviations: WHO, World Health Organization; AAF, alcohol-attributable Consumption of certain substances, such as coffee and tea, may
fraction; ASDR, age-standardized death rate; HCV, hepatitis C virus; HBV,
hepatitis B virus; GDP, gross domestic product; BMI, body mass index.
provide a protective effect against the development of liver

Journal of Hepatology 2016 vol. 65 j 9981005

 JOURNAL OF HEPATOLOGY
disease [18,19]. For all variables discussed, the potential for sig- pure alcohol, which can be found in 350 ml (12 oz.) of beer (5% alcohol),
nificant geographical variation exists, due to cultural and socioe- 150 ml (5 oz.) of wine (12% alcohol), or 45 ml (1.5 oz.) distilled spirits (40% alco-
hol) [28]. We applied these guidelines to the population level, defining 21 g/day
conomic differences in exposure to risk or protective factors. The or greater of daily consumption as heavy drinking (12 standard drinks per day
economics and policies of alcohol sale can also be expected to across men and women). Henceforth, this country classification will be referred
vary considerably across the globe and to influence the way it to as the daily intake designation. Recognizing that other classifications to
is consumed [20]. define heavy drinking exist, we performed an exploratory analysis with other
published cut-off points (Supplementary Tables 4 and 5). To consider annual con-
On a population level, an association between alcohol con-
sumption, we converted data expressed in liters/year to standard drinks per/year
sumption and cirrhosis mortality has been well documented, using the conversion of 71.4 standard drinks per liter. In standard drinks,
reflecting the risk relationship demonstrated on the individual 1.5 drinks/day corresponds to 547 drinks/year and 7.7 liters/year. To assess
level [2124]. Most population level studies have assessed the GDP, countries were stratified according to the World Bank income categories
effect of per capita drinking on the death rate from cirrhosis, pri- of gross national income (low income: $1,045, lower-middle: $1,046-$4,125,
upper-middle: $4,126-$12,745, high-income: >$12,745). Supplementary Table 1
marily in Europe and the U.S. One study compared the effect of describes all variables included in the database and their sources.
per capita consumption on cirrhosis mortality in European coun-
tries that predominantly consume spirits as opposed to wine and Statistical analysis
beer [21]. Moreover, it has been hypothesized that differences in
population drinking patterns (e.g., heavy episodic vs. daily drink- Global patterns in AAF and ASDR were examined by comparing mean values
ing) shape the relationship between alcohol consumption and across the WHO Global Burden of Disease regional classifications (African region,
cirrhosis mortality [22]. However, to our knowledge, there are region of the Americas, Eastern Mediterranean region, European region, South-
no studies assessing the influence of daily drinking patterns on East Asian region, and Western Pacific region; http://www.who.int/choice/
demography/regions/en/). When appropriate, USA and Canada were considered
cirrhosis mortality at the population level, or on the burden of
separately as high-income North America, and Eastern Europe was considered
alcoholic cirrhosis specifically. In this study, we aimed to identify separately from Europe. We performed multiple imputation, in order to take
the main determinants of cirrhosis mortality and the weight of advantage of the full data set for the regression analyses (described in the Supple-
alcohol in the global cirrhosis burden. We used recent data mentary materials and methods). We conducted exploratory univariate linear
regression analyses to investigate the potential individual association between
reported in the WHO 2014 Global Status Report on Alcohol and
each parameter and AAF and ASDR. Parameters that showed a significant associ-
Health to evaluate associations between population drinking, cir- ation in the univariate analysis were included in a multivariate linear regression
rhosis cofactors, and economic indicators and the burden of alco- model. Comparisons between study groups were performed by Students t test or
holic cirrhosis. Chi- squared test when appropriate, and Pearson coefficients were used for
regressions. Data were analyzed using the IBM SPSS Statistics 21 statistical pack-
age, and p <0.05 was considered to be statistically significant.

Materials and methods

Database development
Results

We obtained detailed data on 193 countries from the WHO Global Information Geographical variation in alcohol consumption, cirrhosis mortality
System on Alcohol and Health, accessed July, 2014 (http://apps.who.int/gho/- and contribution from alcohol
data/node.main.GISAH?lang=en). From this data repository, we abstracted
country-specific data on levels of alcohol consumption, drinking patterns, and We first analyzed the geographical patterns of alcohol consump-
our main outcome variables: the fraction of cirrhosis attributable to alcohol use
tion and cirrhosis mortality. The characteristics of the global cir-
(alcohol-attributable fraction, AAF), and the age-standardized death rate from cir-
rhosis (ASDR). The WHO collected data through household surveys, primarily the rhosis burden and its potential determinants are described in
WHO global survey on alcohol and health, which was last conducted in 2012 in Table 1. On a per capita basis, the highest alcohol consumption
collaboration with all WHO regional offices and the European Commission. Preva- was observed in Europe with 10 liters/year (3677 standard
lence data on potentially predisposing factors for cirrhosis, including cigarette
drinks/year). In contrast, the rate was 1.1 liters/year (79 standard
smoking and obesity, were abstracted from the WHO Global Health Observatory
(http://www.who.int/gho/en/). For hepatitis C virus (HCV) and B (HBV) virus drinks/year) in the Eastern Mediterranean. There, an average of
seroprevalence, we obtained published data from systematic reviews of studies 6.5% of the population reported drinking, in contrast to 66%
reporting HBV and HCV infection [25,26]. Data on per capita gross domestic pro- reported in Europe. When considering alcohol intake among
duct (GDP) were extracted from the World Bank (http://data.worldbank.org/). those who drink, the average daily consumption ranged from
24 g per day in the Western Pacific region to 36 g in the African
Parameters on cirrhosis burden and potential determinants region.
The pattern of geographical variation in the percent of cirrho-
Our primary outcome was the AAF of cirrhosis, the proportion of cirrhosis attri-
sis attributable to alcohol use (AAF) differed from all-cause cir-
butable to alcohol use in each country. The WHO derived this value using relative
risk functions of cirrhosis for alcohol consumption in former and current drinkers.
rhosis mortality (Fig. 1A). The AAF rates were highest in the
The mathematical definition is provided as Supplementary Fig. 1. The secondary regions of USA/Canada and Europe, suggesting that in wealthier
outcome of cirrhosis mortality was assessed through the ASDR. Death rates were countries, alcohol makes a larger contribution to cirrhosis than
standardized to account for differences in age distributions across populations to in developing countries. The AAF values were: 62 in USA/Canada;
make values comparable. As potential determinants of AAF and ASDR, we consid-
60 in Europe; 53 in Latin America; 50 in the Western Pacific; 48
ered the following variables: average daily drinking among drinkers; annual per
capita consumption of alcohol; type of alcohol (wine, beer, spirits, other); heavy in Africa; 33 in South-East Asia; and 14 in the Eastern Mediter-
episodic drinking (consumption of 60 g or more of pure alcohol on at least one ranean. Notably, a significantly higher percent of cirrhosis is attri-
occasion in the last 30 days); obesity (body mass index (BMI) of 30 kg/m2 or butable to alcohol in Eastern Europe compared to Europe as a
greater); prevalence of cigarette smoking; seroprevalence of HCV and HBV; and whole, with 67% in Eastern Europe. Fig. 2 presents AAF patterns
GDP. We classified countries by heavy or moderate daily drinking based on guide-
lines from the U.S. Department of Agriculture [27]. The guidelines state that >2
in world map form.
standard drinks per day for men and >1 standard drink per day for women is con- Cirrhosis mortality (assessed through ASDR) was shown to
sidered heavy drinking. In the US, one standard drink contains roughly 14 g of vary significantly by region (Fig. 1B). In deaths per 100,000 within

Journal of Hepatology 2016 vol. 65 j 9981005 999

Research Article
Table 1. Characteristics of outcome variables and their potential determinants by WHO global region.a

Parametersb AFR AMR EMR EUR SEAR WPR

AAF of cirrhosis, % 48.3 (2.6) 53.7 (1.3) 14.3 (1.8) 60.8 (1.6) 33.1 (4.7) 49.6 (4.2)
ASDR, per 100,000 36.4 (2.2) 21.6 (2.7) 20.6 (4.3) 23.9 (2.9) 26.8 (3.6) 18.0 (4.2)
Daily consumption, g/day 36.4 (2.4) 29.1 (1.6) 32.6 (4.4) 33.6 (1.5) 33.7 (5.7) 24.3 (2.2)
Per capita consumption, L/year 5.1 (0.5) 7.2 (0.3) 1.1 (0.2) 10.0 (0.5) 2.3 (0.7) 5.1 (0.7)
Per capita consumption, SD/year 3677 (36) 514 (21) 79 (14) 714 (36) 164 (50) 364 (50)
Drinkers among population, % 29.7 (2.1) 54.9 (1.5) 6.5 (0.6) 65.6 (2.6) 12.4 (2.4) 42.5 (3.2)
HED among drinkers, % 19.1 (2.5) 19.5 (2.2) 2.9 (1.8) 26.6 (2.0) 8.2 (2.9) 21.6 (3.8)
Obesity prevalence, % 8.5 (1.0) 24.9 (1.1) 23.7 (2.6) 21.3 (0.5) 5.0 (1.3) 26.9 (4.4)
Cigarette smoking, % 10.0 (0.8) 17.0 (1.7) 17.4 (1.6) 28.4 (1.0) 19.5 (3.0) 26.4 (1.9)
HCV seroprevalence, % 2.4 (0.1) 1.8 (0.1) 3.4 (0.1) 2.7 (0.1) 2.7 (0.2) 2.4 (0.1)
GDP per capita, x103 USD 2.7 (0.64) 11.2 (2.0) 18.1 (6.1) 24.5 (3.4) 2.8 (0.7) 14.1 (4.0)
Values are means (standard errors).
a
AFR, African region; AMR, region of the Americas; EMR, Eastern Mediterranean region; EUR, European region; SEAR, South-East Asian region; WPR, Western Pacific region.
b
AAF, alcohol-attributable fraction; ASDR, age-standardized death rate from cirrhosis; SD, standard drinks; HED, heavy episodic drinking; HCV, hepatitis C virus sero-
prevalence estimates; GDP, gross domestic product; USD, United States dollars.

Identification of associations and independent predictors of the

Africa
contribution by alcohol to cirrhosis burden
Latin America
World region

USA & Canada The univariate associations between selected parameters indica-
Europe tive of alcohol intake, drinking patterns, cirrhosis cofactors, and
economic status and the outcome variable of AAF are shown in
Eastern Mediterranean
Table 2. To capture regional variation in the predictors of AAF,
Southeast Asian
we conducted complete analyses by WHO global region for both
Western Pacific AAF and ASDR, shown in Supplementary Tables 2 and 3. The fol-
lowing parameters of alcohol consumption showed positive and
0 20 40 60 80
significant associations with AAF (p <0.05 for all parameters,
Alcohol-attributable
fraction of cirrhosis (%) Table 2): average daily alcohol intake among drinkers, per capita
annual consumption, and the percent of population that drinks.
Africa Of note, AAF increased by 28% for a univariate transition from
Latin America the moderate to heavy daily intake designation. This led to our
hypothesis that daily drinking levels may influence the alcoholic
World region

USA & Canada

cirrhosis burden in a country. Regarding drinking patterns, the
Europe percent of total consumption represented by wine intake and
Eastern Mediterranean the percent of heavy episodic drinkers in the population and
Southeast Asian among drinkers were significantly associated with AAF in the
univariate analysis. Prevalence of cigarette smoking and of daily
Western Pacific
smoking, as well as GDP, were also positively associated with
0 10 20 30 40 50 AAF. Furthermore, per capita consumption increases with GDP;
Age-standardized death rate wealthier countries tend to consume more alcohol and have a
from cirrhosis (per 100,000)
higher proportion of cirrhosis caused by alcohol. The seropreva-
lence of HCV and HBV showed negative associations with AAF.
Fig. 1. Mean AAF and ASDR of cirrhosis by WHO world region. Horizontal bars
are standard errors. (A) The countries with the highest alcohol-attributable As expected, in countries where viral hepatitis is more prevalent,
fractions are in the developed world (Europe, US & Canada). Fractions range from its contribution to the cirrhosis burden is elevated relative to
14% (Eastern Mediterranean) to 60% (US & Canada). (B) Cirrhosis mortality is alcohol.
highest in Africa and Asia, with respective rates of 36.4 deaths per 100,000
All variables that showed significant univariate associations
population and 26.8, in contrast to 9.4 in USA/Canada.
were included in a series of multivariate models developed to
identify the independent determinants of AAF. The final model
included HCV seroprevalence, the daily intake designation, per
capita annual consumption, the percent of population that drinks,
the adult population, the rates were: 36 in Africa; 27 in South- and GDP per capita (Table 3). AAF increased independently with
East Asia; 23 in Europe; 23 in Latin America; 20 in the Eastern the following parameters (p <0.01 for all parameters): daily
Mediterranean, 18 in the Western Pacific; and 9 in USA/Canada. intake designation, per capita annual consumption and the per-
Interestingly, a significantly greater cirrhosis death rate was cent of population that drinks. Independent negative associations
observed in Eastern Europe compared to Europe as a whole, with were captured for HCV seroprevalence and GDP. The negative
a rate of 36 in Eastern Europe. Fig. 3 presents ASDR global pat- relationship with GDP contrasts with the positive univariate
terns in world map form. association, suggesting that the relationship between AAF and

1000 Journal of Hepatology 2016 vol. 65 j 9981005

 JOURNAL OF HEPATOLOGY

Alcohol-attributable fraction (%)

18.85 75.00

Fig. 2. Global AAF values by country. The countries with the highest alcohol-attributable fractions are in the developed world (Europe, United States and Canada).
Fractions range from 14% (Eastern Mediterranean) to 60% (United States and Canada). Data are not available for countries in white.

economic status is confounded by the level of alcohol consump- the presence of heavy daily drinking, are predictive of the burden
tion, which increases with GDP. Taken together, this model of alcohol in cirrhosis mortality.
demonstrates the independent effects of overall and daily drink-
ing, economic condition, and viral hepatitis on the contribution of Parameters not independently associated with the alcohol-
alcohol to the cirrhosis burden. attributable fraction of cirrhosis (AAF)

Usefulness of daily drinking classification in predicting alcohol- The percentages of total consumption represented by each type
attributable fraction of cirrhosis of alcohol (wine, beer, spirits, other) were included in our
exploratory analyses based on previous studies that have investi-
To best capture the effect of daily drinking on AAF, we classified gated the influence of alcohol type on cirrhosis risk [5,11,12]. The
countries by heavy or moderate daily drinking using published percentages of beer, liquor, and other alcohol did not show signif-
guidelines (see Materials and methods) [27]. We applied another icant associations in the univariate analyses. While wine con-
widely accepted definition of heavy drinking (40 g/day for men; sumption was positively associated with AAF in a univariate
20 g/day for women) and found results consistent with our own model, the significance of this parameter was lost in multivariate
(Supplementary Tables 4 and 5). Our stratification, referred to models. Taken with the independent relationships presented
in this publication as the daily intake designation, showed a above, these findings indicate that the amount of alcohol con-
strongly positive and significant independent relationship with sumed, but not the type, is the main factor influencing the contri-
the AAF of cirrhosis. A transition from the moderate to heavy des- bution of alcohol to cirrhosis mortality.
ignation corresponded with an increase of 11% in AAF (p <0.01). Similarly, the percentages of the total population and of drin-
The mean AAF for countries in the heavy drinking designation kers who met criteria for heavy episodic drinking (consumption
(54%, n = 134) was significantly higher than for countries in the of 60 g or more of alcohol on at least one occasion in the last
moderate designation (26%, n = 34; p <0.001) (Fig. 4). These 30 days) were positively associated with AAF in the univariate
results indicate that not only overall drinking levels, but also models but lost significance in multivariate models. Prevalence

Journal of Hepatology 2016 vol. 65 j 9981005 1001

Research Article

Age standardized death rate (per 100,000)

8.25 95.05

Fig. 3. Global ASDR values by country. Cirrhosis mortality is highest in Africa and Asia, with respective rates of 36.4 deaths per 100,000 population and 26.8, in contrast to
9.4 in United States/Canada. Data are not available for countries in white.

Table 2. Values and univariate regression coefficients (95% confidence

of cigarette smoking showed a positive univariate association,
interval) associated with AAF, n = 168 countries.
but no independent effect in multivariate models, whereas obe-
Parameter Regression coefficient sity (BMI >30 kg/m2) showed no significant association in the
(95% CI) univariate analysis. These potentially predisposing factors there-
Daily consumption, g/day 0.5 (0.32-0.69)* fore do not show a clear influence on the weight of alcohol in the
Daily intake designation, moderate/heavy 27.55 (21.65-33.46)* cirrhosis burden.
Per capita consumption, L/year 3.93 (3.60-4.27)*
Wine consumption, % 0.23 (0.08-0.39)* Identification of associations and independent predictors of cirrhosis
Beer consumption, % -0.1 (-0.22-0.02) mortality
Spirits consumption, % -0.1 (-0.21-0.01)
Other alcohol consumption, % 0.1 (-0.01-0.21) Based on the parameters that showed significance in the univari-
HED among population, % 0.91 (0.70-1.13)* ate analyses (Supplementary Table 3), we developed a series of
HED among drinkers, % 0.6 (0.44-0.77)* multivariate models to identify independent determinants of
Drinkers among population, % 0.59 (0.53-0.66)* all-cause cirrhosis mortality (assessed by ASDR). The final model
Obesity, % prevalence 0.01 (-0.28-0.30) included the following parameters: percentage of the population
Overweight, % prevalence 0.07 (-0.09-0.23) that drinks, GDP per capita, seroprevalence of HCV, and sero-
Cigarette smoking, % prevalence 0.61 (0.28-0.93)* prevalence of HBV. All parameters had a significant and positive
Daily cigarette smoking, % prevalence 0.56 (0.18-0.93)* regression coefficient with ASDR, with the exception of GDP per
GDP per capita, USD 0.19 (0.04-0.34)* capita, which had a significant negative effect (Table 4). As pre-
HCV seroprevalence, % -11.51 (-15.43--7.60)* dicted, population level drinking and prevalence of viral hepatitis
HBV seroprevalence, % -3.86 (-6.98--0.73)* are associated with higher mortality from cirrhosis. In contrast,

higher economic status of a country is associated with lower

p <0.05. HED, heavy episodic drinking; GDP, gross domestic product; USD, United
States dollars; HCV, hepatitis C virus; HBV, hepatitis B virus. mortality.

1002 Journal of Hepatology 2016 vol. 65 j 9981005

 JOURNAL OF HEPATOLOGY
Table 3. Factors independently associated with AAF. sion that daily drinking and overall per capita consumption inde-
pendently influence the contribution of alcohol to a countrys
Parameter Regression coefficients (95% CI)*
burden of cirrhosis. Remarkably, a novel classification of coun-
HCV seroprevalence, % -4.12 (-5.96--2.28)
tries by moderate or heavy daily drinking, referred to in this pub-
Daily intake designation, moderate/ 11.11 (7.57-14.66)
heavy
lication as the daily intake designation, was shown to be a strong
Per capita consumption, L/year 2.04 (1.45-2.63)
predictor of the AAF outcome. Although we acknowledge that the
use of this classification is a simplification, this association is
Drinkers among population, % 0.26 (0.16-0.36)
robust, since it did not significantly, vary if different thresholds
GDP per capita, USD -0.11 (-0.18--0.04)
or cut-off points are used to define countries with heavy drinking.
Adjusted regression coefficients (95% confidence intervals) for AAF of cirrhosis,
globally (n = 168 countries). In contrast, parameters such as heavy episodic drinking, the type

p <0.01 for all parameters; HCV, hepatitis C virus; GDP, gross domestic product; of alcohol consumed, and the prevalence of cigarette smoking
USD, United States dollars. and obesity, were shown to have no independent effect.
Geographical patterns of cirrhosis mortality were consistent
with a previous systematic review that showed the highest age-
80 standardized mortality in regions of Africa, South-East Asia, and
Eastern Europe [29]. The same review showed that alcohol is
attributable for the greatest proportion of cirrhosis in Western
Alcohol-attributable fraction of cirrhosis (%)

Europe and high-income North America, as confirmed by our

findings. Moreover, the influence of per capita consumption on
60
cirrhosis mortality has been well documented; in populations
with higher annual per capita alcohol consumption, the death
rate from cirrhosis is elevated [2124]. In this study, we focused
on the burden of alcoholic cirrhosis relative to other causes of cir-
40 rhosis, and found a positive independent effect of per capita
intake on AAF. We also investigated the effect of daily drinking,
which has been shown to increase individual risk for the develop-
ment of cirrhosis [6,7,9]. Here, the focus was placed on popula-
tion level drinking patterns and their effects on the burden of
20
alcoholic cirrhosis. The individual level association between daily
drinking and cirrhosis risk was mirrored on the population level,
as discussed below. While some studies have supported an influ-
ence of alcohol type on an individuals risk for cirrhosis, our find-
0 ings did not confirm this relationship on a population level [6,12].
Moderate Heavy
Indeed, this negative result is consistent with the finding of an
Fig. 4. Mean alcohol-attributable fraction (AAF) of countries classified by individual level study that showed no effect of wine consumption
heavy or moderate daily drinking. A transition from the moderate to heavy on cirrhosis risk among heavy drinkers [13]. Further large
classification corresponded with an 11% increase in AAF. Horizontal bars
represent the means: 26.0% for moderate; 53.5% for heavy (n = 34 and 134,
prospective studies should be undertaken to investigate this
respectively). important issue. Interestingly, GDP had a negative effect on AAF
in the multivariate model accounting for alcohol consumption,
in contrast to a positive association shown in the univariate anal-
ysis. Further studies specifically focused on the influence of eco-
Table 4. Factors independently associated with ASDR. nomic factors on cirrhosis burden and mortality, are needed to
Parameter Regression coefficients (95% CI) unmask the effects of these parameters.
Drinkers among population, % 0.155 (0.04-0.27)
This study was based on the most comprehensive worldwide
data available at this time, the WHO 2014 Global Status Report on
HCV seroprevalence, % 4.90 (1.40-8.37)
Alcohol and Health [2]. Data on cirrhosis mortality, the contribu-
HBV seroprevalence, % 7.52 (4.42-10.61)
tion from alcohol, drinking levels and patterns, and potentially
GDP per capita, USD -0.34 (-0.48--0.20)
predisposing factors were obtained through household surveys
Adjusted regression coefficients (95% confidence intervals) for ASDR of cirrhosis,
globally (n = 168 countries). and collaboration with all WHO regional offices. The 2012 global

p <0.01 for all parameters; HCV, hepatitis C virus; HBV, hepatitis B virus; GDP, survey on alcohol and health, the primary survey used for the
gross domestic product; USD, United States dollars. WHO publication, addressed alcohol consumption, policy, and
national monitoring and surveillance. It had a 91% response rate
from WHO member states and associate members, covering 97%
Discussion of the worlds population. The study was strengthened by the
inclusion of 193 countries in its analyses, which were conducted
This study presented the geographical patterns of the cirrhosis both on regional and global levels in order to capture geograph-
burden worldwide and the contribution by alcohol to that burden ical variation and global trends.
(alcohol-attributable fraction of cirrhosis, AAF). We investigated Our conclusions were drawn on a global scale; however, it is
potential relationships between population parameters of alcohol important to note that regional variation in the relationships
intake, drinking patterns, and predisposing factors for cirrhosis, investigated may be significant. For this reason, the analyses for
and the outcome variable of AAF. The results support the conclu- each predictor of ASDR and AAF were conducted separately by

Journal of Hepatology 2016 vol. 65 j 9981005 1003

Research Article
region and included in the Supplementary material. Moreover, Conflict of interest
the influence of unrecorded and unmarketed alcohol consump-
tion may limit the validity of conclusions that involve parameters The authors who have taken part in this study declared that they
of alcohol intake. The WHO report partially accounted for this do not have anything to disclose regarding funding or conflict of
limitation by conducting an additional survey of experts in coun- interest with respect to this manuscript.
tries with significant unmarketed alcohol consumption, in con-
junction with a systematic literature review on intake in those
countries. Finally, cirrhosis mortality data available from WHO Author contribution
were expressed in terms of 100,000 in population aged 15 or
older. This may introduce bias due to the low prevalence of alco- ES participated in the conception and design of this study, inter-
holic cirrhosis at age 15 and the variability in disease patterns by pretation, and manuscript writing. JA, MCL and JGA performed
age across the world. the statistical analysis and DC contributed to the interpretation
Our results revealed contrasting patterns in the burden of cir- of the results. ES, NN, JA, MCL, RB, and JGA participated in the col-
rhosis mortality worldwide vs. the contribution by alcohol to that lection, analysis and interpretation of the data and revision of the
burden. The highest rates of cirrhosis mortality were shown to final version of the manuscript. RB was involved in the concep-
occur in Africa and South-East Asia. In contrast, the highest pro- tion, design and supervision of this paper, analysis and interpre-
portion of cirrhosis mortality is attributable to alcohol in the tation of the data and participated in drafting and critically
wealthier regions of Europe and North America (Western Europe revising the final version of the paper.
and USA/Canada). These differing patterns may point to various
influences of socioeconomic conditions on the causes of cirrhosis, Acknowledgements
including the ability of wealthier populations to afford more alco-
hol, as well as poorer healthcare services, higher seroprevalence We would like to thank Anne Green for her numerous contribu-
of viral hepatitis, and greater vulnerabilities to the harms from tions to the study effort, as well as Gemma Odena, Jiegen Chen,
alcohol consumption in developing countries. The multivariate Veronica Massey, and Jaeyoun Cheong for their constant support
model (Table 4) assessing predictors of cirrhosis mortality and help.
showed a strong positive relationship between viral hepatitis
prevalence rates and cirrhosis mortality, suggestive of the persis-
tent disease burden of HCV and HBV, particularly in the develop-
Supplementary data
ing world.
In regards to daily drinking, countries with an average intake
Supplementary data associated with this article can be found, in
greater than 21 g/day of pure alcohol (1.5 standard drinks) were
the online version, at http://dx.doi.org/10.1016/j.jhep.2016.06.
shown to have a significantly higher proportion of cirrhosis
018.
mortality attributable to alcohol. The daily intake designation
was based on guidelines published by the U.S. Department of
Agriculture, which define heavy drinking as greater that 1 stan-
References
dard drink per day for women and 2 drinks per day for men
[27]. The increase in AAF associated with a transition from the [1] Rusyn I, Bataller R. Alcohol and toxicity. J Hepatol 2013;59:387388.
moderate to heavy drinking designation supports the usefulness [2] World Health Organization. Global status report on alcohol and health, 2014.
of this tool in predicting the relative burden of alcoholic [3] Corrao G, Bagnardi V, Zambon A, Arico S. Exploring the dose-response
relationship between alcohol consumption and the risk of several alcohol-
cirrhosis.
related conditions: a meta-analysis. Addiction 1999;94:15511573.
In conclusion a populations daily drinking levels appear to [4] Becker U, Deis A, Sorensen TI, Gronbaek M, Borch-Johnsen K, Muller CF, et al.
have a strong influence on the weight of alcohol in its liver cirrho- Prediction of risk of liver disease by alcohol intake, sex, and age: a
sis burden. The daily intake designation of countries employed in prospective population study. Hepatology 1996;23:10251029.
this study was shown to be an effective indicator parameter for [5] Klatsky AL, Armstrong MA. Alcohol, smoking, coffee, and cirrhosis. Am J
Epidemiol 1992;136:12481257.
the percent of cirrhosis attributable to alcohol. In contrast, drink- [6] Askgaard G, Gronbaek M, Kjaer MS, Tjonneland A, Tolstrup JS. Alcohol
ing patterns such as alcohol type and heavy episodic drinking do drinking pattern and risk of alcoholic liver cirrhosis: A prospective cohort
not appear to have an independent effect. Health policies should study. J Hepatol 2015;62:10611067.
target heavy daily drinking as a means of reducing the burden of [7] Kamper-Jorgensen M, Gronbaek M, Tolstrup J, Becker U. Alcohol and
cirrhosis: dose response or threshold effect? J Hepatol 2004;41:2530.
alcoholic cirrhosis on a population level.
[8] Savolainen VT, Liesto K, Mannikko A, Penttila A, Karhunen PJ. Alcohol
consumption and alcoholic liver disease: evidence of a threshold level of
effects of ethanol. Alcohol Clin Exp Res 1993;17:11121117.
[9] Sorensen TI, Orholm M, Bentsen KD, Hoybye G, Eghoje K, Christoffersen P.
Financial support
Prospective evaluation of alcohol abuse and alcoholic liver injury in men as
predictors of development of cirrhosis. Lancet 1984;2:241244.
This work was supported by the National Institute on Alcohol [10] Hatton J, Burton A, Nash H, Munn E, Burgoyne L, Sheron N. Drinking patterns,
Abuse and Alcoholism (NIAAA) (1U01AA021908-01) and P30 dependency and life-time drinking history in alcohol-related liver disease.
DK34987. JA wishes to express his gratitude to the Mexican Addiction 2009;104:587592.
[11] Mathurin P, Deltenre P. Effect of binge drinking on the liver: an alarming
National Council of Science and Technology (CONACYT, Mexico public health issue? Gut 2009;58:613617.
City, Mexico) for partially supporting his predoctoral stay at [12] Becker U, Gronbaek M, Johansen D, Sorensen TI. Lower risk for alcohol-
IDIBAPS. induced cirrhosis in wine drinkers. Hepatology 2002;35:868875.

1004 Journal of Hepatology 2016 vol. 65 j 9981005

 JOURNAL OF HEPATOLOGY
[13] Pelletier S, Vaucher E, Aider R, Martin S, Perney P, Balmes JL, et al. Wine [22] Norstrom T, Ramstedt M. Mortality and population drinking: a review of the
consumption is not associated with a decreased risk of alcoholic cirrhosis in literature. Drug Alcohol Rev 2005;24:537547.
heavy drinkers. Alcohol Alcohol 2002;37:618621. [23] Smart RG, Mann RE, Suurvali H. Changes in liver cirrhosis death rates in
[14] Altamirano J, Bataller R. Cigarette smoking and chronic liver diseases. Gut different countries in relation to per capita alcohol consumption and
2010;59:11591162. Alcoholics Anonymous membership. J Stud Alcohol 1998;59:245249.
[15] Dam MK, Flensborg-Madsen T, Eliasen M, Becker U, Tolstrup JS. Smoking and [24] Singh GK, Hoyert DL. Social epidemiology of chronic liver disease and
risk of liver cirrhosis: a population-based cohort study. Scand J Gastroenterol cirrhosis mortality in the United States, 19351997: trends and differentials
2013;48:585591. by ethnicity, socioeconomic status, and alcohol consumption. Hum Biol
[16] Naveau S, Giraud V, Borotto E, Aubert A, Capron F, Chaput JC. Excess weight 2000;72:801820.
risk factor for alcoholic liver disease. Hepatology 1997;25:108111. [25] Ott JJ, Stevens GA, Groeger J, Wiersma ST. Global epidemiology of hepatitis B
[17] Halimi C, Deny P, Gotheil C, Trinchet JC, Mal F, Scavizzi M, et al. Pathogenesis virus infection: new estimates of age-specific HBsAg seroprevalence and
of liver cirrhosis in alcoholic patients: histological evidence for hepatitis C endemicity. Vaccine 2012;30:22122219.
virus responsibility. Liver 1991;11:329333. [26] Mohd Hanafiah K, Groeger J, Flaxman AD, Wiersma ST. Global epidemiology
[18] Ruhl CE, Everhart JE. Coffee and tea consumption are associated with a lower of hepatitis C virus infection: new estimates of age-specific antibody to HCV
incidence of chronic liver disease in the United States. Gastroenterology seroprevalence. Hepatology 2013;57:13331342.
2005;129:19281936. [27] U.S. Department of Agriculture. Dietary Guidelines for Americans, 2010.
[19] Masteron GS, Hayes PC. Coffee and the liver: a potential treatment for liver [28] National Institute on Alcohol Abuse and Alcoholism. Drinking Levels
disease? Eur J Gastroenterol Hepatol 2010;22:12771283. Defined: http://www.niaaa.nih.gov/alcohol-health/overview-alcohol-con-
[20] Tighe A, editor. Statistical handbook. London: Brewing Publications Limited; sumption/what-standard-drink.
2003. [29] Mokdad AA, Lopez AD, Shahraz S, Lozano R, Mokdad AH, Stanaway J, et al.
[21] Ramstedt M. Population drinking and liver cirrhosis mortality: is there a link Liver cirrhosis mortality in 187 countries between 1980 and 2010: a
in Eastern Europe? Addiction 2007;102:12121223. systematic analysis. BMC Med 2014;12:145.

Journal of Hepatology 2016 vol. 65 j 9981005 1005