Alcohol consumption

Burden of disease and injury attributable to alcohol consumption in the Middle East and North Africa region, 1990-2019


The Institute for Health Metrics and Evaluation (IHME) manages the GBD project, which has so far estimated the level and trends associated with 369 diseases and injuries and 87 risk factors, from 1990 to 2019, in 204 countries and territories.8.9. This study reports the burden of disease attributable to alcohol consumption in the countries that make up the MENA region, from 1990 to 2019. The MENA region includes: Afghanistan, Algeria, Bahrain, Egypt, Iran, Iraq, Jordan, Kuwait, Lebanon, Libya, Morocco, Oman, Palestine, Qatar, Saudi Arabia, Sudan, Syrian Arab Republic, Tunisia, Turkey, United Arab Emirates and Yemen. The total population of the MENA region was estimated at 608.7 million in 2019ten. The comparative risk assessment approach was used in GBD 2019 to estimate the burden of disease and injury attributable to alcohol consumption. Detailed information on each analytical step and the methodology used to estimate the burden of disease, injury and risk factors can be found in a previous article8. Data for fatal and non-fatal estimates are available at and

Definition of alcohol consumption

Alcohol consumption (exposure) was defined as the number of grams of pure alcohol consumed per day by current drinkers. The level of exposure was estimated using the following indicators: (a) the proportion of current drinkers, i.e. people who have consumed one or more alcoholic beverages (or an approximation) during the previous year. (b) alcohol consumption, which was the average number of grams of alcohol consumed each day by current drinkers over a period of one year, and (c) liters of alcohol stock per capita, which was the number of liters of pure alcohol per inhabitant over a period of one year. In addition, to account for several types of bias, the following variables were also included: (a) number of tourists, which is the total number of people who visited a place in the previous year. (b) length of stay of tourists, which is the total number of days tourists stayed in the host country, and (c) unrecorded alcohol stock, which is the estimated proportion of the stock alcohol produced outside established markets8.

Information source

A systematic literature review was conducted by IHME to identify and extract data on key indicators. Population-level survey data, which contained participant-level information, was searched using the Global Health Exchange (GHDx) and the Global Health Data Online Database. ‘IHME, to create the required alcohol consumption indicators on current drinkers and alcohol consumption. Appropriate information has been extracted from data sources8but only data from representative samples were included.

The estimated prevalence of current drinkers was divided by age and sex. First, a meta-regression tool (MR-BRT) was used to estimate region-specific sex ratios to separate reported prevalence by sex. Spatio-temporal Gaussian process regression (ST-GPR) was then used to separate estimates into standard five-year GBD age groups, in studies that did not report estimates in these groups. The MR-BRT was also used to adjust data from studies that did not use the standard definition of alcohol consumption8.

Since GBD 2017, substantial improvements have been made to supply-level data modeling. The raw data included the domestic (WHO|Global Information System on Alcohol and Health (GISAH); Food and Agriculture Organization of the United Nations (FAO)) and retail supply (Euromonitor) of pure ethanol consumed (number of liters). Domestic supply was defined as the sum of production and imports minus exports. WHO and FAO sources were combined, with FAO data only used when there was no WHO data available for that year of location, since figures from the WHO already take FAO values ​​into account. Outliers, implausible data points, or data that created implausible model fluctuations have been removed, a detailed description of this process having been previously reported8.

Modeling strategy

Although population-based studies provide accurate estimates regarding the prevalence of current drinkers, they also underestimate actual levels of alcohol consumption.8.11. Therefore, the liters per capita variable was thought to provide a more accurate estimate of the overall volume of alcohol consumption, although it does not produce the estimates of consumption by age and sex needed to estimate disease burden. and alcohol-related injuries. Therefore, population-based survey data and total volume of alcohol consumed in a location (from FAO, GISAH and Euromonitor) were used to model consumption patterns by age and age. sex.8.

The survey data was used by ST-GPR to calculate estimates for each location/year/age/gender, as well as to model liters of alcohol per capita (LPC) and total number of tourists. LPCs have been adjusted to account for unrecorded consumption and tourist consumption. In addition, ST-GPR was also used to generate estimates by gender and age using the estimated percentage of current drinkers in a location/year/gender/age and previously estimated drinking trends. All estimates were expressed in grams/day8.

Estimated relative risk data

The Theoretical Minimal Risk Exposure Level (TMREL) to alcohol was calculated, i.e. the level of exposure that minimizes the chances of suffering a load from any cause related to alcohol. alcohol. IHME conducted a systematic review of the literature to find all cohort and case-control studies that reported a relative risk, hazard ratio, or odds ratio for any alcohol-disease or alcohol-injury pair covered by GBD 20198. The systematic review only included studies that were representative of their specific population, reported a categorical or continuous dose of alcohol consumption, and reported measures of uncertainty. The studies were then used to calculate a dose-response model using the Disease Modeling Ordinary Differential Equation Solver (DisMod ODE) to fit nonlinear Bayesian meta-regressions. A DisMod ODE model was preferred over conventional mixed effects meta-regression because it is able to estimate nonparametric splines on doses (i.e. there is a nonlinear dose-response relationship ) and to include heterogeneous doses through dose integration research normally reports doses categorically in broad ranges and DisMod ODE estimates specific doses when categories overlap between studies, using a step of integration). The results of the meta-regression were then used to estimate a nonparametric curve for all doses between zero and 150 g/day, along with their corresponding relative risks. The relative risk was considered the same for all ages and for men and women8.

The systematic review8 found that the following diseases and injuries were associated with alcohol consumption: cardiovascular disease, diabetes and kidney disease, digestive disease, neoplasms, neurological disorders, respiratory infections and tuberculosis, self-harm and interpersonal violence, substance use disorders substances, transport-related injuries and unintentional injuries .

Estimation of the proportion of cancers attributable to alcohol consumption

The population attributable fraction (PAF) was used to estimate the burden of alcohol-related disease and injury by country, age, sex and year, using the following formula:

$$PAF(x) = frac{{P_{A} + int_{0}^{150} {P(x)*RR_{C} (x)dx – 1} }}{{P_{A} + int_{0}^{150} {P(x)*RR_{C} (x)dx} }};;;;;;P(x) = P_{C} * Gamma (p)$$

where: Pvs denotes the prevalence of current drinkers, Pa is the prevalence of abstainers, RRVS(x) is the relative risk function for current drinkers, and p are the parameters determined by the mean and the standard deviation of the exposure. This equation was used to produce 1000 runs of the exposure and relative risk models and the PAF was used to calculate the attributable burden8.

Deaths and DALYs attributable to alcohol consumption for each country, age, sex, year and illness/injury were calculated by multiplying the FAPs by the estimated number of deaths or DALYs for each country, age, sex, year and illness/injury. . The number of deaths was estimated using the Cause of Death Ensemble (CODEm) model in GBD 2019. CODEm generates a number of individual models to identify the best model fit using all available data and covariates. The predictive validity of the different models, or combinations of models, was assessed and the one with the best predictive validity (out of sample) was chosen. In addition, DALYs for illnesses/injuries were estimated in three steps: (a) years lived with disability (YLD) were calculated by multiplying the severity-specific disability weights by the prevalence of each category of severity for each illness/injury; (b) the number of deaths in an age group was then multiplied by the remaining life expectancy in that age group, taken from the standard GBD life table, to produce the years of life lost (YLL) for each illness/injury; (c) YLL and YLD were then summed for each disease/injury to calculate DALYs. A full description of the methods used to calculate the total number of deaths and DALYs has already been reported.8. Estimates were all reported as age-standardized counts, proportions (PAFs), and rates per 100,000, along with their corresponding 95% IUs. The UIs were calculated by running 1,000 repetitions at each calculation step and including uncertainty from multiple sources (i.e. input data, measurement error, and estimates of the residual non-sampling error). User interfaces were defined as the 25th and 975th values ​​of the ordered draws8.

The relationship between the burden of disease/injury attributable to alcohol consumption and SDI has also been studied. Smoothing spline models12 were used to determine the shape of the association between the burden attributable to alcohol consumption, in terms of DALYs, and SDI for the 21 MENA countries. The GDI is a multifactorial measure of socio-economic development that contains income per capita distributed by lag, the level of education of the population aged > 15 years and the total fertility rate 13.14.

Ethics approval

The present study was approved by the ethics committee of Shahid Beheshti University of Medical Sciences (IR.SBMU.RETECH.REC.1401.070). This study is based on publicly available data and reflects the views of its authors only and not of the Institute for Health Metrics and Evaluation.