A systematic comparison of the global comparative risk assessments for alcohol

ABSTRACT Aims To compare systematically the alcohol‐attributable mortality and burden of disease estimates for 2016 from a recent study by Shield and colleagues and the Global Burden of Disease study 2017 (GBD). Method This study compared estimates of alcohol‐attributable mortality and disability adjusted life years (DALYs) lost for 2016 with regards to absolute and relative differences, by region and by cause of disease or injury. Relative differences between the two studies are reported herein as percentage (%) differences. A difference of 10% or more was considered meaningful. Results The studies estimated similar global levels of overall alcohol‐attributable mortality for 2016 (Shield and colleagues estimated 5.1% more alcohol‐attributable mortality than the GBD study) but not alcohol‐attributable DALYs lost (18.3% difference). There were marked differences by region and cause of disease or injury. Compared with the results from Shield and colleagues, the GBD study estimated a lower alcohol‐attributable burden in Eastern Europe by 252 770 alcohol‐attributable deaths (45.2% difference) and 6.1 million alcohol‐attributable DALYs lost (32.9% difference) and in Western sub‐Saharan Africa by 124 200 alcohol‐attributable deaths (55.7% difference) and 7.0 million alcohol‐attributable DALYs lost (63.4% difference), and estimated a higher alcohol‐attributable burden in East Asia by 227 100 alcohol‐attributable deaths (48.0% difference) and 2.2 million DALYs lost (11.0% difference). With regard to the cause of disease or injury, Shield and colleagues attributed an overall detrimental effect to alcohol on ischaemic heart disease mortality, whereas the GBD study attributed a net beneficial effect. The GBD study, as compared with Shield and colleagues’ study, estimated a lower alcohol‐attributable mortality because of liver cirrhosis and injuries by 262 500 (44.6% difference) and 398 800 (46.2% difference), respectively. Conclusions Differences in estimates of the alcohol‐attributable burden of disease in two recent studies indicate the need to improve the accuracy of underlying data and risk relations to obtain more consistent estimates and to formulate, advocate for, and implement alcohol policies more effectively.