Aims Alcohol is believed to be an important factor behind the sharp rise in mortality during the period 1990–94 in Russia. However, the rise in the standard alcohol consumption proxy does not seem to be sufficient to explain all the increase in mortality. This study adopts a novel approach to exploring the role of the alcohol factor in the increased mortality by investigating whether the mismatch between trends in mortality and recorded alcohol consumption is due to an underestimation of the consumption increase.
Design and measurements First, the alcohol effect on the male accident rate was estimated using data for the period 1959–89. Next, the estimated alcohol effect and the observed accident mortality rate for the period 1990–98 were used to backcast alcohol consumption during that period. Thirdly, the backcasted alcohol series was used to predict trajectories in alcohol poisoning mortality, the homicide rate and all-cause mortality during the period 1990–98.
Findings There was a markedly stronger increase in the backcasted consumption proxy than in the standard alcohol consumption proxy during the period 1990–98. There was a substantial gap between the observed mortality rates and the rates predicted from the standard alcohol consumption proxy, whereas the predictions from the backcasted alcohol proxy were much closer to the target.
Conclusions Much of the rise in Russian mortality in 1990–94 appears to have been due to the increase in population drinking, but this increase is grossly underestimated by the commonly used consumption proxy combining alcohol sales, estimation of illicit alcohol production and proportion of alcohol-positive violent deaths.