BACKGROUND AND AIMS: Mortality from alcohol-attributable causes is patterned by income. We study analysed the income trajectories 17-19 years prior to death in order to determine: 1) whether income levels and trajectories differ between those who die of alcohol-attributable causes, survivors with similar sociodemographic characteristics, all survivors and those dying of other causes; 2) whether the income trajectories of these groups differ by education; and 3) whether there are differences in income trajectories between Finland and Sweden - two countries with differing levels of alcohol-attributable mortality but similar welfare-provision systems.

DESIGN: Retrospective cohort study using individual-level longitudinal register data including information on income, cause of death and socioeconomic position.

SETTING: Finland and Sweden Participants: The subjects comprised an 11-percent sample of the Finnish population in 2006-07 and the total population of Sweden aged 45-64 in 2007-08.

MEASUREMENTS: Median household income trajectories by educational group were calculated by cause of death and population alive during the respective years. Additionally, propensity score matching was used to match the surviving population to those dying from alcohol-attributable causes with regard to sociodemographic characteristics.

FINDINGS: The median income 17-19 years prior to death from alcohol-attributable causes was 92% (Finland) and 91% (Sweden) of survivor income: one year prior to death the respective figures were 47% and 57%. The trajectories differed substantially. Those dying of alcohol-attributable causes had lower and decreasing incomes for substantially longer periods than survivors and people dying from other causes. These differences were more modest among the highly educated. The baseline sociodemographic characteristics of those dying of alcohol causes did not explain the different trajectories.

CONCLUSIONS: In Finland and Sweden, income appears to decline substantially before alcohol-attributable death. Highly educated individuals may be able to buffer the negative effects of extensive alcohol use on their income level. Income trajectories are similar in Finland and Sweden despite marked differences in the level of alcohol-attributable mortality.