Marital transitions and drinking patterns among old and very old people in Sweden and Finland

Abstract Background Changes in marital status are major life transitions that may affect older adults’ physical and mental health, support network, and coping behaviors such as alcohol use. This study examined the effects of baseline and longitudinal marital status on alcohol use, gender interaction, and whether depressive symptoms and loneliness mediate these associations. Methods Data from the GERDA study in Northern Sweden and Western Finland included 5,071 adults aged 65-85 years (53.6% women; 77.4% partnered) surveyed in 2016 and 2021. Transitions were categorized as partnered (stable/entered partnership) or unpartnered (stable/divorced/widowed). Alcohol use in 2021 was assessed using AUDIT-C (non-drinker, low-risk, and hazardous drinking and binge drinking). Multinomial and binary logistic regressions assessed associations between marital status and alcohol use, adjusting for baseline sociodemographic, psychosocial, and health factors. Gender-stratified causal mediation analyses tested indirect effects via depression and loneliness. Inverse probability weights adjusted for attrition. Results Being unpartnered at baseline were associated with higher odds of abstaining versus partnered (aRRR=1.39, 95%CI=1.06-1.83 and aRRR=1.47, 95%CI=1.14-1.91 respectively). No associations with hazardous or binge drinking were found. Between 2016 and 2021, unpartnered individuals had higher odds of abstaining. Partnered women were less likely than partnered men to drink at hazardous level (aRRR=0.36, 95%CI=0.28-0.47) and unpartnered women had even lower odds (interaction aRRR=0.56, 95% CI = 0.33-0.95). No significant mediation via depression or loneliness was found. Among women, being unpartnered directly predicted lower odds of hazardous and binge drinking. Conclusions Older unpartnered adults are more likely to abstain and no mediation via depression loneliness was found. Alcohol use in later life appears shaped more by partnership status than by psychological distress after marital loss. Key messages • In later life, drinking is shaped more by social dynamics within partnerships than emotional distress after relationship loss. Prevention should target social drinking norms, especially among women. • Future studies should improve causal inference by using repeated measures over time and address selection bias through strategies like inverse probability weighting based on baseline factors.