Objective: To investigate the age, period, and cohort effects of polypharmacy in older adults using Swedish register data covering the period 2006 to 2020.

Design: Repeated cross-sectional study using routinely-collected health care data.

Setting: Nationwide, Sweden.

Participants: A 10% random sample was drawn each year from the source population of all adults aged ≥ 65 in Sweden, 2006–2020 (cumulative n > 3,000,000).

Measurement: Polypharmacy was defined as the use of ≥ 5 medicines. Drug data were extracted from the National Prescribed Drug Register (NPDR) Medication use was assessed on 1st January (1-day point prevalence) each year based on the drug duration episodes. Age-Period-Cohort analysis was conducted to explore these effects.

Results: Overall, 32.8% of older adults was exposed to polypharmacy in 2020 compared to 28.6% in 2006. This increase was more pronounced among individuals aged 85 to 89, from while the prevalence remained relatively steady among those aged 65 to 84. In the formal Age-Period-Cohort analysis, the cohort differences were weak for polypharmacy, but more prevalent for specific medication classes.

Conclusion: Polypharmacy is mainly influenced by age and period effects, but not by cohort. The rise in polypharmacy is primarily propelled by an increased pace in medication use among individuals aged 75 to 89 years. These findings can provide valuable insights for making effective strategies aimed at reducing polypharmacy.

Objectives: The educational gradient in late-life health is well established. Despite this, there are still ambiguities concerning the role of underlying confounding by genetic influences and gene-environment (GE) interplay. Here, we investigate the role of educational factors (attained and genetic propensities) on health and mortality in late life using genetic propensity for educational attainment (as measured by a genome-wide polygenic score, PGS_Edu) and attained education.

Methods: By utilizing genetically informative twin data from the Swedish Twin Registry (n = 14,570), we investigated influences of the educational measures, familial confounding as well as the possible presence of passive GE correlation on both objective and subjective indicators of late-life health, that is, the Frailty Index, Multimorbidity, Self-rated health, cardiovascular disease, and all-cause mortality.

Results: Using between-within models to adjust for shared familial factors, we found that the relationship between educational level and health and mortality later in life persisted despite controlling for familial confounding. PGS_Edu and attained education both uniquely predicted late-life health and mortality, even when mutually adjusted. Between-within models of PGS_Edu on the health outcomes in dizygotic twins showed weak evidence for passive GE correlation (prGE) in the education-health relationship.

Discussion: Both genetic propensity to education and attained education are (partly) independently associated with health in late life. These results lend further support for a causal education-health relationship but also raise the importance of genetic contributions and GE interplay.