Abstract Aims To evaluate temporal trends, across three decades, in the population attributable fractions (PAFs) of modifiable risk factors for 5-year risk of cardiovascular diseases (CVDs). Methods and results Within population-based Rotterdam Study, we defined three time groups of individuals without established CVD at ‘baseline’ with a mean age of 70 years, and followed for five years: Epoch 1990s (1989–93, n = 6195), Epoch 2000s (1997–2001, n = 5572), and Epoch 2010s (2009–14, n = 5135). The prevalence of risk factors and related relative risks were combined to quantify PAFs. The PAF of the six risk factors combined for global CVD was 0.57 [95% confidence interval (CI) 0.47–0.65], 0.52 (0.39–0.62), and 0.39 (0.18–0.54) in three respective epochs. Hypertension contributed the highest PAF to global CVD in Epoch 1990s (0.37, 95% CI: 0.28–0.44) and 2000s (0.34, 95% CI: 0.22–0.43), while smoking was the largest contributor in Epoch 2010s (0.20, 95% CI: 0.06–0.32). Dyslipidaemia changed population-level coronary heart disease risk over time. For stroke, hypertension became a less significant contributor over time, but smoking became a larger contributor. For heart failure, all risk factors showed non-significant PAFs in Epoch 2010s. PAFs related to individual risk factor varied among women and men. Conclusion Six modifiable risk factors to population-level global CVD risk decreased over time, but still explained 39% of total CVD in the latest decade. PAFs changed considerably for hypertension, dyslipidaemia, and smoking. Risk factors had different PAFs for different CVDs with pronounced sex differences.

Trends in population attributable fraction of modifiable risk factors for cardiovascular diseases across three decades