Aims: To calculate the population-attributable risk (PAR) of coronary events (CE) from 10 risk factors, during long-term follow-up.
Methods: We used both case-cohort and case-control analyses for calculation of PAR in relation to 10 baseline risk factors. First CE (fatal or nonfatal, n=3072) in 22,444 males and 10,902 females was recorded during a mean follow-up of 20 years by use of national registers.
Results: Using a Cox regression analysis in a case-cohort design, smoking (prevalence in men 49%, women 37%) was the strongest risk factor, RR 2.29 (95% CI 2.09-2.52; PAR 39%), followed by hypercholesterolaemia, RR 1.70 (95% CI 1.56-1.86; PAR 18%), and diabetes, RR 1.67 (95% CI 1.41-1.99; PAR 3%). For women the strongest risk factors were smoking, RR 3.16 (95% CI 2.50-3.98; PAR 44%), diabetes, RR 2.59 (95% CI 1.78-3.76; PAR 6%), and hypertension, RR 2.47 (95% CI 1.94-3.14; PAR 23%). In men, smoking was the strongest predictor both after 10 years [RR 2.69 (95% CI 2.23-3.24)] and 20 years [RR 2.45 (95% CI 2.15-2.79)], followed by hypercholesterolaemia (RR 2.16-1.63), hypertension (RR 2.04-1.51), and diabetes (RR 1.85 -1.47). The case-control design gave very similar results. Total PAR varied from 74% (fully adjusted Cox regression, case-control, in men) to 116% in women (case-cohort).
Conclusion: Smoking is the most important long-term risk factor for CE in both genders, based on data from a population with a high proportion of smokers. Ten measured variables explained almost all variation in risk and could be used as a basis for intervention programmes.