Table 4. Population Attributable Risks of Incident Atrial Fibrillation Risk Factors Across Time Periods.
| Decade of Atrial Fibrillation Onset Age- and Sex-Adjusted Hazards Ratios (95% CI) |
Trend | |||||
|---|---|---|---|---|---|---|
| 1958-1967 | 1968-1977 | 1978-1987 | 1988-1997 | 1998-2007 | ||
| Current smoking | Not contributing | Not contributing | Not contributing | Not contributing | Not contributing | |
| Alcohol consumption | Not contributing | Not contributing | Not contributing | Not contributing | Not contributing | |
| Body mass index | 12•0 | 12•9 | 14•1 | 15•4 | 16•9 | ↑ |
| Systolic blood pressure | 47•3 | 33•0 | 19•9 | 8•3 | -2•1 | ↓ |
| Hypertension treatment | 9•8 | 12•0 | 14•5 | 17•0 | 19•5 | ↑ |
| Diabetes | 3•2 | 3•7 | 4•4 | 5•1 | 5•9 | ↑ |
| Electrocardiographic LVH | 10•4 | 6•9 | 4•5 | 2•9 | 1•8 | ↓ |
| Significant heart murmur | 21•9 | 15•4 | 10•1 | 6•1 | 3•1 | ↓ |
| Heart failure | 7•8 | 5•6 | 3•9 | 2•5 | 1•4 | ↓ |
| Myocardial infarction | 3•6 | 3•6 | 3•6 | 3•6 | 3•6 | - |
For risk factors without trend in prevalence, the average predicted prevalence from logistic models with age and sex as predictors (Table 2) were used in population attributable risk calculation; for factors with trend in prevalence, the average predicted prevalence from logistic models with age, sex and period as predictors were used in population attributable risk.
Risk factors with non-significant hazards ratios (p-value >0.05) were deemed not contributing (Supplementary Table 3-1). For risk factors without trend in hazards ratio, the constant hazards ratio estimated from Cox models with age, sex, period and risk factors as predictors (Supplementary Table 3-1) were used in population attributable risk calculation; for factors with trend in hazards ratio, the hazards ratios from Cox models with age, sex, period, risk factors, and period and risk factor interaction as predictors (Supplementary Table 3-2) were used in population attributable risk.