Table 2.
Dementia types and mortality outcomes in the dementia cohort dividing by a history of atrial fibrillation or not.
| Baseline characteristics | Non-AF history (N = 6,716) | With AF history (N = 1,679) | P-value |
|---|---|---|---|
| Dementia types (N, %) | <0.001a | ||
| Alzheimer disease + senile dementia | 5,768 (85.9%) | 1,312 (78.1%) | <0.001a |
| Vascular dementia | 1,145 (17.0%) | 482 (28.7%) | <0.001a |
| Other dementia | 930 (13.8%) | 261 (15.5%) | 0.08a |
| Mortality outcomes | |||
| All-cause deaths (N, %) | 2,549 (38.0%) | 641 (38.2%) | 0.87a |
| Incidence rate per 1,000 person-years (95% CI) | 78.2 (72.0–84.4) | 85.2 (81.9–88.5) | |
| Model 0: crude effect (HR, 95% CI) | 1 (reference) | 1.162 (1.139–1.187) | <0.001b |
| Model 1: adjusted effect* (HR, 95% CI) | 1 (reference) | 1.208 (1.142–1.277) | <0.001b |
| Age < 65 years* (HR, 95% CI) | 1 (reference) | 5.549 (1.148–26.81) | 0.033b |
| Age ≥ 65 years* (HR, 95% CI) | 1 (reference) | 1.194 (1.128–1.263) | <0.001b |
| P for interaction between age and AF: | 0.003c | ||
| Cardiovascular deaths (N, %) | 721 (10.7%) | 255 (15.2%) | <0.001a |
| Incidence rate per 1,000 person-years (95% CI) | 24.1 (22.3–25.8) | 31.6 (27.7–35.6) | |
| Model 0: crude effectƗ (SHR, 95% CI) | 1 (reference) | 1.210 (1.077–1.359) | <0.001b |
| Model 1: adjusted effect*Ɨ (SHR, 95% CI) | 1 (reference) | 1.185 (1.033–1.358) | <0.001b |
| Age < 65 years* (SHR, 95% CI) | 1 (reference) | 1.632 (1.007–55.29) | <0.001b |
| Age ≥ 65 years* (SHR, 95% CI) | 1 (reference) | 1.177 (1.026–1.349) | 0.020b |
| P for interaction between age and AF: | 0.038c | ||
AF, atrial fibrillation; CI, confidence interval; HR, hazards ratio; N, number; SHR, sub-distributional hazard ratio.
Model 0, crude effect.
Methods of statistical tests: aChi-square test, bCox proportional hazard regression, cinteraction analyses.
*Model 1 was adjusted for age, sex, CHA2DS2-VASc, chronic kidney disease, chronic obstructive pulmonary disease, beta-blockers, antiarrhythmic drugs, novel oral anticoagulants, and warfarin.
^{\hyphen\hskip-3pt{\rm I}}For estimating the SHR, competing risk of cardiovascular death was evaluated using Fine-Gray subdistribution hazard model: Non-death (no death event: 0) vs. cardiovascular deaths (main event: 1) vs. non-cardiovascular deaths (other death event: 2).