Table 2.
Regression coefficients derived from generalised Cox regression for final model
| Variable | Coefficient (SE) | P-Value |
|---|---|---|
| Baseline Hazard 1 | −11.20 (3.95) | 0.01 |
| Baseline Hazard 2 | −7.88 (3.82) | 0.04 |
| Baseline Hazard 3 | −7.595 (3.83) | 0.05 |
| NPH (cogd, time):1 | −1.597 (1.93) | 0.41 |
| NPH (cogd, time):2 | 2.53 (0.70) | <0.001 |
| NPH (cogd, time):3 | −3.28 (1.03) | 0.002 |
| NPH (cogd, time):4 | 2.04 (0.97) | 0.04 |
| Cognitively inactive (yes vs. no) | 0.29 (0.23) | 0.21 |
| Physically inactive (yes vs. no) | −0.10 (0.24) | 0.68 |
| BMI | 0.02 (0.03) | 0.55 |
| SES (Low vs. High) | 0.15 (0.24) | 0.52 |
| Gender (Female vs. Male) | −0.33 (0.23) | 0.15 |
| Years of education | −0.03 (0.06) | 0.64 |
| Age at baseline | 0.06 (0.05) | 0.19 |
| Stroke ever (yes vs. no) | 0.02 (0.32) | 0.96 |
| High BP ever (yes vs. no) | −0.04 (0.23) | 0.86 |
| Diabetes ever (yes vs. no) | 0.34 (0.31) | 0.27 |
| Zygosity (yes vs. no) | −0.12 (0.22) | 0.59 |
| ApoEe4 (yes vs. no) | 0.62 (0.24) | 0.01 |
Note: NPH- nonproportional hazards; cogd – cognitive dispersion; BMI – body mass index; SES – social economic status; cogd was modelled with an NPH spline with one knot positioned at 4 (study time). Statistically significant values are shown in bold.