Table 2. Stroke Population Risk Tool (SPoRT)–Model.
| Hazard Ratio (95% CI) | ||
|---|---|---|
| Male Model* | Female Model* | |
| Age | 1.11 (1.09–1.13) | 1.11 (1.09–1.12) |
| Age spline (65 years) | 0.97 (0.95–0.99) | |
| Age time (per year) | 0.997 (0.995–0.9999) | 0.996(0.993–0.997) |
| SPoRT Behaviour Score** (per unit) | 1.12 (1.07–1.17) | 1.15 (1.11–1.19) |
| Hypertension | ||
| No | 1.0 [Reference] | 1.0 [Reference] |
| Yes | 1.37(1.16–1.60) | 1.39 (1.20–1.61) |
| Missing | 0.80 (0.11–5.88) | 1.53 (0.24–9.84) |
| Heart Disease | ||
| No | 1.0 [Reference] | 1.0 [Reference] |
| Yes | 1.36 (1.14–1.63) | 1.44 (1.22–1.71) |
| Diabetes | ||
| No | 1.0 [Reference] | 1.0 [Reference] |
| Yes | 1.29 (1.06–1.57) | 1.74 (1.45–2.09) |
| Missing | –- | |
| Survey cycle | ||
| 3.1 (2005) | 1.0 [Reference] | 1.0 [Reference] |
| 2.1 (2003) | 1.03 (0.85–1.26) | 1.05 (0.87–1.26) |
| 1.1 (2001) | 1.26 (1.04–1.53) | 1.18 (0.99–1.42) |
| Model Assessment | ||
| Discrimination | ||
| C-stat (95% CI) | 0.85 (0.83–0.86) | 0.87 (0.85–0.88) |
| Ratio of 75 to 25 risk percentile (5-year risk range) | 13.3 (0.11 to 1.40) | 14.0 (0.08 to 1.07) |
| Ratio of 95 to 5 risk percentile | 149.7 (0.03 to 4.79) | 179.2 (0.026 to 4.70) |
| Calibration | ||
| Subgroup differences No. (%) | 2 (3.0) † | 4 (5.7) ‡ |
*The full model was calibrated to survey cycle year
**0–9 for males, 0–11 for females
† Observed versus predicted estimates were compared for 67 subgroups—selected based on meeting the criteria of having more than 5% of total observed stroke events (i.e., more than 22 events). We report the number of subgroups where there was a clinically important difference (predefined as ≥ 20% difference) in observed versus predicted number of events. The 67 subgroups were: deciles of predicted risk (4), local health networks (9), age (7), body mass index (4), physical activity (3), alcohol consumption (6), smoking (4), diet (3), self perceived stress (4), ethnicity (1), family income (7), family education (4), high blood pressure (2), diabetes (2), heart disease (2) SPoRT Behaviour Score (5)
‡ Observed versus predicted estimates were compared for 65 subgroups—selected based on meeting the criteria of having more than 5% of total observed stroke events (i.e., more than 23 events). We report the number of subgroups where there was a clinically important difference (predefined as ≥ 20% difference) in observed versus predicted number of events. The 65 subgroups examined were: deciles (4), local health networks (8), age (6), body mass index (5), physical activity (3), alcohol consumption (5), smoking (4), diet (3), self perceived stress (4), ethnicity (1), family income (7), family education (3), high blood pressure (2), diabetes (2), heart disease (2), SPoRT Behaviour Score (6).