Cut point | No. of participants | Incidence of SSO, n (%)* | % Threshold sensitivity (95% CI)† | %Threshold specificity (95% CI)† | %Threshold admission† |
---|---|---|---|---|---|
Current practice‡ | 1415 | 9.5 | 51.9 | 55.8 | 45.0 |
OCRS score§ | 1415 | ||||
0 | 614 | 28 (4.6) | 100 (0.98–1.0) | 0 (0–0) | 100 |
1 | 123 | 10 (8.1) | 79.3 (0.72–0.85) | 45.8 (0.45–0.46) | 56.6 |
2 | 346 | 40 (11.6) | 71.9 (0.64–0.79) | 54.6 (0.54–0.55) | 47.9 |
3 | 120 | 11 (9.2) | 42.2 (0.34–0.50) | 78.5 (0.78–0.79) | 23.5 |
4 | 147 | 31 (21.1) | 34.1 (0.27–0.42) | 87.0 (0.86–0.88) | 15.0 |
5 | 40 | 10 (25.0) | –¶ | –¶ | –¶ |
6 | 13 | 3 (23.1) | – | – | – |
7 | 10 | 1 (10.0) | – | – | – |
8 | 1 | 0 (0.0) | – | – | – |
10 | 1 | 1 (100.0) | – | – | – |
Note: CI = confidence interval, SSO = short-term serious outcome.
↵* Incidence of SSO if individual patient has this total score (e.g., a patient with a score of 3 has a 9.2% probability of a SSO).
↵† Estimated proportion for sensitivity, specificity and hospital admission if admission threshold was equal to or greater than the specific total score (e.g., admission 47.9% at a point total threshold ≥ 2).
↵‡ Estimates based on actual SSO and hospital admission rates at the 6 study sites.
↵§ Potential scores range from 0 to 16.
↵¶ Thresholds > 5 were not clinically reasonable because of poor sensitivity.