Table 3:
External validation of clinical prediction rules in terms of diagnostic accuracy and use of rapid antigen detection testing (n = 676)
| Clinical prediction rule | Area under the ROC curve of the score (95% CI) | Diagnostic accuracy of the rules-based strategy | Rapid antigen detection testing* | |||
|---|---|---|---|---|---|---|
|
|
|
|||||
| Sensitivity, % (95% CI) | Specificity, % (95% CI) | C–index (95% CI) | % (95% CI) | p value† | ||
| Breese39 | 0.60 (0.55–0.64) | 88 (84–92) | 82 (78–86) | 0.85 (0.82–0.88) | 76 (73–80) | 0.01 |
|
| ||||||
| Wald42 | 0.58 (0.53–0.62) | 94 (91–97) | 63 (58–68) | 0.78 (0.76–0.81) | 65 (61–69) | < 0.001 |
|
| ||||||
| Edmond38 | NA | 66 (61–72) | 86 (82–89) | 0.76 (0.73–0.79) | 48 (44–52) | < 0.001 |
|
| ||||||
| McIsaac40 | 0.56 (0.52–0.60) | 94 (92–97) | 54 (49–59) | 0.74 (0.71–0.77) | 52 (48–56) | < 0.001 |
|
| ||||||
| Attia43 | 0.62 (0.58–0.66) | 87 (83–91) | 88 (85–91) | 0.87 (0.85–0.90) | 86 (84–89) | 0.99 |
|
| ||||||
| Joachim37 | 0.59 (0.55–0.63) | 88 (84–92) | 40 (35–45) | 0.64 (0.61–0.67) | 24 (21–27) | < 0.001 |
Note: CI = confidence interval, NA = not available.
*
Use of rapid antigen detection testing corresponds to the number of patients that would undergo testing after the application of the clinical prediction rule.
†
One-sided exact binomial probability test of whether absolute use of rapid antigen detection testing was < 80%.