Table 3. Clinical consequences of using prediction models to guide antibiotic prescription.
| Rotterdam, n = 248 | Coventry, n = 301 | |
|---|---|---|
| Observed antibiotic prescription, n (%) | 51 (21%) | 105 (35%) |
| Predictions by Nijman's model | ||
| Threshold 10% | Rotterdam | Coventry |
| Number of children below threshold (low-risk group) | 130 (52%) | 193 (64%) |
| Expected antibiotic prescription when guided by threshold (benefit) | 35 (14%) | 49 (16%) |
| Expected under treatment when prescription was guided by threshold (harm)a | 5 (2%) | 15 (5%) |
| Threshold 15% | ||
| Number of children below threshold | 167 (67%) | 229 (76%) |
| Expected antibiotic prescription when guided by threshold | 28 (11%) | 36 (12%) |
| Expected under treatment when prescription was guided by thresholda | 8 (3%) | 22 (7%) |
| Predictions by Oostenbrink's model | ||
| Threshold 10% | Rotterdam | Coventry |
| Number of children below threshold | 69 (28%) | 94 (31%) |
| Expected antibiotic prescription when guided by threshold | 44 (18%) | 77 (26%) |
| Expected under treatment when prescription was guided by thresholda | 0 (0%) | 8 (3%) |
| Threshold 15% | ||
| Number of children below threshold | 110 (44%) | 178 (59%) |
| Expected antibiotic prescription when guided by threshold | 35 (14%) | 51 (17%) |
| Expected under treatment when prescription was guided by thresholda | 2 (1%) | 13 (4%) |
| Predictions by Irwin's model | ||
| Threshold 10% | Rotterdam | Coventry |
| Number of children below threshold | 100 (40%) | 155 (51%) |
| Expected antibiotic prescription when guided by threshold | 38 (15%) | 64 (21%) |
| Expected under treatment when prescription was guided by thresholda | 5 (2%) | 15 (5%) |
| Threshold 15% | ||
| Number of children below threshold | 120 (48%) | 198 (66%) |
| Expected antibiotic prescription when guided by threshold | 33 (13%) | 48 (16%) |
| Expected under treatment when prescription was guided by thresholda | 8 (3%) | 22 (7%) |
a Number of children with a bacterial infection who were treated with antibiotics, but who were classified as low-risk according to the used prediction model and threshold