Precision and rationale of thresholds on the Predicted Risk Of New Event (PRONE) score for three indicative interventions
| Intervention | Tolerance for missing practitioners who will recur in short-term | Tolerance for netting practitioners who will not recur in short-term | Rationale | Implications for threshold on PRONE score | Nominal thresholds on PRONE score | Sensitivity/specificity (%) |
|---|---|---|---|---|---|---|
| Advise doctor of future complaint risk | Low | High | Prefer lower false negative rate (quality-of-care considerations); false positives not problematic (cheap to implement, minimally confronting) | High sensitivity desirable, specificity level subordinate | ≥3 | 94/19 |
| Exhortation or compulsion to undertake CME | Moderate | Moderate | Prefer lower false negative rate (quality-of- care) and lower false positive rate (absorbs resources, may waste practitioners’ time) | Balance between sensitivity and specificity | ≥5 | 69/58 |
| Refer to regulator for further action | High | Low | Must minimise false positive rate (natural justice); false negative rate undesirable (quality-of-care) but trumped | High specificity essential, sensitivity level subordinate | ≥12 | 13/99 |
CME, continuing medical education.