Introduction
The QUiPP app is a clinical risk-prediction tool that we developed and validated (Kuhrt et al.) for the prediction of spontaneous preterm birth (sPTB). The algorithm combines results from predictive tests (quantitative fetal fibronectin and cervical length) with clinical risk factors to generate a predicted risk value for delivery <30, <34 and <37 weeks’ gestation, and within 7 or 14 days.
The app was created using an asymptomatic cohort of women at high risk of sPTB, some of whom had received prophylactic intervention (cerclage / progesterone). Our objective was to establish whether the app accurately predicts risk, or underestimates risk for women who do not receive intervention, secondary to the treatment paradox.
Methods
We performed a retrospective sub-analysis of women who did not receive intervention, at two time points (the first visit between 18+0-21+6 and 22+0–27+6 weeks’). Risk of delivery before 34 weeks’ was calculated with the algorithm, and using a clinically important threshold of 5% (<5% represented ‘low-risk’ and >5% ‘high-risk'). Predicted event rates were compared with actual rates of sPTB (%).
Results
All women who delivered early in our ‘high-risk’ group (risk >5%) were appropriately identified [early gestation: 14.2 v 6.0% (22/367), p=0.000001; late gestation 13.4 v 7.6% (29/383), p=0004]. The number of women who delivered before 34 weeks’ gestation (actual event rate) was lower than that predicted by the app.
The event rate in the ‘low-risk’ group (risk <5%) was the same or lower than predicted [early gestation: 2.5 v 0.8% (4/473), p=0.01; late gestation: 2.2 v 1.7% (13/788), p=0.39].
Conclusion
The QUiPP app is a safe and accurate risk-prediction tool in women with and without intervention. If anything, it slightly overestimates risk of sPTB. The treatment paradox does not influence accuracy, and it can be relied upon to safely target care.