DIGIROP-birth is an individualized prediction model for the risk of ROP treatment developed by Aldina Pivodic. It is freely available at www.digirop.com.
Swedish National Patient Registry data from infants screened for ROP (2007–2018) were analyzed with Poisson regression for time-varying data (gestational age (GA), postnatal age, birth weight, sex and important interactions), to develop an individualized predictive model for ROP treatment, DIGIROP-Birth. The model was validated, internally, and externally (US and European cohorts), and compared to four published prediction models.
The studied outcome was ROP treatment. The measures are estimated momentary and cumulative risks, hazard ratios with 95% confidence intervals, area under the receiver operating characteristic curve (AUC), sensitivity, specificity, positive (PPV) and negative predictive value (NPV).
Among 7,609 infants, GA 28.1 (SD 2.1) weeks, weight 1119 (SD 353) g, 3454 (45.4%) girls, 442 (5.8%) were treated for ROP, 142 (40.1%) of those born <24 gestational weeks. Irrespective of GA, the risk for receiving ROP treatment increased during postnatal weeks 8–12 (HR, 1.54/week; 95% CI, 1.39 to 1.70); thereafter, it decreased (HR, 0.70/week; 95% CI, 0.67 to 0.74). Validations of DIGIROP-Birth for GA 24–30 weeks showed high predictive ability for the model overall (AUC: internal 0.90, temporal 0.94, US external 0.87, European external 0.90), by calendar periods and by race/ethnicity. The sensitivity, specificity, PPV and NPV were numerically at least as high as those obtained from CHOP-ROP, OMA-ROP, WINROP and CO-ROP, models requiring more complex data.
We validated an individualized prediction model, GA 24-30 weeks, enabling early risk prediction of ROP treatment based on birth characteristics data. Postnatal age rather than postconceptional age was a better predictive variable for the outcome. The model is generalizable, accessible through an application (www.digirop.com) and has at least as good test statistics as other models requiring longitudinal neonatal data not always available for ophthalmologists.