Assessing the risk of early unplanned rehospitalisation in preterm babies: EPIPAGE 2 study

BMC Pediatrics

Table 3 Predictive performance measures for logistic regression models constructed on 2707 eligible, complete-case babies in the EPIPAGE 2 cohort and validated using 10-fold cross-validation

Model	AUROC	95% CI	Sensitivity	Specificity	Tjur’s Coefficient	Hosmer-Lemeshow
One predictor	0.60	0.57–0.62	0.91	0.23	0.015	0.003
Ten predictor	0.62	0.59–0.65	0.77	0.42	0.019	<0.001
20 predictor	0.62	0.58–0.72	0.72	0.46	0.020	<0.001
Ten predictor (imputed data)^a	0.63 (IQR 0.004)	–	0.75 (IQR 0.01)	0.45 (IQR 0.01)	0.019 (IQR 0.001)	<0.001

^aPerformance measured over 20 imputed data sets and measures reported as median and inter-quartile range (IQR)

ISSN: 1471-2431