Table 5.

Comparison of predictive models for respiratory syncytial virus hospitalization in infants born 32/33–35 wGA

	FLIP [91]	FLIP 2 [93]	PICNIC [92]	RISK [10]	RISK-II [11]	PONI [12]
Risk factors	7 Birth ± 10 weeks of season start Birth weight Breast feeding ≤2 months Number of siblings ≥2 years Number of family members with atopy Number of family members with wheeze Sex	4 Birth ± 10 weeks of season start School-age siblings or day care attendance Mother smoking during pregnancy Sex	7 Small (<10th percentile) GA Sex Born during RSV season (Nov–Jan) Family history without eczema Subject or siblings attending day care >5 individuals in the home, including the subject >1 smoker in the household	4 Born Aug 14th to Dec 1st Presence of siblings or subject day care attendance Breast fed ≤2 months or not Atopy in 1st degree family member	5 Birth between Aug 14th and Dec 1st Day care attendance and/or siblings Neonatal respiratory support Breastfeeding ≤4 months Maternal atopic constitution	6 Age on 1st October ≤3 months Smoking of family members Age of mother at delivery ≤25 years Children 4–5 years old present Smoking of mother during pregnancy Subject day care attendance
Sensitivity/specificity	0.72/0.71	0.062/0.99	0.68/0.72	0.46/0.79	Low risk (1% hospitalization): 0.90/0.35 High risk (13% hospitalization): 0.32/0.90	NR
ROC AUCa	0.791	0.687	0.762	0.703	0.72	0.755

GA gestational age, NR not reported, ROC AUC area under the receiver operating characteristic curve

^aROC curves are constructed by plotting the sensitivity (true positives; number of RSV hospitalized infants predicted to be hospitalized) against the specificity (false positives; number of non-hospitalized infants predicted to be RSV hospitalized), with areas closer to one representing better predictive accuracy