Comparison of different neonatal illness severity scores in predicting mortality risk of extremely low birth weight infants / 浙江大学学报·医学版

ABSTRACT

To compare different illness severity scores in predicting mortality risk of extremely low birth weight infants (ELBWI). From January 1st, 2019 to January 1st, 2020, all ELBWI admitted in the Children's Hospital of Nanjing Medical University, Nanjing Maternity and Child Health Care Hospital and the First Affiliated Hospital of Nanjing Medical University were included in the study. ELBWI with admission age ≥1 h, gestational age ≥37 weeks and incomplete data required for scoring were excluded. The clinical data were collected, neonatal critical illness score (NCIS), score for neonatal acute physiology version Ⅱ (SNAP-Ⅱ), simplified version of the score for neonatal acute physiology perinatal extension (SNAPPE-Ⅱ), clinical risk index for babies (CRIB) and CRIB-Ⅱ were calculated. The scores of the fatal group and the survival group were compared, and the receiver operating characteristic (ROC) curve was used to evaluate the predictive value of the above illness severity scores for the mortality risk of ELBWI. Pearson correlation analysis was used to analyze the correlation between illness scores and birth weight, illness scores and gestational age. A total of 192 ELBWI were finally included, of whom 114 cases survived (survival group) and 78 cases died (fatal group). There were significant differences in birth weight, gestational age and Apgar scores between fatal group and survival group (all <0.01). There were significant differences in NCIS, SNAP-Ⅱ, SNAPPE-Ⅱ, CRIB and CRIB-Ⅱ between fatal group and survival group (all <0.01). The CRIB had a relatively higher predictive value for the mortality risk. Its area under the ROC curve (AUC) was 0.787, the sensitivity was 0.678, the specificity was 0.804, and the Youden index was 0.482. The scores of NCIS, SNAP-Ⅱ, SNAPPE-Ⅱ, CRIB and CRIB-Ⅱ were significantly correlated with birth weight and gestational age (all <0.05). The correlation coefficients of CRIB-Ⅱ and CRIB with birth weight and gestational age were relatively large, and the correlations coefficients of NCIS with birth weight and gestational age were the smallest (0.191 and 0.244, respectively). Among these five illness severity scores, CRIB has better predictive value for the mortality risk in ELBWI. NCIS, which is widely used in China, has relatively lower sensitivity and specificity, and needs to be further revised.