66 Evaluating Illness Severity for Very Low Birth Weight Infants: Crib or Crib-II ?

Abstract

Background: The clinical risk index for babies (CRIB) is a risk-adjustment instrument used worldwide to determine illness severity in infants born at less than 31 weeks' gestation, or having 1500 g birth weight or lower. However, the appropriateness of CRIB has been questioned, and CRIB has very recently been updated and simplified into a five-items scoring system (CRIB-II) for infants born at less than 32 weeks' gestation, in order to improve its accuracy. Here, we compared the accuracy of CRIB and CRIB-II scores in predicting mortality for VLBW infants in a small unit. Methods: A total of 147 NICU-admitted VLBW infants with data available for both CRIB and CRIB-II were enrolled in the study (M:70, F:77; mean ± SD: 28.6 ± 2.5 weeks; birth weight: 1070 ± 325 g). Major non-lethal malformations were present in 3 (2.04%) infants and minor birth defects were observed in 7 (4.76%). Extremely low birth weight infants were 59 (40.15% of total; M:27, F:32). Newborns with lethal congenital abnormalities, metabolic disorders or undergoing complex cardiac surgery were excluded. Both CRIB and CRIB-II scores were calculated for each newborn. Neonatal mortality (death before hospital discharge) was selected as the outcome measure. Results: The median CRIB and CRIB-II scores were 2.0 (range:0–16), and 8.0 (range:3–20), respectively. A significant positive correlation between CRIB and CRIB-II was observed (P <0.0001), with a mortality rate of 11.56% (17/147). Newborns with unfavourable outcome (M:8, F:9) had a significantly lower birth weight (696±205 g vs. 1130±300; p<0.0001), gestational age (25.7±2.9 weeks vs. 29.08±2.1 weeks; p<0.0001), CRIB (median [25th-75th percentiles]: 8.0 vs. 2.0, P<0.0001) and CRIB-II (14.0 vs. 8.0, P <0.0001) than surviving infants. Mean AUCs for CRIB, CRIB-II, gestational age and birth weight in identifying neonatal mortality in VLBW infants ranged from 0.924 (CRIB) to 0.869 (gestational age). No significant differences were observed for the AUCs of CRIB vs. CRIB-II (P=0.34), CRIB vs. gestational age (P=0.18), CRIB vs. birthweight (P=0.24), CRIB-II vs. gestational age (P=0.62), or CRIB-II vs. birth weight (P=86). Conclusion: Our main findings suggest that CRIB and CRIB-II have comparable accuracy in predicting neonatal mortality in VLBW infants. Although our observed mortality rate is higher than that reported for the original CRIB-II population, differences between the two proportions are not statistically significant (P =0.16).