The CRIB score (CS) is a risk-adjustment statistical model derived from minimum and maximum FiO2, maximum base deficit and congenital malformation data, as well as gestational age (GA) and birthweight (BW). In the UK CS at≤ 12 hrs of age has been proposed as a better predictor of neonatal mortality (Mort) and morbidity than BW alone.

Objective: To compare the utility of CS to that of BW alone in predicting Mort and intraventricular hemorrhage (IVH) in the USA.

Design: CS data were collected prospectively for infants with BW 401-1500 gms born 7/94-6/95 who were admitted to the 12 Network centers. Logistic regression models were developed to compare the ability of CS and BW alone to predict risk of Mort and of IVH.

Results: CS were obtained on 1891/2502 infants (76%). 186/611 infants with incomplete CS died before 12 hrs of age (30%) and 306 (50%) were missing base deficit data. CS distribution was 0-1=31% with 1% Mort, 2-4=25% with 6.1% Mort, 5-9=24% with 13.5% Mort and 10-20=20% with 46.1% Mort. The observed percent Mort and IVH in the lowest and highest predicted risk groups based on logistic estimates were: Table

Table 1
Full size table

For the entire cohort BW alone was a better predictor of Mort than CS. If only infants with CS are considered, there was a small increase in the ability of CS compared to BW alone to predict Mort. Neither was very discriminating in predicting the risk of IVH.

Conclusion: Because the derivation of CS relies on physiologic data, its utility in predicting Mort is confined to infants who survive long enough for these data to be obtained. For large cohorts of VLBW infants BW alone remains the most useful predictor of Mort. Neither of the models is a good predictor of IVH.