A comparison of two mortality risk scores for very low birthweight infants: clinical risk index for babies and Berlin score

Abstract

Objective. To compare the performance of two mortality risk scores for very low birthweight (VLBW) infants.

Design and setting. Cohort study in two university-associated neonatal tertiary care units.

Patients. The clinical risk index for babies and the Berlin score were assessed in 343 VLBW infants (below 1500 g) admitted to one center and 257 infants admitted to a second center between 1992 and 1996. An additional 572 VLBW infants admitted at the former center during 1978–1987 and 294 during 1988–1991 were studied for changes in risk-adjusted mortality over time.

Measurements and results. Goodness of fit was excellent for both scores in both centers. Ability of discrimination was similarly high for both scores. The area under the receiver operating characteristic curves for all 600 infants was 0.84 for the clinical risk index for babies, 0.82 for the Berlin score, and 0.77 for birthweight alone. Both scores discriminated less well in the present samples than they did in the populations in which they were developed. Applying the Berlin score in three periods revealed a continuous decrease in risk-adjusted mortality from 1978 to 1996, indicating improvement in neonatal care.

Conclusions. Both scores predicted death in hospital with high accuracy and interhospital reliability. Decreasing mortality in VLBW infants during the past 20 years requires regular reevaluation of existing scoring systems to avoid overestimation of mortality risk.