Defining quality of care indicators for neonatal intensive care units independent of maternal risk factors.

Observed and birthweight-specific neonatal mortality rates have been used for assessing quality of neonatal care, but these are crude and affected by risk characteristics of the population served. Even when neonatal mortality rate is corrected for four risk factors, race, sex, birthweight, and multiple births, (California Data Research Facility, Santa Barbara, CA) it is possible that the corrected neonatal mortality rate is not comparable among institutions because of population differences not corrected for, eg, prenatal care. To analyze whether our high neonatal mortality rate is primarily dependent on population risk or quality of neonatal care, we used contemporaneous data collection by senior physicians and a microcomputer database system to construct indices of quality of care that are based on diagnoses graded according to disease severity. For the 1987/1988 academic year, we found: neonatal intensive care unit nosocomial infection rate, 20%; severe intraventricular hemorrhage per 100 very low birthweight infants (1500 g), 20%; bronchopulmonary dysplasia per 100 cases of severe respiratory distress syndrome, 27%; necrotizing enterocolitis per 100 neonatal intensive care unit discharges, 5%; air leak per 100 cases of severe respiratory distress syndrome, 21%; and neonatal mortality rate per very low birthweight delivery rate, 0.4. We propose that microcomputer, hospital-based analyses will improve comparisons of neonatal intensive care unit quality of care if appropriate indices can be sufficiently well-defined and shared.