Neonatal Intensive Care Units (NICUs) represent a very effect array of technologies, yet marked variation is evident in mortality, morbidity and resource utilization. Although this variation may represent differences in NICU populations, adjustments for birth weight, race, and outborn status reduce but do not eliminate it. It is important to distinguish whether these differences are due to severity of illness or to ineffective application of NICU technologies. No direct whether comparison have been performed due to lack of validated severity indices for neonatal intensive care. We have recently demonstrated that it is possible to measure illness severity and using the newly validated Score for Neonatal Acute Physiology (SNAP) and to quantify medical resource use using the Neonatal Therapeutic Intervention Scoring System (NTISS). We hypothesize that controlling for difference in admission illness severity between NICUs will eliminate some, but not all NICU differences in outcome. Specifically we anticipate no differences in mortality rates but that rates of NICU complications such as bronchopulmonary dysplasia (BPD, neonatal chronic lung disease), intraventricular hemorrhage (IVH), practice styles. Moreover we expect that therapeutic intensity and length of stay (and by inference costs), will vary markedly reflecting aggressive or conservative practice styles. Our goals are to demonstrate the amount of inter-NICU variation attributable to differences in population risk (birth weight and illness severity) and that due to differences in practice style. A second goal is to identify variation in practice that may increase costs without improving outcomes. We propose to examine variations in 1) mortality, 2) morbidity and 3) resource use in a multi-center prospective cohort study of illness severity among 7 NICUs in 2 adjacent states. We will enroll an newborns 1500 gm birth weight during a 2-year period, expected to total 2000, and 300 deaths. Study personnel will assess illness severity and resource use on days 1, 3, and 14 using SNAP and NTISS respectively. We will also collect length of stay (LOS) and 4 selected morbidities, BPD, IVH, NEC and nosocomial bacteremia. For each outcome we will document differences i crude incidence rates among NICUs,and then adjust for baseline population risk (birth weight, sex, race, illness severity). Persistent variation will provide evidence of differences in effectiveness of the NICU """"""""package"""""""" """"""""Further adjustment for known treatment-related risk factors (invasive catheters, mechanical ventilation, etc.) will identify the sources of disparity in outcomes. A second set of analyses will focus on variations in length of stay (LOS) and medical resource use when adjusted for birth weight and illness severity. Residual variation due to differences in complication rates and discharge policies will be documented. Finally, we will perform in depth analyses of 5 selected technologies for evidence of ineffective application or excessive use. The significance of this research will be the demonstration of important inter-institutional variation, in morbidity, and medical resource use that cannot be attributed to differences in NICU populations or severity of illness. This will facilitate efforts to improve outcomes and to reduce cost by targeting technologies or practices for trials of efficacy or efficiency.
