admissions are for mildly ill infants, meaning that a significant proportion of overall spending is for a less severely ill newborn population.5 Understanding variation in practice patterns and outcomes across both high- and low-risk newborns helps clinicians, health systems, and other partners and policymakers identify opportunities for both better care and higher value. A REPORT OF THE DARTMOUTH ATLAS PROJECT 7 Newborn Study Populations in this Report Very low birth weight newborns Very low birth weight (VLBW) newborns, defined as those with a birth weight less than 1,500 grams (about 3 pounds 5 ounces), are commonly premature (< 37 weeks gestation), but this group also includes older infants that are born small for gestational age. These smallest and, typically, most premature newborns are at the greatest risk for complications and need the highest level of care in a NICU. Immediate respiratory support is required at birth, usually with a ventilator. They are cared for in an isolette to maintain their body temperature. Cardiopulmonary monitoring is also essential, as apnea (pauses in breathing) and bradycardia (drops in the heart rate) are common. Nutritional needs cannot be met through bottle or breastfeeding, and VLBW newborns initially require intravenous fluids and nutrition with slow advancement of oral feedings to avoid serious, and sometimes fatal, bowel complications. Finally, as these newborns grow, vision and development must be assessed; these services may not be available at lower level NICUs. Level III/IV NICUs, with neonatal nurses, neonatologists, and other sub-specialty physicians, provide the highly specialized care necessary for VLBW newborns. These units also have readily available specialized equipment and support services such as pharmacy, radiology, and respiratory therapy. This combination of specialized clinicians and services improves outcomes for these tiny patients, as shown in a recent research paper demonstrating improved outcomes in VLBW newborns delivered in hospitals with Level III/IV NICUs.1 1. Lasswell S, Barfield WD, Rochat R, Blackmon L. Perinatal regionalization for very low birth weight and very preterm infants: a meta-analysis. JAMA. 2010;304[9]:992–1000. Late preterm newborns The typical length of human gestation is 40 weeks, with those born before the 37th week considered premature. Although much is known about infants born very or extremely premature, those born just a few weeks early, designated as late preterm (34-36 weeks gestation), are not well studied. This group of patients is at lower risk of major complications but frequently has important medical needs. The developing fetus gains skills required for transition to extra-uterine life in the third trimester. These skills include the ability to regulate their temperature, to safely feed by mouth, and to breathe without assistance. Like all developmental milestones, these are achieved at varying times within a range of normal development. Therefore, an infant born in this late preterm window may have all or none of the skills needed for a safe transition to the extra-uterine environment. This uncertainty at the time of birth leads to the need for flexibility in the care of these newborns. Some may need intensive monitoring and respiratory support, while others will be able to breathe, feed, and stay warm without any assistance. Provision of respiratory support generally requires a Level III/IV NICU, but support for feeding and temperature regulation can be provided in less specialized units such as a Level II NICU. The wide variation in normal development and transition, and therefore medical need, for late preterm infants often creates uncertainty regarding the best location and level of care required. Other newborn groups VLBW and late preterm newborns are two distinct newborn groups, one with high risk of mortality and morbidity, the other with low risk. This report also uses other groupings of risk that depend on the topic of interest and the availability of data. The national studies using Vital Records information identify differing newborn risk by multiple categories of birth weight and gestational age. The Norwegian studies use populations defined by gestational age. The analysis of Anthem-insured newborns includes a group termed “low risk,” which is defined as those newborns with an absence of serious medical diagnoses and surgical procedures. A Report of the Dartmouth Atlas Project 8 DARTMOUTH ATLAS OF NEONATAL INTENSIVE CARE A Framework for Interpreting Variation in Health Care Measuring variation in health care is motivated primarily by an interest in understanding the causes and consequences of differences in the performance of health care providers and systems. Health care is expected to vary to the extent that populations differ in their needs and preferences for health care. Unwarranted variation is the variation that cannot be explained by population needs or preferences, but rather is due to differences in health system performance. Over the past two decades, a classification system for unwarranted variation was developed by Wennberg and colleagues,37 with variation in health care utilization categorized into three types: effective care, preference-sensitive care, and supply-sensitive care. Variation in health care capacity, such as hospital beds and physicians, is a fourth category.