clinicians.22 The care required to improve outcomes of high-risk pregnancies can range from monitoring the mother’s and fetus’ wellbeing to delivering in a tertiary center that can provide the highest level of care necessary for the mother and newborn. Just A REPORT OF THE DARTMOUTH ATLAS PROJECT 5 as levels of care have been defined for pregnant women, hospital newborn units are generally classified as Level I (well newborn nursery), Level II (special care nursery), Level III (neonatal intensive care unit), and Level IV (regional neonatal intensive care unit), reflecting their capabilities for providing medical and surgical care.23 Ideally, pregnant women and newborns receive care in the most appropriate setting, aided by systems of maternal and newborn transport. This concept of regionalization24 has been stressed by two trends: the expansion of NICUs and the high financial margins for obstetrical and NICU services. While the growth and dissemination of NICUs from academic tertiary centers to hospitals in community settings affords better access to specialized care closer to home, it is not a guarantee that newborns receive care in the most appropriate unit level. Table 1.2 shows the expansion of NICU beds and neonatologists in recent decades. In many cities, there are multiple NICUs within the same area, leading to competition and, in some instances, relatively low patient volumes. This is a particular concern for very premature infants, as newborn mortality is lower in high-volume Level III or IV units.25 The second and related trend is the commercialization of maternal and NICU care. Not all very premature newborns are born in higher-level NICU hospitals, and some have suggested that the loss of revenue by the sending hospital acts as a disincentive to transfer pregnant women.26 On the newborn side, most NICUs are “high-margin” services.27,28 This is a strong incentive for further building and expansion of NICUs and for keeping beds full, potentially leading to overuse of services, especially in lower-risk newborns. Finally, when premature newborns are stable and growing, there are missed opportunities to transfer back to a hospital closer to home. The lack of insurance reimbursement for “back transfers” of newborns is an important barrier to family-centered care for many families.29 Table 1.2. Change in the supply of NICU beds and neonatologists, 1995 to 2013 Intensive NICU beds per 1,000 live births 1995 2013 Percent change All live births 3.4 5.7 69% ≥ 2,500 grams 0.08 0.13 66% < 2,500 grams 46.0 72.6 58% < 1,500 grams 280.6 466.0 66% Live births per neonatologist 1981 1996 2013 Percent change 1996-2013 All live births 7,201 1,687 965 -43% ≥ 2,500 grams 6,712 1,565 888 -43% < 2,500 grams 490 123 77 -37% < 1,500 83 23 14 -40% Sources: AMA Masterfile, AHA Survey, U.S. Vital Records. The number of NICU beds increased almost 70% per newborn during the 18-year period from 1995 to 2013. As the number of neonatologists increased from 1981 to 2013, the number of neonatologists for each newborn fell dramatically. A Report of the Dartmouth Atlas Project 6 DARTMOUTH ATLAS OF NEONATAL INTENSIVE CARE The Opportunity to Achieve the Triple Aim with the Guidance of Population-Based Newborn Analysis The Triple Aim—better care, better health, and lower costs—provides a useful framework for understanding opportunities to improve newborn care.30-32 Better care can be understood as both high technical quality and better patient and family experiences. Higher quality is associated with better health: higher survival, fewer adverse events, and improved long-term outcomes. Providing the right care at the right place at the right time limits waste and lowers costs.33,34 Joined with better outcomes, efficient use of medical services is the basis of high-value care. How does analysis of population-based health care data help achieve these aims? Descriptions and investigation of regional and hospital variation provide rich information about the delivery of newborn care, particularly neonatal intensive care, at different hospitals by different providers; geographic analysis can reveal the practice styles of hospitals and physician groups within each region. Information about specific NICUs can be obtained using measurement at the hospital level (although the hospitals are not identified), the most important locus for care improvement. In turn, these findings can raise questions about the reasonableness of current practice patterns, stimulating provider engagement as well as public discussion. The measures can also show what is attainable in terms of quality and efficiency and can offer benchmarks to guide clinical improvement and policy development. The findings often generate ideas regarding both the causes and effects of the variations, and, in turn, the data can then be used to test hypotheses as to the best ways to deliver care. Finally, the measurement set can support public reporting of performance measures, which accelerates the pace of improvement.35 Good care can be expensive, but using health care resources efficiently is also an important aim.33 In a California study, very low birth weight (VLBW) newborns represented less than 1% of all births but accounted for 36% of total newborn hospital payments.36 While the sickest newborns (e.g., VLBW) have the highest payments, most NICU