PEDIATRICS Vol. 108 No. 2 August 2001, pp. 426-431

Are Neonatal Intensive Care Resources Located According to Need? Regional Variation in Neonatologists, Beds, and Low Birth Weight Newborns

David C. Goodman, MD, MS^{*, §}, Elliott S. Fisher, MD, MPH^{, §,} , George A. Little, MD^*, Thérèse A. Stukel, PhD^§, and Chiang-hua Chang, MS^§

From the ^* Department of Pediatrics,  Department of Medicine, ^§ Department of Community and Family Medicine and Center for the Evaluative Clinical Sciences, Dartmouth Medical School, Hanover, New Hampshire; and  VA Outcomes Group, White River Junction, Vermont.

	ABSTRACT

Top Abstract Methods Results Discussion References

Objective.  Despite marked growth in neonatal intensive care during the past 30 years, it is not known if neonatologists and beds are preferentially located in regions with greater newborn risk. This study reports the relationship between regional measures of intensive care capacity and low birth weight infants using newly developed market-based regions of neonatal intensive care.

Design.  Cross-sectional small-area analysis of 246 neonatal intensive care regions (NICRs).

Data Sources.  1996 American Medical Association and American Osteopathic Association masterfiles data of clinically active neonatologists; 1999 American Academy of Pediatrics Section on Perinatal Pediatrics survey of directors of neonatal intensive care units in the United States with 100% response rate; 1995 linked birth/death data.

Results.  The number of total births per neonatologist across NICRs ranged from 390 to 8197 (median: 1722) and the number of total births per intensive care bed ranged from 72 to 1319 (median: 317). The associations between capacity measures and low birth weight rates across NICRs were statistically significant but negligible (R²: 0.04 for neonatologists; 0.05 for beds). NICRs in the quintile with the greatest neonatologist capacity (average of only 863 births per neonatologist) had very low birth weight (VLBW) rates of 1.5% while those in the quintile of lowest neonatologist capacity (average of 3718 births per neonatologist) had VLBW rates of 1.3%; a similar lack of meaningful difference in VLBW rates was noted across quintiles of intensive care bed capacity. Including midlevel providers and intermediate care beds to the analyses did not alter the findings.

Conclusions.  Neonatal intensive care capacity is not preferentially located in regions with greater newborn need as measured by low birth weight rates. Whether greater capacity affords benefits to the newborns remains unknown.  Key words:  neonatal intensive care, physicians, small-area analysis, low birth weight infants.

Neonatal intensive care has experienced tremendous growth in the past 30 years. This expansion in capacity grew out of neonatology's proven effectiveness in improving birth weight specific outcomes and was coupled to efforts to regionalize care through a system of unit classification and maternal and newborn transport.^1-5 Success in caring for sick newborns stimulated increases in neonatology training positions which, in turn, resulted in sustained growth in the number of neonatologists. Although in 1981 there were 83 very low birth weight ([VLBW]; <1500 g) infants per neonatologist in the United States, by 1996 there were 23, and the number of extremely low birth weight (< 1000 g) infants per neonatologist in 1996 was only 11.⁶ The ever larger number of neonatologists also permitted an accompanying growth in units and beds as neonatal intensive care units (NICUs) diffused from academic medical centers into smaller and smaller community hospitals.^7-9

The expansion in neonatal intensive care capacity has not occurred evenly across the United States.⁷ In 1983, Merenstein et al¹⁰ reported a 1.7-fold variation in the number of neonatologists per newborn across the nine census regions. More recently, Schwartz et al⁷ studied newborn special care beds across the same regions and found a 1.8-fold variation. Because these census regions are very large, they are likely to minimize the extent of variation. Using newly developed and much smaller health service areas termed neonatal intensive care regions ([NICRs]; N = 246), we have found a more than fourfold regional variation in clinically active neonatologists.¹¹

Regional variation in the capacity of neonatal intensive care raises the important question of whether areas with a greater number of ill newborns also have proportionately more intensive care resources. In this study, we describe the relationship between 2 measures of regional neonatal intensive care capacityneonatologists and bedsand the major determinant of neonatal medical need and mortality: the occurrence of VLBW births.

	METHODS

Top Abstract Methods Results Discussion References

Defining NICRs

As previously described,¹¹ we used standard methods of small-area analysis^12,13 to define 246 regions of neonatal intensive care. These areas differ from other geographic units used for workforce analysis (States, counties, metropolitan statistical areas, or Health Care Financing Administration-defined Health Service Areas) in that they were defined to specifically represent geographic markets for neonatal intensive care. Briefly, counties were assigned to regions with at least 1 neonatal intensive care unit based on the location of birth of VLBW infants. These NICRs were adjusted to minimize border crossing for care; as a result, the intensive care resources are highly matched to the newborn populations and per capita capacity measures in NICRs (neonatologists per newborn and intensive care beds per newborn) are unbiased measures of resource inputs.

Data Sources and Capacity Measures

The sources of neonatologist counts were the January 1, 1996 American Medical Association and American Osteopathic Association Masterfiles that represent a census of US physicians irrespective of association membership. Three thousand one hundred seventy-four physicians self-designated themselves as neonatologists. To arrive at an estimate of the clinically active neonatology workforce, we excluded neonatologists working the majority of their professional time in medical teaching (N = 97), administration (N = 100), or research (N = 232) and those working <20 hours per week (N = 118) leaving 2627 neonatologists. Residents and clinical fellows (N = 377) were included but were assigned a weight of 0.35 to account for their lower clinical productivity.^14,15 The median number of neonatologists per NICR was 6 with a range of 1 to 110 (interquartile range: 3-12).

To measure NICU beds, a 1999 survey of NICU directors was conducted in conjunction with the American Academy of Pediatrics (AAP) Section on Perinatal Medicine. Unit directors reported the number of neonatal intensive care beds, neonatal intermediate care beds, and neonatology physician extenders (eg, neonatal nurse practitioner, neonatal nurse clinician, registered neonatal clinician, physician assistant, etc). Each unit director received up to 3 mailings; nonresponders were contacted by telephone and fax with a final response rate of 100%.

The number of births by NICR and birth weight category was derived from the National Center for Health Statistics 1995 Linked Birth/Death Data Set.¹⁶ This file includes all US newborns and their birth weight as reported on birth certificates.

Lorenz Curves and GINI Indices

The GINI Index and its graphic counterpart the Lorenz curve were used to measure the extent of variation in neonatal intensive care capacity in relation to newborn populations.^17,18 The latter is a plot of the cumulative population of US newborns (across NICRs) by the cumulative number of neonatologists (or beds) per newborn in each NICR ordered from the lowest neonatologists (or beds) per newborn to the highest. If physicians (or beds) were equally distributed across NICRs in relation to the newborn population, the curve would be a linear diagonal. Inequality or variation in the distribution results in a concave line; greater variation results in greater deviation from the straight line. The GINI Index is equivalent to the ratio of 1) the area between the linear diagonal and the Lorenz curve, and 2) the total area under the diagonal. It varies from 0, where physicians are distributed in perfect evenness (the Lorenz curve is concordant to the diagonal), to 1, where the physician distribution is maximally uneven (also the condition of maximum Lorenz curve deviation).

	RESULTS

Top Abstract Methods Results Discussion References

Variation in Neonatal Intensive Care Capacity

Neonatal intensive care capacity varied substantially across NICRs (Fig 1) The number of total births per neonatologist ranged from 390 to 8197 (interquintile range of 1167 [20th percentile] to 2818 [80th percentile]; median: 1722) while the number of births per intensive care bed ranged from 72 to 1319 (interquintile range: 209-526; median: 317). Across NICRs, there was a low correlation between the number of newborns per neonatologist and the number of newborns per intensive care bed (Spearman 0.32; P < .01).

View larger version (64K): [in this window] [in a new window]   Fig. 1.   Total births per neonatologist and intensive care bed.

Relationship of Regional Capacity to Low Birth Weight Newborns

Figure 2 shows that there was a slight association across NICRs between beds and neonatologists per newborn and low birth weight rates. Although these are statistically significant (P < .01), only 4% of NICR variation in neonatologists (R² = 0.04) and 5% of the variation in beds (R² = 0.05) were explained by differences in low birth weight rates.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   VLBW infants (<1500 g) per 10 000 live births versus intensive care capacity per live births in NICRs (N = 246).

Similarly, the low and VLBW infant birth rates differed little across the quintiles of capacity (Table 1). For example, the NICRs within the high quintile of capacity had on average 863 births per neonatologist and a VLBW rate of 1.5% while the very low quintile (N = 49) had 3718 births per neonatologist and a VLBW rate of 1.3%.

These analyses were extended to other types of neonatal intensive care capacity. Adding neonatal midlevel providers to the number of neonatologists did not change the lack of meaningful relationship between clinician capacity and low or VLBW rates. Nor did adding intermediate to intensive care beds alter the lack of rate differences across bed quintiles (Table 1).

GINI indices and Lorenz curves (Fig 3) confirmed high degrees of disparities between intensive care capacity and the newborn population. GINI indices were 0.56 for both neonatologists and beds with respect to total live births. The indices were virtually unchanged after recalculating these with respect to VLBW infants (0.55 for both neonatologists and beds).

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Proportion of neonatal intensive care capacity by proportion of newborn population.

	DISCUSSION

Top Abstract Methods Results Discussion References

Regional variation in neonatal intensive care capacity is not explained by differences in patient need as manifest by low birth weight rates. Neither neonatologists nor intensive care beds varied substantially in accordance with the proportion of low birth weight infants. With neonatal capacity expressed as the number of VLBW infants per bed or neonatologist, the variation across the 246 NICRs remained more than fourfold.

Differences across small areas in the supply and utilization of medical care resources have been repeatedly demonstrated for both adults and children.^1318-21 Linking these resources to patient need has been more difficult given the general lack of patient-level data that directly accounts for illness levels. In this respect, studying neonatal intensive care offers important advantages in understanding the variation of medical services provided to children, because the needs of the birth cohort within each NICR can be accurately and completely ascertained through birth certificate data. We choose birth weight as a measure of need because it is both the single most important predictor of neonatal mortality and the birth characteristic that is used most extensively for perinatal planning purposes.²²

Is it possible that the regional variation in neonatal intensive care resources that we observed could be explained by errors in measurement? Although any data has noise, we believe that the data used in this project are better than most. In general the NICRs developed for this study reflect the geography of care well enough that there is little boundary crossing of VLBW infants from region of maternal residence to region of birth, or to region of terminal care.¹¹ In addition, although not reported in this article, we have also developed measures of neonatologists from a 1998 survey conducted by the AAP Section on Perinatal Pediatrics. The extent of regional variation is similarly high. Nor did we find that clinician capacity was better related to birth weight risk when neonatal midlevel providers were added to the neonatologists. We have previously reported that the regional dominance of academic medical centers staffed with neonatal fellows or neonatologists engaged predominantly in research or teaching does not explain the differences in neonatologists.¹¹ Although definitional problems remain in classifying beds as intensive care, the magnitude of variation was no lower when intermediate care beds from the AAP survey were included.

An alternative, though unlikely, explanation for regional variation is that capacity is located in accordance with patient needs that are unaccounted for by birth weight. Birth weight is generally used in epidemiologic studies as the more reliably ascertained proxy for prematurity. Other newborn illnesses that often require intensive care services include congenital anomalies, birth asphyxia, infection, and extreme hyperbilirubenemia. The regional incidence of these conditions may vary although the extent is not known for small areas and is generally unavailable for planning purposes. Given the more than four-fold variation across regions of similar need based on birth weight, an easily measured variable, it is hard to understand how neonatal intensive care capacity could have been planned and built rationally in accordance with these unmeasured illness levels.

It seems likely that intensive care units and neonatologists are disproportionately located in some areas for the same factors that are associated with greater capacity of other medical care resources. Physicians tend to settle close to where they are trained, in urban settings, and where household income and wealth are higher.²³ Hospital-based specialty services are more likely in teaching hospitals but also in community hospitals that seek to provide the full spectrum of clinical care associated with academic medical centers. The weak association between the number of neonatologists and the number of beds suggest that these factors have varying influence from NICR to NICR. Although none of these reasons are surprising or unreasonable by themselves, the end result is a remarkable level of resource variation.

Regional variation in pediatric care capacity that is poorly related to population disease burden should stimulate further examination of the provision of pediatric care by institutions and clinicians. Neonatal intensive care is effective at improving outcomes in premature infants.^3-5 It logically follows that there is a minimum level of capacity required for effective care, but it is not known if any regions of the United States fall below this level. If there are underserved regions, then addressing undercapacity should be a priority for state maternal and child health programs and medical communities. If some or many regions exceed the level of capacity needed to provide effective care, then it is important to understand what, if any, additional benefits derives from these higher levels of resource input.

Neonatal intensive care is, by nature, expensive; its value can only be understood in the context of ongoing unmet perinatal health care needs. What is clear is that the 2 most significant preventable birth outcomes, the occurrence of low birth weight and prematurity itself, cannot be affected by neonatal intensive care. The full value of present and future resources expended on neonatal intensive care can be more completely evaluated when the alternatives of improving preconceptual and prenatal care are also considered.

A greater understanding of the implications of neonatal intensive care capacity will come from studies that consider the broad range of newborn risk. Equally important would be analyses that relate capacity to newborn outcomes adjusted for risk. Results from these studies would make a substantial contribution to understanding whether regions with fewer resources are underserved, or whether regions of greater resources are at risk of over treatment.

	ACKNOWLEDGMENTS

This research was supported by a grant from the Robert Wood Johnson Foundation.

We would like to express our appreciation for the collaborative efforts of the AAP Section on Perinatal Pediatrics, in particular Dr Dilip Bhatt, in the survey of NICU directors.

	FOOTNOTES

This article has been presented, in part, at the 2000 Annual Meeting of the American Academy of Pediatrics; October 28-November 1, 2000; Chicago, IL.

Received for publication Jan 22, 2001; accepted Mar 30, 2001.

Reprint requests to (D.C.G) Department of Pediatrics and Department of Community and Family Medicine and Center for the Evaluative Clinical Sciences, Dartmouth Medical School, 7251 Strasenburgh Hall, Hanover, NH 03755. E-mail: david.goodman{at}dartmouth.edu

	ABBREVIATIONS

VLBW, very low birth weight; NICU, neonatal intensive care unit; NICR, neonatal intensive care region; AAP, American Academy of Pediatrics.

	REFERENCES

Top Abstract Methods Results Discussion References

1. Committee on Perinatal Health. (TIOP I) Toward Improving the Outcome of Pregnancy: Recommendations for the Regional Development of Maternal and Perinatal Health Services. White Plains, NY: March of Dimes Birth Defects Foundation; 1976
2. Little G, Merenstein G Toward improving the outcomes of pregnancy, 1993: perinatal regionalization revisited. Pediatrics. 1993; 92:611-612 [Abstract/Free Full Text]
3. Williams R, Chen P Identifying the sources of the recent decline in perinatal mortality rates in California. N Engl J Med 1982; 306:207-214 [Abstract]
4. Richardson D, Gray J, Gortmaker S, Goldmann D, Pursley D, McCormick M Declining severity adjusted mortality: evidence of improving neonatal intensive care. Pediatrics 1998; 102:893-899 [Abstract/Free Full Text]
5. Hack M, Klein N, Taylor H Long-term developmental outcomes of low birth weight infants. Future Child. 1995; 5:176-196 [CrossRef][Medline]
6. Goodman D, Little G General pediatrics, neonatology, and the law of diminishing returns. Pediatrics. 1998; 102:396-399 [Free Full Text]
7. Schwartz R, Kellog R, Muri J Specialty newborn care: trends and issues. J Perinatol 2000; 20:520-529 [CrossRef][Medline]
8. Chance G Regionalized care established by neonatal intensive care units. Pediatr Ann 1995; 25:519-528
9. Gagnon D, Allison-Cooke S, Schwartz R Perinatal care: the threat of deregionalization. Pediatr Ann 1988; 17:447-452 [Medline]
10. Merenstein G, Rhodes P, Little G Personnel in neonatal pediatrics: assessment of numbers and distribution. Pediatrics. 1985; 76:454-456 [Abstract/Free Full Text]
11. Goodman D, Fisher E, Little G, Stukel T, Chang C. The uneven landscape of newborn intensive care services: variation in the neonatology workforce. Effective Clinical Practice. In press
12. Wennberg J, Gittelsohn A Small area variations in health care delivery. Science. 1973; 182:1102-1108 [Abstract/Free Full Text]
13. Wennberg J, Cooper M, series eds. The Dartmouth Atlas of Health Care in the United States. Chicago, IL: American Hospital Association; 1996
14. Graduate Medical Education National Advisory Committee. Report of the Graduate Medical Education National Advisory Committee to the Secretary, Department of Health and Human Services, Volume 1. Washington, DC: Office of Graduate Medical Education, Health Resources Administration, Public Health Service, US Department of Health and Human Services; 1980
15. Jacobsen S, Rimm A The projected physician surplus reevaluated. Health Aff. 1987; 6:48-56 [CrossRef][Medline]
16. National Center for Health Statistics. Linked Birth/Death Data Set. Hyattsville, MD: Centers for Disease Control; 1995
17. Todaro M. Economic Development of the Third World. New York, NY: Longman; 1989
18. Chang RK, Halfon N Geographic distribution of pediatricians in the United States: an analysis of the fifty states and Washington, DC. Pediatrics. 1997; 100:172-179 [Abstract/Free Full Text]
19. Perrin J, Homer C, Berwick D, Woolf A, Freeman J, Wennberg J Variations in the rates of hospitalization of children in three urban communities. N Engl J Med 1989; 320:1183-1187 [Abstract]
20. Goodman DC, Fisher ES, Gittelsohn A, Chang CH, Fleming C Why are children hospitalized? The role of non-clinical factors in pediatric hospitalizations. Pediatrics. 1994; 93:896-902 [Abstract/Free Full Text]
21. Goodman D, Fisher E, Bubolz T, Mohr J, Poage J, Wennberg J Benchmarking the US physician workforce: an alternative to needs-based or demand-based planning. JAMA. 1996; 276:1811-1817 [Abstract/Free Full Text]
22. Paneth N The problem of low birth weight. Future Child. 1995; 5:19-34 [CrossRef][Medline]
23. Wennberg J, Cooper M, series eds. The Dartmouth Atlas of Health Care1998. 2nd ed. Chicago, IL: American Hospital Association; 1997
24. Seifer SD, Vranizan K, Grumbach K Graduate medical education and physician practice location. JAMA. 1995; 274:685-691 [Abstract/Free Full Text]