[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 34.203.245.76. Please contact the publisher to request reinstatement.
[Skip to Content Landing]
Figure 1.
Range of Neonatal Intensive Care Unit (NICU) Inborn Admission Rates for All Infants and Infants With a Gestational Age of 34 Weeks or More
Range of Neonatal Intensive Care Unit (NICU) Inborn Admission Rates for All Infants and Infants With a Gestational Age of 34 Weeks or More

Unit of analysis is the individual NICU. Shaded rectangle contains interquartile range and median (blue dot); lines above or below the box identify values extending further by 1.5 times the interquartile range; the symmetric curves to either side estimate the density of the distribution of values (relative number of units at each admission rate value).

Figure 2.
Variation in Percent of Admissions Born at a Gestational Age of 34 Weeks or More That Met High Illness Acuity (California Perinatal Quality Care Collaborative [CPQCC]) Criteria
Variation in Percent of Admissions Born at a Gestational Age of 34 Weeks or More That Met High Illness Acuity (California Perinatal Quality Care Collaborative [CPQCC]) Criteria

Findings stratified by neonatal intensive care unit (NICU) level of care. Shaded rectangle contains interquartile range and median (blue dot); lines above or below the box identify values extending further by 1.5 times the interquartile range; the symmetric curves to either side estimate the density of the distribution of values (relative number of units at each admission rate value). CCS indicates California Children’s Services.

Figure 3.
Correlation Between Inborn Admission Rate and Percentage of Admissions Meeting High Illness Acuity (California Perinatal Quality Care Collaborative) Criteria
Correlation Between Inborn Admission Rate and Percentage of Admissions Meeting High Illness Acuity (California Perinatal Quality Care Collaborative) Criteria

Findings stratified by neonatal intensive care unit (NICU) level of care. Admissions represent all neonates at a gestational age of 34 weeks or more for regional (A), community (B), intermediate (C), and non–California Children’s Services (CCS) (D) NICUs.

Figure 4.
Correlation Between Inborn Admission Rates by Gestational Age (GA)
Correlation Between Inborn Admission Rates by Gestational Age (GA)

Correlation of a particular GA stratum and overall inborn admission rate for all infants not of that GA stratum for GA of 34 to 36 weeks (A), 37 to 38 weeks (B), and 39 weeks or more (C).

Table.  
Study Population Sample and Sample Subgroups During 2015 Study Perioda
Study Population Sample and Sample Subgroups During 2015 Study Perioda
1.
Harrison  W, Goodman  D.  Epidemiologic trends in neonatal intensive care, 2007-2012.  JAMA Pediatr. 2015;169(9):855-862.PubMedGoogle ScholarCrossref
2.
Carroll  AE.  The concern for supply-sensitive neonatal intensive care unit care: if you build them, they will come.  JAMA Pediatr. 2015;169(9):812-813.PubMedGoogle ScholarCrossref
3.
American Academy of Pediatrics Committee on Fetus and Newborn.  Levels of neonatal care.  Pediatrics. 2012;130(3):587-597.PubMedGoogle ScholarCrossref
4.
Hynan  MT, Mounts  KO, Vanderbilt  DL.  Screening parents of high-risk infants for emotional distress: rationale and recommendations.  J Perinatol. 2013;33(10):748-753.PubMedGoogle ScholarCrossref
5.
Moolenaar  RL, Crutcher  JM, San Joaquin  VH,  et al.  A prolonged outbreak of Pseudomonas aeruginosa in a neonatal intensive care unit: did staff fingernails play a role in disease transmission?  Infect Control Hosp Epidemiol. 2000;21(2):80-85.PubMedGoogle ScholarCrossref
6.
Polin  RA, Denson  S, Brady  MT; Committee on Fetus and Newborn; Committee on Infectious Diseases.  Epidemiology and diagnosis of health care-associated infections in the NICU.  Pediatrics. 2012;129(4):e1104-e1109.PubMedGoogle ScholarCrossref
7.
Wigert  H, Johansson  R, Berg  M, Hellström  AL.  Mothers’ experiences of having their newborn child in a neonatal intensive care unit.  Scand J Caring Sci. 2006;20(1):35-41.PubMedGoogle ScholarCrossref
8.
Fegran  L, Helseth  S, Fagermoen  MS.  A comparison of mothers’ and fathers’ experiences of the attachment process in a neonatal intensive care unit.  J Clin Nurs. 2008;17(6):810-816.PubMedGoogle ScholarCrossref
9.
Wennberg  J.  Tracking Medicine. New York, NY: Oxford University Press; 2010.
10.
Grady  D, Redberg  RF.  Less is more: how less health care can result in better health.  Arch Intern Med. 2010;170(9):749-750.PubMedGoogle ScholarCrossref
11.
Lipitz-Snyderman  A, Bach  PB.  Overuse of health care services: when less is more … more or less.  JAMA Intern Med. 2013;173(14):1277-1278.PubMedGoogle ScholarCrossref
12.
Schulman  J, Dimand  RJ, Lee  HC, Duenas  GV, Bennett  MV, Gould  JB.  Neonatal intensive care unit antibiotic use.  Pediatrics. 2015;135(5):826-833.PubMedGoogle ScholarCrossref
13.
California Department of Health Care Services. Provider Standards. http://www.dhcs.ca.gov/services/ccs/Pages/ProviderStandards.aspx#nicu. Updated 2017. Accessed July 17, 2014.
14.
California Perinatal Quality Care Collaborative. California Perinatal Quality Care Collaborative (CPQCC). https://www.cpqcc.org/perinatal-programs/cpqcc-data-center/downloads. Accessed August 15, 2017.
15.
California Perinatal Quality Care Collaborative Data Center. CPQCC Data Center. https://www.cpqcc.org/perinatal-programs/cpqcc-data-center. Accessed August 15, 2017.
16.
Vermont Oxford Network. Vermont Oxford Network Database. Release 19.0. https://public.vtoxford.org/wp-content/uploads/2014/11/Manual_of_Operations_Part2_v19.pdf. Published October 2014. Accessed October 5, 2017.
17.
Bhatt  DR, Adams  M, Clearly  J,  et al. NICUs & neonatologists in California. California Association of Neonatologists. District IX Section on Neonatal-Perinatal Medicine (Sonpm). https://canneo.groupsite.com/main/summary. Accessed October 5, 2017.
18.
Hamilton  BE, Martin  JA, Osterman  MJ.  Births: preliminary data for 2015.  Natl Vital Stat Rep. 2016;65(3):1-15.PubMedGoogle Scholar
19.
Stata Statistical Software. Version 14. College Station, Texas: Stata Press; 2015.
20.
Richardson  DK, Zupancic  JA, Escobar  GJ, Ogino  M, Pursley  DM, Mugford  M.  A critical review of cost reduction in neonatal intensive care: II; strategies for reduction.  J Perinatol. 2001;21(2):121-127.PubMedGoogle ScholarCrossref
21.
Fisher  ES, Wennberg  JE.  Health care quality, geographic variations, and the challenge of supply-sensitive care.  Perspect Biol Med. 2003;46(1):69-79.PubMedGoogle ScholarCrossref
22.
The Dartmouth Atlas of Health Care. Supply-sensitive care. http://www.dartmouthatlas.org/keyissues/issue.aspx?con=2937. Accessed May 4, 2017.
23.
Howell  EM, Richardson  D, Ginsburg  P, Foot  B.  Deregionalization of neonatal intensive care in urban areas.  Am J Public Health. 2002;92(1):119-124.PubMedGoogle ScholarCrossref
24.
Goodman  DC, Fisher  ES, Little  GA, Stukel  TA, Chang  C-H.  Are neonatal intensive care resources located according to need? regional variation in neonatologists, beds, and low birth weight newborns.  Pediatrics. 2001;108(2):426-431.PubMedGoogle ScholarCrossref
25.
Freedman  S.  Capacity and utilization in health care: the effect of empty beds on neonatal intensive care admission.  Am Econ J Econ Policy. 2016;8(2):154-185.PubMedGoogle ScholarCrossref
26.
Profit  J, McCormick  MC, Escobar  GJ,  et al.  Neonatal intensive care unit census influences discharge of moderately preterm infants.  Pediatrics. 2007;119(2):314-319.PubMedGoogle ScholarCrossref
27.
March of Dimes Perinatal Data Center. Special Care Nursery Admissions. https://www.marchofdimes.org/peristats/pdfdocs/nicu_summary_final.pdf. Published 2011. Accessed May 5, 2017.
Original Investigation
January 2018

Association Between Neonatal Intensive Care Unit Admission Rates and Illness Acuity

Author Affiliations
  • 1California Department of Health Care Services, California Children’s Services, Sacramento
  • 2Kaiser Permanente, Southern California, Panorama City
  • 3Perinatal Epidemiology and Health Outcomes Research Unit, Division of Neonatology, Department of Pediatrics, Stanford University School of Medicine, Stanford, California
  • 4Neonatal Intensive Care Unit, Lucile Packard Children’s Hospital, Palo Alto, California
  • 5California Perinatal Quality Care Collaborative, Palo Alto
JAMA Pediatr. 2018;172(1):17-23. doi:10.1001/jamapediatrics.2017.3913
Key Points

Question  For neonates born at 34 weeks’ gestation or more, how do neonatal intensive care unit inborn admission rates and percentage of high illness acuity admissions vary and are they related?

Findings  A cross-sectional study of 358 453 live births accounted for 79% of all inborn neonatal intensive care unit admissions. Admission rates varied 34-fold; high illness acuity accounted for 12% of admissions and varied 40-fold; the correlation was negative.

Meaning  High illness acuity fails to explain the variation in inborn admission rates for neonates born at 34 weeks’ gestation or more; 88.0% of these admissions do not meet high illness acuity criteria.

Abstract

Importance  Most neonates admitted to a neonatal intensive care unit (NICU) are born at gestational age (GA) of 34 weeks or more. The degree of uniformity of admission criteria for these infants is unclear, particularly at the low-acuity end of the range of conditions warranting admission.

Objectives  To describe variation in NICU admission rates for neonates born at GA of 34 weeks or more and examine whether such variation is associated with high illness acuity or designated facility level of care.

Design, Setting, and Participants  Cross-sectional study of 35 921 NICU inborn admissions of GA at 34 weeks or more during calendar year 2015, using a population database of inborn NICU admissions at 130 of the 149 hospitals in California with a NICU. The aggregate service population comprised 358 453 live births. The individual NICU was the unit of observation and analysis. The analysis was stratified by designated facility level of care and correlations with the percentage admissions with high illness acuity were explored. The hypothesis at the outset of the study was that inborn admission rates would correlate positively with the percentage of admissions with high illness acuity.

Exposures  Live birth at GA of 34 weeks or more.

Main Outcomes and Measures  Inborn NICU admission rate.

Results  Of the total of 358 453 live births at GA of 34 weeks or more, 35 921 infants were admitted to a NICU and accounted for 79.2% of all inborn NICU admissions; 4260 (11.9%) of these admissions met high illness acuity criteria. Inborn admission rates varied 34-fold, from 1.1% to 37.7% of births (median, 9.7%; mean [SD], 10.6% [5.8%]). Percentage with high illness acuity varied 40-fold, from 2.4% to 95% (median, 11.3%; mean, 13.2% [9.9%]). Inborn admission rate correlated inversely with percentage of admissions with high illness acuity (Spearman ρ = −0.3034, P < .001). Among regional NICUs capable of caring for patients with the highest degree of illness and support needs, inborn admission rate did not significantly correlate with percentage of admissions with high illness acuity (Spearman ρ = −0.21, P = .41).

Conclusions and Relevance  Percentage of admissions with high illness acuity does not explain 34-fold variation in NICU inborn admission rates for neonates born at GA of 34 weeks or more. The findings are consistent with a supply-sensitive care component and invite future investigation to clarify the lower-acuity end of the range of conditions considered to warrant neonatal intensive care.

Introduction

Newborns in the United States at all birthweights are increasingly likely to be admitted to a neonatal intensive care unit (NICU).1 Recent work has raised questions about the uniformity of criteria for admission to a NICU and possible overuse of neonatal intensive care for some newborns.1,2

The American Academy of Pediatrics designates 4 levels of care for neonates.3 Level I nurseries care for well newborns. Level II facilities are variously called intermediate care nurseries, special care nurseries, or level II NICUs and provide care to neonates with noncritical illness and support needs. Levels III and IV facilities care for neonates with progressively greater degrees of illness and support needs, along with neonates with lower acuity of illness. A population-based time-trend analysis of admission rates to level III and IV NICUs showed a substantial increase between 2007 and 2012 in NICU admission rates after adjustment for maternal and newborn characteristics.1 More than half of admitted newborns were at least 2500 g birthweight; NICU admissions were increasingly likely to be full term and normal birthweight.1 The causes of the patterns are unclear. In addition, little is known about whether or how a facility’s level of care influences inborn admission rates.

For some neonates, birthweight or gestational age (GA) can be a noncontroversial proxy measure identifying care needs beyond the capability of a level I nursery. For example, even without coexisting pathophysiology, all infants born at a GA below 34 weeks require monitoring and support solely due to their developmental immaturity. Thus, it may be generalized that neonates delivered under medical supervision at less than 34 weeks’ GA are admitted to a NICU.

For neonates born at GA of 34 weeks or more, the degree of uniformity of NICU admission criteria is unclear. More precisely, information is lacking about the low-acuity end of the range of conditions considered to warrant neonatal intensive care. This information gap matters because NICU care is costly, stressful, and entails risk of iatrogenesis.4-8 When considering health care for adults, laypersons and professionals alike have often assumed that more care is better.9 However, the evidence indicates no such clear relationship10,11; the perinatal and pediatric contexts merit similar scrutiny.

Our group recently found 40-fold variation in NICU antibiotic prescribing practice across 127 NICUs with similar burdens of proven infection and other factors unambiguously warranting antibiotic exposure.12 This evidence that a considerable portion of antibiotic use lacks clear warrant invites similar investigation of the broader justification for NICU resource use.

To explore whether practice variation exists for NICU admission of neonates born at GA of 34 weeks or more, we measured GA-stratified inborn NICU admission rates and examined associations with high illness acuity, designated NICU level of care, and overall inborn NICU admission rate. We hypothesized that inborn admission rates would correlate positively with the percentage of high illness acuity admissions.

Methods

California Children’s Services (CCS), within the California Department of Health Care Services, confers state approval for 3 levels of NICU care: regional, community, and intermediate,13 generally corresponding to American Academy of Pediatrics levels IV, III, and II, respectively.

The CCS standards include a requirement for annual data reporting of specific variables. All CCS-approved NICUs submit their data to the California Perinatal Quality Care Collaborative (CPQCC),14 which prepares an annual report for each NICU and submits an aggregate data set to the CCS. The CPQCC collects clinical data prospectively by using an expanded version of the Vermont Oxford Data set15,16 and a supplemental CCS data form. All CCS-approved NICUs must complete the supplemental CCS data form; additional CPQCC NICUs, not CCS-approved, typically do but are not required. The collected data constitute a single overarching database.

Of 149 NICUs in California,17 139 currently belong to CPQCC and 122 are CCS-approved; 135 participated in CPQCC during calendar year 2015, the study period. The overarching CPQCC/CCS data set describes NICU care and outcomes for 369 307 inborn live births of the 491 487 total live births in California in 201518 (Table). The 122 180 California live births (24.9% of total) born at the 90 hospitals known to the Department of Health Care Services and not reporting to the CPQCC include neonates subsequently transferred to CPQCC NICUs (outborn admissions, excluded from this study) or admitted to 1 of 14 NICUs at these 90 hospitals. We conducted the study analysis using the CPQCC/CCS data set for calendar year 2015. This study was approved by the Stanford University Institutional Review Board. The data set contained no individual patient identifiers, so informed consent was waived.

Study Population

All infants born at the sampled hospitals (accounting for 369 307 of the 491 487 total live births in California during the study period) with reported GA of 34 weeks or longer were included in the analysis. The Table describes the sampling from the overall California birth cohort.

Study Variable Definitions

Inborn admission rate is the percentage of all live births at a hospital who were admitted to the NICU. The rate is computed by counting in the numerator the number of live births at a hospital who were admitted to the NICU and in the denominator the total number of live births at the hospital. We express this proportion as a percent value. The corresponding GA-stratified inborn admission rate is computed similarly, using numerator and denominator counts of live births within the following strata: 34 to 36 weeks, 37 to 38 weeks, and 39 weeks or more.

The CPQCC high illness acuity admission percentage is the percentage of all inborn admissions for a particular GA stratum that meet CPQCC criteria for high acuity of illness. Although all infants of 1500 g or less birthweight are described in detail in the overarching CPQCC/CCS database (ie, all such infants are considered to have high acuity of illness), infants weighing more than 1500 g only are described with comparable detail in the database if they have 1 or more explicit high illness acuity criteria. In either case, all such information is derived by primary data collection from the medical record. Infants with more than 1500 g birthweight who do not meet high illness acuity criteria are represented in the database only in each facility’s count of live births and NICU admissions. For the present study, high illness acuity admissions were GA of 34 weeks or more and met at least 1 of the following criteria:

  1. Death,

  2. Intubated- or nonintubated-assisted ventilation for 4 hours or more,

  3. Early bacterial sepsis,

  4. Major surgery requiring anesthesia,

  5. Acute transport out of the NICU,

  6. Suspected encephalopathy or suspected perinatal asphyxia, and

  7. Active therapeutic hypothermia.

Among the 135 reporting hospitals, it was not possible to compute an inborn CPQCC admission percentage for 13 of the facilities: 1 did not report this variable and 12 were self-designated freestanding NICUs whose overarching organization did not include a maternal delivery service. For 8 of these hospitals, however, data quantifying the birth service population—located in the same building as the NICU—were available and used to compute inborn admission rates.

Statistical Analysis

The unit of observation and unit of analysis was the individual NICU. The primary outcome was inborn admission rate. Before the study began, we hypothesized that inborn admission rates would correlate positively with percentage of high illness acuity admissions and with NICU level of care. We examined, as a post hoc comparison, correlation between inborn admission rate for specific GA strata with overall inborn admission rate for all other GA strata (including GA<34 weeks but excluding those in the stratum of interest).

We estimated the magnitude of linear correlation by Spearman rank correlation. To compare high illness acuity admission percentage across designated NICU levels of care we used the nonparametric Kruskal-Wallis equality-of-populations rank test. Hypothesis tests were 2-sided with level of significance at P ≤ .05. We used Stata, version 14 (StataCorp)19 for all computations and graphic displays.

Results

Our data set reported on care exposures for 369 307 of the 491 487 live births in California (Table). Of the 130 birth hospitals, 18 were regional-level NICUs, 82 were community-level NICUs, 14 were intermediate-level NICUs, and 16 were non–CCS-approved NICUs. Inborn NICU service population births ranged from 606 to 9679 (median, 2500; mean, 2738). Of the total live births on which we had data, 358 453 (97.1%) infants were born at a GA of 34 weeks or more. Of these infants, 35 921 (10.0%) were admitted to a NICU and accounted for 79.2% of all 45 334 inborn NICU admissions at all GAs. Of inborn NICU admissions at a GA of 34 weeks or more, only 4260 (11.9%) infants met the criteria for a high illness acuity (CPQCC) admission.

As shown in Figure 1, inborn admission rates for neonates born at a GA of 34 weeks or more varied 34-fold across NICUs, from 1.1% to 37.7% of births (median, 9.7%; mean [SD], 10.6% [5.8%]). The distribution of values was similar to that for overall inborn admission rates of all GAs. Among the 8 extreme high outlier values for inborn admission rate for neonates at a GA of 34 weeks or more, only 4 represented regional-level NICUs.

The percentage of admissions born at a GA of 34 weeks or more that met high illness acuity (CPQCC) criteria varied 40-fold across NICUs, from 2.4% to 95.0% (median, 11.3%; mean, 13.2% [9.9%]) (Figure 2). The Kruskal-Wallis test indicated no significant differences across levels of care.

There was an inverse correlation between inborn admission rate and the percent of high-acuity (CPQCC) admissions (Spearman ρ = −0.3034, P < .001). Stratifying this association by NICU level of care (Figure 3) revealed that the inverse correlation was largely driven by the community-level NICUs (Spearman ρ = −0.39, P < .001) and intermediate-level NICUs (Spearman ρ = −0.57, P = .06). The pattern for intermediate-level NICUs suggested the possibility that high values for percentage of high-acuity (CPQCC) admissions might reflect small numbers of admissions (low denominator values). However, only 1 of 7 intermediate-level NICUs with more than 10% of high-acuity (CPQCC) admissions had fewer than 90 admissions, and the mean number of admissions across these 7 NICUs was 202. Variation in inborn admission rate did not correlate significantly with percentage of high-acuity (CPQCC) admissions among regional-level NICUs (Spearman ρ = −0.21, P = .41) and non-CCS NICUs (Spearman ρ = −0.02, P = .94).

Figure 4 shows that inborn admission rates for specific GA strata correlated strongly with overall inborn admission rates for all other GA strata (including GA<34 weeks but excluding those in the stratum of interest): for 34 to 36 weeks, ρ = 0.61, P < .001; for 37 to 38 weeks, ρ = 0.78, P < .001; and for 39 weeks and greater, ρ = 0.68, P < .001. Except as shown, stratification of analytical questions by NICU level of care did not alter the main inferences.

Discussion

For the service population of neonates born at a GA of 34 weeks or more, NICUs vary 34-fold in their reported inborn admission rates and 40-fold in reported percentage of high illness acuity. As is apparent in Figure 2, at all levels of care—including regional NICUs—typically fewer than 20% of admissions are characterized by high illness acuity. Considering that neonates at a GA of 34 weeks or greater account for 79.2% of all inborn NICU admissions and, overall, only 11.9% of these are characterized by high illness acuity, 88.0% of inborn NICU admissions at 34 weeks’ gestation or greater and 70.0% of all inborn NICU admissions are not characterized by high illness acuity.

Although we hypothesized that inborn admission rates would correlate positively with high illness acuity, we found an inverse association. That is, in hospitals with relatively higher admission rates, the percentage of these admissions with high illness acuity tends to be relatively low. In hospitals with relatively lower admission rates the percentage of these admissions with high illness acuity tends to be relatively high. In addition, a hospital’s inborn admission rate for a GA substratum strongly and positively correlates with its overall inborn admission rate for all other GA strata. That is, a NICU’s inborn admission rate for a GA substratum accurately predicts its inborn admission rate for neonates in all other GA strata.

If level of NICU care and a service population with correspondingly high illness acuity largely explain higher inborn admission rates, then regional NICUs (and plausibly, many community-level NICUs) should show a strong positive correlation between these factors. However, this is not the case. Many NICUs with high inborn admission rates were not regional-level NICUs and were not serving populations with burdens of high illness acuity. Hence, if the wide variation in inborn admission rates is not explained by a high burden of illness acuity—especially for regional-level NICUs—then what factors drive most of these NICU admissions?

The CPQCC high illness acuity criteria do not identify all newborns with high medical needs. Examples of such excluded circumstances include some congenital cardiac anomalies, dysmorphic conditions, seizures, and some cases of neonatal abstinence syndrome. In addition, it is well known that NICUs often serve functions other than provision of critical or intensive care. It has been estimated that between 4% and 8% of newborns experience difficulties of perinatal transition that may last only hours but require evaluation and observation not available in the well-baby area of a hospital.20

The present study is unable substantially to explain the wide variation in NICU resource use, the absence of a positive correlation between NICU use and high illness acuity, and the strong positive correlation between a NICU’s inborn admission rate for a GA substratum and its inborn admission rate for neonates in all other GA strata. However, the findings are consistent with others1 that have raised concerns for supply-sensitive care.21,22 Supply-sensitive care has little evidence base; instead, it reflects available service capacity and payment systems that incentivize service provision.22 In this circumstance, availability of NICU beds becomes a determinant of NICU care. To promote access to NICU care at the maternal delivery facility and ongoing care close to home, increasing numbers of NICUs have been established, a phenomenon sometimes referred to as “deregionalization.”23,24 For such units to remain in service, they must stay busy enough to pay for their costs.2 NICUs often are important contributors to a hospital’s overall financial balance.2 To this point, recent work using data from California and New York finds that an additional empty NICU bed the day before an infant’s birth increases the probability that the infant will be admitted to the NICU, particularly for infants of higher birthweight.25 NICU discharge also correlates with bed availability.26

The ability to predict from a NICU’s admission rate for a GA substratum of the study population, the inborn admission rate for neonates of all other GAs—and the negative correlation of inborn admission rate with percent of high illness acuity—raises concern regarding the degree of NICU resource use that may be considered discretionary. It is unclear what decision rules guide admission for neonates without high illness acuity and the degree to which such rules vary across NICUs.

We emphasize that available data do not demonstrate clear overuse of NICU services. However, our findings identify the need for future investigation to clarify the lower-acuity end of the range of conditions considered to warrant neonatal intensive care. It may be preferable to manage some of these conditions in a transitional nursery facility that provides the evaluation and monitoring needed in the first few hours of life.

Limitations

A main limitation of this study is the inability to discriminate whether the observed wide variation in admission rates primarily reflects discretionary practice variation or clinically significant unmeasured case mix differences. This inability results from limited information currently collected for databases on NICU admissions that do not meet high illness acuity criteria. It is understandable that the focus of NICU-level data collection to date has been on infants who meet high illness acuity criteria. However, our study reveals important knowledge gaps regarding NICU resource use for 70% of all inborn NICU admissions. Data elements, such as admitting diagnoses and length of NICU stay (and from this, aggregate NICU patient days), could help discriminate discretionary practice variation from clinically significant case mix differences among NICUs, as well as characterize the proportion of all NICU patient days accounted for by this large patient subgroup. Another limitation relates to our hospital sample describing inborn NICU admission for 75% of all California live births and 93% of California NICUs. Among the other 25% of live births, infants admitted to any of the 7% of California NICUs that did not report to the CPQCC/CCS are not represented in this study. All other NICU admissions from this subgroup are described in the outborn NICU admission portion of the CPQCC/CCS database. Any unknown biases that may operate at nonreporting facilities thus affect a small proportion of the overall service population. Since California provides substantial NICU regulatory oversight and guidance, particularly via CCS and CPQCC, it is possible that our findings underestimate the range of practice variation in less-regulated areas. In addition, the CPQCC/CCS data set comprises self-reported data for which a comprehensive audit of accuracy and completeness is not feasible. Our findings are consistent, however, with a 2011 report from the March of Dimes on NICU admissions for 183 030 newborns.27 This study also assumes that the number of neonates with high acuity illness in California not admitted to a NICU is negligible; data to answer the question are not readily available.

Conclusions

Our study of inborn NICU admissions for neonates born at a GA of 34 weeks or more found that this subpopulation accounts for 79.2% of all inborn NICU admissions and that reported inborn admission rates vary 34-fold across NICUs. This wide variation is unexplained by the reported percentage of admissions with high illness acuity, comprising only 11.9% of inborn NICU admissions born at a GA of 34 weeks or more. Requisite data are currently unavailable to discriminate whether the observed wide variation in admission rates primarily reflects discretionary practice variation or clinically significant unmeasured case mix differences. The paucity of NICU database elements describing 70% of all inborn NICU admissions suggests that such information has been assumed to be not worth the collection effort. Our findings indicate that this assumption merits further consideration.

Back to top
Article Information

Accepted for Publication: September 5, 2017.

Corresponding Author: Joseph Schulman, MD, MS, California Department of Health Care Services, California Children’s Services, 1515 K St, Ste 400, PO Box 997413, MS 8100, Sacramento, CA 95899 (joseph.schulman@dhcs.ca.gov).

Published Online: November 27, 2017. doi:10.1001/jamapediatrics.2017.3913

Author Contributions: Dr Schulman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Schulman, Braun, Lee, Profit, Dimand, Gould.

Acquisition, analysis, or interpretation of data: Schulman, Braun, Lee, Profit, Duenas, Bennett, Dimand, Jocson.

Drafting of the manuscript: Schulman, Braun, Profit.

Critical revision of the manuscript for important intellectual content: Schulman, Lee, Profit, Duenas, Bennett, Dimand, Jocson, Gould.

Statistical analysis: Schulman, Jocson, Gould.

Obtained funding: Dimand.

Administrative, technical, or material support: Schulman, Braun, Lee, Profit, Duenas, Bennett, Dimand.

Study supervision: Schulman, Duenas, Dimand.

Conflict of Interest Disclosures: None reported.

Funding/Support: Dr Profit’s effort was supported in part by grants R01 HD083368-01 and R01 HD08467-01 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development.

Role of the Funder/Sponsor: The Eunice Kennedy Shriver National Institute of Child Health and Human Development had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the Eunice Kennedy Shriver National Institute of Child Health and Human Development or the National Institutes of Health.

References
1.
Harrison  W, Goodman  D.  Epidemiologic trends in neonatal intensive care, 2007-2012.  JAMA Pediatr. 2015;169(9):855-862.PubMedGoogle ScholarCrossref
2.
Carroll  AE.  The concern for supply-sensitive neonatal intensive care unit care: if you build them, they will come.  JAMA Pediatr. 2015;169(9):812-813.PubMedGoogle ScholarCrossref
3.
American Academy of Pediatrics Committee on Fetus and Newborn.  Levels of neonatal care.  Pediatrics. 2012;130(3):587-597.PubMedGoogle ScholarCrossref
4.
Hynan  MT, Mounts  KO, Vanderbilt  DL.  Screening parents of high-risk infants for emotional distress: rationale and recommendations.  J Perinatol. 2013;33(10):748-753.PubMedGoogle ScholarCrossref
5.
Moolenaar  RL, Crutcher  JM, San Joaquin  VH,  et al.  A prolonged outbreak of Pseudomonas aeruginosa in a neonatal intensive care unit: did staff fingernails play a role in disease transmission?  Infect Control Hosp Epidemiol. 2000;21(2):80-85.PubMedGoogle ScholarCrossref
6.
Polin  RA, Denson  S, Brady  MT; Committee on Fetus and Newborn; Committee on Infectious Diseases.  Epidemiology and diagnosis of health care-associated infections in the NICU.  Pediatrics. 2012;129(4):e1104-e1109.PubMedGoogle ScholarCrossref
7.
Wigert  H, Johansson  R, Berg  M, Hellström  AL.  Mothers’ experiences of having their newborn child in a neonatal intensive care unit.  Scand J Caring Sci. 2006;20(1):35-41.PubMedGoogle ScholarCrossref
8.
Fegran  L, Helseth  S, Fagermoen  MS.  A comparison of mothers’ and fathers’ experiences of the attachment process in a neonatal intensive care unit.  J Clin Nurs. 2008;17(6):810-816.PubMedGoogle ScholarCrossref
9.
Wennberg  J.  Tracking Medicine. New York, NY: Oxford University Press; 2010.
10.
Grady  D, Redberg  RF.  Less is more: how less health care can result in better health.  Arch Intern Med. 2010;170(9):749-750.PubMedGoogle ScholarCrossref
11.
Lipitz-Snyderman  A, Bach  PB.  Overuse of health care services: when less is more … more or less.  JAMA Intern Med. 2013;173(14):1277-1278.PubMedGoogle ScholarCrossref
12.
Schulman  J, Dimand  RJ, Lee  HC, Duenas  GV, Bennett  MV, Gould  JB.  Neonatal intensive care unit antibiotic use.  Pediatrics. 2015;135(5):826-833.PubMedGoogle ScholarCrossref
13.
California Department of Health Care Services. Provider Standards. http://www.dhcs.ca.gov/services/ccs/Pages/ProviderStandards.aspx#nicu. Updated 2017. Accessed July 17, 2014.
14.
California Perinatal Quality Care Collaborative. California Perinatal Quality Care Collaborative (CPQCC). https://www.cpqcc.org/perinatal-programs/cpqcc-data-center/downloads. Accessed August 15, 2017.
15.
California Perinatal Quality Care Collaborative Data Center. CPQCC Data Center. https://www.cpqcc.org/perinatal-programs/cpqcc-data-center. Accessed August 15, 2017.
16.
Vermont Oxford Network. Vermont Oxford Network Database. Release 19.0. https://public.vtoxford.org/wp-content/uploads/2014/11/Manual_of_Operations_Part2_v19.pdf. Published October 2014. Accessed October 5, 2017.
17.
Bhatt  DR, Adams  M, Clearly  J,  et al. NICUs & neonatologists in California. California Association of Neonatologists. District IX Section on Neonatal-Perinatal Medicine (Sonpm). https://canneo.groupsite.com/main/summary. Accessed October 5, 2017.
18.
Hamilton  BE, Martin  JA, Osterman  MJ.  Births: preliminary data for 2015.  Natl Vital Stat Rep. 2016;65(3):1-15.PubMedGoogle Scholar
19.
Stata Statistical Software. Version 14. College Station, Texas: Stata Press; 2015.
20.
Richardson  DK, Zupancic  JA, Escobar  GJ, Ogino  M, Pursley  DM, Mugford  M.  A critical review of cost reduction in neonatal intensive care: II; strategies for reduction.  J Perinatol. 2001;21(2):121-127.PubMedGoogle ScholarCrossref
21.
Fisher  ES, Wennberg  JE.  Health care quality, geographic variations, and the challenge of supply-sensitive care.  Perspect Biol Med. 2003;46(1):69-79.PubMedGoogle ScholarCrossref
22.
The Dartmouth Atlas of Health Care. Supply-sensitive care. http://www.dartmouthatlas.org/keyissues/issue.aspx?con=2937. Accessed May 4, 2017.
23.
Howell  EM, Richardson  D, Ginsburg  P, Foot  B.  Deregionalization of neonatal intensive care in urban areas.  Am J Public Health. 2002;92(1):119-124.PubMedGoogle ScholarCrossref
24.
Goodman  DC, Fisher  ES, Little  GA, Stukel  TA, Chang  C-H.  Are neonatal intensive care resources located according to need? regional variation in neonatologists, beds, and low birth weight newborns.  Pediatrics. 2001;108(2):426-431.PubMedGoogle ScholarCrossref
25.
Freedman  S.  Capacity and utilization in health care: the effect of empty beds on neonatal intensive care admission.  Am Econ J Econ Policy. 2016;8(2):154-185.PubMedGoogle ScholarCrossref
26.
Profit  J, McCormick  MC, Escobar  GJ,  et al.  Neonatal intensive care unit census influences discharge of moderately preterm infants.  Pediatrics. 2007;119(2):314-319.PubMedGoogle ScholarCrossref
27.
March of Dimes Perinatal Data Center. Special Care Nursery Admissions. https://www.marchofdimes.org/peristats/pdfdocs/nicu_summary_final.pdf. Published 2011. Accessed May 5, 2017.
×