Abstract
Video Abstract
BACKGROUND: Increased admissions of higher birth weight and less acutely ill infants to NICUs suggests that intensive care may be used inappropriately in these populations. We describe variation in use of NICU services by gestational age and NICU type.
METHODS: Using the Vermont Oxford Network database of all NICU admissions, we assessed variation within predefined gestational age categories in the following proportions: admissions, initial NICU hospitalization days, high-acuity cases ≥34 weeks’ gestation, and short-stay cases ≥34 weeks’ gestation. High acuity was defined as follows: death, intubated assisted ventilation for ≥4 hours, early bacterial sepsis, major surgery requiring anesthesia, acute transport to another center, hypoxic-ischemic encephalopathy or a 5-minute Apgar score ≤3, or therapeutic hypothermia. Short stay was defined as an inborn infant staying 1 to 3 days with discharge from the hospital.
RESULTS: From 2014 to 2016, 486 741 infants were hospitalized 9 657 508 days at 381 NICUs in the United States. The median proportions of admissions, initial hospitalized days, high-acuity cases, and short stays varied significantly by NICU types in almost all gestational age categories. Fifteen percent of the infants ≥34 weeks were high acuity, and 10% had short stays.
CONCLUSIONS: There is substantial variation in use among NICUs. A campaign to focus neonatal care teams on using the NICU wisely that addresses the appropriate use of intensive care for newborn infants and accounts for local context and the needs of families is needed.
- VON —
- Vermont Oxford Network
What’s Known on This Subject:
Recent evidence of increasing admissions of higher birth weight and less acutely ill infants to NICUs suggests that intensive care may be inappropriately used in these populations.
What This Study Adds:
Variation in use indicates that evidence-based standards for the appropriate use of ICUs for newborn infants are needed.
Data collected by organizations such as Vermont Oxford Network (VON),1–3 National Institute of Child Health and Development Neonatal Research Network,4 Pediatrix,5 Children’s Hospitals Neonatal Database,6 and California Perinatal Quality Care Collaborative7 have contributed greatly to our understanding of the characteristics and outcomes of infants born <1500 g or <29 weeks in the United States.
Less is known about infants admitted to neonatal intensive care who are not very low birth weight.8 From 2007 to 2012, admissions to neonatal intensive care increased in all birth weights but particularly among larger and less-premature infants.9 In California, infants ≥34 weeks’ gestational age accounted for nearly 80% of all inborn neonatal intensive care admissions, with wide variation based on center type.10 However, only 12% of those infants met high-acuity illness criteria, such as assisted ventilation for at least 4 hours, bacterial infection in the first 3 days from birth, suspected encephalopathy or perinatal asphyxia, or active therapeutic hypothermia; and the proportion of high-acuity infants varied at the center level from 2.4% to 95%.10 At a group of 19 hospitals, the percentage of infants 35 to 42 weeks’ gestation admitted to a NICU without a prespecified, identifiable high-acuity cause ranged from 0% to 59%; such infants accounted for 9.1% of total NICU days with lower lengths of stay than other infants.11
As hospitals and insurers work to control costs and improve the value of care,12 use of health care resources is under increased scrutiny. The Choosing Wisely in Newborn Medicine campaign identified 5 tests and treatments that could potentially be used less frequently in newborn care.13 However, the campaign did not ask: Which infants are being admitted to your NICU, how long are they staying, and why are they there?
The next important focus for neonatology is using the NICU wisely. To support that new focus, we assessed the proportions of admissions, initial hospital days, high-acuity cases ≥34 weeks, and short-stay cases ≥34 weeks to identify patterns and variation in the use of NICU services at US VON members.
Methods
VON is a voluntary worldwide community of practice dedicated to improving the quality, safety, and value of care through a coordinated program of data-driven quality improvement, education, and research.
Members participating in the Expanded Database submitted data on all infants admitted to a NICU, defined as any location within the center in which newborn infants received continuous positive airway pressure or intermittent mandatory ventilation, not including areas where these modalities of respiratory support were used only for brief periods of stabilization before transfer to another location. Local staff collected infant data using uniform definitions14 until death, discharge from the hospital, or transfer to other centers. All data underwent automated checks for quality and completeness at the time of submission. The University of Vermont Committee on Human Research determined that the use of the VON Research Repository for this analysis was not human subjects research. All data were deidentified.
Center Definitions
Members completed surveys of center characteristics (response rate = 100%). NICU type was derived from responses to the 2016 survey or the last, most complete survey. Centers were divided into 4 groups by using responses to whether the center was required by state regulation to transfer infants to another center for assisted ventilation on the basis of the infant characteristics or duration of ventilation required or to whether 1 of 8 surgeries was performed at the center (omphalocele repair, ventriculo-peritoneal shunt, tracheoesophageal fistula or esophageal atresia repair, bowel resection or reanastomosis, meningomyelocele repair, cardiac catheterization, cardiac surgery requiring bypass). The 4 groups were as follows: ventilation restrictions, no surgery, surgery except cardiac requiring bypass, and all surgery. NICUs with ventilation restrictions correspond to level II under the current American Academy of Pediatrics levels of neonatal care. NICUs that do not perform surgery on site and NICUs that perform all surgery except cardiac requiring bypass correspond to level III. NICUs that perform all surgery correspond to level IV.15
Infant Measures
Any initial resuscitation included oxygen, face mask or endotracheal tube ventilation, nasal continuous positive airway pressure, epinephrine, cardiac compression, or surfactant immediately after birth. Ventilation after initial resuscitation included conventional or high frequency. Hypoxic-ischemic encephalopathy was measured only among infants at least 36 weeks’ gestational age. Other definitions can be found in the VON Manual of Operations.14
Extremely preterm was defined as 22 to 25 weeks’ gestational age, very preterm as 26 to 29 weeks, moderate preterm as 30 to 33 weeks, late preterm as 34 to 36 weeks, early term as 37 to 38 weeks, term as 39 to 40 weeks, and postterm as 41 weeks or older. The early-term, term, and postterm categories were adapted from the American College of Obstetricians and Gynecologists Committee on Obstetric Practice and Society for Maternal–Fetal Medicine guidelines.16
Usage Measures
Adapting a definition from Schulman et al10 to fit VON data items, high acuity was defined for infants born ≥34 weeks’ gestational age as follows: death, intubated assisted ventilation for 4 hours or more, early bacterial sepsis, major surgery requiring anesthesia at the reporting center, transport to another center for surgical or medical services, hypoxic-ischemic encephalopathy or a 5-minute Apgar score of <4, or therapeutic hypothermia.
Length of stay was measured as the number of days from the date the infant was admitted until the date of initial discharge from the hospital, transfer, or death. Short stays were defined as infants who were inborn at the reporting center ≥34 weeks’ gestational age who stayed 1 to 3 days and were discharged from the hospital.
Analyses
These analyses include members in the United States only. All tables and figures include 486 741 infants admitted to 381 NICUs from January 1, 2014, to December 31, 2016. To describe admissions at the infant level, we removed infants deemed by probabilistic match to be duplicates. We used Kruskal-Wallis tests to assess differences in medians between NICU types. All analyses were conducted by using SAS 9.4 (SAS Institute, Inc, Cary, NC).
Results
From 2014 to 2016, 381 centers submitted data on all NICU admissions; 16% of centers had ventilation restrictions, 32% did not perform surgery, 37% performed surgery except cardiac requiring bypass, and 16% performed all types of surgery. Infants admitted to centers that performed all surgery were most likely to be outborn, to have a congenital anomaly, and to have lower 1-minute Apgar scores (Table 1). These highest-level centers were most likely to report performing ventilation after initial resuscitation with the longest durations of ventilation, therapeutic hypothermia, and major surgery requiring anesthesia.
Characteristics of Infants and Interventions Received by NICU Type
The median (range) annual number of NICU admissions was 159 (37–609) infants at centers with ventilation restrictions, 275 (27–1308) infants at centers that did not perform surgery, 380 (14–1853) infants at centers that performed surgery except cardiac requiring bypass, and 664 (42–1786) infants at centers that performed all surgery. Of the 486 741 infants admitted to the 381 NICUs, the median (range) proportions of NICU admissions in the predefined gestational age categories were as follows: 2% (0%–23%) for extremely preterm, 5% (0%–46%) for very preterm, 16% (3%–33%) for moderate preterm, 29% (1%–45%) for late preterm, 19% (0%–33%) for early term, 24% (0%–44%) for term, and 3% (0%–13%) for postterm (Fig 1). The median proportions of admissions by gestational age category differed by NICU type (Supplemental Fig 5). P values testing the equality of the medians between NICU types within gestational age categories were <5%, except in the late term category.
Proportion of admissions by gestational age category. The line represents the median (50th percentile). The borders of the box represent the 25th and 75th percentiles. The whiskers represent the minimum and maximum percentiles.
Infants were hospitalized 9 657 508 days through initial discharge during the 3-year study period, of which 368 970 (4%) were at centers with ventilation restrictions, 1 904 070 (20%) were at centers that did not perform surgery, 3 790 817 (39%) were at centers that performed surgery except cardiac requiring bypass, and 3 593 651 (37%) were at centers that performed all types of surgery. The median (range) proportions of initial hospitalized days were as follows: 9% (0%–30%) for extremely preterm, 21% (0%–46%) for very preterm, 27% (10%–43%) for moderate preterm, 21% (5%–41%) for late preterm, 10% (3%–28%) for early term, 10% (2%–28%) for term, and 1% (0%–6%) for postterm (Fig 2). The median proportions of hospitalized days by gestational age category differed by NICU type (Supplemental Fig 6). P values testing the equality of the medians between NICU types within gestational age categories were <1%, except in the late term category.
Proportion of days hospitalized by gestational age category. The line represents the median (50th percentile). The borders of the box represent the 25th and 75th percentiles. The whiskers represent the minimum and maximum percentiles.
Extremely and very preterm infants represented 7% of admissions and 30% of hospitalized days. The median proportion of hospitalized days in this population was 9% at centers with ventilation restrictions, 26% at centers that did not perform surgery, 33% at centers that performed surgery except cardiac requiring bypass, and 37% at centers that performed all surgery (P < .0001). Infants ≥34 weeks’ gestational age represented 74% of admissions and 69% of hospitalized days. The median proportion of hospitalized days in this population was 59% at centers with ventilation restrictions, 44% at centers that did not perform surgery, 40% at centers that performed surgery except cardiac requiring bypass, and 40% at centers that performed all surgery (P < .0001).
Fifteen percent of the infants ≥34 weeks’ gestational age were high acuity, which ranged from 0% to 100% at the center level. High-acuity infants accounted for 31% of the initial hospital days. The median proportion of high-acuity infants was 6% at centers with ventilation restrictions, 8% at centers that did not perform surgery, 12% at centers that performed surgery except cardiac requiring bypass, and 37% at centers that performed all surgery (P < .0001) and varied by gestational age categories and NICU types (Fig 3). P values used to test the equality of the medians between NICU types within gestational age categories were <1% for all gestational age categories.
Proportion of high-acuity infants born ≥34 weeks’ gestational age by gestational age category and NICU type. The line represents the median (50th percentile). The borders of the box represent the 25th and 75th percentiles. The whiskers represent the minimum and maximum percentiles.
Of infants ≥34 weeks’ gestational age, 10% had short stays, ranging at the center level from 0% to 100%. Short stays accounted for 3% of the initial hospital days for infants ≥34 weeks’ gestation. Of infants with short stays, 22% had mothers with chorioamnionitis compared with 9% of infants without short stays. The median proportion of high-acuity infants ranged from 12% at centers with ventilation restrictions, 8% at centers that did not perform surgery, 7% at centers that performed surgery except cardiac requiring bypass, and 5% at centers that performed all surgery (P < .0001) and varied by gestational age categories and NICU types (Fig 4). P values used to test the equality of the medians between NICU types within gestational age categories were <5%, except in the late term category.
Proportion of short stays among infants born ≥34 weeks’ gestational age by gestational age category and NICU type. The line represents the median (50th percentile). The borders of the box represent the 25th and 75th percentiles. The whiskers represent the minimum and maximum percentiles.
Discussion
From 2014 to 2016, VON members reported nearly half a million NICU admissions. In our cohort, 74% of the infants admitted were ≥34 weeks’ gestational age, similar to the 80% reported in California.10 Fifteen percent of infants ≥34 weeks’ gestational age were high acuity, similar to 12% in California.10 Ten percent of infants in our cohort stayed no more than 3 days and were discharged from the hospital.
The Choosing Wisely in Newborn Medicine campaign used a Delphi process with 1047 survey respondents to identify 5 tests and treatments to avoid during a neonatal admission.13 The stated goal of the Choosing Wisely campaign is to avoid overuse and misuse of medical tests, treatments, and procedures. However, the campaign does not address the overuse and misuse of admissions. NICU admission carries infection risk for infants,17–19 increases acute stress for families,20 and may interfere with breastfeeding.21 We suggest a new campaign, “Using the NICU Wisely.” With our data and those of previous studies9–11 in which substantial variation in the use of NICU admissions and stays is indicated, we suggest that this new campaign should focus neonatal care teams on defining and implementing guidelines for the appropriate use of the NICU. Although our data cannot be used to determine if the patterns we observe represent overuse or underuse of the NICU in different populations of newborns, NICUs should monitor admissions by gestational age, acuity, and associated lengths of stay to understand how these factors contribute to potential overuse in the local context. Policies and protocols for NICU admission should be reviewed in light of NICUs’ efforts to increase family-centered and family-engaged care.22
Fifteen percent of infants ≥34 weeks’ gestational age were high acuity. Our definition of high acuity, which was modeled after Schulman et al,10 does not include all types of acute issues that may require NICU admission. Ziegler et al11 identified 16 reasons, ranging from infection to brain injury to respiratory issues. Infants are admitted for many nonacute reasons, such as hypoglycemia and rule-out sepsis, and for specific interventions such as antibiotic administration or line placement.23 More practically, bed supply24,25 and supply-sensitive care26–28 may play roles in admission rates. In recent research, authors have pointed to possible overuse of NICU care, particularly among larger infants.9,24 Quality improvement approaches can reduce potentially unnecessary NICU admissions, such as those for hypoglycemia,29 elective delivery before 39 weeks’ gestation,30 neonatal abstinence syndrome,31 or rule-out sepsis.32
Not surprisingly, the distribution of days of stay differed within and between NICU types. Short stays of 1 to 3 days also varied by NICU type, occurring more frequently when infants’ mothers had chorioamnionitis. Use of a sepsis risk calculator to divide newborns into observation, evaluation, and treat strata32 and the clinical monitoring of well-appearing late-preterm and term infants exposed to chorioamnionitis33 may help reduce unnecessary admissions for rule-out sepsis. Financial incentives may account for some of the observed variation in length of stay.34 As hospitals focus on the value of health care, examining admission practices for early-term, term, and postterm infants and common characteristics shared by short-stay infants could alleviate some length of stay burden as well as improve the birth hospitalization experience for infants and their families.
From 2014 to 2016, infants stayed nearly 10 million days during the initial hospitalization. Using a rough estimate of $3000 per day,35 these NICU admissions cost over $28.9 billion, or over $9.6 billion per year. On the basis of the number of births and an average NICU admission rate of 10%, we estimate that our sample represents ∼40% of NICU admissions in the United States. Under these assumptions, the approximate annual cost for the initial days of hospitalization for NICU stays in the United States may be around $24 billion. This figure represents an increase from the $16.9 billion in medical care estimated in 200536 and deserves more rigorous study.
This cohort of centers comprises a convenience sample that is not population based. Results from this sample may not be generalizable to all US NICUs.
This study cannot be used to explain why infants are admitted to NICU care. We do not know the proportion of births at each hospital that are admitted to a NICU. We have no information on center-level protocols for NICU admission or bed supply.24,25 Information on regional distribution of perinatal care and state policies governing perinatal care would not be appropriately used on our convenience sample. Further research to identify the cases of variation is needed.
Our NICU types differ from the current types proposed by the American Academy of Pediatrics.15 NICUs with ventilation restrictions are most like level II in the current guidelines, whereas NICUs that do not perform surgery or perform surgery except cardiac surgery requiring bypass are most like level III, and NICUs that perform all surgery are most like level IV. The American Academy of Pediatrics is developing a certification program based on its current classification.37 This process may help address the variation in NICU services and use.
Conclusions
Variation in the proportion of admissions by gestational age and use between and within NICU types suggests that a campaign for using the NICU wisely and addressing the appropriate use of ICUs for newborn infants is required. These standards must address local context and the needs of families.
Acknowledgments
We thank our medical and nursing colleagues who submit data to VON on behalf of infants and their parents. Participating centers are listed in Supplemental Table 2.
Footnotes
- Accepted August 15, 2018.
- Address correspondence to Erika M. Edwards, PhD, MPH, Director of Data Science, Vermont Oxford Network, 33 Kilburn St, Burlington, VT 05401. E-mail: eedwards{at}vtoxford.org
FINANCIAL DISCLOSURE: Dr Edwards receives salary support from Vermont Oxford Network, and Dr Horbar is an employee of Vermont Oxford Network.
FUNDING: Supported by Vermont Oxford Network.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
References
- Copyright © 2018 by the American Academy of Pediatrics
