In this inaugural effort, we determined there were 1424 NICUs in the US that vary in level, size, and characteristics. Higher level and larger NICUs are more prevalent in academic medical centers in addition to being more likely to be housed within a children’s hospital. There are higher-level NICUs and NICUs with a larger number of beds in areas that have high population density, have a higher proportion of individuals identifying as a racial or ethnic minority, and a higher proportion of individuals experiencing poverty. After multivariable analysis, number of NICU beds was related to percentage of the population that consists of minorities, type of hospital (academic), geographic region (Northwest), whether part of a children’s hospital, and NICU level. After multivariable analysis, level of NICU remained associated with number of NICU beds, whether part of a children’s hospital, type of hospital (academic), geographic region (Northeast or Southeast) and population density. Although not significant on univariate analysis, CON legislation was discovered to be related to both NICU level and number of beds when controlling for other variables.

Here we highlight the significant variation in NICU levels, size, and other physical characteristics. While NICU level will define whether a NICU has the ability to provide a full range of respiratory support, NICUs at the same acuity level can also vary in terms of services offered and available specialties; such as use of extracorporeal membrane oxygenation, care for infants with diaphragmatic hernias, cardiac care, surgical intervention, and availability of associated follow-up clinics. Policies and procedures can vary across NICUs, including nurse-to-patient ratio, family presence policies, whether there are physical therapists, occupational therapists, and speech-language pathologists on staff in the NICU, and whether they are constructed of single-patient rooms or open bays. Further, culture within each NICU differs, including whether the NICU uses developmental care strategies, provides specialized parental education regarding caring for babies in the NICU, and whether evidence-based feeding practices such as cue-based feeding are integrated. The Vermont Oxford Network (VON), California Perinatal Quality Care Collaborative (CPQCC) and other statewide perinatal quality collaboratives, and The Children’s Hospitals Neonatal Consortium (CHNC) have moved the needle on aligning with collecting specific outcome measures across NICUs and have become central in working to improve the quality of NICU care. However, with a large number of NICUs and the significant variability among NICUs, parents are not able to identify these differences and often are faced with receiving care in the hospital that is part of their network, close to home, or known to them. By defining metrics of US-based NICUs, we can start to benchmark to improve access to care for all patients.

Preterm birth disproportionately affects women and families of color [34]. Further, infants with public insurance or living in rural areas are more likely to be born preterm [35]. With higher rates of preterm birth necessitating NICU care, the need for more NICUs with specialized higher levels of care would seem indicated. However, studies have demonstrated poor resource allocation and poor access to necessary medical care among these populations, leading to health inequities [36]. Unequal care within the NICU setting has been identified, and the downstream effects of systemic racism can have significant social impacts [36]. This study found there are more hospitals with higher numbers of NICU beds in areas that have higher percentages of individuals from racial and ethnic minorities and who are living in poverty. However, the number of NICU beds increases linearly with increased number of minorities until 50% minorities and then plateaus without any further increase in beds with increases in minorities in the community beyond 50%. This could potentially signal no additional increase in beds to account for the potential increase in utilization within communities with high density of minorities (50–100%). This study does not provide information regarding the quality-of-care metrics at each hospital. Other studies have focused on the differences in quality of care based on geographic locations and have found that despite the presence of healthcare resources, there are still disparities in the quality of care provided [36,37,38,39]. Simply having proximity to a higher level NICU was not necessarily an indicator of accessing those services, as healthcare choices are affected by other considerations such as insurance type [40]. Admissions to the NICU, where specialized care can be received, are higher when resources, such as insurance, exist [11]. Neonatal mortality is higher among certain groups of infants, such as the uninsured [41].

There appear to be geographical influences on types of NICUs present, and CON legislation was observed to relate to the types (both in terms of acuity level as well as number of beds) of NICUs in US locations. High-risk infants tend to have a longer length of stay as compared with other patient populations, so the presence of a NICU can potentially be profitable to a hospital. Having an extra layer of CON that ensures there is a need for additional NICU beds within a region before new NICUs can be opened would seem to discourage smaller units. This is consistent with our findings in which we identified relationships between CON legislation and type of NICU, with states who have CON legislation having a larger number of hospitals with higher acuity level and NICU beds. Interestingly, we did find that NICUs in the Northeast have NICUs with higher numbers of beds. NICUs in the Northeast and Southeast were more likely to have higher acuity level. Whether this reflects concentrated urban medical centers dedicated to regional care across large areas of land is unknown.

Another important concept underlying the designation of NICUs at different levels of care is regionalization. Regionalization is one strategy to reduce maternal and neonatal morbidity and mortality by facilitating early identification of high-risk pregnancies and establishing systems of care optimized for resource allocation [42]. Improvements in neonatal care were observed when the March of Dimes first designated different levels of NICU care and recommended that mothers be referred to the appropriate level of NICU based on their risk factors and the hospital’s commensurate resources and personnel [12, 42]. The benefits of regionalized care systems highlight improved competency of providers with consistent, high-volume exposure to high-risk neonates, as well as reductions in neonatal morbidity and mortality [43,44,45]. Despite consistent evidence of the benefits of regionalization, in the past few decades there has been increases in the number of smaller-volume NICUs as patient volumes and number of practicing neonatologists increases, maternal and neonatal medical technology advances and becomes more readily available, and hospitals strive to offer comprehensive services [44, 46]. There is some evidence that the growth in the number of NICUs in the last 30 years has contributed to de-regionalization of NICU care in many parts of the US [44, 46], with many Level II NICUs potentially accepting and caring for infants at high-risk with needs beyond the scope of care for that NICU [12]. A better understanding of how this has impacted neonatal mortality and morbidity is warranted. Having higher acuity NICUs with larger numbers of NICU beds in more locations could be a result of more regionalization in care, but this also requires further investigation.

This study had several limitations. It should be noted that univariate analyses were used without controlling for how the different variables reported influence each other. In addition, no adjustments to significance levels were made to account for multiple comparisons.

The nature of healthcare facilities, specifically NICUs, proved a rapidly changing target, with NICUs closing or expanding during the timeframe of this study. Data including level of NICU and number of beds were at risk of inconsistencies due to the ever-changing nature of the hospital care systems where NICUs often change number of beds (usually increasing capacity, but sometimes decreasing due to areas of the hospital closing). We suspect these changing numbers may be especially volatile during and following the COVID-19 pandemic, in which many units reallocated staffing and closed NICUs to provide space for other patients to accommodate hospital-wide demands [47]. Although this may represent a temporary impact on NICU beds, it is unclear how the pandemic may have shaped bed allocations temporarily or permanently. Inconsistencies in the multiple sources of information also existed. Phone calls directly to hospital systems proved helpful, but discrepancies in data from these reports also were observed and reconciled as needed. Survey and phone responses are also subject to bias and error, and there is a high likelihood of error due to reconciling multiple sources of information, including the use of websites that could be outdated.

The data available did not allow further differentiation of NICUs located in free-standing children’s hospitals associated with separate but proximately located adult hospitals as compared to those not associated with adult hospitals. These different ways that NICUs can exist within hospital systems is another important area for future inquiry. Additionally, there is a lot of noise in the way that beds can be reported. This study collected information on the number of approved NICU beds to define NICU size/volume, which does not account for NICU census. The number of approved beds were also put in context of the highest level of acuity at that hospital, with no differentiation of number of beds at each level of acuity within each hospital. In addition, our data is complicated by states having different classifications of NICU levels, and those with different requirements for each level [8]. Although the AAP created a new classification system to reduce variability across the United States [14], several states may not have adopted the new classifications at the time of data collection. In addition, a change in classification may have occurred during data collection, leading to inconsistencies in the labeling of NICU acuity level. Further, while we investigated the hospital characteristics in context of CON legislation in each state [48], we did not investigate findings in context of Centers of Excellence or other markers of quality, which is an important area for future inquiry.

The population statistics were derived from the city or county where the hospital was located; however, if a NICU resided near a geographic boundary between states, counties, or cities, the correlations calculated may not properly account for true patient population in a particular NICU. Further, the population surrounding the hospital (in its city or county) may not relate to the geographic markets for newborn or NICU care within the geographic area surrounding the hospital, meaning the data may not necessarily represent where the NICU population was drawn from. Because we investigated each NICU as a unit, specific geographical areas with a large density of NICUs may confound the results, as the characteristics of one unit in the area may influence the others. Further, this study did not account for regionalization, which could have impacted the number, size, and type of NICUs in a given area. Although we were able to demonstrate relationships between higher population of minorities and people living in poverty associated with a greater number of NICU beds, this does not imply greater access to NICU services, and warrants further investigation.