Materials and Methods

A descriptive analysis of current US perinatal resources was used to determine spatial relationships between the population of women of reproductive age (ages 15–44 years), individual perinatal subspecialists (maternal fetal-medicine and neonatology), and obstetric and neonatal level III and higher facilities in the United States. Data are presented by state and US Department of Health and Human Services (DHHS) region. We used geographic visualization and conducted surface interpolation, nearest neighbor, and proximity analyses (described in the following text).

All 50 states and the District of Columbia were included. The 2010 US Population Census was used to determine the number of women of reproductive age nationally and by US DHHS region. We assumed pregnant women were equally distributed across the women of reproductive-age population. Therefore, the proportion of pregnant women within a geographic area who had access to a perinatal resource was used as a proxy for access among women of reproductive age. The number of pregnant women was approximated by using the number of live births from the 2012 National Center for Health Statistics Natality File.

Individual perinatal subspecialists included practitioners who have subspecialty board certification in maternal-fetal medicine (MFM) or neonatal perinatal medicine (referred to as neonatologists) according to the October 2013 National Provider Index. Only subspecialists listed as active (currently practicing medicine) were included.

Obstetric critical care unit (OCCU) refers to facilities with a level III obstetric unit as identified in the 2012 American Hospital Association (AHA) annual survey data. AHA defines a level III obstetric unit as one that provides services for all serious illnesses and abnormalities and is supervised by a full-time maternal-fetal specialist; neonatal critical care unit (NICU) refers to a facility with a level III or higher NICU as identified in the 2011 American Academy of Pediatrics (AAP) directory data. NICUs were linked to the AHA database using Link Plus software. Unlinked NICUs from the AAP list because of name or address inaccuracies were resolved using the hospital’s web site.

The Homeland Security Infrastructure Program (HSIP) Gold 2010 hospital data set was used to confirm hospital locations. The HSIP also provides locations for individual facilities within a hospital system. Records unable to be linked to the HSIP database because of address inaccuracies were geocoded manually using physical addresses listed on the facility’s web site.

A surface raster was created to visualize the women of reproductive-age population using inverse distance weighting spatial interpolation of census block populations. Inverse distance weighting is a deterministic interpolation of values that assumes each measured point has a local influence that diminishes with distance. Raster and vector map layers were built for all women of reproductive age. Point layers for each of the different types of perinatal critical care providers were then created.

Lists of MFM subspecialists and neonatologists were extracted from the National Provider Index using taxonomy codes specific for their primary practice. Subspecialists were then geocoded to their listed office location using Centrus Desktop version 6.0. Subspecialist layers were used to compare the number and distance of subspecialty units and practitioners using nearest neighbor and proximity analyses.

Subspecialist layers were also used to create zonal layers for access to care proximity analyses. Zonal layers were formed using Euclidean distance buffers placed around each hospital at 10 and 50 miles and each individual subspecialist at 10 miles. Euclidean distance buffers use straight line distance (as the crow flies) and appear on a map as perfect circles around a point. Zones that overlapped were merged to form confluent zones, and zones that crossed state boundaries were clipped by the state boundaries, allowing zones to be analyzed by state.

Thiessen polygons were generated around each hospital for the entire study area (continental United States, Alaska, and Hawaii) to link women of reproductive age with their nearest critical care hospital; the size and shape were dependent on the proximity to neighboring hospitals. Polygons that included parts of more than 1 state were divided into 1 or more sections, with 1 section for each state represented. Women of reproductive age residing in sections with concordant state locations (section and nearest hospital in the same state) were considered in-state; women of reproductive age with discordant state locations (section and nearest hospital not in the same state) were defined as out of state.

To describe access to critical care, population layers were overlaid with zonal layers and analyzed for spatial proximity. The population located within 50 miles of a critical care facility were considered to have access; those located more than 50 miles from a critical care facility were considered not to have access. Distance is a well-established barrier to accessing care; 50 miles was selected based on both the golden hour and because it approximates the farthest distance most people appear willing to travel for specialized medical care. Proportions of women of reproductive age with and without access to OCCUs and/or NICUs were calculated. In addition, zonal overlap was assessed to estimate the number of women who had access to both critical care for herself and her newborn.

Distance between the nearest complementary OCCU and NICU was evaluated using a nearest neighbor analysis to estimate the percentage of facilities requiring transfers of critically ill mothers or infants. OCCUs and NICUs less than 0.25 miles apart were considered the same facility. Transfers of less than 10 miles, and those of at least 10 miles, were assessed separately. The number of facilities within 10 miles of each other was also used to describe clustering of critical care hospitals. The 10 mile distance was based on examining the distribution of the data. The median distance between subspecialists’ primary office location and a critical care unit was approximately 0.5 miles. However, we decided to use a more generous (although somewhat arbitrary) cutoff of 10 miles, which was between 1 and 2 SDs of the mean.

Proximity of MFM practice locations to obstetric units, and neonatologist practice locations to NICUs, was assessed using a nearest-neighbor analysis. All analyses were conducted using SAS versions 9.2 and 9.3 (SAS Institute, Cary, NC) and ArcGIS version 10.3 (ESRI, Redlands, CA).

Results

In October 2013 there were 2.5 neonatologists for every 1 MFM specialist in the United States (4797 neonatologists; 1888 MFM specialists). Similarly, in 2012 there were 1.4 hospitals with an NICU for every 1 hospital with an OCCU (848 NICUs; 599OCCUs). The aforementioned information corresponds to approximately 6 neonatologists per NICU (4797/848), compared with approximately 3 MFM specialists per OCCU (1888/599). In 2012, there were 4.8 MFM specialists and 12.0 neonatologists per 10,000 live births, respectively ( Table 1 ).

Nearly all obstetric (95%) and NICU (99%) units were located in urban areas. Therefore, whereas large geographic areas in the United States were not covered by either of the perinatal facility zones (46%), the vast majority of the population did have access to (defined as living within 50 miles of) one or both types of critical care units (94%). Specifically, 91% of the US population of women of reproductive age had access to an OCCU, 93% had access to an NICU, and 87% had access to both.

However, access to specialized obstetric and neonatal care varied by state and region ( Table 2 ). In all but 2 states within DHHS regions 1–5 (primarily states east of the Mississippi River), at least 80% of the women of reproductive age had access to an obstetric critical care unit, and in all states at least 80% of the women of reproductive age had access to an NICU.

Table 2

Percentage of women of reproductive age (ages 15–44 years) living within 50 miles of perinatal critical care services, by subspecialty, 2012

HHS region	State/district name	Level III NICU	Level III OB unit	Difference
				OB-NICU	Both are >80%	Either is <20%
I	United States	92.8	90.5	–2.4	√
	Connecticut	99.8	99.8	0.0	√
	Maine	89.4	83.4	–6.1	√
	Massachusetts	98.6	98.6	0.0	√
	New Hampshire	97.1	97.4	0.3	√
	Rhode Island	98.6	98.8	0.2	√
	Vermont	89.3	92.9	3.6	√
II	New Jersey	99.1	97.5	–1.6	√
	New York	97.1	96.9	–0.2	√
III	Delaware	80.5	97.1	16.6	√
	District of Columbia	99.0	99.0	0.0	√
	Maryland	95.5	96.8	1.2	√
	Pennsylvania	97.9	94.6	–3.3	√
	Virginia	96.6	97.5	0.9	√
	West Virginia	80.3	87.4	7.2	√
IV	Alabama	87.1	75.7	–11.4
	Florida	96.7	92.8	–3.9	√
	Georgia	93.5	96.5	3.0	√
	Kentucky	84.0	86.4	2.4	√
	Mississippi	81.4	69.1	–12.3
	North Carolina	97.9	98.7	0.9	√
	South Carolina	94.2	94.8	0.6	√
	Tennessee	94.9	93.0	–1.9	√
V	Illinois	96.4	91.6	–4.7	√
	Indiana	95.6	95.1	–0.5	√
	Michigan	94.7	91.1	–3.6	√
	Minnesota	84.8	79.9	–5.0
	Ohio	95.8	96.5	0.7	√
	Wisconsin	91.0	84.9	–6.1	√
VI	Arkansas	66.3	60.1	–6.2
	Louisiana	97.8	91.8	–5.9	√
	New Mexico	59.7	63.7	4.0
	Oklahoma	75.6	84.4	8.8
	Texas	94.6	85.8	–8.7	√
VII	Iowa	73.8	91.9	18.1
	Kansas	76.2	77.9	1.7
	Missouri	84.9	80.4	–4.5	√
	Nebraska	71.7	71.6	–0.1
VIII	Colorado	89.5	94.3	4.7	√
	Montana	45.4	45.4	0.0
	North Dakota	73.8	22.0	–51.8
	South Dakota	57.8	43.4	–14.4
	Utah	93.3	91.4	–1.9	√
	Wyoming	16.1	16.2	0.2		√
IX	Arizona	89.2	89.6	0.4	√
	California	97.8	96.2	–1.6	√
	Hawaii	64.1	63.9	–0.1
	Nevada	95.4	75.2	–20.2
X	Alaska	55.8	1.9	–53.9		√
	Idaho	76.4	58.6	–17.8
	Oregon	90.3	72.5	–17.8
	Washington	90.4	95.2	4.8	√

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

1. The role of aspirin dose on the prevention of preeclampsia and fetal growth restriction: systematic review and meta-analysis
2. Evaluation of microscopic hematuria and risk of urologic cancer in female patients
3. Trends and racial and ethnic disparities in the prevalence of pregestational type 1 and type 2 diabetes in Northern California: 1996–2014
4. Longitudinal fetal heart rate changes are not generally disregarded but should be supplemented by meaningful categorization of decelerations

Stay updated, free articles. Join our Telegram channel

Join