Pre-Natal Care & Birth Outcomes in Atlanta, Georgia

In a state with high rates of low birth-weights, how relevant are accessibility to prenatal care and the neighborhood women live in?

Introduction

Georgia's maternal mortality rate is one of the highest in the nation, and there is a marked disparity in risk between Black women and women of other racial backgrounds, with the most marked difference in metropolitan Atlanta. (CDC, 2018; Platner et al, 2016). Additionally, Georgia consistently performs poorly in rankings of birth outcomes, such as low birth weights (4th) and pre-term births (6th) (CDC, n.d.), which are risk factors for babies' health and development.

Georgia is part of a nation-wide problem. The United States has the highest maternal mortality rates among other developed countries and there are stark racial disparities in maternal health outcomes. Many of these maternal and child deaths are considered preventable with appropriate medical interventions. Access to and use of prenatal care are critical to identifying high-risk pregnancies, preventing maternal deaths, and improving birth outcomes.

This project seeks to understand how accessible pre-natal health care is in the city of Atlanta, Georgia, and identify spatial patterns of poor maternal-child health outcomes across the city. We would like to understand if there is correlation between neighborhoods in which pre-natal healthcare is less accessible, and those neighborhoods in which poor maternal and infant health outcomes are higher. Additionally, we are interested in understanding whether these patterns are consistent across majority white, majority black, and racially mixed neighborhoods. We expected to find disparities in majority black neighborhoods, based on national trends (Platner et al., 2016).

Related work

There is a significant body of work on questions of prenatal care access and use, the spatial distribution of low-birth rates, and the influence of demographic and socio-economic factors in these outcomes. Moreover, there is a notable body of work that uses geospatial analysis to understand these questions, and a few authors (Yin, 2017, Tu et al., 2012) that focus their analyses on Georgia. This literature is supplemented by grey literature from government and healthcare organizations.

An article by Yin (2017) found that women in Georgia with better spatial access are less likely to have adequate care, which contradicts existing literature and indicates that this relationship is both “complicated and context specific.” This geospatial analysis used a multimodal, two-step floating catchment area method to estimate special accessibility, and Yin recognizes that incorporating detailed bus routes would improve the spatial accessibility estimation An article by Tu et al (2012) seeks to understand the relationship between low birth weights and six factors, including the adequacy of prenatal care, in their analysis of spatial variations in birth weights across Georgia. They found that the positive impacts of prenatal care were higher in lower income, lower educational attainment areas. In their study, these areas tended to be rural, and it is unclear if the relationship exists in urban areas. Although rural areas are often highlighted in discussions of maternity care deserts, a 2020 report by March of Dimes found that 1 in 3 women of childbearing age live in urban maternity care deserts. However, this analysis, which covers the entire United States, looks at care available at the county level, and does not explore how access may differ within counties or urban areas.

We have chosen Atlanta as our study site because it is highly diverse and highly segregated. According to Nate Silver and Brown University’s American Communities Project (2015), Atlanta is a very highly segregated U.S. city. Atlanta’s citywide index is 56.8%, its neighborhood diversity index is 30.7%, and its integration/segregation index is -14.5%. The city’s poverty rate (20.8%) is double the 2019 national average. The city is 40.9% white, 51% black, and 4.4% Asian (notably a majority minority city) and has a population of 488,800. Georgia also has poor maternal-child health outcomes. The state ranks 4th in the low birth weight rate and 6th in preterm birth rates (CDC, n.d.).

Research Questions

  1. How many women are able to access prenatal care providers in Atlanta using public transportation?
  2. What are the spatial patterns in poor birth outcomes in Atlanta?

Data Sources


Methodology

Pre-processing for spatial statistics analysis

Workflow for Global Moran's I and Getis-Ord Gi* Hot Spot Analysis

Create a Geolocator and Geocode Prenatal Care Provider Addresses

Preprocessing Layers

Create Network Dataset, Run Origin-Destination Cost Matrix and Service Area Analysis

Create a Single Layer from Service Area Analysis and Calculate Population in Service Areas


Results

Network Analysis : Who Can Reasonably Access Prenatal Care by Bus

Spatial Statistics on Low Birth Weight

General statistical trends can be viewed in these maps, showing that the census block groups with the highest proportion of African Americans is in the southwest of the city, and the median household income is lower than the median in many of those tracts as well. This shows the density of the vulnerable population (and does correlate to the regions in the hot spot map of high incidence of low birth weight shown below). The relevance of spatial patterns for rates of low birth weight require the thoroughness of the subsequent spatial statistical analysis.

Incremental Spatial Autocorrelation was used to determine a suitable distance band, and then Global Moran's I was run on the low birth weight data to determine that this data is indeed highly clustered.

This output shows the strong spatial autocorrelation for percentage of low birth weights in Atlanta. Low birth weight is strongly correlated to location.

This information was used to create a map of hot and cold spots (with Getis-Ord Gi*) of low birth weight rates across the city.

The two maps show a strong correlation to the regions of the city that face high rates of low birth weight and are majority African Americans and out of bus service areas for prenatal care.

The Getis-Ord Gi* hot spot analysis marks areas of high positive z-score as intense clusters of high values (so more incidences of low birth weight) and areas and low negative z-scores as intense clusters of low values (so less incidences of low birth weight). Low birth weight infants (under 2,500 grams) have a higher risk for mortality than babies born at regular weights.


Limitations & Future Study

This research analysis was limited by the available data. While originally we were interested in examining maternal mortality, the data was not publicly available at the intended study scale, likely due to privacy concerns. Potential further study with sensitive data would be enlightening but have to be displayed in a limited manner in publication to ensure respect for study subjects and their families.

There are indications that maternal-baby health outcomes have worsened over time in the U.S., so an Emerging Hotspot Analysis might provide insight into spatial patterns of change, especially in conjunction with changes in other risk factors (race, poverty, healthcare access). This analysis would require temporally and spatially-detailed data.

Future analysis could replicate our approach in other racially and economically diverse cities to better understand if our results are unique to Atlanta or reflective of urban socio-economic dynamics.


References

  • CDC, 2018. Maternal Mortality by State, 2018 [online]. NCHS, National Vital Statistics System. Available from https://www.cdc.gov/nchs/maternal-mortality/MMR-2018-State-Data-508.pdf [Accessed 15 February 2021].
  • CDC. N.d. National Center for Health Statistics: Georgia [online]. CDC. Available from: https://www.cdc.gov/nchs/pressroom/states/georgia/ga.htm[Accessed 26 January 2021]
  • DataUSA. n.d. Atlanta, GA. Available from https://datausa.io/profile/geo/atlanta-ga#housing [Accessed February 15 2021].
  • March of Dimes. 2018. Health Moms. Strong Babies [online]. March of Dimes. Available from:https://www.marchofdimes.org/peristats/tools/ReportFiles/HMSB/Healthy%20Moms%20Strong%20Babies_Georgia.pdf [Accessed 29 January 2021.]
  • Osterman, M. J.K., and Martin, J.A., 2016. Timing and Adequacy of Prenatal Care in the United States, 2016 [online]. Centers for Disease Control. Available from: https://www.cdc.gov/nchs/data/nvsr/nvsr67/nvsr67_03.pdf [Accessed 29 January 2021].
  • Platner, Marissa, MD, Tammy L. Loucks, MPH, DrPH, Michael K. Lindsay, MD, MPH, and Jane E. Ellis, MD, PhD, 2016. Pregnancy-Associated Deaths in Rural, Nonrural, and Metropolitan Areas of Georgia. Obstetrics & Gynecology, Vol. 0, No. 0, Month 2016, 1-8.
  • Silver, Nate, 2015. The Most Diverse Cities are Often the Most Segregated [online]. FiveThirtyEight. Available from https://fivethirtyeight.com/features/the-most-diverse-cities-are-often-the-most-segregated/ [Accessed 31 January 2021].
  • Tu, J., Tu, W. and Tedders, S.H. 2012. Spatial variations in the associations of birth weight with socioeconomic, environmental, and behavioral factors in Georgia, USA. Applied Geography, 34, 331-344.
  • Yin. P. 2018. Does Better Spatial Access Lead to Better Use of Prenatal Care? A Population Study in Georgia. The Professional Geographer, 70(3), 363-373.

Final project for IDCE 388 Advanced Vector GIS. Wintersession 2021. Clark University. Thanks to Yelena Ogneva-Himmelberger and Ellen Pappalardo.

Pre-processing for spatial statistics analysis

Workflow for Global Moran's I and Getis-Ord Gi* Hot Spot Analysis

Create a Geolocator and Geocode Prenatal Care Provider Addresses

Preprocessing Layers

Create Network Dataset, Run Origin-Destination Cost Matrix and Service Area Analysis

Create a Single Layer from Service Area Analysis and Calculate Population in Service Areas

This output shows the strong spatial autocorrelation for percentage of low birth weights in Atlanta. Low birth weight is strongly correlated to location.