Our project maps out areas that lack the proper infrastructure to ensure a healthy community. We did this by identifying “healthcare deserts,” or areas across the U.S. where people lack adequate access to six key healthcare services: (1) pharmacies, (2) primary care providers, (3) hospitals, (4) hospital beds, (5) trauma centers, and (6) low-cost health centers. We then looked at how socioeconomic barriers, such as low income, lack of insurance, and limited internet access, can make it even harder to access care in a healthcare desert and perpetuate health disparities.
Since release, our work has been shared in the Washington Post, and the Philidelphia Inquirer among others, showcasing the shortage of healthcare that many American’s face. Our data insights have also helped inform fieldwork for practitioners as well. Rural pharmacy and medical students have used our data to learn about the challenges their patients face and to develop community health projects. Additionally, we are working with researchers who are innovating in areas such as affordable housing and public insurance design to incorporate our healthcare deserts data into their work.
Using Python, we wrangled data from multiple sources to build our measure of healthcare deserts. To create data visualizations, we needed to tie the data together to a common geographic unit, so we geolocated facility addresses to latitude and longitude coordinates and built Census tract-level and county-level data. Since several of the clinically-based criteria for healthcare deserts were based on transportation time, we calculated driving times to healthcare facilities utilizing a Euclidean distance measure combined with a validated detour index and location- and vehicle-specific average driving speeds.
We also brought in population data from the Census to understand the real magnitude of who is impacted by lack of healthcare access. We estimated the size of the populations living in healthcare deserts as well as looked at the socioeconomic characteristics of those populations. To summarize the Census tract- and county-level data and look at the national picture, we calculated statistics on the distribution of the population (and population characteristics) across different healthcare desert dimensions and levels of healthcare desert intensity.
What was the hardest part of this project?
One of the greatest challenges was coming up with a meaningful definition of a “healthcare desert.” Rather than use a systematic definition based on arbitrary thresholds (such as looking within a 10-mile radius or using relative measures such as percentiles), we conducted an extensive review of existing health services literature in order to develop criteria based on real data and clinical research.
Healthcare access is a multi-faceted problem, and we wanted to come up with a measure of “healthcare deserts” that translated all of these important dimensions into an easy-to-understand measure.
What can others learn from this project?
Our project was impactful because it distilled a lot of complex information into a simple metric that makes sense and is grounded in research and data. We believe other journalists can benefit from tapping into the wealth of existing clinical and academic research to inform their data stories and create more nuanced data visualizations.