Mapping Health Care Access in Malawi: Update 2

Following my literature review of the health care system in Malawi, I have now spent several days on my second phase of research: familiarizing myself with ArcGIS, a geographic information system that works with spatial data. Basically, ArcGIS uses data with latitudes and longitudes to create maps for visual analysis. My faculty advisor, Professor Carrie Dolan, provided me with an ArcGIS training program that she helped create while working with AidData. AidData is a research lab at William & Mary that works to present data in a clear, useful way in order to equip policymakers and international development organizations to make better-informed decisions. ArcGIS is a necessary tool within the world of AidData and public health data analysis. Learning how to navigate this software has given me a new appreciation for tools that communicate findings that are otherwise buried in statistics and spreadsheets. Below is the citation for the training program I completed.

 

Dolan CB, Delcher C, Harris, C., Decatur, A. Introduction to ArcGIS using AidData. Williamsburg, VA: College of William and Mary; July 2013.

 

Working through this training manual is the logical next step in my research because ArcGIS will be a crucial tool in my investigation of the association between distance to health care and under 5 mortality in Malawi. Before I can create my own maps to analyze this relationship, I needed to learn how to maneuver through the software, make use of the built-in features, and manipulate data sets.

 

Along with ArcGIS basics and spatial analysis jargon, the training included several exercises that allowed me to follow along, play with data, and create maps that I could cross reference with screenshots of the actual products provided by the training manual. The data used in this training was from Malawi and ranged from population data collected in a census, physical boundaries of Malawi’s 28 districts, locations of HIV/AIDS projects around the country, and health funding information. As an exercise in manipulating data, I learned how to combine health funding and population data into one table. These two variables were originally measured by two separate studies, but they were both calculated at the district levels. I was, therefore, able to create a table that listed both the level of health funding and the population values for each district. Once these variables were in the same table, I was able to create a new variable that related health funding and population values: I plugged in an equation that divided the monetary value of health funding by the total population by district to determine the amount of money disbursed to each district for health funding per person.

 

Using this health funding per person variable, I was then able to create choropleth maps that visually conveyed the differences in health funding distribution around Malawi. Choropleth maps use shading and patterns to represent proportions of a measurement such as population density. The variations in color signal variations in the measured statistic. To create these maps, I had to learn how to switch on the “symbology” function, choose how to separate the range of funding into several separate color distinctions (for these maps, I experimented with 3-4 breaks in the range), and include a legend. The first map included below (“Health funding per person and HIV/AIDS projects in Malawi”) shows the variations in health funding per person (from the lowest at $4.10 to the highest at $76.36) by district through the different shades of teal. In this map, the boundary lines separate Malawi’s 28 different districts, allowing comparisons to be made across districts. The second layer on this map shows the location of each reported HIV/AIDS project in Malawi. With both variables – health funding per person and HIV/AIDS project location – present, it is easier to draw conclusions about which districts receive the least health funding as well as where HIV/AIDS interventions are found most sparingly. Studying this map could help inform future proposals about health funding distribution as well as where to plant HIV/AIDS projects in order to most effectively expand access to health and address poor health outcomes.

Health Funding_Malawi_Districts

The second map (“Health funding per person and HIV/AIDS projects”) includes the same data as the first map, however different boundary lines are drawn. Instead of comparing health funding across districts, this map allows comparisons across regions. There are three regions in Malawi: the Northern, Central, and Southern Regions. To compute each region’s health funding per person, I discovered new tools in ArcGIS that allowed me to aggregate the data for all of the districts within the same region. For example, I combined the population of the thirteen districts that comprise the Southern Region in order to get the total population of the Southern Region. So, this second map gives a different perspective on the same data used in map 1.

Health Funding_Malawi_Regions

The third map (“Hospitals and HIV/AIDS projects in Malawi”) is the product of different data sets and different ArcGIS tools. The first layer on the map displays all of the locations of hospitals in Malawi. As evidenced by this map, the region with the most funding – the Southern Region – not only has the most HIV/AIDS projects but also the most hospitals. This unequal distribution of hospitals, funding, and health interventions is most apparent in the comparison of the Northern and Southern Regions. This map also builds in an analysis of how far HIV/AIDS projects are from hospitals with the use of red dot symbols that vary in size in proportion to distance. The third layer in this map uses the same symbology tool (varying shades of teal) as maps 1 and 2 to represent the differences in the number of people per hospital bed per district. Between the completely white colored districts that indicate zero hospital beds in the entire district and the dark teal districts that have between two and four thousand per bed, it is clear that Malawi’s health care system does not currently have the capacity to effectively serve its population.

People per bed_Malawi

While Professor Dolan’s training involved using data from Malawi, the relevance of the maps produced during the training program was coincidental. I enjoyed playing around with ArcGIS and discovering some key features and tools that will help me create maps for my own analysis. Although I ran into complications with the licensing and installation of this software, it was fairly user-friendly once it was up and running. The training provided me with a great introductory understanding of the system, however the system’s capabilities are too vast and complex to be covered in entirety in a 100 page tutorial. I am looking forward to using my data on under 5 mortality and health facility location to create maps for my research, but I anticipate needing to rely on my faculty advisor, Google searches, and trial and error in order to complete the next stage of my research.