Mapping Health Care Access in Malawi: Update 3

After completing both my literature review to establish context for the poor health outcomes in Malawi and introductory training for ArcGIS (a geographic information system that uses spatial data to create maps for analysis), I just wrapped up the final stage of my research project: creating my own map that relates under 5 mortality rates in Malawi to the distances children travel to receive health care and summarizing my findings in a science journal-style paper. This last portion of my research had the largest learning curve due to the many challenges and setbacks I encountered. I also worked with my faculty advisor, Professor Dolan, to narrow the focus of my research project.

 

While attempting to create my own map using my newly acquired basic ArcGIS skills and several data sets compiled for this project, I ran into several technological obstacles. First, my ArcGIS licensing did not enable me to use spatial analysis extensions which meant I was not equipped to calculate distances between two points (such as health facilities and Malawian children). Thanks to Robert Rose of the Center for Geospatial Analysis at William & Mary as well as a successful reinstallation of the software, this problem was quickly fixed. Once I had fully functioning software, my data sets failed to import due to their bulky size. At this stage, I also discovered which file types ArcGIS can and cannot work as I observed that .shp files alone enable spatial representation.

 

Sorting through this series of technological issues, I learned about various aspects of data management. I used several national surveys conducted by the Demographic Health Survey Program which collects data on several population, health, and nutrition indicators around the world in order to equip decisionmakers with information to effectively manage and implement programs and policies. These surveys – Malaria Indicator Survey (MIS), Demographic Health Survey (DHS), and Service Provision Assessment (SPA) – included responses to hundreds of health-related questions, some of which were beyond the scope of my research. In order to reduce the size of the files to make them useable in ArcGIS, I opened up the survey responses in Excel, learned how to decode the column headings using the proper edition of the DHS Recode Manual, and discarded unnecessary information in order to work strictly with the data that is relevant to my project. For the 2014 MIS, I kept only the reported locations of Malawian 2,171 children under 5 years old. For the 2013/2014 SPA, I kept the locations, types, funding structure, and diagnostic capacities of the 956 documented health facilities. For the 2015 DHS, I kept only the under 5 mortality rates by region. While the additional information available from these data sets was fascinating and could help implicate confounding variables in this study, it was very helpful to open up the files, better understand the data I worked with, and keep only the necessary data.

 

Thanks to Professor Dolan, I also learned about the process of maneuvering data sets through several different software systems in order to have appropriate file types. While cleaning up the survey response spreadsheet, I worked in a .csv file. After narrowing down the number of responses contained in the .csv file, this file was then opened in Q (a data analysis software) and converted into a .shp file which could be opened in ArcGIS. At this point, I also learned the importance of projection and identifying the projection of my data set because my 2D map would inevitably be a distorted representation of the spherical globe.

 

Once my data files were cleaned up and converted to the correct file types, I was able to apply ArcGIS skills picked up during training. I layered three different data sets: the region-wide under 5 mortality rates formed the base of the map because this data set also drew the national and regional boundaries of Malawi. The locations of health facilities and children under 5 years old were then represented as points on this map of Malawi. The “Near” tool in ArcGIS allowed me to calculate the distance from each child to the nearest health facility. I then programmed circles whose sizes were proportional to the distance from a health facility (the farther away a child was from a health facility, the larger the circle representing the child’s location on the map). Finally, I created a choropleth map which used gradient shades of turquoise to show differences in regional under 5 mortality rates.

Distance_Under 5 Mortality_Malawi

During the process of constructing this map, I spoke with Professor Dolan about the roadmap of my research. The natural next steps for this project would be to carry out statistical analysis with the software system STATA in order to determine the strength of association between distance to health care and under 5 mortality. I would also do tests of statistical significance to confirm or deny any apparent association. While these next steps were part of my original plan for the summer, the process of completing ArcGIS training and learning to manage data became time-consuming focal points in my research. Professor Dolan and I agreed that it was necessary to narrow the aims of my research in order to stick to the time table of this research project. Therefore, the products of my research project are my map (“Distance to Health Care and Under 5 Mortality in Malawi”) and research paper which compile all of the data needed to perform future statistical analysis. These next steps –learning how to use STATA and executing the necessary statistical tests – could form the basis of a proposal for another research project in the near future. Statistical analysis is needed in order to draw conclusions from this initial research about the effect of distance to health care on mortality.

 

The second product of my research is a research paper which synthesizes findings from my literature review with details about the process and materials involved in producing my map. This paper was an exercise in writing scientifically and mimicking the format of a scientific research journal, much like those that I studied during my literature review. I separated my research summary and findings into four categories: background (which establishes context for the variables being studied), methods (which explains the methods used to acquire data), results (which provides the research findings), and discussion (which draws conclusions from the findings, discusses limitations, and makes suggestions for future research).

 

This research project gave me significant practice with reviewing academic scholarship, utilizing software that is frequently used in the public health field, and presenting my findings in the standard format of a scientific journal. I also learned how to overcome several technological challenges, manage large data sets, and reflect on my progress in order to modify the aims of my research project. I enjoyed the freedom to continually ask and explore questions that were both relevant to my big, driving research question about the relationship between distance to health care and mortality rates and fascinating to me, such as how the prevalence of corruption in Malawi affects health outcomes.

Speak Your Mind