Epidemiologists are key to MSF teams responding to disease outbreaks. By gathering information about who has got the disease, epidemiologists can look for patterns and help teams understand how fast a disease is spreading, and how it can be stopped.
As an epidemiologist, or “epi” working as part of an MSF response, I expected – and was prepared – to face challenges, and potentially some serious obstacles.
What didn’t occur to me was that my first hurdle would be doing something as familiar to me as data management and analysis.
Data management is one of the cornerstones of epidemiology – we need to have all our information organised before we can start using it to get insights about an outbreak.
During my handover with my predecessor I was not willing to waste any precious time discussing the statistical software I used for data analysis. Nor did it even cross my mind to do so. As any epi will understand; I was comfortable with one particular type of data analysis software. At that moment, in the midst of an outbreak, I wasn’t really prepared to consider switching to a new software for data analysis purposes.
However, my predecessor’s data analysis software “loyalty” lay elsewhere. That meant that I inherited some data analysis scripts in an unfamiliar software suite and that I had to spend most of my time trying to re-write these into the software scripts that I was familiar with.
I also naively presumed that I would be able to mix-and-match between the two data analysis platforms to be able to visualize the cases associated with the hepatitis E outbreak in maps. Maps are really useful when you’re trying plan a response as they can help the team see where they might need to focus their efforts: they can help us get to the source of an outbreak, faster, or help us see places that aren’t affected yet but are at risk, so we can get preventative measures in place, like organising a vaccination campaign.
These issues consumed my first two weeks in Am Timan. Losing time at such an early stage in my field placement had knock-on , not only for me, but also for the previous epi, who I was forced to contact for assistance.
Troubleshooting this together, while being in different places in the world was far from ideal, due to a poor internet connection; relying on intermittent dialogue and email screenshots back and forth.
Making a difference
For these reasons, I think the R for Epis (R4Epis) initiative will be a turning point for epis operating across MSF (and beyond), none of whom can afford to lose time in this way. By developing a library of standard, context-specific analysis tools, situations like mine should become avoidable, and transitioning between epis during outbreak and emergency responses should become much smoother.
And what can we do with this extra time? Well the obvious answer is: dive straight into analysis, spend more time focusing on careful interpretation, and be more innovative in our recommendations and implementation for health interventions supported by the data.
Also, as epis will be less swamped with data and with the task of trying to find a path through the analysis. They will have time (and energy) to try and look beyond a simple exploratory analysis of their data, and be able to conduct a much more in-depth analysis.
As I gain more time from initiative like R4Epis, I look forward to using the additional capacity created to review more literature around the topic, and to strengthen the recommendations I am able to generate for the project.
Considering there are a wide variety of computer programs and web-based tools available to assist in epidemiological investigations, why choose R-software?
The most important reason is that R is open source software – in other words, it’s free and transparent.
Freely available software packages are limited in number and readiness of use and they are not always user friendly. In the past I have been using software that requires a licence, which is expensive and requires several levels of approval.
Open source, free software can be used both at work and on personal computers – providing the opportunity to finalise reports after assignments or rerunning analyses for the purposes of an article or conference.
One R feature I am especially looking forward to is the ability to also do spatial analysis with the data you have – in short, making a map. This is a valuable tool in monitoring any outbreak or illustrating the geographical distribution of the disease in question.
R also detects the main language of the operating system in the computer and tries to enable menus and dialog boxes in that language – a note-worthy and valuable function in an international organisation such as MSF.
Making the change
The list of features that I’ve highlighted is by no means exhaustive. Also, don’t assume that I am an expert in R just because I am excited about this project. Far from it; switching to R will require a lot of effort on my part, including extensive training in person, watching youtube videos and learning from the scripts that are generated through the R4Epis project.
Also, probably more importantly, switching to R will really require a positive and open attitude to change my attitude towards new, unfamiliar software. Therefore, it will be some time before I can give a more comprehensive list of benefits (and drawbacks).
However, despite some inevitable apprehension, the potential of the R4Epis initiative is clear and promising. The future for epidemiological investigations in the field is looking more bright – it offers homogeneity of analysis, time for in-depth investigations, and extensive analytical capabilities. Ultimately this will result in stronger and better targeted interventions allowing us to better serve the populations we want to help.