News‎ > ‎

Open access, georeferenced epidemiological data

posted Feb 3, 2012, 8:30 AM by Nicola Wardrop   [ updated Nov 30, 2012, 1:15 AM ]
I was recently asked to prepare a hands-on GIS practical for some of the distance learning MSc students at the University of Edinburgh's Global Health Academy. What was needed was a worked through example of a risk mapping exercise. Predictive mapping (or risk mapping) of disease requires georeferenced data of observed disease prevalence or occurrence and absence, along with environmental datasets relevant to the distribution of the disease in question (e.g. temperature or landcover). The environmental datasets tend to be fairly easy to obtain...there are many different free sources of such data online. The problem was in obtaining suitable georeferenced epidemiological data.

The data I have acquired and used through my own research are not really suitable for the practical. Ideally what you need is a very clear spatial pattern, which can be modelled fairly simply with a few environmental covariates at a medium spatial resolution. I have spatially modelled my own data and produced predictive maps, but the modelling process was more complex than I would want to use as an example, so I set about trying to find some open access, geo-referenced, epidemiological data, which was freely available for to use for practicals etc.

What I found was that there are actually very few sources of such data. Many maps are provided, but the raw data is generally not supplied online. However, I did find a couple of sources which are ideal for my requirements, so I thought I'd share them:

The Global Neglected Tropical Diseases database has plenty of spatially referenced schistosomiasis survey data
The original John Snow Cholera analysis has been digitised by a PhD student in Geography and Environment at the University of Southampton