The Effect of Electrical Load Shedding on Pediatric Hospital Admissions in South Africa

The Republic of South Africa (SA) faced repeated episodes of temporary power shutdowns in 2014/2015, but also in the years thereafter. Based on my co-author’s experience at the burns unit at Red Cross children’s hospital, we had the hypothesis that this may have caused an increase in pediatric hospital admissions. After collecting publicly available data from Twitter and Facebook, establishing co-operations with the city of Cape Town, acquiring weather data, developing a directed acyclic graph and under the consideration of various statistical angles, we concluded that load shedding (i.e. the temporary power shutdowns) did indeed increase the number of hospital admissions (by 10%; 95% confidence interval: 4%-15%).

What I like about this study is that it shows how to creatively answer questions when it is difficult to obtain data; in our case from ESKOM, the national monopoly power supplier. And how to integrate causal thinking routinely into data analysis: co-operations and discussion on various levels helped us to derive a well thought through DAG, to gain confidence into our results and interpretations. There are many interesting angles in the supplementary material: for example the use causal inference in time series, the usefulness of model averaging when deciding for time trends, and a lot of clinical and spatial nuances not covered in the main text.

Those who are interested in the paper, may favour to watch the video abstract before jumping to the pdf…