Mapping the Global Burden of Disease project’s Summary Exposure Values by state for 2019

As part of a current project with Calvin Isch and Richard Brown, I’ve been looking at the Global Burden of Disease project’s Summary Exposure Values (SEVs). The SEV is a really useful measure for us, because we’re interested in the extent to which people’s perceptions of risk are associated with objective measures of exposure to health risks. Luckily, the Global Burden of Disease project have done some incredibly detailed work to try to quantify exposures to certain risks.

The GBD describe the SEV as follows:

“A measure of a population’s exposure to a risk factor that takes into account the extent of exposure by risk level and the severity of that risk’s contribution to disease burden. SEV takes the value zero when no excess risk for a population exists and the value one when the population is at the highest level of risk; we report SEV on a scale from 0% to 100% to emphasize that it is risk-weighted prevalence.”

I would recommend this excellent Lancet paper for more details on the construction of this measure.

Since I’m particularly interested in perceptions about uncontrollable mortality risks (risk exposures which are not impacted by individual behaviour), I’ve been using the SEV values for environmental and occupational risks, which is a combined index of those risks not classified by the GBD project as being related to behaviour (see https://ghdx.healthdata.org/ for more on the data).

This all seemed like an excellent excuse to make a new map, this time using Leaflet, an excellent R package, which allows you to create and customise interactive web maps without knowing any JavaScript. Here’s the resulting map. To see the interactive version, please click on the image link and view the map in RPubs, as WordPress won’t allow me to embed it using an iframe.

A map of the Global Burden of Disease (GBD) Project’s summary exposure values (SEV) by state for the USA in 2019.

Looking to convert various statistics to correlation coefficients? Here’s a script I made earlier

In the process of producing a recent meta-analysis on telomeres as markers of exposure to stress and adversity, we needed to convert various effect types (e.g. standardised betas, F-ratios and T-statistics) to correlation coefficients, ready to be meta-analysed. The script is fairly straightforward, but it took a little while to write the script and format the data entry to work smoothly with the script. So, to save others some time and energy, I’ve made a generic version of the script, along with a template input file–just to make things easy to use. You can download the files here, along with a guide on how to correctly enter your data into the template input file, so that the correlation converter script can read it.

Quick-start guide:
1. Enter your data into this template csv, using the table in this document as a guide.
2. Make sure the template csv is in the same folder as this R script.
3. Run the script in R. If you don’t have R, you can learn more and download it here.
4. Once you’ve run the script, a new file, entitled “MetaAnalyisCommonAssociations.csv” should appear in the same folder. It should contain all the data in your original csv, plus two new columns containing the “CommonEffect” (the correlation coefficient) and the “CommonEffectVariance” (the measure of variance).

What is the script doing? The script contains comments explaining what each piece of code does, but for those who find diagrams and equations easier to follow, this flow chart shows the operations it performs:

EffectConversionFlowChart