I am using country-level incidence estimates from the GBD study by country in BenMAP format which are available here: https://www.epa.gov/benmap/benmap-ce-regional-datasets. I have problem in adding the health impact function, I used the predefined function by Krewski et al. (2009); unfortunately, Benmap does not detect the suggested health impact function. Could you please help me in this regard. Thanks a lot in advance!
The answer below assumes that your question refers to the Krewski (2009) all-cause mortality function. To clarify, what exactly do you mean when you say “BenMAP does not detect the suggested health impact function”? Do you mean that it just will not work when you try to run an analysis?
The reason why the Krewski function will not work is that its health endpoint (mortality, all cause) is not in your incidence dataset. The GBD study attributes only five health risks to ambient particulate matter: cerebrovascular disease (stroke), ischemic heart disease (IHD), lung cancer, chronic obstructive pulmonary disorder (COPD), and acute lower respiratory infections. The first four are in your incidence dataset.
The Krewski et al. (2009) study provides concentration-response relationships for three mortality causes: all cause, IHD, and lung cancer. You can see this by viewing the EPA Standard Health Impact Functions in the US setup:
To use these health impact functions, you need to use the database export and database import tools under the “Tools” dropdown menu. And, you need incidence datasets that exactly match the endpoints (see “Endpoint” column above). You have mortality incidence rates for IHD and Lung Cancer, BUT you actually need to retype the endpoint name from “Mortality, Mortality, Lung Cancer” to “Mortality, Lung Cancer.” From there, you should be able to select the IHME incidence dataset in the “Health Impact Functions” window in your analysis:
However, this is a somewhat incomplete picture of mortality attributable to air pollution, as you are ignoring other health endpoints (e.g., stroke, COPD). We are currently working to incorporate the GBD study’s Integrated Exposure Response (IER) functions, which will cover all four health outcomes in your incidence dataset. These functions should be publicly available within a few months.
In the meantime, you may consider downloading all-cause mortality from the GBD results tool for use with the Krewski all-cause function. You may use your existing incidence data as a template for formatting. Note that the results tool will give you deaths per 100,000 individuals. You must divide this value by 100,000 to generate deaths per person per year. Here is a guide to a few of the more confusing settings on the results tool:
• Context: Cause
• Measure: Deaths
• Cause: All causes
• Metric: Rate
I hope you find this information helpful.
Thank you so much for your response. I deeply appreciate you for providing such helpful information.
To answer your question, yes, I meant that the health impact analysis does not work when I try.
I followed the steps you told me, thanks a lot for your guidance.
I downloaded the GBD incidence values, and I divided the values by 100,000.
Is it reasonable that I use the resulting incidence values for the city of Tehran, by multiplying the incidence values of each specific group age to the population of each grid?
I did so, I gained a health impact result like the following figure. I would be grateful if you could tell me the resulting graph shows the change in the number of mortality or the number of mortality after running the scenario?
Thank you so much!
So what if I want to have the baseline incidence rate per day, because my CRF is daily metric (same day, short-term exposure), can I divide the value of deaths per person per year by 365?
Is it reasonable to do it?
GBD result tools gives three value for each section named “Value” ,“Upper” and “Lower”.