C Case Study - WHO guidelines

The provided pipeline uses a relative risk of 1.06 per 10 µg/m³, with lower and upper confidence intervals (lci, uci) 1.02 and 1.08 respectively. The counterfactual is set as the minimum PM_2.5 by state.

Updated WHO 2021 recommendations specify a target for annual PM_2.5 of 5 µg/m³. Interim targets are set at 35, 25, 15 and 10 µg/m³. The review which contributed to the setting of these target PM_2.5 levels used a hazard ratio of 1.08 per 10 µg/m³.

To explore the counterfactual scenario as described by the WHO recommended target, the calculation of counterfactual scenario and the health impact function must be modified.

---

Counterfactual scenario

The counterfactual exposure is calculated in the R target combined_exposures, The function call to do_env_counterfactual can be replaced with a block of code that adds the counterfactual exposure and delta to the output of target data_env_exposure_pm25.

However, the do_env_counterfactual already includes functionality to add these columns based on an absolute value rather than using the minimum exposure by state. To use a counterfactual scenario based on the WHO’s recommended target of 5 µg/m³, alter the cf_mode argument from "min" to "abs", indicating the counterfactual is an absolute value. Provide a third argument cf_value as a numeric set to the WHO recommendation, 5.

Your code will look similar to the following, in place of the combined_exposures target:

  ...
# Provide counterfactual scenario and calculate delta
  tar_target(
    combined_exposures,
    do_env_counterfactual(data_env_exposure_pm25,
                          "abs",
                          5),
    pattern = map(data_env_exposure_pm25),
  ),
  ...

---

Health impact response

Next, modify the R target health_impact_function which returns a function to calculate the attributable number. In the do_health_impact_function, the argument exposure_response_func takes a three element numeric vector, representing the relative risk, lower confidence interval and upper confidence interval (in that order). Change the relative risk to that used by WHO, 1.08. If no confidence intervals are available, they may be set to NA.

Your code should look similar to the following:

  ...
  tar_target(health_impact_function,
             do_health_impact_function(
               case_definition = 'crd',
               exposure_response_func = c(1.08, NA, NA),
               theoretical_minimum_risk = 0
             )
  ),
  ...