Author: Lauren Bishop

As a conclusion to this section on epidemiological measures of association and frequency, we want to briefly introduce one measure of public health impact, which can provide important context about the burden of different health outcomes on population health.

Attributable proportion measures the quantity of the health outcome in the exposed group that can be attributed to the exposure. 

In other words, the attributable proportion represents the proportion of health events that hypothetically would be reduced if the exposure did not exist or could be removed from the equation.

Attributable proportion has a key assumption: if the number of health events in the unexposed group is the expected risk (baseline) for that event, then the difference in risk between the exposed and unexposed groups can be attributed to, or caused by, the exposure.

This also means that the attributable proportion should only be calculated for one exposure, or risk factor, that causes the health outcome.

The attributable proportion is calculated by subtracting the risk for the unexposed group from the risk for the exposed group, dividing the difference by the risk for the unexposed group, and then multiplying the quotient by 100.

Example

0.56	0.06

A classic example of attributable proportion is the relationship between smoking and mortality attributable to lung cancer. Our study population of male British doctors in the 20th century contains daily smokers and non-smokers. If the mortality rate for lung cancer among daily smokers is 0.56 deaths per 1,000 persons per year and the mortality rate for lung cancer among non-smokers is 0.06 deaths per 1,000 persons per year, what is the attributable proportion? The attributable proportion is: (risk for the exposed group – risk for the unexposed group)/risk for the exposed group, x 100. Attributable proportion: (0.56 – 0.06)/0.56 x 100 = 89.3%. If the assumptions for calculating the attributable proportion hold, and assuming the two groups of doctors are comparable, these results would indicate that about 89% of deaths due to lung cancer within this group could be attributed to daily smoking. This would mean that if smoking did not exist, or we could remove smoking as an exposure, the mortality rate attributable to lung cancer would hypothetically decrease by 89.3%. However, the remaining 10.7% of deaths from lung cancer in this group would still occur.