2. Measurements of incidence
2.2. Incidence proportion as risk
A useful way to think about incidence proportion is that it is the probability of developing disease over a stated period of time; as such, it is an estimate of risk. Since all of the persons with new cases of disease are also represented in the denominator, a risk is also a proportion.
Risk is a measure that is usually easily understood. When risk is misunderstood, it is often because the ‘specified time’ over which risk is measured is not expressed.
Examples of risk include:
- In a town of 100 people, 20 of them develop influenza during the influenza season. The risk is 20 out of 100 or 20% to develop influenza during that season.
- In a population of 10,000 males at 60 years of age, 900 of them will develop heart disease during the next 10 years. In this population, the risk is 9% to develop heart disease during the 10 year period.
This diagram shows how the calculation of risk depends on the time period over which it is measured. The orange crosses indicate the occurrence of disease in five people who are being followed. The green lines indicate different times at which the follow-up might stop. If the follow-up time ended at the first green line the risk would be zero, because no cases occurred before that time. If the trial continued to the second green line, the risk would be 20% (one case in five people), and it would be 100% if the follow-up continued to the third green line.
Interpreting a risk
It is difficult, if not impossible, to interpret a statement such as this (taken from a newspaper article):
‘Men at age 60 have a 7% risk of a heart attack.’
The problem with this particular statement is that there is no time period mentioned. A risk of 7% would be very high if it occurred during one year, but it would be low if it referred to a 30-year period.
It is common to see risk measures reported without a time period, but such measures cannot be interpreted. Sometimes the time measure is implied, but vague. For example, one might see a statement such as: ‘The risk of dying from open-heart surgery is 2%.’ But does this refer to the risk during the first day following surgery, the first week, or longer? The value of the risk would be different for each of those time intervals.
Therefore, the incidence proportion (also called cumulative incidence) must specify a time period as an integral part of the result reported.