The measure of risk requires that all persons at risk at t0 are followed for the entire period t1 during which the risk is being measured (Figure).

Then the risk over the period t1 is defined as b/c. However, as risk relates to a cohort, a specific static population, it has an important drawback. Depending on the length of the follow-up period, a smaller or larger number of people will die from other causes than the one under study (competing risks). In addition, some will be lost to follow-up. When the numbers in these groups are high, the risk of the health outcome of interest will be underestimated.

To address this problem epidemiologists often use incidence rate (synonyms: incidence density, hazard). Just as in the case of risk, the numerator of the incidence rate is the number of new cases; the denominator, however, is the total of all time periods at risk of disease for the subjects followed (mostly expressed in person years). Incidence rate is the reciprocal of the average waiting time; when the incidence rate is 5 per person year, one has to wait 0,2 person year for an event to occur. In contrast to risks, incidence rates can exceed 1. Theoretically, they can become as great as infinity, depending on the number of cases and the units of time used. When renal registries present an RRT incidence rate of 147 per million population, they mean to say that in a million person years lived by the general population 147 persons started RRT. As general populations are dynamic, this million person years were not lived by exactly one million people, but by a slightly higher number of people, as some died during the year of study and were replaced by newborns. Incidence rates are sometimes said to express the ‘force of morbidity’.

The prevalence proportion, or prevalence, does not measure disease onset, but disease status. It is the proportion of people in a population that has disease at a particular point in time. Prevalence is affected both by incidence and by disease duration. Diseases with a short duration may have a low prevalence even if the incidence rate is high. The longer the duration of a disease once it occurs, the higher the prevalence. Prevalence is not as useful as incidence for studying the causes of disease. It is, however, extremely useful for measuring the disease burden on a population. The prevalence of a disease reflects the need for health care resources. Most national and regional renal registries predict the future prevalence of dialysis patients from the current prevalence, incidence rates and patient survival. Health authorities can then use these data for the informed planning of dialysis facilities.

For further reading
Rothman K. Epidemiology: an introduction. Oxford University Press, 2002.
Victora CG. What’s the denominator? The Lancet 1993;342:97-99.