The two main ways by which the frequency of disease is measured are incidence and prevalence. These can be distinguished by differences in the time of disease onset.
- counts new cases of the disease (or outcome)
- counts new and existing cases of the disease (or outcome)
Incidence quantifies the development of disease. Incidence can be estimated using data from a disease registry data or a cohort trial. There is an implicit assumption of a period of time, such as new cases within a month (or a year).
A summary incidence rate can estimate the risk (e.g., probability of disease in an individual) if the risk is constant across the summarized groups.
As defined, incidence is a count of new cases. However, it is often expressed as a proportion of those at risk. The denominator includes all persons at risk for the disease or condition, i.e. disease-free or condition-free individuals in the population at the start of the time period. Persons in the denominator, those at-risk, should be able to appear in the numerator. Obviously, the denominator would not include persons who already have the disease or condition. Incidence can also be expressed in terms of person-time at risk.
Rates are usually expressed per 100, 1,000, or 100,000 persons. In a strict application, "rate" should only be used when the denominator is an estimate of the total person-time at risk. (You will find the term "rate" used inconsistently in epidemiologic reports. It is better to seek the source of the numbers than to rely on the nomenclature.)
Two Common Measures of Incidence
- Cumulative Incidence
- The cumulative incidence consists of the number of persons who newly experience the disease or studied outcome during a specified period of time divided by the total population at risk. This calculation assumes all persons in the denominator contribute an equal amount of time to the measure.
- Incidence Density Rate
- Incidence density rate (also known as incidence rate; person-time rate) is the number of persons who newly experience the outcome during a specified period of time divided by the sum of the time that each member of the population is at-risk.
Since prevalence counts both new and existing cases, the duration of the disease affects the prevalence. Diseases with a long duration will be more prevalent than those with a shorter duration. Chronic, non-fatal conditions are more prevalent than conditions with high mortality. The prevalence of disease is directly related to the duration of the disease. Prevalence is not an apt descriptor of an acute condition.
Similar to incidence, persons included in the denominator must have the potential for being in the numerator, i.e. at-risk for the disease or condition. Prevalence is often expressed after multiplication by 100 (%), 1000, or 100,000.
The prevalence pool is the subset of the population with the condition of interest. The prevalence pool is not generally useful for hypothesis-driven epidemiologic research because these are not new cases, but can be useful in tracking the natural history of the disease, evaluating effects of treatments, or disease burden.
For most etiologic research, incidence is the more appropriate measure. Studying the incidence of a rare condition, however, poses a challenge. Given a small number of new cases, it can be preferable to estimate prevalence instead of incidence in these situations. For example, birth defect rates reported as the number of cases/live births is a prevalent measure. Similarly, an autopsy rate is a prevalent measure.
Two common measures of prevalence
The difference is whether the estimate is made over a period of time or at one specific time as illustrated below:
- Point prevalence
- Prevalence of condition of interest at a specific time.
Number of existing cases on a specific date/ Number in the defined population on this date
Point prevalence ranges from 0 to 100. (%)
Point prevalence can be estimated from a cross-sectional survey or disease registry data by calculating the percentage with a particular disease or condition on a particular date.
- E.g. what percentage had a particular type of flu on 1/17/2009?
- Period prevalence
- Prevalence of outcome of interest during a specified period of time.
Less frequently used.
Number of cases that occurred in a specified period of time/ Number in the defined population during this period
Period prevalence generally ranges from 0 to 100 %. (Theoretically, period prevalence can exceed 100% if you allow individuals who had the disease more than once to be counted for each case of the disease within the reporting period.)
E.g. What percentage of the population had an episode of flu between October and May within the most recent flu season?