Incidence in epidemiology is the rate of new cases of a disease or condition occurring in a specific population over a defined period of time. The key word is “new.” Incidence only counts people who freshly develop a condition, not everyone currently living with it. This distinction makes incidence one of the most important tools for understanding how quickly a disease is spreading or how risky a particular exposure might be.
Why “New Cases” Matters
Incidence exclusively tracks the moment someone goes from healthy to sick. If 500 people in a city are diagnosed with diabetes for the first time during 2024, those 500 are the incidence. The thousands of people already managing diabetes from previous years don’t count. This focus on new occurrences makes incidence especially useful for spotting outbreaks, evaluating whether a prevention program is working, or comparing risk across different groups.
Prevalence, by contrast, captures everyone who currently has a condition, whether they were diagnosed last week or ten years ago. The two measures are mathematically linked: prevalence roughly equals incidence multiplied by the average duration of the disease. A condition with low incidence but long duration (like well-managed HIV) can have high prevalence. A condition with high incidence but short duration (like the common cold) tends to have lower prevalence at any given moment.
Defining the Population at Risk
Calculating incidence requires a clearly defined denominator: the population actually capable of developing the disease. This sounds obvious, but errors here are common. If you’re calculating the incidence of uterine cancer, for example, only females belong in the denominator. Including the entire population would artificially deflate the rate and give a misleading picture of risk.
People who already have the disease are also excluded from the at-risk population. Someone who was diagnosed with Type 2 diabetes last year can’t be a “new case” this year. Getting the denominator right is just as important as accurately counting the new cases in the numerator.
Two Ways to Measure Incidence
Epidemiologists use two main approaches, depending on the type of study and the data available.
Cumulative Incidence (Incidence Proportion)
This is the simpler version. You take the number of new cases during a time period and divide by the total population at risk at the start of that period. The result is a proportion, often expressed as a percentage or as cases per 1,000 or per 100,000 people. For example, if 125 patients undergo colon cancer surgery and 34 die within five years, the five-year cumulative incidence of death is 34 out of 125, or about 27 per 100 individuals.
Cumulative incidence works best in closed populations where everyone is followed for the same length of time and nobody drops out. In practice, this is hard to achieve, which is where the second method becomes useful.
Incidence Rate (Person-Time Rate)
In longer studies, people inevitably drop out, die from unrelated causes, or join at different times. The incidence rate handles this by building time directly into the denominator. Instead of dividing new cases by the number of people, you divide new cases by the total person-time of observation.
Person-time works like this: one person followed for five years without getting sick contributes five person-years. Another person followed for two years before developing the disease contributes two person-years. You add up all the individual observation times to get total person-years, then divide the number of new cases by that sum. The result is typically reported as cases per 100,000 person-years. For instance, the incidence rate of prostate cancer in the United States from 2015 to 2019 was 112.7 cases per 100,000 men per year.
This approach gives a more accurate picture when people contribute unequal amounts of follow-up time, which is the norm in real-world research.
How Incidence Numbers Are Reported
Raw incidence figures are usually tiny fractions, so they’re multiplied by a standard number to make them easier to read and compare. The most common multipliers are per 1,000, per 10,000, or per 100,000 population. The choice depends on how rare the disease is. For a relatively common condition, per 1,000 might work. For cancer or HIV, per 100,000 is standard.
The global incidence of HIV in 2021, for instance, was about 19.0 cases per 100,000 population, calculated from roughly 1.5 million new infections worldwide that year. In the United States in 2022, over 1.85 million new cancer cases were reported across all types. Presenting that figure as a rate per 100,000 allows meaningful comparisons between countries, age groups, or time periods regardless of population size.
Open Versus Closed Populations
The type of population being studied affects how incidence is calculated. A closed (or fixed) population has a defined membership that doesn’t change. Think of a clinical trial: everyone enrolls at the start, and no new participants join midway through. Cumulative incidence is straightforward here.
An open (or dynamic) population has people constantly entering and leaving. The residents of a city in a given year aren’t a fixed group. People move in, move out, are born, and die throughout the year. The population is really a flow of people, and the average number present during the year, multiplied by the time period, gives the person-years. Incidence rates using person-time are the standard tool for these dynamic populations. An occupational cohort offers a useful example: it can look fixed if you define it by employment start date, but it becomes dynamic when viewed across calendar time, with workers joining and leaving at different points.
What Incidence Tells You That Prevalence Cannot
Incidence answers the question “how fast is this disease occurring?” while prevalence answers “how much of this disease exists right now?” Both are valuable, but they serve different purposes. If you want to know whether a new environmental exposure is causing illness, incidence is the measure that reveals it, because it isolates the transition from healthy to sick. Prevalence can’t separate new cases from old ones, so a rising prevalence could mean more people are getting sick or simply that existing patients are surviving longer.
This relationship, where prevalence equals incidence multiplied by disease duration, also explains why curing a disease faster reduces prevalence even if incidence stays the same, and why a disease with a stable incidence can appear to grow more common as treatments improve and patients live longer. Understanding this interplay is one of the core skills in reading health statistics accurately.