How this instrument works
Incidence rate counts only new cases of a condition that appear during a defined stretch of time, then divides by the total observation time contributed by everyone at risk — commonly expressed as person-years. Scaling the result to a round denominator, typically per 100,000 person-years, lets a small study and a large one report numbers that mean the same thing, rather than raw counts that only make sense next to their own population size.
The person-years denominator solves a specific comparison problem: a study following 1,000 people for one year and a study following 500 people for two years each contribute 1,000 person-years of observation, even though neither followed the same number of people for the same length of time. Incidence rate treats those two studies as equivalent exposure, which is exactly the point — the rate reflects risk per unit of time at risk, not an artifact of how the study happened to be sized or timed.
It is worth keeping incidence separate from prevalence, since the two get mixed up constantly. Incidence counts new cases arising over a period — it is a measure of how fast a condition is spreading. Prevalence counts every existing case, new and old, at one point in time — it is a snapshot of how much disease is currently present. A condition can have low incidence but high prevalence if it develops slowly and people live with it for years, or the reverse if it strikes fast but resolves or is fatal quickly.
- Enter New cases — only cases newly diagnosed during the study window, not existing ones carried over from before it.
- Enter Population at risk — the group eligible to develop the condition, excluding anyone who already has it if the condition cannot recur.
- Enter the Time period in years the population was followed.
- Read the rate per 100,000 person-years; the working block shows the division and the scaling step separately.
Worked example — 50 new cases, 10,000 at risk, 1 year
A clinic tracks 10,000 people at risk for one year and records 50 new diagnoses. That is 50 divided by 10,000 person-years, times 100,000, giving a rate of 500 per 100,000 person-years.
Now compare a second study that follows a different-sized group for a different length of time: 120 new cases among 20,000 people tracked for 2 years. The denominator is 20,000 × 2 = 40,000 person-years, so the rate is 120 ÷ 40,000 × 100,000 = 300 per 100,000 person-years — directly comparable to the first result, even though neither the population size nor the follow-up length matched. A third, smaller study — 10 new cases among 5,000 people over 1 year — comes to 10 ÷ 5,000 × 100,000 = 200 per 100,000 person-years, again on the same scale as the other two.
Questions
What is the difference between incidence and prevalence?
Incidence counts only new cases that develop during a defined follow-up period — it measures how fast a condition is appearing. Prevalence counts every existing case, old and new, at a single point in time — it measures how much disease is currently present in a population. A slow-developing, long-lasting condition can have low incidence but high prevalence; a fast-striking, quickly resolved one can show the reverse pattern.
Why does the formula multiply population by years instead of just dividing by population?
Because studies rarely follow identical numbers of people for identical lengths of time, and dividing by population alone would make a one-year study and a ten-year study look artificially comparable. Multiplying population by years produces person-years, a unit of total observation time, so two studies with different sizes and different durations can be measured against the same standard.
Why per 100,000 instead of a percentage?
Because most conditions this measure is used for are rare enough that a percentage would round to a string of zeros and lose all readability. Per 100,000 person-years keeps the numbers in a legible range and is the convention most public health agencies and journals already report in, which keeps figures comparable across sources.
Should someone who already has the condition count in the population at risk?
Only if the condition can recur or affect them again. For a one-time, non-recurring diagnosis, existing cases are usually excluded from the at-risk population, since they can no longer contribute a new case to the numerator. Leaving them in the denominator without excluding them from future counting understates the true rate among those genuinely still at risk.
Can incidence rate exceed the size of the population?
The scaled rate, per 100,000 person-years, absolutely can exceed 100,000 for a fast-moving condition with a short observation window — it is a rate projected onto a standard denominator, not a literal count bounded by the real population size. What cannot happen is more new cases than there were people at risk to develop them within the actual study.
References
Read this first: This instrument computes a screening figure from population formulas — it is not a diagnosis, and it cannot see the whole picture a clinician can. Use it to inform a conversation, not to replace one.