Incidence Rate Calculator

What incidence rate counts

The incidence rate measures how often new cases of a disease or injury occur in a population, expressed explicitly per unit of time. The CDC's glossary defines it as the number of new cases over a specified period divided either by the average population or by the cumulative person-time the population spent at risk (CDC, Principles of Epidemiology, 3rd edition, 2012). Two words in that sentence do the heavy lifting: new and time.

Only cases that begin during the window count. Someone who has lived with a condition for years is a prevalent case, not an incident one, and mixing the two groups into one numerator is the fastest way to wreck the number. The time unit sits in the denominator on purpose: 20 cases in a town of 8,000 means one thing per week and another thing per year, so a rate is never complete until it says what base and what span it uses.

The reason the definition insists on a time frame is that rates without one cannot be compared. A county with 40 new cases in a year and a county with 40 new cases in a decade are the same count and tell wildly different stories, and only the denominator with its time unit tells them apart.

Person-time and population denominators

The CDC teaches the rate with two denominator styles, and both appear below (CDC, Principles of Epidemiology, Lesson 3, 2012).

A person-year is one person observed for one year. Ten people followed for one year each contribute 10 person-years, and so does one person followed for 10 years. That flexibility is the point: people join cohorts late, drop out early, or die, and person-time credits each of them for exactly the disease-free time they genuinely contributed.

Worked example: a cohort follow-up study

The CDC course works this example: 16 new cases of disease arise during 6,400 person-years of follow-up. The person-time rate is 16 / 6,400 = 0.0025 cases per person-year, which is 2.5 cases per 1,000 person-years or 250 cases per 100,000 person-years (CDC, 2012).

The same lesson explains how to treat participants who leave early. A person lost to follow-up halfway through a year of observation is conventionally credited with half a year at risk, on the assumption that losses happen on average mid-year. So a participant followed for one full year and then lost half way through year two contributes 1.5 person-years to the denominator. It is an assumption, and the CDC notes the trade-off openly: person-time methods assume the chance of disease stays constant over the observed time, so 10 people followed for one year and one person followed for 10 years are treated as equal even when that is not biologically true.

One more piece of bookkeeping catches people: the numerator counts cases, while the denominator counts time at risk. A participant who develops the disease stops contributing disease-free time on the day of onset. So if the 16 cases above arose across the follow-up window, the denominator already reflects the shrinking pool of people still at risk, and the two numbers stay in step.

Worked example: an annual surveillance rate

Health departments usually lack person-time data, so annual reports use the mid-period population instead. The CDC's textbook example: the United States recorded 44,232 new AIDS cases in 2003, and the estimated mid-year population was 290,809,777. The rate is (44,232 / 290,809,777) x 100,000 = 15.21 new AIDS cases per 100,000 population for 2003 (CDC, 2012).

The same machinery scales down to a county. As a worked illustration, suppose a county of 412,000 residents reports 38 new tuberculosis cases in a year: 38 / 412,000 x 100,000 = 9.22 cases per 100,000 residents per year. That per-100,000 convention is what makes a county of 400,000 and a state of 30 million comparable on one chart, which is exactly why surveillance programs standardize on it (CDC, Field Epidemiology Manual, 2024).

Picking the multiplier

A raw rate per person-year is usually a small decimal, so epidemiologists scale the denominator to keep the number readable. The multiplier changes only the display, never the underlying rate.

Move the decimal one place per power of ten: multiply the raw rate by 1,000 for a per-1,000 figure and by 100,000 for a per-100,000 figure. State the base next to every rate you publish, because 2.5 per 1,000 and 2.5 per 100,000 differ by a factor of 100.

Health departments print the base into the wording, not just the number, so a reader cannot miss it: 9.22 cases per 100,000 population per year, or 250 cases per 100,000 person-years of follow-up. When you compare two figures, check the base first and the number second.

Incidence rate, risk and prevalence

Three measures get confused constantly, and each answers a different question.

The CDC's course uses the 4-of-18 example for risk: 4 new cases among 18 people at risk is (4 / 18) x 100, or 22 cases per 100 population over that period (CDC, 2012). Risk has no time in the denominator, which is why a risk without a stated span means little. Prevalence is a snapshot; incidence is a flow. A short-lived illness can have high incidence and low prevalence at any moment, while a lifelong condition can do the opposite.

Mistakes that skew the number

• Mixing incident and prevalent cases. Counts from surveillance systems arrive as either new events or existing ones, and the CDC's field manual tells analysts to check which before doing anything else (CDC, Field Epidemiology Manual, 2024). The CDC's HIV surveillance glossary makes the same point, defining incidence as new cases over a specified period divided by the average population or by cumulative person-time at risk (CDC HIV Surveillance Terms, 2012). Prevalent cases in an incidence numerator inflate the rate.
• Using the total population when part of it cannot be at risk. If 18 of a town's 10,000 residents were under active follow-up, a rate against 10,000 understates the rate by more than 500-fold.
• Dropping the time unit. A figure written as 2.5 with no base and no span cannot be compared with anything.
• Counting the same person twice across a window. A person can become an incident case once for one condition; re-reporting the same diagnosis in the same year double counts.
• Reading a rate from a small count as precise. A county with 3 cases and a rate of 9.22 per 100,000 carries wide uncertainty; small numerators swing hard from year to year.

What the calculator asks for

Pick the denominator style first. The person-time mode wants new cases and total person-years (or person-months, converted to years) and reports the raw rate per person-year plus per-1,000, per-10,000 and per-100,000 person-year figures. The population mode wants new cases and the mid-period population and reports the raw rate per person per year plus per-10,000, per-100,000 and per-million figures, in the surveillance style health departments publish.

Enter counts as whole numbers and the denominator without commas; the results format themselves. Every computation runs in your browser, and nothing you type leaves the page.

Finally, keep the numerator and the denominator from different places. Using case counts from one registry with population estimates from another year quietly produces a rate that no source would defend.