Audit Person-Time Before Calculating an Incidence Rate

Separate people from time at risk

An incidence rate relates new events to accumulated time at risk. It is not simply the percentage of enrolled participants with an event. Different follow-up lengths make the distinction especially important: 100 people followed for a month do not contribute the same observation time as 100 people followed for a year.

CDC's archived Principles of Epidemiology lesson explains person-time denominators and follow-up contributions. Use the definition appropriate to your endpoint and record when observation starts and stops.

Build an auditable denominator

For a first-event analysis, identify each person's eligible entry date and the earliest applicable stop date, such as the first event, loss to follow-up, death, or study end. Document how the design handles delayed entry, temporary gaps, and changes in exposure. A recurrent-event endpoint needs a different explicit rule for time and events after the first occurrence.

Use the person-time audit worksheet to retain the unit, eligible time window, counting rule, and source. Check for negative durations, overlapping duplicate records, events before entry, and person-time after a stopping event. Reconcile summed time against the participant-level dataset rather than inferring it from enrollment alone.

Calculate a fictional rate

Suppose a study records 12 qualifying first events during 600 person-years at risk. The rate is 12/600 = 0.02 events per person-year, or 20 events per 1,000 person-years. These invented figures demonstrate the calculation; they do not describe a real disease population.

The person-time calculator allows a person-years or person-months label. It does not convert an entered denominator automatically. If the same exposure is represented as 7,200 person-months, the rate is about 1.667 events per 1,000 person-months. Keep the unit visible when copying the output.

Review what the rate can support

A crude rate is a summary over the observed follow-up. It does not by itself describe how hazard changes over time, or give an individual's probability of an event within a specified period. Different populations and follow-up patterns can produce different crude rates even without a simple causal explanation.

If comparing groups, check that event ascertainment, time origins, eligibility, and counting rules align. Consider whether age, exposure changes, clustering, competing events, or other design features require a model. Save the denominator audit with the analysis so a later revision can reproduce the same rate without guessing how time was accumulated.

Source checked on 30 October 2026; numerical example independently recalculated.