Skip to content

    What does a rate per 100 person-years mean?

    What does a rate per 100 person-years mean?

    A study reports three events per 100 person-years. Is that the same as saying three out of every 100 people had the event? Not necessarily. The phrase combines the number of people with the amount of time each person was observed.

    Person-years help researchers describe event rates when participants are followed for different lengths of time. The idea is straightforward once you separate people from time—and it comes with limits worth keeping in mind.

    Add up the time people contribute

    One person followed for one year contributes one person-year. Two people each followed for six months together contribute one person-year. Ten people followed for one year each contribute ten person-years, assuming the relevant time is counted for each person.

    Now imagine a fictional study in which participants together contribute 100 person-years of observation and researchers record three qualifying events. The rate is three events per 100 person-years.

    The number doesn't tell you there were exactly 100 participants, or that everyone was followed for one year. The same total could come from different combinations of people and follow-up time. It also doesn't, by itself, predict what will happen to any one person.

    Why not just count the participants?

    In many studies, people enter at different dates or remain in follow-up for different lengths of time. Someone observed for two years contributes more observation time than someone observed for three months. A person-time measure accounts for those differences in the denominator.

    Exactly when the clock starts and stops depends on the study and outcome. For a first occurrence, researchers may stop counting a person's time at the event, the end of follow-up or another specified point. Read the study's definition; don't assume every rate uses the same rules.

    The event count matters too. Some analyses count a person's first event, while others may count repeated events. Those rates answer different questions even if they use the same phrase “per 100 person-years.”

    A rate is not automatically a one-year risk

    In a simple situation with stable rates and uncommon events, a rate and a one-year proportion can sometimes look numerically similar. But three events per 100 person-years is not automatically a 3% chance for each person in a year. The risk over a period depends on the study population, how rates change over time, competing events and how the analysis was conducted.

    Think of the rate as a measure of events relative to observed time. A risk or cumulative incidence answers a related question about the proportion experiencing an event over a specified period. Ask which measure the report provides before translating it into a personal probability.

    If the headline leaves out the follow-up details, look for the methods or a table in the study report. A clinician can help explain how the result relates to your own health history.

    Compare rates only after checking the setup

    Suppose another fictional group has six events per 100 person-years. That rate is twice three per 100 person-years, but the comparison is useful only if the events, time counting and populations are sufficiently comparable.

    Were participants similar at the start? Was one group older or living with different conditions? Were they followed for the same purposes and were events identified the same way? An observed rate difference doesn't automatically establish that a treatment caused it, particularly in an observational study.

    Researchers may adjust for measured differences, but adjustment cannot always account for everything. Look at the study design, the estimate and its confidence interval as well as the two rates.

    Small event counts can be uncertain

    Three events and 100 person-years produce a neat-looking rate, but three events provide limited information about uncommon outcomes. The confidence interval may be wide. A report that gives only a rounded rate can hide how much uncertainty remains.

    Ask how many events occurred, how much follow-up time was available and whether participants were lost to follow-up. Also check whether an event was defined and measured consistently. More time observed can improve information, but it doesn't fix every problem with a study.

    For a real GLP-1 example, an analysis of 126 trials found 6.4 cases of a rare optic-nerve condition called NAION per 100,000 person-years in people taking GLP-1 medicines, compared with 3.8 per 100,000 person-years in those receiving a placebo or another type of glucose-lowering medicine. The rates summarize observed time; they don't predict your individual risk. Sudden vision loss needs urgent assessment. Our guide to eye changes during GLP-1 treatment explains when not to wait.

    Questions for the next headline

    When you encounter a person-year rate, try asking:

    • What event was counted, and was it the first event or any event?
    • How many events and how much total follow-up produced the rate?
    • When did observation start and stop?
    • How precise was the estimate?
    • Are the groups and their follow-up comparable?

    If you're discussing a treatment, ask your clinician how the result fits with other evidence, benefits and potential harms. Don't convert a rate into your own predicted risk without the context needed to do so.

    Person-years are a way to keep the amount of observation in view. Once you know what was counted and for how long, the number becomes far more informative than the headline alone.