Which factors are needed to compute sample size for a trial with time-to-event outcomes?

Which factors are needed to compute sample size for a trial with time-to-event outcomes?

• A. Expected hazard ratio, baseline survival, accrual time, follow-up time, alpha, power, and anticipated censoring
• B. Only sample size and effect size
• C. Number of events, measurement error, and dropout rate
• D. Only alpha and power

Answer: (A)

In time-to-event studies, the number of observed events, not just the number of participants, largely drives statistical power. To plan the sample size, you need the expected treatment effect on hazard (the hazard ratio) and how common the event is at baseline (baseline survival or hazard). You also must specify how long you will recruit participants (accrual time) and how long each participant is followed after accrual ends (total follow-up). These timing factors determine when events occur and how many you’ll observe by the study’s end. Alpha and desired power set the statistical requirements, and anticipated censoring accounts for people who drop out or are not followed long enough to have an event, which reduces the number of observed events. All of these inputs together define the required sample size because they shape the expected event count. Other options miss essential pieces or rely on elements not central to this calculation, such as focusing only on overall sample size and effect size, including measurement error, or using only alpha and power.

In time-to-event studies, the number of observed events, not just the number of participants, largely drives statistical power. To plan the sample size, you need the expected treatment effect on hazard (the hazard ratio) and how common the event is at baseline (baseline survival or hazard). You also must specify how long you will recruit participants (accrual time) and how long each participant is followed after accrual ends (total follow-up). These timing factors determine when events occur and how many you’ll observe by the study’s end. Alpha and desired power set the statistical requirements, and anticipated censoring accounts for people who drop out or are not followed long enough to have an event, which reduces the number of observed events. All of these inputs together define the required sample size because they shape the expected event count.

Other options miss essential pieces or rely on elements not central to this calculation, such as focusing only on overall sample size and effect size, including measurement error, or using only alpha and power.