Skip to content
Omnikalc

Events needed calculator: how many events does a survival trial need?

Enter the hazard ratio, significance level, power, allocation ratio and expected event rate to get the number of events and the total number of participants.

Updated on
Your inputs

Value out of range

Value out of range

Value out of range

Results

Events needed

331

Total participants needed
828
Participants in group 1
414
Participants in group 2
414
Sum of the z values (α and power)
3.242

Survival trials compare the time until an event, such as death or relapse, between two groups. The events needed calculator takes the hazard ratio, the significance level, the power, the allocation ratio and the expected proportion with the event, and gives the number of events, the total participants and the participants in each group.

How the calculation works

  • Decide the hazard ratio you want to detect, the two-sided significance level and the power. The hazard ratio is the relative rate of the event in one group compared with the other.
  • In a log-rank comparison, the number of events needed is given by Schoenfeld’s formula: d = (z(1−α÷2) + z(power))² × (1 + r)² ÷ (r × (ln HR)²), where r is the allocation ratio.
  • The study’s power depends on the number of events, not on the number of people. To find the total number of participants, divide d by the expected proportion with the event by the end of follow-up.
  • Round up. The participants are split between the two groups according to the allocation ratio.

Example

A trial wants to detect a hazard ratio of 0.70 with a two-sided α of 0.05, 90 % power and equal groups. The z values add up to 1.960 + 1.282 = 3.242. Then d = 10.508 × 4 ÷ (0.357)² = 330.4, so 331 events are needed. If 40 % of participants are expected to have the event, 828 people are needed, with 414 in each group.

Why it matters

Planning a survival study starts with the number of events. This calculator applies Schoenfeld’s formula to show how the hazard ratio and the power change that number, and then turns it into a recruitment target.

Good to know

Schoenfeld’s formula assumes proportional hazards, which means a constant hazard ratio over time, and a log-rank test. Accrual, drop-out and non-proportional effects need more detailed planning in specialized software. The expected event proportion should come from the follow-up time you plan.

The sample size calculator covers other designs, and the median survival calculator summarizes survival from data.

Frequently asked questions

Why does a survival trial count events and not people?

In time-to-event analysis, a person who stays event-free during follow-up adds little information about the difference in rates. The information comes from the events, so the power of the log-rank test depends on how many events occur. Enrolling more people helps only by making events happen sooner.

What is a hazard ratio?

It compares the instantaneous rate of the event between two groups. A hazard ratio of 0.70 means that at any moment the group on the new treatment has 30 percent lower rate of the event than the other group. It is the effect size used in log-rank and Cox models.

What does the allocation ratio do?

It sets how many participants go to group 2 for each participant in group 1. Equal allocation needs the fewest events. Unequal allocation, such as 2 to 1, needs more events in total but may help recruitment or give more safety data on the new treatment.

How do I find the expected proportion with the event?

Estimate it from earlier studies of a similar population, using the survival at the planned follow-up time in each group, and average the two. This step is the most uncertain, so many protocols plan for a range, and may use an interim review to check the event rate.

Was this calculator helpful?

Thanks for your feedback!

Results are estimates provided for information purposes. They do not replace professional advice.