- Published
- Updated
No observed events in one arm does not establish that its event risk is exactly zero. Ratios based on small counts can be unstable. Present event counts, observation coverage and uncertainty together rather than interpreting a dramatic numerical output as conclusive evidence.
Define the question and assumptions
Check numerators, denominators, event definitions and follow-up duration first. Some ratio calculations become undefined or extreme with a zero cell. A statistician should choose suitable estimation and interval methods; do not accept a software default correction without review. Distinguish an observed absence of events from missing event information and consider each group opportunity for observation.
Hypothetical study example
In a hypothetical small trial, one arm has two events and the other none. The team does not present an infinite ratio as a certain risk difference. It reports counts, an appropriate uncertainty summary and limited follow-up without claiming that safety has been completely established in the event-free arm.
Design review checks
- Check whether zero reflects complete observation or missing event documentation.
- State the correction and interval method used for the reported estimate.
- Assess differences in follow-up duration and intensity of event collection.
Record the decision and its limits
Accepted interpretation explains the limitations of small counts. If additional methods are used, show their influence transparently. Zero events do not promise exclusion of an uncommon risk. Avoid decimal precision beyond the information available and discuss clinical importance, event details and observation time alongside the numerical summary.
Frequently asked questions
- Does zero events mean zero risk?
- No. Risk can remain uncertain, especially in small samples with short follow-up.
- Is an infinite ratio a definite conclusion?
- No. Explain computational instability and the selected estimation approach.