Fragility Index Calculator
Enter event and non-event counts for a 2x2 trial table. This calculator uses a two-sided Fisher exact test and reports a classic fragility index plus fragility quotient.
What Is the Fragility Index?
The fragility index (FI) is a simple way to judge how stable a statistically significant result is in a randomized trial with a binary outcome (event vs non-event). It answers this question:
How many patient outcomes would need to flip for the result to lose statistical significance?
If the answer is very small, the finding may be statistically fragile even when the p-value is below 0.05.
How This Calculator Works
Inputs
- Treatment events
- Treatment non-events
- Control events
- Control non-events
- Alpha level (default 0.05)
Statistics Used
This tool computes a two-sided Fisher exact p-value for the 2x2 table you enter. If the result is significant, it then computes a classic fragility index by adding events to the group with fewer events (assuming the event is an adverse outcome) until significance is lost.
It also reports the fragility quotient (FQ), defined as:
FQ = Fragility Index / Total Sample Size
How to Interpret the Result
- FI = 0: The study is already non-significant at your alpha threshold.
- Small FI (e.g., 1–3): Very few outcome flips can change the conclusion.
- Larger FI: The finding tends to be more numerically stable.
- FQ: Helps compare fragility across studies of different sample sizes.
Worked Example
Suppose a trial reports:
- Treatment: 1 event, 49 non-events
- Control: 9 events, 41 non-events
If this table is significant, the calculator asks how many additional events in the lower-event group would remove significance. If only 1 or 2 flips are needed, that result is considered fragile.
Important Limitations
1) FI is not a replacement for clinical judgment
A low fragility index does not automatically invalidate a treatment effect. It is one lens among many.
2) Depends on model choice and alpha threshold
Using Fisher exact test vs other tests, or alpha 0.05 vs 0.01, can change the FI.
3) Best for binary outcomes
The classic FI framework is designed for two-group, binary-outcome trial tables and may not transfer directly to time-to-event or continuous outcomes.
Best Practices for Researchers and Readers
- Report FI together with p-value and confidence intervals.
- Also report loss to follow-up, since missing outcomes can exceed FI.
- Use FI as a complement to effect size, not a replacement.
- Consider fragility quotient when comparing studies with different sample sizes.
Use this calculator as a fast, transparent way to explore result stability in clinical trial summary tables.