The Fragility Index in Multicenter Randomized Controlled Critical Care Trials

Area of Science:

• Critical care medicine
• Clinical trial methodology
• Biostatistics

Background:

• Frequentist analysis and p-values may be insufficient for reporting clinical trial outcomes.
• The fragility index offers a measure of statistical significance sensitivity.

Purpose of the Study:

• To calculate the fragility index for critical care trials reporting statistically significant mortality effects.
• To assess the relationship between trial characteristics and fragility index.

Main Methods:

• Literature search of PubMed/MEDLINE for multicenter randomized controlled trials in critical care.
• Identified 56 eligible trials from 862 initially found.
• Calculated fragility index and analyzed correlations with sample size and p-value.

Main Results:

• Median fragility index was 2 (IQR, 1-3.5); over 40% of trials had an index ≤1.
• 12.5% of trials had loss to follow-up exceeding their fragility index.
• Larger sample size correlated positively, while p-value correlated negatively, with fragility index.

Conclusions:

• Critical care trial findings on mortality frequently hinge on a small number of events.
• Clinicians should exercise caution with trials exhibiting a low fragility index.
• Reporting fragility index in future critical care trials is recommended for enhanced interpretation and decision-making.