AUTHOR=Li Huiyi , Liang Zhenyu , Meng Qiong , Huang Xin TITLE=The Fragility Index of Randomized Controlled Trials for Preterm Neonates JOURNAL=Frontiers in Pediatrics VOLUME=10 YEAR=2022 URL=https://www.frontiersin.org/journals/pediatrics/articles/10.3389/fped.2022.876366 DOI=10.3389/fped.2022.876366 ISSN=2296-2360 ABSTRACT=Background

As a metric to determine the robustness of trial results, the fragility index (FI) is the number indicating how many patients would be required to reverse the significant results. This study aimed to calculate the FI in randomized controlled trials (RCTs) involving premature.

Methods

Trials were included if they had a 1:1 study design, reported statistically significant dichotomous outcomes, and had an explicitly stated sample size or power calculation. The FI was calculated for binary outcomes using Fisher’s exact test, and the FIs of subgroups were compared. Spearman’s correlation was applied to determine correlations between the FI and study characteristics.

Results

Finally, 66 RCTs were included in the analyses. The median FI for these trials was 3.00 (interquartile range [IQR]: 1.00–5.00), with a median fragility quotient of 0.014 (IQR: 0.008–0.028). FI was ≤ 3 in 42 of these 66 RCTs (63.6%), and in 42.4% (28/66) of the studies, the number of patients lost to follow-up was greater than that of the FI. Significant differences were found in the FI among journals (p = 0.011). We observed that FI was associated with the sample size, total number of events, and reported p-values (rs = 0.437, 0.495, and −0.857, respectively; all p < 0.001).

Conclusion

For RCTs in the premature population, a median of only three events was needed to change from a “non-event” to “event” to render a significant result non-significant, indicating that the significance may hinge on a small number of events.