Statistical Fragility of Findings From Randomized Phase 3 Trials in Pediatric Oncology
Author Information
Author(s): Hannah E. Olsen, PeiāChi Kao, Caleb Richmond, David S. Shulman, Wendy B. London, Steven G. DuBois
Primary Institution: Boston Children's Hospital and Harvard Medical School
Hypothesis
The fragility index (FI) has not been studied in phase 3 trials in pediatric oncology.
Conclusion
Pediatric oncology trials are similarly or less statistically fragile than adult oncology trials, and statistical fragility has not changed over the last four decades.
Supporting Evidence
- The median FI for trials with a binary outcome was 4.5, indicating a small number of patients can change statistical conclusions.
- Statistical fragility appears to have remained constant over the four decades evaluated.
- Despite sample size limitations, pediatric cancer trials are not more statistically fragile than adult oncology trials.
Takeaway
In pediatric cancer trials, only a small number of patients can change the results of a study, showing that these trials can be fragile.
Methodology
The study analyzed phase 3 pediatric oncology trials published between 1980 and 2020, calculating fragility indices for trials with binary and time-to-event outcomes.
Potential Biases
Potential bias due to the limited number of analyzable trials from earlier decades and the reliance on reported data.
Limitations
The study's sample size was limited by the availability of sufficient data from trials, and many trials did not report critical information for analysis.
Participant Demographics
Trials included pediatric and adolescent populations with various cancer diagnoses.
Statistical Information
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website