(0414) The Fragility Index: How Robust Are Randomized Controlled Trials Supporting Guidelines for Thyroid Cancer

Introduction: Guidelines for thyroid cancer are informed by the best available evidence which are classically randomized controlled trials (RCTs), but these trials may not be as robust as they claim. The Fragility Index (FI) helps determine the robustness of RCTs by determining the minimum number of patients needed to switch from an event to nonevent outcome to render results nonsignificant. This study reports the fragility index of RCTs informing thyroid cancer guidelines.

Methods: Three reviewers independently screened RCTs cited by the American Thyroid Association (ATA) for thyroid cancer guidelines. Articles included were those reporting a statistically significant primary dichotomous outcome. The FI, fragility quotient (FQ), and the FI minus number lost to follow-up (FI-LTF) were assessed. The FQ is defined as the FI divided by the sample size.

Results: The ATA has published 4 guidelines regarding thyroid cancer. From these, there were a total of 55 RCTs that were identified. Upon full-text review, 12 articles were identified as those with a statistically significant primary dichotomous outcome. Overall, among all studies, the median (IQR) FI was 8 (3,51) with a range of 2-83. Three trials (25%) had an FI ≤ 3, indicating very low robustness. Median FQ was 7% (2.4%, 15.1%) with a range of 1.4%-19.8%. Median FI-LTF was 16 (2.25, 36) with a range of -15 to 65. Three trials had an FI < number of patients lost to follow-up.

Conclusions: The RCTs supporting thyroid cancer guidelines have low robustness and are generally fragile with a median FI of 8. This is further supported by an FQ of 7%, a metric that accounts for sample size. Additionally, a quarter of the trials had more patients lost to follow-up than their FI, suggesting different results could have occurred if the LTF patients were included. Overall, these findings emphasize the need for caution in the application of these trials and further highlight the need for more robust RCT designs in future studies to strengthen the clinical guidelines for the management of thyroid cancer.