Abstract
Randomized clinical trials (RCTs) on vertebroplasty are crucial for guiding the treatment of vertebral compression fractures, but their overlooked statistical fragility can undermine clinical reliability. Minor outcome changes may overturn significant findings, risking unreliable evidence, and impacting patient care. This study assessed the fragility of significant outcomes in vertebroplasty RCTs, hypothesizing high sensitivity to such changes. PubMed, Embase, and MEDLINE were searched for RCTs on vertebroplasty reporting dichotomous outcomes. The fragility index (FI) and reverse FI quantified the number of outcome reversals needed to change statistical significance for significant and nonsignificant results, respectively. The fragility quotient (FQ) was calculated as the FI divided by the study sample size. Subgroup analysis was conducted by outcome category. A total of 276 outcomes from RCTs were analyzed. The median FI was 5 (interquartile range [IQR]: 4-5), with a FQ of 0.053 (IQR: 0.019-0.088). Statistically significant outcomes (n = 36) had a median FI of 3 (IQR: 2-4) and FQ of 0.034 (IQR: 0.018-0.051), whereas nonsignificant outcomes (n = 240) showed a median FI of 5 (IQR: 4-5) and FQ of 0.062 (IQR: 0.021-0.088). Fracture-related outcomes were the most robust (FI: 5, FQ: 0.088), whereas cement leakage was the most fragile (FI: 3, FQ: 0.041). Pain outcomes had an FI of 5 (FQ: 0.062), and complications and vertebroplasty versus kyphoplasty outcomes were more robust (FI: 5, FQ: 0.013). Patients lost to follow-up exceeded the FI in 79% of outcomes. The statistical findings in vertebroplasty RCTs are fragile and warrant cautious interpretation. A small number of outcome reversals or consistent postoperative follow-up can shift the significance of the results. Standardized reporting of P values alongside FI and FQ metrics is recommended to help clinicians evaluate the robustness of study findings.