Robustness of Randomized Control Trials Supporting Current Neurosurgery Guidelines

Abstract

BACKGROUND: Treatment guidelines in neurosurgery are often based on evidence obtained from randomized controlled trials (RCTs). OBJECTIVE: To evaluate the robustness of RCTs supporting current central nervous tumor and cerebrovascular disease guidelines by calculating their fragility index (FI)—the minimum number of patients needed to switch from an event to nonevent outcome to change significant trial primary outcome. METHODS: We analyzed RCTs referenced in the Congress of Neurological Surgeons and American Association of Neurological Surgeons guidelines on central nervous tumor and cerebrovascular disease management. Trial characteristics, finding of a statistically significant difference in the primary endpoint favoring the experimental intervention, the FI, and FI minus number lost to follow-up were assessed. RESULTS: Of 312 RCTs identified, 158 (50.6%) were published from 2000 to 2010 and 106 (34%) after 2010. Sixty-three trials (19.2%) were categorized as surgical trials, and the rest studied medical treatment (82.0%) or percutaneous intervention (8.33%). The trials had a median power of 80.0% (IQR 80.0-90.0). Of these, 120 trials were eligible for FI calculation. The median FI was 7.0 (IQR 2.0-16.25). Forty-four (36.6%) trials had FI ≤ 3 indicating very low robustness. After adjusting for covariates, recently published trials and trials studying percutaneous interventions were associated with significantly higher FI compared with older trials and trials comparing surgical approaches, respectively. Trials limited to single centers were associated with significantly lower FI. CONCLUSION: Trials supporting current guidelines on neuro-oncological and neurovascular surgical interventions have low robustness. While the robustness of trials has improved over time, future guidelines must take into consideration this metric in their recommendations.