Between-centre differences and treatment effects in randomized controlled trials: a case study in traumatic brain...

Area of Science:

• Clinical Trials Methodology
• Traumatic Brain Injury Research
• Statistical Analysis in Medicine

Background:

• Randomized controlled trials (RCTs) in Traumatic Brain Injury (TBI) exhibit significant between-centre outcome variations.
• Clinicians often believe these variations impact treatment effect estimations in TBI RCTs.

Purpose of the Study:

• To assess the influence of between-centre outcome differences on the estimated treatment effect in a large TBI RCT.
• To evaluate the impact of centre-specific variations on treatment efficacy in TBI.

Main Methods:

• Utilized data from the MRC CRASH trial on corticosteroid infusion efficacy in TBI patients.
• Employed fixed and random effects logistic regression models to analyze 14-day mortality and treatment effects.
• Quantified between-centre outcome and treatment effect variations using odds ratios (OR) and 95% ranges.

Main Results:

• Analyzed 9978 patients across 237 centres; 14-day mortality was 19.5%.
• Corticosteroid treatment showed increased mortality (OR = 1.22, p = 0.00010).
• Random effects models revealed substantial between-centre outcome variations (95% range: 0.27-3.71) but minimal impact on the overall treatment effect (OR = 1.24, p = 0.00003).
• Limited, statistically significant between-centre variation in treatment effect was observed (95% OR range: 1.17-1.26).

Conclusions:

• Substantial between-centre outcome differences in TBI RCTs do not necessarily confound the overall estimated treatment effect.
• Findings challenge the prevailing clinical belief regarding the impact of centre variations on treatment effect estimation in TBI research.