Many multicenter trials had few events per center, requiring analysis via random-effects models or GEEs

Many multicenter trials contain few events per center. Adjustment for center using random-effects models or GEE with model-based (non-robust) standard errors may be beneficial in these scenarios.