This JAMA Guide to Statistics and Methods discusses analytical approaches to accounting for differences in treatment effect by study center when randomized trials enroll patients and administer interventions at multiple sites.
It is common for treatments to be evaluated in clinical trials that involve many sites or centers, primarily because one center rarely can enroll sufficient numbers of patients to complete the trial.1 The use of multiple clinical sites introduces complexity because outcomes at different sites may be systematically different, eg, due to differences in patient populations, ancillary treatment practices, or other factors. Thus, appropriate statistical analyses of multicenter clinical trials consider these center effects to yield a better understanding of the overall mean treatment effect and the variability in treatment effects and patient outcomes among sites.1
In an article in JAMA, Dodick et al2 published the results of a clinical trial that compared migraine prevention by 2 different dosing regimens of fremanezumab vs placebo. The number of migraine days were recorded during a 28-day baseline period and a 3-month treatment period. The primary outcome for the study was the change from baseline in the mean number of monthly migraine days during treatment. The 875 participating patients were recruited from 123 centers in 9 countries. Using a primary analysis that accounted for each patient’s mean number of migraines during the baseline period,2,3 treatment, country (US vs non-US), and other factors, the authors reported a difference with monthly dosing vs placebo of −1.5 days (95% CI, −2.01 to −0.93 days; P < .001) and with single higher dosing vs placebo of −1.3 days (95% CI, −1.79 to −0.72 days; P < .001). They also conducted a post hoc sensitivity analysis that accounted for effects of the specific country of enrollment.2
ESTIMATING TREATMENT EFFECTS IN MULTICENTER CLINICAL TRIALS
Why Are Differences Between Centers Considered When Estimating Treatment Effects?
The goals of the statistical analysis of a multicenter clinical trial include providing a valid estimate of the treatment effect (ie, the mean difference in outcomes between patients treated in the 2 groups) and understanding and quantifying the remaining uncertainty or precision in the estimated treatment effect.1 Patients treated at different centers may differ in their overall prognoses but experience the same relative benefit of a treatment compared with standard care. Alternatively, patients at different centers may differ in both their overall prognoses and in the treatment effect. Only the first case is considered in this article.
The randomization of patients to treatments in multicenter trials is usually stratified by center to achieve balance in the numbers of patients receiving each treatment within each center and, in what follows, it is assumed this has been done.4 Balance improves the statistical efficiency of the trial, increasing precision in the estimation of treatment effects given a particular sample size. It also reduces the risk in modestly sized trials that chance imbalance in treatment allocation at centers with smaller numbers of patients results in a bias if that center has better or worse outcomes on average than other centers....