Drugmakers planning to assign clinical trial participants to a separate arm of the study based on baseline monitoring must report those plans to the European Medicines Agency, including which factors — called covariates — they plan to rely on to make those assignments.
Often, these covariates are measures taken early in a trial, such as a baseline measurement, that would prompt sponsors to stratify patients into different treatment arms, according to guidance released last week.
The guideline specifies not only that a small number of covariates should be included in the primary analysis but also that sponsors should justify which ones they include. That’s because separating patients into different treatment arms based on these covariates may become overwhelming if there are too many — especially if the trial in question is small, the EMA says.
Companies are free to choose whether to use the raw outcome number or the change from baseline as the primary endpoint. Regulators do, however, want to see the baseline measure listed as a covariate in the primary analysis, even if the company chooses to use the change from baseline. The EMA notes that when the baseline is included as a covariate in a standard linear model, the estimated treatment effects are identical for change from baseline and raw outcome.
The agency discourages including anything that might be affected by events that occur after randomization, such as duration of treatment, compliance levels or needs for rescue medications.
Multicenter trials will generally separate patients based on the trial site, the guideline says. If this is not the case, then sponsors should pick other criteria, such as country, region or similarities in co-medications or palliative care.
The guideline takes effect Sept. 1. Read it at www.fdanews.com/03-30-15-covariates.pdf. — Lena Freund