Run-in bias in randomised trials: the case of COPD medications.

Most randomised trials of newer and established medications in chronic obstructive pulmonary disease (COPD) enrol patients who have already been treated, often with drugs of the same class or even with the very same drugs that are to be studied. Some trials such as UPLIFT evaluated the effect of a drug as an add-on, with patients continuing their existing treatments [1]. Other trials, such as OPTIMAL and the Salford Lung Study, discontinued the regular maintenance treatment and replaced it with a randomly allocated active treatment [2, 3]. Finally, trials such as TORCH and SUMMIT had patients discontinue their regular maintenance treatment and directly receive the randomly allocated treatment, which included placebo [4, 5]. This last approach, however, has been shown to complicate the interpretation of results, particularly under the placebo group after the discontinuation of maintenance therapy [6, 7].