Skip to main content

Table 1 Examples of reactive revisions

From: Changes to aspects of ongoing randomised controlled trials with fixed designs

1. A reactive revision to the choice of primary endpoint occurred in the FIELD trial. The FIELD trial randomised 9795 patients with type 2 diabetes mellitus to fenofibrate (a drug that modifies the lipid profile) or placebo [3]. The original primary endpoint of coronary heart disease (CHD) death was changed to ‘any coronary event’ on the basis of a blinded (pooled) review of interim data that indicated a lower than expected CHD death rate that was inadequate for reaching the statistical power target.

2. A reactive revision to the analysis set composition arose in the CO17 randomised controlled trial of cetuximab in patients with pre-treated metastatic epidermal growth factor receptor (EGFR)-positive colorectal carcinoma in response to external evidence suggesting that K-ras wild-type was a predictive biomarker. The effect of cetuximab was indeed found to be largely restricted to this subset of randomised patients [4, 5].

3.The Sentinel Node Biopsy versus Axillary Clearance (SNAC1) trial changed the primary analysis from a comparison of proportions (of patients experiencing lymphoedema) to a comparison of means (of arm swelling) to take advantage of the greater statistical power of parametric analysis methods [6].