BACKGROUND: Strict clinical remission endpoints in ulcerative colitis (UC) trials produce low remission rates and do not reflect the good outcomes of UC therapy. We proposed the use of the VWF (Voting With their Feet) endpoint, the percentage of subjects leaving a randomized controlled trial (RCT) arm for lack of efficacy). The aims were 1) to determine if the VWF endpoint can be extracted from 5-aminosalicylate (5-ASA) RCTs in UC; 2) to perform meta-analyses of VWF and clinical remission (CR) endpoints; and 3) to determine the statistical power of the VWF endpoint. METHODS: Fixed effects meta-analysis and power calculations were used. RESULTS: In 5 studies, including 1048 subjects, 9.5% of patients left 5-ASA study arms for lack of efficacy, versus 28.3% leaving placebo. The rate of failure to achieve CR was 68.2% with 5-ASA, versus 86.9% with placebo. The relative risk (RR) of treatment failure for 5-ASA using the VWF endpoint was 0.33 (95% confidence interval [CI] 0.24-0.44), which was significantly smaller than with the CR endpoint (RR 0.81, 95% CI 0.76-0.88). The statistical power of VWF was slightly greater than CR. CONCLUSIONS: VWF is inexpensive, intuitive, and has similar statistical power to CR. The VWF endpoint can confirm the validity of outcome measures in clinical trials, and estimate real-world clinical efficacy.
BACKGROUND: Strict clinical remission endpoints in ulcerative colitis (UC) trials produce low remission rates and do not reflect the good outcomes of UC therapy. We proposed the use of the VWF (Voting With their Feet) endpoint, the percentage of subjects leaving a randomized controlled trial (RCT) arm for lack of efficacy). The aims were 1) to determine if the VWF endpoint can be extracted from 5-aminosalicylate (5-ASA) RCTs in UC; 2) to perform meta-analyses of VWF and clinical remission (CR) endpoints; and 3) to determine the statistical power of the VWF endpoint. METHODS: Fixed effects meta-analysis and power calculations were used. RESULTS: In 5 studies, including 1048 subjects, 9.5% of patients left 5-ASA study arms for lack of efficacy, versus 28.3% leaving placebo. The rate of failure to achieve CR was 68.2% with 5-ASA, versus 86.9% with placebo. The relative risk (RR) of treatment failure for 5-ASA using the VWF endpoint was 0.33 (95% confidence interval [CI] 0.24-0.44), which was significantly smaller than with the CR endpoint (RR 0.81, 95% CI 0.76-0.88). The statistical power of VWF was slightly greater than CR. CONCLUSIONS: VWF is inexpensive, intuitive, and has similar statistical power to CR. The VWF endpoint can confirm the validity of outcome measures in clinical trials, and estimate real-world clinical efficacy.