Letters Section Editor: Robert M. Golub, MD, Senior Editor.
In Reply: In response to Dr Yuan and colleagues, the primary aim of this phase 1 study was to assess the effects of 4 active treatments relative to placebo on log of maximum ALT, AST, and total bilirubin via point estimates, confidence intervals, and 2-sided tests. Analyses of power and precision focusing on ALT were primary considerations in choosing the target sample size of 165 participants. Subsequently, 147 eligible participants were successfully recruited, randomized, and studied simultaneously. This was not a sequential study and there were no interim analyses. The power analysis assumed α = .05 and relied on an estimated SD of 0.65 (95% confidence interval, 0.54-0.79) on the log scale, based on the earlier study described in the second paragraph of our article. Estimated power curves were plotted as functions of fold difference for 7 choices of N; eg, for N = 165 and any true 1.4-fold difference, the power estimate was 0.80 (95% confidence interval, 0.62-0.92). We note that power calculations, including those used in the planning stage, provide no valid guidance in the interpretation of study results.1- 3 The confidence interval locations and widths we reported provided the appropriate guidance for interpretation of study results.
Watkins PB, Kaplowitz N, Colucci SV, Stewart PW, Harris SC. Acetaminophen and Aminotransferase Elevations—Reply. JAMA. 2006;296(23):2798-2799. doi:10.1001/jama.296.23.2798-b