Sample Size, Power, Statistical Analyses, and Study Outcomes

Details of the study design have been previously published (5). The primary outcomes were plasma IL-6 and walking speed over 400 m at usual pace at 12 months. The primary comparisons of losartan and fish oil were made using contrasts at 12 months from a mixed model using all three strata with adjustment for the baseline level of the outcome, visit, strata, and clinical site. Inflammation markers were log-transformed prior to analysis. All participants with baseline and at least one follow-up visit were included in the analysis according to their randomized group, consistent with the intention to treat principle. We report here point estimates and 95% confidence intervals of main effects as well as p-values of two-sided tests. The target sample size of 300 was based on marginal comparisons (135 and 165/group) of each active intervention and placebo using one-sided tests at the 10% level. We did not expect definitive evidence; our goal was to exclude small effects with little clinical value. For IL-6, we had 91% power to detect an effect if the difference (on the natural log scale) was at least 0.1625; this is equivalent to the difference between 4.20 and 3.57 pg/mL (or a 15% effect). We had 66% power for a 10% effect and 99% power for a 20% effect. For 400 m walk speed, we had >99% power to detect a difference of 0.095 m/s (a substantial meaningful change), and 86% power for a difference of 0.038 m/s (a small meaningful change).