Figure 1.

Coprimary composite outcomes and main secondary outcomes from the FHN Daily Trial. Kaplan–Meier curves are shown for the composite outcomes of death or change in LV mass (A) and death or change in the PHC score from the RAND 36-item health survey (B). For each value for the coprimary composite outcome on the horizontal axis, the Kaplan–Meier curve indicates the proportion of patients in the respective treatment groups with an equal or more favorable outcome. The horizontal distance between the Kaplan–Meier curves at the 50% value on the vertical axes indicates the median composite outcome results. Median outcomes for the composite outcome of death or change in LV mass correspond to a reduction in LV mass of 12.3 g in the frequent hemodialysis group, as compared with a reduction of 2.2 g in the conventional dialysis group (difference in medians, 10.1 g). The greater separation in the two curves on the right side of the graph of the change in LV mass is because nine patients had reductions in LV mass of at least 60 g; all of them were in the frequent hemodialysis group. The median results for the composite outcome of death or change in PHC score correspond to an increase in the PHC score of two points in the frequent hemodialysis group as compared with no change in the conventional dialysis group (difference in medians, two points). Changes in LV mass ranged from a decrease of 51.2 g to an increase of 68.8 g in the conventional dialysis group and from a decrease of 174.5 g to an increase of 61.9 g in the frequent hemodialysis group. Changes in the PHC score ranged from a decrease of 27 points to an increase of 22 points in conventional dialysis group, and from a decrease of 28 points to an increase of 29 points in the frequent hemodialysis group. The standardized effect sizes for the main secondary outcomes (C) were calculated as follows: the mean differences in LV mass, PHC score (in which higher scores indicate better physical health), beck depression inventory score (in which higher scores indicate more severe depression), albumin concentration before dialysis, phosphorus concentration before dialysis, and SBP before dialysis were divided by the baseline SD; the mean difference in log dose of ESA was divided by the SD of the log baseline ESA dose; the log risk ratio for failure to complete the trail making test part B was divided by square root ([1−p]/p), where p is the fraction of participants who did not complete the test within 5 minutes at baseline; the log hazard ratio for hospitalization unrelated to vascular access or death was divided by square root (1/p), where p is the fraction of patients with a hospitalization unrelated to vascular access or death. ESA doses of <5000 erythropoietin equivalent units were set to 5000 before log transformation. Reprinted from ref. 15, with permission. CI, confidence interval; ESA, erythropoiesis-stimulating agent; FHN, Frequent Hemodialysis Network; LV, left ventricular; PHC, physical health composite; SBP, systolic BP.