Figure 2.

Median change in 24-h systolic blood pressure and median percentage change in plasma volume and red cell mass in substudy participating subjects in the placebo, dapagliflozin and hydrochlorothiazide groups. Because of the small sample size, data are reported as median (25th and 75th percentile) in order to exclude the influence of few extreme values that drive mean changes. In two patients plasma volume change could not be measured because an end of study measurement was not available. As a result, the sample size was too small to show 2.5–97.5% percentile. P, placebo; D, dapagliflozin; H, hydrochlorothiazide.