Figure 3. Comparison of the Primary Composite Outcome in the Trial involving Adolescents and Adults and the Trial involving Children.

Shown are the prespecified comparisons of the percentages of patients with a superior response. Each panel shows a comparison of a similar step-up in therapy for children and for adolescents and adults. Shown are responses at 14 weeks in adolescents and adults and in children who at baseline had poorly controlled asthma while receiving twice-daily treatment with a low-dose inhaled glucocorticoid (fluticasone propionate) (50 μg in children and 100 μg in adolescents and adults). In children, the step-up trial treatments included doubling the dose of fluticasone to 100 μg (FP100, the double-fluticasone group); doubling the dose of fluticasone to 100 μg and adding salmeterol at a dose of 50 μg (FP100/SM50, the salmeterol–double-fluticasone group); quintupling the dose of fluticasone to 250 μg (FP250, the quintuple-fluticasone group); or quintupling the dose of fluticasone to 250 μg and adding salmeterol at a dose of 50 μg (FP250/SM50, the salmeterol–quintuple-fluticasone group). In adolescents and adults, the step-up interventions included adding salmeterol to the baseline dose of fluticasone (FP100/SM50, the fluticasone–salmeterol group), increasing the dose of fluticasone by a factor of 2.5 (FP250, the 2.5-fluticasone group), increasing the dose of fluticasone by a factor of 2.5 and adding salmeterol (FP250/SM50, the salmeterol–2.5-fluticasone group), or quintupling the dose of fluticasone (FP 500). A superior response was determined with respect to the hierarchical composite outcome that incorporated asthma exacerbations, asthma-control days, and the change in the FEV₁. The numbers in each bar represent the percentage of patients who had a superior response to that specific treatment as compared with the alternative treatment. Gray bars indicate the percentage of patients in whom one treatment was not superior to the other. The two groups of patients (patients from the trial involving children and those from the trial involving adolescents and adults) were compared to identify interactions between the two groups and the composite superiority outcome. T bars indicate 95% confidence intervals.