Abstract
Objective outcomes for pediatric community-acquired pneumonia (CAP) are lacking. The desirability of outcome ranking (DOOR) and response adjusted for duration of antibiotic risk (RADAR) outcome encompass clinical benefit and adverse effects, while also accounting for antibiotic exposure. We evaluated DOOR and RADAR (DOOR/RADAR) through simulations and compared sample-size considerations to noninferiority designs in a hypothetical trial comparing antibiotics and no antibiotics (ie, placebo) for children with mild CAP. We also evaluated a trial comparing different durations of antibiotic therapy. Three scenarios were considered: 1 with no difference in DOOR between the 2 groups, 1 in which placebo is more efficacious, and another in which amoxicillin is more efficacious than placebo. The power to detect a difference between arms was greater using DOOR/RADAR compared with DOOR alone. Assuming a sample size of 200, DOOR had 2.5%, 50%, and 65% power to detect a statistical difference between arms for scenarios 1-3, respectively, significantly less than DOOR/RADAR. Importantly, DOOR/RADAR incorrectly identified placebo as superior in scenario 3, where amoxicillin was truly efficacious. Sample size requirements for noninferiority designs were larger to achieve similar levels of power as DOOR and DOOR/RADAR. DOOR/RADAR has the potential to lead to an incorrect conclusion declaring placebo superior when amoxicillin is efficacious.