Coenrollment in a randomized trial of high-frequency oscillation: prevalence, patterns, predictors, and outcomes

Abstract

Objective: Enrollment of individual patients into more than one study has been poorly evaluated. The objective of this study was to describe the characteristics of patients, researchers and centers involved in coenrollment, studies precluding coenrollment, and the prevalence, patterns, predictors, and outcomes of coenrollment in a randomized clinical trial. Design, Setting, Methods: We conducted an observational study nested within the OSCILLation for Acute Respiratory Distress Syndrome Treated Early Trial, which compared high-frequency oscillatory ventilation to conventional ventilation. We collected patient, center, and study data on coenrollment in randomized patients. Multilevel regression examined factors independently associated with coenrollment, considering clustering within centers. We examined the effect of coenrollment on safety and the trial outcome. Interventions: None. Measurements and Main Results: Overall, 127 of 548 randomized patients (23.2%) were coenrolled in 25 unique studies. Coenrollment was reported in 17 of 39 centers (43.6%). Patients were most commonly coenrolled in one additional randomized clinical trial (76; 59.8%). Coenrollment was less likely in older patients (odds ratio, 0.87; 95% CI, 0.76–0.997), and in ICUs with greater than 26 beds (odds ratio, 0.56; 95% CI, 0.34–0.94), and more likely by investigators with more than 11 years of experience (odds ratio, 1.73; 95% CI, 1.06–2.82), by research coordinators with more than 8 years of experience (odds ratio, 1.87; 95% CI, 1.11–3.18) and in Canada (odds ratio, 4.66; 95% CI, 1.43–15.15). Serious adverse events were similar between coenrolled high-frequency oscillatory ventilation and control patients. Coenrollment did not modify the treatment effect of high-frequency oscillatory ventilation on hospital mortality. Conclusions: Coenrollment occurred in 23% of patients, commonly in younger patients, in smaller centers with more research infrastructure, and in Canada. Coenrollment did not influence patient safety or trial results