CMStatistics 2019: Start Registration
View Submission - CMStatistics
B1994
Title: Evaluating the effect of interventions on recurrent event outcomes with multiple competing risks Authors:  Elizabeth Colantuoni - Johns Hopkins University (United States) [presenting]
Abstract: Among older adults who are critically ill and receiving care in an intensive care unit (ICU), delirium is very common (>70\%) and strongly associated with long-term cognitive impairment that is similar to Alzheimers Disease and Related Dementias (ADRD). Hence, there is a growing number of clinical trials evaluating interventions to prevent or treat delirium, with the goal of reducing associated long-term cognitive impairment/ADRD in critical illness survivors. Evaluating the impact of interventions on delirium is complicated due to the time-varying nature of delirium, the fact that delirium can frequently re-occur, and patients may experience death or be discharged from the ICU prior to completing the pre-planned duration of patient follow-up (e.g. 14 days). In addition, it is often hypothesized that interventions that may reduce delirium onset or duration would also reduce mortality and duration of ICU stay. Current approaches to evaluate the delirium endpoint rely on aggregating the data into a simple binary summary (ever vs. never delirium) or a composite endpoint (delirium free days to 14 days). These approaches will be contrasted to a novel extension of an existing method, the joint recurrent and terminating event model, that allows for the multiple terminating events of death and ICU discharge.