Temporal Improvements in COVID-19 Outcomes for Hospitalized Adults

• Activity Number: 388 - Statistical Methods in Handling Longitudinal Data and Temporal Effect
• Type: Contributed
• Date/Time: Wednesday, August 10, 2022 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Gail Potter* and Tyler Bonnett and Kay Tomashek and Pablo Tebas
• Companies: National Institutes of Health / NIAID and National Institutes of Health/ NIAID and National Institutes of Health/ NIAID and University of Pennsylvania
• Keywords: platform trial; COVID-19; propensity scores; length-biased sampling; concurrent controls

Abstract

The Adaptive COVID Treatment Trial (ACTT) is an adaptive platform trial to conduct a series of phase 3, randomized, double-blind, placebo-controlled trials for COVID-19 treatments. The first three stages took place sequentially from Feb to Nov 2020, and each included a remdesivir arm: it was the treatment arm in ACTT-1 and the control arm in ACTT-2 and 3. We compare the hazard of recovery and mortality between sequential remdesivir cohorts to assess whether changes in standard care improved outcomes over time. Cox regression was performed with IPTW weights, using propensity score methodology used to adjust for baseline disease severity and risk factors. A separate model, also using propensity scores, estimates the change in odds of ventilation from ACTT-1 to ACTT-2. Oxygen use was tracked over time, but we modeled baseline ventilation to remove confounding from changes in other standards of care. We explore length-biased sampling by simulation and show this bias is higher during periods with higher caseloads. We found temporal improvements in recovery and mortality. Our results support the exclusion of non-concurrent controls in analysis of data from platform trials.