Development and application of statistical models to evaluate potential treatment effects in observational COVID-19 studies

• Applicant: Professor Dr. Martin Wolkewitz
• Subject Area: Epidemiology and Medical Biometry/Statistics
• Term: since 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 458593554

The urgent need to find effective COVID-19 treatment options led to an extensive research on repurposing of available drugs (such as Remdesivir or Hydroxychloroquine) in observational studies. However, in contrast to randomized trials, observational studies require advanced statistical modelling to acknowledge time-dependent treatment allocation, confounding, different time-scales and heterogeneity. In addition, the clinical endpoints of COVID-19 patients are very complex due to multiple oxygen-support states (such as mechanical ventilation or extracorporeal membrane oxygenation) and death and discharge as competing outcomes. In preliminary work, we studied potential pitfalls and survival biases in pandemic settings of influenza A (H1N1pdm09). In a close collaboration with national (e.g., Freiburg, n=213 COVID-19 patients) and international clinical partners (e.g., Isfahan, n>600 COVID-19 patients), we are currently applying multistate methodology to study time complex outcomes of COVID-19 patients. In this project we aim to further extend these methods to study the possible benefit of time-dynamic treatment regimes. Therefore, we will 1) perform a literature review to evaluate the methodological quality of published article regarding common time-related types of bias, 2) develop appropriate statistical models which address the aforementioned challenges and 3) apply advanced methodology to observational COVID-19 data and thus improve evidence-based decision making.

• DFG Programme: Research Grants