[Skip to Navigation]
Sign In
Comment & Response
March 26, 2020

Double Counting of Patients in Meta-analyses of Observational Studies—Reply

Author Affiliations
  • 1Department of Stem Cell Transplantation, Medical Center Hamburg-Eppendorf, Hamburg, Germany
JAMA Oncol. 2020;6(5):787-788. doi:10.1001/jamaoncol.2020.0173

In Reply We would like to thank Fingrut and Chen for highlighting the limitations of conventional meta-analysis approaches of retrospective studies. We agree that it would require meta-analyses of patient-level data to provide the most unbiased approach possible for synthesizing retrospective evidence. In the field of allogeneic stem cell transplantation, prospective randomized evidence is unfortunately scarce, and yet consensus mostly relies on evidence from retrospective studies and expert experience. The synthesis of retrospective studies is prone to bias, such as reporting bias and missing data, key dimensions in the assessment of risk of bias, as highlighted in the updated version of the Cochrane Handbook for Systematic Reviews of Interventions (available at http://www.training.cochrane.org/handbook). Therefore, we had to make some choices and assumptions to conduct our analyses.1 This was done in accordance with recommended methods and after careful consideration of information available in the studies that we relied on.

Add or change institution