Hazard ratios in the Summary of Findings table: experience of the Cochrane Haematological Malignancies Group

Article type
Authors
Herbst C, Bauer K, Brillant C, Skoetz N, Engert A
Abstract
Background: Summary of Findings tables with the GRADE approach have been recommended in Cochrane reviews since 2008. In reviews concerning haematological malignancies the primary outcome is often overall survival and analysed using hazard ratios. Methods: A Summary of findings (SoF) table was produced using the GRADE Profiler for a review examining combined modality treatment versus chemotherapy alone for patients with early stage Hodgkin’s lymphoma. The SoF was produced by the main author and discussed with other authors. Results: Hazard ratios (HRs) were used for the outcomes overall survival and progression-free survival and a dichotomous outcome for complete response rate. The review has a HR of ∼0.6 for overall survival, indicating improved survival for the group receiving combined modality treatment. The HRs were entered automatically and marked as HR. The assumed risks had to be entered as the mortality risk (i.e. negative events) and not as the risk of survival (i.e. positive events). All involved authors believed this to be counter-intuitive. Conclusions: In the SoF tables, outcomes evaluated using HR switch between the presentation of the outcome (name and HR) and presenting calculated risk based on the inverse of the outcome (i.e. negative instead of positive events). Since overall survival is a very important outcome in many reviews, we would prefer a more consistent presentation in the SoF table.