Rethinking Meta-Analysis: Addressing Problems of Non-Transportability When Combining Treatment Effects Across Patient Populations

• Activity Number: 309 - Advances in Causal Inference
• Type: Contributed
• Date/Time: Tuesday, July 30, 2019 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Tat Thang Vo* and Stijn Vansteelandt and Raphael Porcher
• Companies: Ghent University and Ghent University and Centre de Recherche Épidémiologie et StatistiqueS Université de Paris (CRESS-UMR1153)
• Keywords: causal inference; meta-analysis; non-transportability; outcome regression; inverse probability weighting; non-transportability

Abstract

Lack of transportability of effect measures across studies complicates most meta-analyses. As a result of this, treatments that are equally effective on patient subgroups may appear to have different effectiveness on patient populations with different case mix. It is therefore important that meta-analyses be explicit for what patient population they describe the treatment effect. To achieve this, we develop novel approaches for meta-analysis of randomized clinical trials, which use individual patient data (IPD) from all trials to infer the treatment effect for the patient population in a given trial, based on direct standardization using either outcome regression (OCR) or inverse probability weighting (IPW). Accompanying random-effect meta-analysis models are developed. The new approaches enable disentangling heterogeneity due to case-mix inconsistency from that due to beyond case-mix reasons, using statistical tests, (generalized) I2 statistics or (generalized) prediction interval. The proposed strategy is also plausible when the individual patient data is available for some but not for all trials in the meta-analysis.