Meta Analysis of Adverse Events - Cervarix, August 12, 2007
To: Mr. Matt Whitman
File: STN 125259/0 CERVARIX
RE: Request for meta-analysis
Date: August 12, 2007
CBER and GSK had a telecon on August 9, 2007 (11:30 AM) to clarify CBER’s request for a meta-analysis of adverse events related to inflammatory neurological disorders and adverse events of potentially autoimmune etiology that are non-neurological from all clinical studies for products that include MPL. As conveyed during the telecon, we have the following comments:
1. In your submission dated August 2, 2007, you proposed to perform “pooled” data analysis in which specified adverse events (AEs) from different studies would be combined without accounting for the study effect. Please note that such pooled analyses are not true meta-analyses and therefore may be subject to biases, notably Simpson’s paradox, less inherent in true meta-analyses. Hence, the findings from the proposed pooled analyses may be an unreliable basis for making regulatory decisions. Therefore, we request that you perform appropriate meta-analyses rather than the proposed pooled analysis. Please respond.
2. The proposed tables of the event rates across all controlled studies in the dataset need to be consistent with the revised analytical plan. Please respond.