Meta-analysis with Peto Odds Ratio method

For combining effect size from several studies, meta-analysis is the accepted method. However, for binary studies based on rare events, the existing effect size measures used in meta-analysis are inefficient. In these conditions, the Peto Odds ratio (POR) has been developed as a relative effect size estimator. in case of large treatment effects or when the group ratio is not balanced, a bias occurs in POR estimates.

POR is employed to pool odds ratios. Petos method works well for rare events as it does not need corrections for zero counts. Further, POR is also employed with dichotomous events based on time-to-event analyses.


Further reading:

Brockhaus, A. Catharina, Ralf Bender, and Guido Skipka. “The Peto odds ratio viewed as a new effect measure.” Statistics in medicine 33.28 (2014): 4861-4874.

Meta-analysis links pain-killers with heart attack

A meta-analysis accepted in BMJ on 10th April, 2017 linked common pain killers with heart attack. The study was based on patient level data from Europe and Canada. Data from 61,460 heart attack patients was compared with 446,763 controls using systematic review and meta-analysis models.

The results of the meta-analysis are shocking. All the NSAIDs (non steroidal anti-inflammatory drugs) like celecoxib, ibuprofen, diclofenac, naproxen and rofecoxib increased the risk of heart attack. Dosing for just one week showed the association of risk. Up to one month dosing, there was a steady increase in risk. After one month, the risk remained constant. The heart attack risk also increased with the dose of these drugs.