CHEST Guidelines-Secondary-Outcomes-and-Peto-OR-in-Meta-analysis

Secondary-Outcomes-and-Peto-OR-in-Meta-analysis_ch

Pdf Summary

The letter from Drs. Anurag Khera and Taramangalam S. Ramakrishnan addresses a June 2014 article by Dong et al. in CHEST, concerning a meta-analysis assessing the risk of tuberculosis (TB) and influenza in patients with Chronic Obstructive Pulmonary Disease (COPD) using inhaled corticosteroids (ICS). The article reported a Peto Odds Ratio (OR) of 2.29 (95% CI, 1.04-5.03), indicating increased risk. However, Khera and Ramakrishnan express skepticism about these findings due to methodological concerns. They note that none of the studies in the meta-analysis were originally designed to assess TB and influenza risk, leading to questions about sample size adequacy. They also critique the use of the Peto OR, mentioning that it is conventionally used under three conditions, only one of which was met in the meta-analysis. Additionally, they highlight the significant influence of individual studies on the overall results, specifically pointing out a study by Calverley et al., which contributed disproportionately (76.55% weight) to the findings.<br /><br />In response, the authors of the original article thank Khera and Ramakrishnan for their comments, acknowledging the primary focus of the trials was not TB or influenza risk assessment. They argue that despite this limitation, the trials’ rigorous follow-up and monitoring might mitigate concerns of under-detection or misclassification. The authors defend their use of the Peto OR, citing its advantages in handling rare events, and suggest potential misclassifications would be nondifferential due to the trials' double-blind design. They acknowledge the challenges of evaluating drug safety within clinical trial settings but assert their approach still provides valuable insights into the risks associated with ICS use in COPD patients.

Keywords

tuberculosis

influenza

COPD

inhaled corticosteroids

meta-analysis

Peto Odds Ratio

methodological concerns

sample size

risk assessment

clinical trials