To the Editor:
I have read with interest the original article entitled “Corticosteroid Therapy for Severe Community-Acquired Pneumonia: A Meta-Analysis” by Cheng et al.1 The authors decided to use the fixed-effects model despite the heterogeneity of the studies included in the meta-analysis (populations, doses, and antibiotics were different in each study). If the random-effects model had been used, which seems more appropriate in this case, the conclusion would have been different (Fig. 1). Another choice that seems inadequate was the decision to use the Peto odds ratio. The Peto odds ratio method usually works well in cases in which the odds ratio is close to 1.2 This is not what happened in 2 studies. Thus, if the odds ratio had been used, even if using the fixed-effects model, there would have been no difference between the corticosteroid and placebo groups (Fig. 2).
Thus, I believe that this meta-analysis should be interpreted with caution.
Footnotes
The author has disclosed no conflicts of interest.
- Copyright © 2014 by Daedalus Enterprises
REFERENCES
- 1.
- 2.