Author + information
- Emanuel Raschi, MD, PhD,
- Francesco Salvo, MD, PhD,
- Elisabetta Poluzzi, PharmD, PhD and
- Fabrizio De Ponti, MD, PhD∗ ()
- ↵∗Department of Medical and Surgical Sciences, Alma Mater Studiorum-University of Bologna, Via Irnerio 48, I-40126 Bologna BO, Italy
A recent meta-analysis by Cheng et al. (1) estimated increased risk of sudden cardiac death/ventricular arrhythmia and cardiovascular death in current users of macrolides. We have read with interest this timely paper, and we take the opportunity to bring to your attention a methodological issue. We note that Cheng et al. (1) erroneously considered our previous work (2) a case-control study, whereas it was a case/non-case study performed in a spontaneous reporting system (SRS) database that used the reporting odds ratio (ROR) to investigate disproportion in reporting torsade de pointe in patients treated with macrolides.
It is important to emphasize the fact that, although odds ratio (OR) and ROR share the same statistical formula, ROR is not a risk measurement but a simple indicator of risk, reflecting imbalance in reporting frequency of a drug-related event in comparison with other drug-related events; thus, “non-exposed” patients are not available, although they are crucial for measuring risk. In other words, ROR cannot estimate risk but only allows identification of potential safety issues, which necessarily requires further investigation (3). For these reasons and for a more reliable pooled risk estimate of macrolide-related proarrhythmia, we believe that our study is not eligible for the present meta-analysis.
Although misclassification of our study and its inclusion in the meta-analysis has no influence on conclusions made by Cheng et al. (1), we would like to express our concern for the incorrect use of studies in SRS (the so-called pharmacovigilance syndrome ), which may result in their erroneous inclusion in safety meta-analyses of observational studies. It is time for a thorough discussion and, hopefully, consensus on the role of disproportionality analyses in measuring drug-related risks. In the meantime, we propose to include disproportionality studies in (qualitative) systematic reviews but keep results separated from pooled risk estimates of (quantitative) meta-analyses.
Please note: The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- 2016 American College of Cardiology Foundation
- Cheng Y.J.,
- Nie X.Y.,
- Chen X.M.,
- et al.
- Bate A.,
- Pariente A.,
- Hauben M.,
- Bégaud B.