ABSTRACT
Title
The Safety of NSAIDs (SOS) project: what can be learned from quantitative reviews of RCTs.
Authors
F. Salvo (1,2), A. Fourrier-Réglat (1,2,3), F. Bazin (1,2), M. Balzarin (4), N. Riera (5), P. Robinson (1,2), M. Haag (6), N. Moore (1,2,3), M.C. Sturkenboom (6), A. Pariente (1,2,3)
1 INSERM, U657, Bordeaux, France
2 Université Victor Segalen, Département de Pharmacologie, Bordeaux, France
3 CHU de Bordeaux, Bordeaux, France
4 Università di Padova, Dipartimento di Pediatria, Padova, Italie.
5 RTI, Health Solution, Barcelona, Espagne
6 Department of Medical Informatic, Erasmus University Medical Centre, Rotterdam, Pais Bas.
1 INSERM, U657, Bordeaux, France
2 Université Victor Segalen, Département de Pharmacologie, Bordeaux, France
3 CHU de Bordeaux, Bordeaux, France
4 Università di Padova, Dipartimento di Pediatria, Padova, Italie.
5 RTI, Health Solution, Barcelona, Espagne
6 Department of Medical Informatic, Erasmus University Medical Centre, Rotterdam, Pais Bas.
Abstract
INTRODUCTION
After rofecoxib withdrawal, a debate on cardiovascular (CV) and gastrointestinal (GI) safety of non-steroidal anti-inflammatory drugs (NSAIDs) was raised. In 2008 the European Commission funded the Safety Of non-Steroidal anti-inflammatory drugs (SOS) project. The first part of SOS was aimed to review and estimate the incidence of selected CV and GI events reported in pooled analyses or meta-analyses (MAs) of randomized clinical trials (RCTs).
METHODS
A literature search was performed using 4 different databases (Medline, Cochrane, ISI, SCOPUS). The events of interest were myocardial infarction (MI), stroke, cerebrovascular events, thromboembolic events, heart failure, GI bleedings, and perforation, ulcer, and bleeding (PUB). Drugs of interest were systemic NSAIDs and aspirin at high dosage. Events incidence were retrieved individually for each MA for the drugs of interest. When the incidence of the CV and GI events of interest in MAs was not reported, it was estimated by pooling data of events reported in each MA with regards to the overall number of exposed persons considered, and overall exposed personyears (PYs) when available.
RESULTS
Overall 1,733 references were identified. After screening titles and abstracts and full text, a total of 30 studies were included, allowing the estimation of 109 and 27 incidences rates for cardiovascular and gastrointestinal events respectively. No data was found on hemorrhagic stroke or lower gastrointestinal bleeding. Coxibs were more studied than traditional NSAIDs (20 meta-analyses vs. 9; one studied both). Many NSAIDs were never considered in meta-analyses. The most frequently evaluated events were MI and PUB. In the selected MAs, cumulative incidences (%) of MI could be estimated only for selective NSAIDs: celecoxib (ranging from 0.12 to 1.35%), etoricoxib (0.24 to 0.66%), lumiracoxib (0.09 to 0.37%), parecoxib/valdecoxib (0.58%), rofecoxib (0.18 to 0.89%), and valdecoxib (0.13 to 0.61%). Cumulative incidences of PUB were estimated for: aspirin (0.00, single intake), etoricoxib (0.11 to 1.24%), meloxicam (0.13 to 0.16%), nabumetone (0.06 to 2.6%), and rofecoxib (0.00 to 0.71%).
CONCLUSION
This systematic review of meta-analyses included information on incidence of cardiovascular and gastrointestinal events for most NSAIDs. However, important knowledge gaps were identified, especially regarding the CV safety of traditional NSAIDs.
After rofecoxib withdrawal, a debate on cardiovascular (CV) and gastrointestinal (GI) safety of non-steroidal anti-inflammatory drugs (NSAIDs) was raised. In 2008 the European Commission funded the Safety Of non-Steroidal anti-inflammatory drugs (SOS) project. The first part of SOS was aimed to review and estimate the incidence of selected CV and GI events reported in pooled analyses or meta-analyses (MAs) of randomized clinical trials (RCTs).
METHODS
A literature search was performed using 4 different databases (Medline, Cochrane, ISI, SCOPUS). The events of interest were myocardial infarction (MI), stroke, cerebrovascular events, thromboembolic events, heart failure, GI bleedings, and perforation, ulcer, and bleeding (PUB). Drugs of interest were systemic NSAIDs and aspirin at high dosage. Events incidence were retrieved individually for each MA for the drugs of interest. When the incidence of the CV and GI events of interest in MAs was not reported, it was estimated by pooling data of events reported in each MA with regards to the overall number of exposed persons considered, and overall exposed personyears (PYs) when available.
RESULTS
Overall 1,733 references were identified. After screening titles and abstracts and full text, a total of 30 studies were included, allowing the estimation of 109 and 27 incidences rates for cardiovascular and gastrointestinal events respectively. No data was found on hemorrhagic stroke or lower gastrointestinal bleeding. Coxibs were more studied than traditional NSAIDs (20 meta-analyses vs. 9; one studied both). Many NSAIDs were never considered in meta-analyses. The most frequently evaluated events were MI and PUB. In the selected MAs, cumulative incidences (%) of MI could be estimated only for selective NSAIDs: celecoxib (ranging from 0.12 to 1.35%), etoricoxib (0.24 to 0.66%), lumiracoxib (0.09 to 0.37%), parecoxib/valdecoxib (0.58%), rofecoxib (0.18 to 0.89%), and valdecoxib (0.13 to 0.61%). Cumulative incidences of PUB were estimated for: aspirin (0.00, single intake), etoricoxib (0.11 to 1.24%), meloxicam (0.13 to 0.16%), nabumetone (0.06 to 2.6%), and rofecoxib (0.00 to 0.71%).
CONCLUSION
This systematic review of meta-analyses included information on incidence of cardiovascular and gastrointestinal events for most NSAIDs. However, important knowledge gaps were identified, especially regarding the CV safety of traditional NSAIDs.