,
8 tweets,
2 min read
Read on Twitter
@ESC_Journals Surprised that a meta-analysis with so many errors made it past peer review: academic.oup.com/eurheartj/adva… . I hope that by raising awareness of these issues, this article can be withdrawn and revised before it makes it into print 1/n
2/n – Here’s the basic intended design: Systematic review with meta-analysis of all randomized controlled trials (RCTs) comparing atorvastatin 80 mg/d (“load”) vs control in patients with ACS treated with PCI on CV events.
3/n – Hasn’t this question been answered? This exact one; yes. However, what is less established is whether statins administered around the time of PCI can reduce post-PCI MI & reduce early (<30 days) CV events.
4/n SECURE-PCI (the largest trial on this topic) was neutral, but if there are other relevant trials, a well-done meta-analysis could be enlightening. But here’s the problem: The authors included a bunch of trials that they shouldn't have.
5/n Authors included a total of 13 studies; at least 5 should have been excluded. The 1st trial that caught my attention was inclusion of a secondary publication of the MIRACL trial. MIRACL specifically included only ACS patients who hadn’t been treated with PCI.
6/n Errors 2 and 3: The authors included not only one, but two secondary publications of the same trial, TNT. TNT included patients with stable CAD - not ACS - and few received atorvastatin anywhere near the time of their prior PCI.
7/n Error 4: The authors included this study: ncbi.nlm.nih.gov/pubmed/28822510, which at least enrolled patients with STEMI, but they were all treated with fibrinolysis rather than primary PCI, and <2% underwent PCI at any point in the trial.
8/n Error 5: 1 of the included studies isn't an RCT: ncbi.nlm.nih.gov/pubmed/24462073. There may be more issues with this review, but it’s moot at this point since we’ve just eliminated >50% of the data from this meta-analysis. which substantially changes the results & conclusion.
