Antiplatelet therapy and atherosclerotic events

Editor—We endorse the response of Baigent and others to Cleland's commentary on the Antithrombotic Trialists' antiplatelet meta-analysis.^1–3 We would like to add some further comments in response to Cleland's article and the editorial in the same issue.⁴

Both suggest that the data in the meta-analysis were revised retrospectively. But the overview methods were planned prospectively. Differences between the data in trial publications and the dataset used for the meta-analysis occurred where trialists provided additional information on the numbers of patients originally randomised, or on unpublished or subsequently available outcomes for small numbers of patients. Minor differences between the current and previous antiplatelet overviews generally relate to additional, unpublished data from a few trials and do not affect any of the results or conclusions.

The claim by Reilly and FitzGerald that the absolute reduction in vascular events with antiplatelet treatment is smaller in acute ischaemic stroke than in other high risk conditions is incorrect. For every 1000 patients treated, about 10 events are prevented in the first month after onset of stroke, and just over one event per month is prevented with long term treatment thereafter.³

Cleland finds it remarkable how seldom trials of antiplatelet agents have shown benefit on their selected primary outcome. Many early trials of antiplatelet treatment were too small to detect moderate benefits reliably, which is why the first meta-analyses were needed. Reilly and FitzGerald suggest that meta-analysis is no longer needed because large enough trials are now being done. This view fails to acknowledge that, firstly, meta-analysis of large trials can assess not just whether a treatment works but also for whom and by how much, and, secondly, trials comparing different antiplatelet regimens have rarely been large enough to detect the small differences expected.

Cleland says that inconvenient trials are ignored in the discussion section of the meta-analysis, citing an unpublished antithrombotic trial, which included fewer than 200 patients and recorded only about 50 vascular events in its comparison between aspirin and control. Including this trial in the meta-analysis would make no difference to the results. Cleland also cites an economic appraisal of aspirin, which he co-authored. Its first sentence, that aspirin is a cheap drug that is effective for the prophylaxis of cardiovascular events, contrasts with the views in Cleland's commentary.

Finally, unlike the Antithrombotic Trialists' meta-analysis, neither the accompanying editorial nor the commentary has been endorsed by hundreds of collaborating trialists worldwide. Furthermore, Cleland's commentary was published despite a reviewer pointing out that his views are maverick, and despite the fact that the conclusions of his article rely on basic errors of fact.²