Editor—Owen’s perception that intravenous β blockers are less commonly given after acute myocardial infarction in the United Kingdom than elsewhere1 is confirmed by data from the European secondary prevention study.2 Clinical management was examined in a representative sample of over 4000 patients admitted to hospital with confirmed acute myocardial infarction in 11 European regions. Intravenous β blockade was given to 13% of patients overall, but this proportion varied from 0.5% (United Kingdom) to 54% (Sweden) in the regional samples. This 100-fold range, larger than the range for any other treatment or procedure studied, is particularly striking for an aspect of management that has been subjected to at least 28 randomised trials in over 27000 patients.
Variation in practice on this scale has important messages for proponents of evidence based medicine and cannot be explained by lack of awareness on the part of clinicians. The key issue is the generalisability of the evidence from trials in highly selected low risk patients. In the largest trial, the first international study of infarct survival, fewer than a third of patients admitted to coronary care units were considered to have been eligible for randomisation.3 The primary end point of vascular mortality in the first week was low in both the atenolol and control groups (3.9% v 4.6%; P<0.04). The pooled estimate of treatment effect on one week mortality in all available trials remains of borderline significance, and definition of the relevant patient group is even more difficult when meta-analysis is used.
No evidence from clinical trials exists to indicate whether intravenous β blockade would be more or less beneficial in haemodynamically compromised patients or those with hypotension induced by thrombolytics. All observational data are invalidated by the confounding of treatment selection and haemodynamic status. The highly significant increase in the use of inotropic support seen after treatment with atenolol in the first international study of infarct survival gives grounds for caution.
The observed variation in practice probably reflects the widely differing judgment of clinicians on the external validity of the published trials of intravenous β blockade. For long term oral β blockade the evidence is stronger and patient selection less problematic. Correspondingly, use of this treatment was more common in the European study (46% at six months after discharge) and rather more consistent between regional samples (range 27-72%).