Pharmacological treatment of systolic heart failure

Abstract

Sir—The results of the MERIT-HF study (June 12, p 2001) add to knowledge on the pharmacological treatment of systolic heart failure. A similar trial of another selective -blocker, bisoprolol, was published earlier this year. The CIBIS-II study has received very little criticism. Now, with the publication of the MERIT-HF trial, we want to question the external validity of the two trials and discuss aspects of the MERIT-HF study. According to CONSORT guidelines for reports of randomised controlled trials, emphasis is put on the situation before randomisation—ie, number of eligible patients, the number of patients not randomised, and the reasons for non-randomisation. In both these heart failure trials no data are given about number of eligible patients. This point may seem trivial, but because the treatment of heart failure with -blockers has side-effects, especially in patients with advanced disease, many patients are probably never regarded as eligible, and therefore not included. How many patients were initially involved and where did they come from? The omission of information on eligible patients is hard to understand, because the external validity depends on the representativeness of those patients included. Some limitations of the two studies were outlined by Norman Sharpe in his commentary, but he did not discuss the presentation of the trials. Furthermore. the investigators of the MERIT-HF study (a third representing the sponsor) emphasise that metoprolol is a “lipid soluble -blocker”, probably inferring that positive data on -blockers in treatment of heart failure are collected from trials on lipidsoluble -blockers only. T h e octanol/buffer partition coefficient of the free form of the drug, which is usually taken as an indication of the lipid solubility of the compound, has not yet proved to have real meaning in this context. If we take data from studies on the effects of -blockers after myocardial infarction, recently extensively reviewed, and plot octanol/buffer coefficients against pooled odds ratios for death, the importance of lipid solubility of blockers (figure) is not supported. Sufficient data are lacking in both situations and should, until confirmed, be regarded more as a redundant message intended for the market.