To the Editor: In a recent observational study, Krum et al concluded that the treatment of heart failure after myocardial infarction in Australian teaching hospitals is suboptimal because angiotensin-converting enzyme (ACE) inhibitors, β-blockers and aldosterone antagonists are underutilised.1 We believe that another explanation, mentioned by the study’s authors, is worth exploring further — for valid clinical reasons, it was not appropriate for certain patients to start or continue taking some of these medications. An understanding of the enrolment criteria of relevant clinical trials is informative.

The large, long-term ACE inhibitor trials quoted by Krum et al — SAVE,2 TRACE and AIRE4 — between them screened 34 037 patients with myocardial infarction and left ventricular dysfunction. Only 5986 patients (18%) met the inclusion/exclusion criteria to be enrolled in one of the trials. Unfortunately, the CAPRICORN5 (β-blocker) and EPHESUS (aldosterone antagonist) trials did not publish the number of patients screened versus the number randomised, but a glance at their exclusion criteria explains why, for some patients, it may not have been appropriate to start these medications during their hospital stay. Some of the exclusion criteria for CAPRICORN were: unstable angina, ongoing therapy with antiarrhythmics (except amiodarone), secondary or tertiary heart block or sick sinus syndrome unless paced, uncontrolled hypertension (> 160/95 mmHg), bradycardia (heart rate, < 60 beats/min), hypotension (systolic blood pressure, < 80 mmHg), requirement for intravenous diuretics or inotropes, chronic obstructive pulmonary disease with ongoing inhaled β2-agonist or steroid therapy, and unstable insulin-dependent diabetes.

Is there any harm in prescribing outside the inclusion/exclusion criteria for clinical trials? A population-based, time-series analysis linking prescription-claims data and hospital admission records of 1.3 million adults in Canada6 showed that hyperkalaemia-related deaths in hospital doubled after the RALES trial (spironolactone) was published in 1999. There was no reduction in re-hospitalisation for heart failure or all-cause mortality. The authors speculated that part of the reason for this was prescription of spironolactone to patients who would have been excluded from the RALES trial.

While we would not advocate prescribing strictly within the boundaries of the inclusion/exclusion criteria of clinical trials, it is important to understand these criteria, so that prescribing in “real world” patients is done with care.

We are reassured that Krum et al’s study suggests there is discretion in the prescribing of drug therapy. Presumably, during ongoing medical assessment, it will be appropriate for some patients to commence some of these medications (potential benefit outweighs potential harm), while others may need to have their medications reviewed because of adverse events.

In reply: We thank Bailey and Naganathan for their thoughtful viewpoint regarding prescribing according to clinical trial criteria. We agree that prescribing in the real world often involves complex decision making, taking into account age, comorbidities, concomitant medications and other factors, whereby guidance regarding individual patients cannot readily be extracted from clinical trial literature. This may certainly contribute to underutilisation of evidence-based drug treatment.1 Nevertheless, several analyses support the contention that physicians who more closely adhere to evidence-based guidelines (which in turn are derived from randomised clinical trials) produce better outcomes for their patients.2,3 Therefore, we would still advocate prescribing as closely as possible to guideline recommendations, while acknowledging that these recommendations may not always be readily applicable to every patient.