In addition to standard features of clinical trial design such as randomisation and double-blinding sensitivity to drug effects is an important consideration when conducting exercise capacity trials in patients with heart failure. Two issues need to be addressed in this context. Firstly, it is important to enrol patients who are potential responders. Patients who have, for their age and sex, normal exercise capacity are unlikely to improve, even when given a drug that has a positive effect on exercise capacity. In addition, those patients who remain clinically stable following withdrawal of their previous drug therapy are unlikely to respond subsequently to an experimental drug with a similar mechanism of action. Secondly, failure to complete scheduled exercise tests during follow-up, prompting a 'per-protocol' analysis of results, may mask the drug's actual effect. To avoid this, an 'intention-to-treat' approach to data collection and analysis, with appropriate allowance made for missing test data, should be adopted.
