In order to facilitate the performance of randomized controlled trials (RCT) in situations where no ''hard'' endpoints can be identified and the patient's subjective impressions about success of failure of a given treatment are paramount, the ''optional crossover design'' is suggested. In this design, patients are randomised to receive either active medication or placebo during phase I. At its end the patient may choose to change to the other treatment arm, if therapy was felt to be unsuccessful (= optional cross-over). In phase II, treatment continues as in phase I except for those patients who have chosen the ''optional cross-over''. Further cross-over points may ensue depending on the particularities of the situation, however, two such options may be adequate for most studies. At the end of the trial period, the numbers of patients in each treatment arm are counted and evaluated statistically. If an optimally successful remedy is being tested, close to 100% of the study population could finish up in the active treatment arm. If, as in most instances, the remedy is not optimally successful, this percentage will be proportionally less. If an ineffective remedy is being tested, the distribution of the total sample within the two treatment arms approaches 50:50%. The ''optional cross-over design'' seems suited for RCT in areas where the complex, unmeasurable and subjective experience of the patient are considered to represent adequate endpoints.
