[Results of large-scale clinical trials and medical practice]

The results of large scale clinical trials are part of the scientific data on which medical and, in particular therapeutic decisions, are base; very concordant data may even be used to define appropriate or inappropriate prescribing behaviour in the context of recommendations or references of good medical practice. Large scale clinical trials, however, have limitations inherent to their method; they give a mean result observed in an average population. The true question is to know if the therapeutic benefit proved in a therapeutic trial will be observed in a given individual in the future. If the aim of treatment is to make a symptom disappear or act on an objective intermediate criterion related to the prognosis, the benefits for the individual are easy to assess. When the aim is to prolong life or prevent a recurrence of a serious illness, the probability of avoiding a complication by treatment is only a statistic. This probability depends o: the amplitude of the therapeutic benefit observed in the trials; the resemblance of the patient to the "mean population" included in trials; the comparison of the treatment of the patient with the "mean treatment" used in the trials. An exact understanding of scientifically proven data then encounters the limits of the capacity of human memory. Probability medicine remains, however, the method which enables, for a given patient, the choice of treatment with the greatest chance of being effective, providing the elementary rules of utilisation of the information are respected.