[Adjuvant chemotherapy of cancer: definition, prognosis, statistical aspects].

Once the need for adjuvant chemotherapy has been defined with precision, it is important to specify when it may be employed. Two factors determine the decision to use this form of therapy: a) the notion of the presence of malignant tumor, sensitive to chemotherapy (breast, bone ovary), b) knowledge of the elements of prognosis. The probability of relapse is estimated on the basis of these two latter points. Adjuvant chemotherapy is prescribed in cases where the risk of relapse is quite important, attaining 40 to 50 per cent. Its effectiveness can be assessed only with rigorous statistical rules. Randomization only permits a comparison of adjuvant chemotherapy to a control group. Certain errors must be avoided in the interpretation of the results: inaccurate analysis of results such as median comparisons; comparison of the number of deaths or the number of relapses; exclusion of patients after randomization. Only the comparison of the remission duration curves by a test adapted to the form of the curves (Log rank test or Weibull's test) seems justified.