Adjuvant chemotherapy for breast cancer: discordance between physicians' perception of benefit and the results of clinical trials.

PURPOSE: Decisions about management of patients depend on the perception of physicians about the results of clinical trials. We therefore determined whether physicians would recommend adjuvant chemotherapy for specified patients with breast cancer, and whether the perceived levels of benefit that led to these decisions were supported by the results of clinical trials. METHODS: We mailed a questionnaire that presented two clinical scenarios to 515 American and European oncologists. Scenario A depicted a premenopausal woman with node-negative, estrogen receptor-negative (ER-) breast cancer, and scenario B depicted a post-menopausal woman with node-positive, ER- disease. Respondents were asked about the evidence that they required from clinical trials to recommend adjuvant chemotherapy for these patients. RESULTS: Replies were received from 307 oncologists. American respondents were more likely to recommend adjuvant chemotherapy than Europeans (A, 93% v 68%; B, 85% v 67%; P < .001 for both comparisons). Opinion was evenly divided as to whether improvement in the relapse-free survival (RFS) rate was sufficient to recommend treatment, or whether improvement in overall survival was necessary. Mean values of 12.9% to 14.6% improvement in RFS at 3 years, or 10.7% to 12.4% improvement in overall survival at 5 years, were required to offer treatment. An overview analysis of clinical trials suggests that adjuvant chemotherapy achieves absolute improvements of 6% to 9% in RFS rates at 3 years and 3% to 4% in overall survival rates at 5 years for patients in these scenarios. CONCLUSIONS: Most oncologists recommended treatment in these scenarios, but require a level of treatment effect that has not been demonstrated in clinical trials. Our results suggest that oncologists' perceptions of the results of clinical trials overestimate the therapeutic gain from use of adjuvant chemotherapy for breast cancer.