Purpose: This project was designed to investigate how American medical oncologists actually use prognostic information to treat primary breast cancer patients, and to study their difficulties in combining complex and sometimes contradictory information.
Methods: A simple 2-page questionnaire was faxed in May and June 1993 to a sample of American medical oncologists who were members of the American Society of Clinical Oncology (ASCO).
Results: When presented with simple case histories of patients with newly diagnosed invasive breast cancer and asked to assess prognosis on the basis of tumor size, number of involved axillary nodes, patient age, estrogen receptor level, and progesterone receptor level, there was a wide divergence of opinions about the probability of disease-free survival at 10 years (both for cases in which the patient received no adjuvant therapy and for those in which the patient did receive such therapy). The use of additional prognostic data (such as S-phase, tumor histologic and nuclear grading, and cathepsin D status) did not refine the estimates, but led to an equal or greater dispersion of estimates of prognosis.
Conclusion: There is a clear need for tools to help oncologists integrate prognostic information for primary breast cancer patients. Such tools might lead to greater accuracy and uniformity of prognostic estimates. Such tools might also help make clear what prognostic tests are worth using for routine clinical practice.