Methods: Survey responses of breast cancer patients from Detroit and Los Angeles SEER registries in 2002 (N=1,800, RR: 77%) were merged with survey responses from their surgeons (N=456, RR: 80%). The final analytic dataset consisted of 1,102 patients of 270 surgeons (excluding missing values and patients with tumors > 5cm). Patients who responded “true” to the question “The chances of being alive five years after surgery is the same for mastectomy as it is for lumpectomy with radiation therapy” were categorized as having survival knowledge. Those answering “false” or “don't know” were categorized as not having survival knowledge. We used logistic regression to evaluate correlates of survival knowledge, controlling for within-surgeon clustering. The first set of independent variables included patient race, education, age and cancer behavior (in situ vs. invasive). We then included surgeon age, gender, and volume, followed by patient treatment location (NCI cancer center, American College of Surgeons cancer program, no cancer program). Finally, we included a measure of patient-surgeon communication about treatment options.

Results: Forty-six percent of respondents had survival knowledge. African American and Hispanic patients and those with lower education were significantly less likely to have survival knowledge than their counterparts (p<0.001). In multivariate regression, neither surgeon characteristics nor treatment location mediated the relationship between race/ethnicity, education and survival knowledge. Surgeon characteristics were not correlates of women's survival knowledge, but women treated at an NCI cancer center were more likely to have survival knowledge than those treated elsewhere (P<0.05). Patient-surgeon communication was significantly associated with survival knowledge (p<0.001), but did not affect the relationship between race/ethnicity, education and knowledge.

Conclusions: Racial/ethnic and education disparities in breast cancer patients' knowledge about surgical treatment options were not mediated by surgeon factors, treatment location or patient-surgeon communication. For non-white and less educated women, being treated at an NCI cancer center may not be sufficient to ensure informed treatment decision making is achieved. Further work exploring the relationship between provider and delivery system characteristics and informed decision making among vulnerable breast cancer patient populations is needed.