Treatment Differences and Other Prognostic Factors Related to Breast Cancer Survival

OBJECTIVE: To determine differences in and relative importance of treatment modalities by hospital type and their effect on survival of breast cancer patients. DESIGN: Cohort of population-based cancer registry breast cancer patients diagnosed from 1984 through 1990. The analysis was done within this cohort, stratified by hospital type and treatment modality. SETTING: Orange County, California, residents diagnosed and treated for breast cancer in 126 hospitals (small community, large community, health maintenance organization HMO, or teaching). PATIENTS: A total of 5892 non-Hispanic white women with no known prior cancer and with localized or regional, histologically confirmed breast cancer. MAIN OUTCOME MEASURES: Effects of treatment modality and hospital type on survival. Adjustments for differences in age, tumor size, number of positive lymph nodes, and histology were included in the analysis. RESULTS: Use of recommended breast-conserving surgery (BCS) was greatest among teaching hospitals, where more than 50% of patients with localized disease received BCS between 1988 and 1990 and 40% to 50% with regional disease received BCS between 1984 and 1990. At nonteaching hospitals, 30% or less of patients received BCS between 1984 and 1989, regardless of stage. Rates of survival after BCS were at least as good as rates of survival after total mastectomy, other factors being equal. Survival rates varied by hospital type for patients with localized disease, with significantly better rates at large community hospitals and significantly worse rates at HMO hospitals in comparison with small hospitals. Patients with regional disease at large hospitals had a significant survival advantage. CONCLUSIONS: Greater use of BCS is strongly urged. Overall, large community hospitals had significantly better survival rates than small community and HMO hospitals. Further follow-up will determine if treatment and survival differences by hospital type persist.