RESUMO
BACKGROUND: The incidence of breast cancer is increasing rapidly in Taiwan. Prognostic factors for survival in Taiwanese breast cancer patients have not been comprehensively studied. METHODS: Registry records of 979 newly diagnosed breast cancer patients who received initial therapy at National Taiwan University Hospital from January 1991 to December 1995 were linked to the national mortality file from 1991 to 1997 using the citizen ID of each patient. The effects of potential prognostic factors were assessed using Cox's regression model. Five-year survival rates of common characteristics were predicted according to the model. RESULTS: Among the 979 patients, 174 died within the 7-year study period (163 died from breast cancer). The overall 5-year survival rate was 81% (95% confidence interval, CI = 78-84%). The adjusted hazard ratio (HR) of mortality for patients aged at least 50 years versus patients aged less than 50 years was 1.49 (95% CI = 1.08-2.05); the HR of mortality for tumors of at least 5 cm in diameter versus those less than 2 cm was 2.45 (95% CI = 1.33-4.51); the HR of mortality for positive versus negative nodes was 3.65 (95% CI = 2.33-5.71); and the HR of mortality for pathology of infiltrating ductal carcinoma versus other types was 2.63 (95% CI = 1.32-5.27). Five-year survival rates for patients without node involvement were higher than 90% regardless of tumor size, while those for patients with positive nodes were significantly lower, except for patients with tumors of less than 2 cm in diameter. CONCLUSIONS: Patients who had a small tumor and involvement of fewer nodes, and who were younger, had a higher probability of survival. Node involvement was more important than tumor size in the prediction of survival.