Generalisability of survival estimates for patients with breast cancer--a comparison across two population-based series.

The purpose of the study was to analyse the generalisability and geographic transportability of survival estimates produced by commonly used prognostic factors. We compared the influence of tumour size, histologic grade, axillary nodal status, oestrogen and progesterone receptor contents, age at diagnosis and two prognostication schemes (the Nottingham Prognostic Index and St. Gallen criteria) in two nationwide cohorts of patients diagnosed with breast cancer in 1991-2, the FinProg (n=2923, Finland) and the SEER series (n=43,249, the United States (US)). Eight-year estimates of breast cancer-specific (84% versus 80%), relative (86% versus 83%), and overall (70% versus 69%) survival were slightly more favourable in the SEER than in the FinProg series, respectively. Despite differences in demographic variables and the frequency of use of adjuvant therapies and mammography screening between the series, the prognostic factors examined produced close to overlapping survival curves with similar shapes. The results suggest that quantitative survival estimates based on frequently used prognostic factors and prognostication schemes are generalisable and transportable between large, unselected cohorts of breast cancer patients.