A multivariate analysis of the contribution of ‘auxometry’ to prognosis in breast cancer

A series of 1419 patients with carcinoma of the breast first diagnosed between 1961 and 1970 were studied to determine if a predictor variable termed ‘auxometry’ by Charlson and Feinstein influenced survival in a fashion that could not be explained by other well known prognostic factors. Methods used included multiple cross-classification, multiple linear regression and Miettinen's multivariate confounding score. Each analysis suggested an independent effect of auxometry, indicating that this variable reflects unique information about host-tumor relationships which should be made use of in the investigation and management of patients with breast cancer.