Hodgkin's disease prognosis: a directly predictive equation.

586 patients with Hodgkin's disease diagnosed between 1970 and 1979 were staged and treated in the same way. Multivariate analysis was used to delineate the prognostic roles of several clinical features at diagnosis. A multiple regression analysis was applied to an exponential model for survival-time distribution, which proved to fit the data accurately. Several clinical characteristics were studied and those that could singly discriminate survival significantly were chosen as predictive variables for the multiple regression. These were: sex, age, stage, histological subtype, presence of constitutional symptoms, mediastinal mass, and erythrocyte sedimentation rate (ESR), and haemoglobin and serum albumin concentrations. ESR, stage, histological subtype, and age proved to be the best prognostic factors, while sex and albumin had minor value. The presence of symptoms, mediastinal bulk, and haemoglobin were not so important. A linear equation for the six variables was derived to calculate the estimated median survival time for any given patient. This equation was validated on an external group of 179 similar patients.