Prediction of survival of patients terminally ill with cancer: Results of an Italian prospective multicentric study

Background. The individualization of prognostic factors in the various stages of cancer facilitates the planning of a therapeutic assistance program aimed at various subsets of patients. The prognostic factors for survival in patients terminally ill with cancer have been investigated in case studies that are often retrospective, monocentric and/or include a mixture of patients in advanced disease stages. The aim of this prospective multicentric study was to verify those clinical factors predictive of survival in a population of patients with terminal cancer. Methods. This prospective and multicentric study was performed on 540 patients with solid tumors in the disseminated phase, no longer subject to specific therapy. Patients were evaluated at study entry and every 4 weeks thereafter. The analysis was performed for 23 clinical parameters. Results. Of 530 assessable patients with a median survival of 32 days, 15 factors were found to be statistically significant prognostic factors. By univariate analysis, 13 factors were found to be indicators of a worse survival: age older than 65 years (P = 0.05); palliative corticosteroid treatment (P <0.001), anorexia (P <0.001); dry mouth (P <0.001); dysphagia (P <0.001); hospitalization (P <0.001); transfusion (P <0.001); weight loss greater than or equal to 10% (P = 0.001); dyspnea (P = 0.01); pain (P = 0.006); increasing amount of pain- killer treatment (P = 0.01); increasing number of symptoms (P <0.001); and worse clinical prediction of survival (P <0.001). Two factors that correlated with a better survival rate were palliative progestin treatment (P = 0.03) and a higher Karnofsky performance status (P <0.001). Multiple regression analysis revealed that only clinical prediction of survival, anorexia, dyspnea, palliative steroidal treatment, Karnofsky performance status, and hospitalization were independent predictors of survival. Conclusions. The importance of certain clinical parameters as prognostic indicators for patients with terminal cancer (clinical experience, physical activity level, clinical symptoms relating to and unrelated to nutritional state) were confirmed; some others possible factors, such as treatment with corticosteroids and hospitalization, also were noted. These may be useful factors in the therapeutic, assistance decision-making process and may eliminate overtreatment and undertreatment resulting from philosophically preconceived attitudes, rather than from considering the patient's true pathologic condition.