Predicting survival in advanced hematologic malignancies: do patient‐reported symptoms matter?

Abstract

To investigate whether patient-reported symptoms provide independent prognostic information for survival in patients with hematological malignancies. Overall 119 patients with various diagnoses were recruited in an observational study and symptoms were assessed with the M.D. Anderson Symptom Inventory (MDASI). Key potential socio-demographic, biomedical, and physician-reported prognostic candidates were also considered. The Cox proportional hazards regression model was used for both univariate and multivariate analyses of survival. Additional sensitivity analysis, based on 500 bootstrap-generated simulation datasets, was also performed to confirm the results obtained with the Cox regression model. The median survival of the entire cohort was 4.8 months (range 0-28 months). The MDASI was completed at baseline by 91% of patients. The final multivariate model retained two parameters as independent prognostic factors for survival: clinical prognostic group and patient's self-reported severity of drowsiness. The following hazard ratios (HR) were found for curable vs. terminal: 0.055 (95% CI, 0.022-0.136; P < 0.001) and 0.193 (95% CI, 0.103-0.362: P < 0.001) for advanced vs. terminal. Patient's self-reported severity of drowsiness independently predicted survival with a HR of 1.801 (95% CI, 1.044-3.107; P = 0.033). Additional sensitivity analysis confirmed the independent prognostic value of variables identified in this study. The results suggest that patients' self-reporting of symptoms provides independent prognostic information for survival in patients with hematologic malignancies. These findings underscore the value of collecting patient-reported symptom data in routine clinical practice.