Comparative multidisciplinary prediction of survival in patients with advanced cancer

The expected survival of patients with metastatic cancer can significantly impact decisions regarding treatment, care setting, and future planning. We evaluated the prognostication ability of a multidisciplinary team (MDT) experienced in providing supportive care and palliative radiotherapy. After clinical assessment of consecutive patients, survival predictions were independently made by each MDT member. Patient demographics, factors influencing predictions, and dates of death were collected. Clinical predictions of survival (CPS) were considered correct if within 30 days of actual survival (AS). Summary statistics and Kaplan-Meier estimates of overall survival were obtained. Correlations between actual and CPS were calculated using Spearman's correlation coefficient. Multivariate logistic regression analysis identified factors associated with prognostication accuracy. A total of 395 predictions (06/2010-07/2012) were made by eight disciplines. Average age was 68 years, 68.3 % of patients were male, and 48.4 % had lung cancer. Median AS was 87 days (95 % CI 66-102 days). Survival was over-estimated 72.4 % (286/395) of the time with r = 0.54 (p < 0.0001) for all predictions across all disciplines. In addition, 30.3 % (36/119) of radiation therapist (RTT) predictions were correct compared to 30.1 % (22/73) of nurses', 28.7 % (43/150) of physicians', and 15.1 % (8/53) of allied health (AH) providers. There were no differences in accuracy by discipline except for the RTT versus AH groups (p = 0.04). Factors most frequently cited as influencing correct predictions were Karnofsky performance status (KPS), extent of disease, and histology. KPS was the only significant variable on multivariate analysis (p a parts per thousand currency signaEuro parts per thousand 0.04). MDT members providing collaborative care for advanced cancer patients utilize similar factors in predicting survival with comparable accuracy.