The Prevalence of Frailty in Cancer Patients and Mortality Prediction With a Novel Frailty Index Based Clinical Algorithm-A Multicenter, Prospective, Observational Study
Abstract Purpose To investigate the prediction capacity and status of frailty in Chinese cancer patients in national level, through establishing a novel prediction algorithm. Methods The percentage of frailty in different ages, provinces and tumor type groups of Chinese cancer patients were revealed. The predictioncapacity of frailty on mortality of Chinese cancer patients was analyzed by FI-LAB that is composed of routine laboratory data from accessible blood test and calculated as the ratio of abnormal factors in 22 variables. Establishment of a novel algorithm MCP(mortality of cancer patients)to predict the five-year mortality in Chinese cancer patients was accomplished and its prediction capacity was tested in the training and validation sets using ROC analysis. ResultsWe found that the increased risk of death in cancer patients can be successfully identified through FI-LAB. The univariable and multivariable Cox regression were used to evaluate the effect of frailty on death. In the 5-year follow-up, 20.6% of the 2959 participants (age = 55.8 ± 11.7 years; 43.5% female) were dead while the mean FI-LAB score in baseline was 0.23 (standard deviation = 0.13; range = 0 to 0.73).Frailty (after adjusting for gender, age, and other confounders) could be directly correlated with increased risk of death, with a hazard ratio of 12.67 (95% confidence interval CI: 7.19, 22.31) in comparison with those without frailty. In addition, MCP algorithm presented an area under the ROC (AUC) of 0.691 (95% CI: 0.659-0.684) and 0.648 (95% CI: 0.613-0.684) in the training and validation set, respectively. Conclusion Frailty is common in cancer patients and FI-LAB has high prediction capacity on mortality. The MCP algorithm is a good supplement for frailty evaluation and mortality prediction in cancer patients.