RESUMEN
OBJECTIVE: We aimed to construct and internally validate a frailty risk prediction model in older adults with lung cancer. METHOD: In total, 538 patients were recruited in a grade A tertiary cancer hospital in Tianjin, and patients were randomly divided into the training group (n = 377) and the testing group (n = 166) at a ratio of 7:3. The Frailty Phenotype scale was used to identify frailty and logistic regression analysis was used to identify the risk factors and establish a frailty risk prediction model. RESULTS: In the training group, logistic regression showed that age, fatigue-related symptom cluster, depression, nutritional status, D-dimer level, albumin level, presence of comorbidities, and disease course were independent risk factors for frailty. The areas under the curve (AUCs) of the training and testing groups were 0.921 and 0.872, respectively. A calibration curve of P = 0.447 validated model calibration. The decision curve analysis demonstrated greater clinical benefit when the threshold probability was >20%. CONCLUSION: The prediction model had a favorable prediction power for determining the risk of frailty, contributing to the prevention and screening of frailty. Patients with a frailty risk score of more than 0.374 should be regularly monitored for frailty and receive personalized preventive interventions.
