Comparing the performance of the CARG and the CRASH score for predicting toxicity in older patients with cancer

Abstract Objectives To compare the CARG (Cancer and Aging Research Group) and CRASH (Chemotherapy Risk Assessment Scale for High-Age Patients) score regarding the predictive performance for severe toxicity in older patients with cancer. Methods We recruited patients ≥70 years and applied the CARG and CRASH score before the start of systemic cancer treatment. The CARG predicts severe overall toxicity; the CRASH additionally predicts hematologic and nonhematologic toxicity. We captured ≥ grade 3 toxicity according to Common Terminology Criteria for Adverse Events (CTCAE) from medical records. Predictive performance was assessed using logistic regression and the area under the receiver operating characteristic curve (ROC-AUC). Results The study cohort comprised 120 patients (50% female, mean age 77.2 years, 57% solid tumors). The median of the CARG (range 0–23) and the combined CRASH (range 0–12) were 9 and 8, respectively. 81% of patients experienced toxicity; 67% showed hematologic toxicity. The predictive performance of the CARG and the combined CRASH was similar for overall toxicity (CARG: Odds ratio per unit increase (OR) 1.266, P = .015; ROC-AUC 0.681, P = .010; combined CRASH: OR 1.337, P = .029; ROC-AUC 0.650, P = .032). For hematologic toxicity, the hematologic CRASH was a significant predictor and showed numerically a higher ROC-AUC than the CARG which was not statistically different (CARG: OR 1.048, P = .462; ROC-AUC 0.564, P = .271; hematologic CRASH: OR 1.602, P = .007; ROC-AUC 0.665, P = .005). Conclusion Both scores exhibited similar predictive performance for toxicity in older patients with cancer.