Journal of Geriatric Oncology · Nov 9, 2026
Geriatric assessment domains, frailty indices, and survival in older adults with breast cancer: a systematic review and meta-analysis
By Jeong SA, Kim HJ, Lim SM et al.
Introduction Geriatric assessment (GA) is recommended for older adults with breast cancer, but the prognostic value of specific GA domains and frailty indices has not been quantitatively synthesized. We evaluated the association between impairments in GA domains and survival outcomes and assessed the prognostic value of validated frailty indices as pragmatic alternatives to full comprehensive geriatric assessment (CGA). Materials and methods PubMed, EMBASE, and CENTRAL were searched from inception to July 4, 2025. Observational cohort studies including patients aged ≥65 years with breast cancer and reporting adjusted hazard ratios (HRs) for overall survival (OS) or breast cancer-specific survival (BCSS) according to GA domains or validated frailty indices were included. Risk of bias was assessed using QUIPS. Fixed-effects models were used to pool adjusted HRs comparing impaired or frail versus non-impaired or fit patients. The certainty of evidence was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach. Results Thirteen studies were included. Impairments in physical function, comorbidity, performance status, cognition, and psychological status were each associated with increased all-cause mortality in separate domain-specific analyses, with pooled HRs ranging from 1.41 to 1.87 and low heterogeneity. Frailty indices were associated with worse BCSS (HR 1.61, 95% confidence interval [CI] 1.00–2.21; I² statistic 52.8%) and OS (HR 1.80, 95% CI 1.60–2.01; I² statistic 71.0%). After excluding studies with the highest risk of bias, frailty indices remained associated with worse OS (HR 1.79, 95% CI 1.59–2.00). Discussion Impairments in CGA domains and frailty by validated indices were associated with poorer survival in older patients with breast cancer. Frailty indices may provide a pragmatic alternative when implementation of a full CGA is limited. Registration This review was prospectively registered in PROSPERO (CRD420251087304).