Brief cognitive screening as a prognostic tool in geriatric oncology: A Brazilian cohort study

Introduction Cognitive impairment is common in older adults with cancer but often remains undetected due to time constraints and limited specialized resources in routine oncology care. We evaluated whether a 2-min cognitive screener predicts all-cause mortality in this population and explored whether functional dependence mediates this association. Material and Methods We conducted a longitudinal study of consecutive patients aged ≥60 years with cancer who were referred for geriatric oncology assessment at a tertiary outpatient cancer center in Brazil. Cognitive status was assessed using the 10-point Cognitive Screener (10-CS), a brief clinical tool that evaluates orientation, memory, and verbal fluency. Cognitive impairment was defined as 10-CS scores ≤5. We used Cox proportional hazards models to examine the association between 10-CS scores and 3-year all-cause mortality, as well as sociodemographic factors, comorbidities, cancer type, metastasis, and sensory deficits. Harrell's C -index was calculated to assess whether adding 10-CS improved mortality discrimination. Mediation analysis incorporated activities of daily living (ADLs) dependence. Results Among 508 patients (mean age = 78.0 ± 7.1 years; female = 42.5%), 224 (44.1%) had cognitive impairment. Of these, 87 (38.8%) had no previous cognitive impairment documented in their medical records. Three-year mortality was higher among patients with impairment (63.6% vs 37.4%, p < 0.001). After adjustment, cognitive impairment remained strongly associated with mortality (HR = 1.87; 95% CI: 1.38–2.54). Adding 10-CS to the model with sociodemographic, comorbidities, and cancer variables improved mortality discrimination (Harrell's C-index 0.72 vs. 0.68; p = 0.01). The 10-CS stratified mortality across subgroups defined by the absence (40.0% vs. 62.9%; p < 0.001) or presence (25.1% vs. 63.6%; p < 0.001) of previous chart-documented cognitive impairment or dementia. Cognitive impairment was associated with dependence in ADL (adjusted prevalence ratio = 4.01; 95% CI = 2.57–6.25). Disability mediated part of the effect on 3-year all-cause mortality, but 62% remained unexplained, indicating that most of the association was independent of functional decline. Discussion Cognitive impairment evaluated with the 10-CS predicted 3-year all-cause mortality in older adults with cancer. This brief clinical tool provided valuable prognostic information beyond standard oncology measures and may assist clinicians in identifying high-risk patients whose cognitive impairment might otherwise go unrecognized.