Accuracy of the G-8 geriatric-oncology screening tool for identifying vulnerable elderly patients with cancer according to tumour site: The ELCAPA-02 study

Abstract Background/Objective G-8 screening tool showed good screening properties for identifying vulnerable elderly patients with cancer who would benefit from a comprehensive geriatric assessment (CGA). We investigated whether tumour site and metastatic status affected its accuracy. Materials and Methods Design: Cross-sectional analysis of a prospective cohort study. Setting: Geriatric-oncology clinics of two teaching hospitals in the urban area of Paris. Participants: Patients aged 70 or over (n = 518) with breast (n = 113), colorectal (n = 108), urinary-tract (n = 89), upper gastrointestinal/liver (n = 85), prostate (n = 69), or other cancers (n = 54). Measurements: Reference standard for diagnosing vulnerability was the presence of at least one abnormal test among the Activities of Daily Living (ADLs), Instrumental ADL, Mini-Mental State Examination, Mini Nutritional Assessment, Cumulative Illness Rating Scale-Geriatrics, Timed Get-Up-and-Go, and Mini-Geriatric Depression Scale. Sensitivity, specificity and likelihood ratios of G-8 scores ≤ 14 were compared according to tumour site and patient characteristics. Results Median age was 80; 48.2% had metastases. Prevalence of vulnerability and abnormal G-8 score was 84.2% (95% confidence interval [95% CI], 81–87.3) and 79.5% (95% CI, 76–83). The G-8 was 86.9% sensitive (95% CI, 83.4–89.9) and 59.8% specific (95% CI, 48.3–70.4). G-8 performance varied significantly (all p values < 0.001) across tumour sites (sensitivity, 65.2% in prostate cancer to 95.1% in upper gastrointestinal/liver cancer; and specificity, 23.1% in colorectal cancer to 95.7% in prostate cancer) and metastatic status (sensitivity and specificity, 93.8% and 53.3% in patients with metastases vs. 79.5% and 63.3% in those without, respectively). Differences remained significant after adjustment on age and performance status. Conclusion These G-8 accuracy variations across tumour sites should be considered when using G-8 to identify elderly patients with cancer who could benefit from CGA.