Comparing frailty prevalence between countries: validation of the Global Burden of Disease study Frailty Index (GBD-FI) in the survey of health, ageing and retirement in Europe

Abstract Background Accurate comparable prevalence proportions are required to better understand the epidemiology of frailty. Estimates in many countries are missing or incomparable. The Global Burden of Disease Frailty Index (GBD-FI) applies the deficit accumulation model to generate frailty scores from items available in the Global Burden of Disease study. Objective To externally validate the GBD-FI. Methods Data were obtained from the Survey of Health Ageing and Retirement in Europe (SHARE). A 20-item modified GBD-FI was compared with established frailty measures: a 70-item frailty index (FI-70), the Clinical Frailty Scale (CFS), Frailty Phenotype (FP) and SHARE-FI. Area under receiver operating characteristic curves (AUC) were fitted to examine diagnostic accuracy for frailty and predictive validity for 2-year mortality. Results In total, 31,624 participants aged ≥50 years from 15 countries were included. Frailty prevalence was 22% using the GBD-FI (ranging from 8% in Switzerland to 41% in Poland). The GBD-FI had good to excellent diagnostic accuracy for frailty, irrespective of approach; the AUC ranged from 0.86 (95% confidence interval: 0.85–0.87) measuring frailty using the CFS to 0.94 (0.93–0.94) with the FI-70. The GBD-FI had similar accuracy for 2-year mortality (AUC 0.71, 0.69–0.74) compared with the CFS (0.73; P = 0.186), FP (0.73; P = 0.392) and SHARE-FI (0.70; P = 0.255) but lower than the FI-70 (0.76; P < 0.001). Conclusion The GBD-FI demonstrated concurrent and predictive validity, suggesting it is a valid measure of frailty. It has the potential to be an efficient, replicable and consistent approach to comparing frailty between countries and regions across time using GBD data.