Measuring frailty in older rural South African population - Construction and validation of a Frailty Index (360G-Wellcome-211749_Z_18_Z)

Data from the HAALSI cohort from rural South Africa will be used to achieve 3 aims: To derive a Rockwood deficit-accumulation Frailty Index (FI) in this population To test whether the FI predicts mortality in the HAALSI population To compare results from the Fried frailty score and the FI A review of existing literature will inform the selection of a minimum of 30 variables for inclusion in the Frailty Index. Examples include (but are not limited to): disease diagnoses (e.g. angina, HIV), functional impairments (e.g. low gait speed, low grip strength), or biochemical/haematological derangements (e.g. low haemoglobin). Absence of a deficit scores zero; presence of a deficit scores 1. Cox proportional hazards models will be used to examine the association between the FI and time to death in the HAALSI data; unadjusted models and models adjusted for age and sex will be developed. Finally, the ability of the FI to predict 1 year mortality will be compared with the ability of the Fried phenotypic score to predict 1 year mortality in the HAALSI population. Receiver-operator characteristic curves will be developed for each score, and the c-statistic for each score compared.