Decline in blood cholesterol and incident disability in activities of daily living in older adults: longitudinal evidence from the CHARLS and ELSA cohorts

Background Cholesterol falls with illness, weight loss and poor nutrition in later life, and these states also lead to disability. We examined the level, long-term mean, within-person variability and change of seven blood lipid indices in relation to incident disability in activities of daily living (ADL). Methods We used two prospective cohorts, the English Longitudinal Study of Ageing (ELSA; 4912 adults without ADL difficulty, 996 incident cases) and the China Health and Retirement Longitudinal Study (CHARLS; 10303 adults, 2015–2018). Cox models gave hazard ratios (HR) per standard deviation (SD) of level, long-term mean, variability and decline for each lipid. The decline hypothesis arose from exploratory analysis of CHARLS and was tested in ELSA, with total cholesterol (TC), non-high-density lipoprotein (non-HDL) cholesterol and the TC/HDL cholesterol (TC/HDL-C) ratio as primary endpoints. Results Each SD of decline in total cholesterol was followed by a higher hazard of ADL disability in ELSA (HR 1.09, 95% confidence interval [CI] 1.01–1.19; P = 0.032; n = 4246) and in CHARLS (1.09, 1.00–1.20; n = 5698), pooled 1.09 (1.03–1.16). For non-HDL cholesterol the HRs were 1.09 (1.00–1.19) and 1.12 (1.01–1.23). The TC/HDL-C ratio gave 1.06 (0.98–1.15) in ELSA and did not meet the replication criterion, and the false discovery rate across the three primary endpoints in ELSA was 0.059. Excluding users of lipid-lowering drugs raised the pooled HR for total cholesterol to 1.14 (1.06–1.23). In ELSA the HR fell from 1.25 to 1.09 across adjustment models, and after full adjustment the P for trend across tertiles was 0.54 in ELSA and 0.45 in CHARLS. The pooled HR per SD higher level of total cholesterol was 0.95 (0.90–1.00). In ELSA, adding the decline to covariates changed the C-index from 0.755 to 0.756. Conclusions Each SD of decline in total cholesterol was followed by a 9% higher hazard of ADL disability in both cohorts, and each SD of decline in non-HDL cholesterol by 9% in ELSA and 12% in CHARLS. A fall in cholesterol may mark declining health, and repeated measures of albumin, inflammation and body composition are needed to test this.