Objective To extrapolate an integrated natural history model that predicts longitudinal changes in metabolic risk factors into older age using the English Longitudinal Study of Ageing (ELSA). Method Growth curve modelling was employed to estimate trajectories for multiple metabolic risk factors from the ELSA dataset, allowing for random variation between individuals. The ELSA models relate age trends in BMI, systolic blood pressure, HbA1c, Total and HDL cholesterol to gender, ethnicity, smoking, and deprivation using Stata 13. The predicted metabolic trajectories in old age ELSA statistical models were simulated and compared with previous analysis of the Whitehall II cohort. Results The statistical models revealed that HbA1c, systolic blood pressure and HDL cholesterol are increasing with age. The BMI curve shows that on average BMI declines in later life and the analysis finds on average total cholesterol does not change with age. Conclusions A simulation study comparing the ELSA data with analyses of the Whitehall II study suggest that individuals have different trajectories in older age. Different statistical models are needed to simulate metabolic trajectories in mid-life and older ages.

Keywords: Dementia, Prevention, Longitudinal analysis, Risk factors