Predicting metabolic syndrome in midlife: a life course analysis using the 1958 British birth cohort

Abstract

Background: Metabolic syndrome (MetS) - 'a clustering of risk factors which includes hypertension, central obesity, impaired glucose metabolism with insulin resistance, and dyslipidaemia' affects nearly a quarter of the world's adult population. Individuals with MetS have higher risk of CVD, T2DM and death from all causes. Identifying those at high risk of MetS via the use of prediction models may guide targeted interventions aimed at reducing the burden of the syndrome. There is a large number of MetS prediction models in the literature, but their usefulness is not known. This makes it difficult for potential users to decide which model to apply in their practice. Therefore, the overall aim of this thesis is to identify and utilise existing risk models to predict MetS in midlife using the 1958 British birth cohort data.The thesis consists of two distinct but inter-related sub-studies. The first study aims to determine the performance of risk models and scores for predicting MetS. The second study aims to develop, and validate existing risk models for predicting MetS in midlife using the 1958 British birth cohort data.Methods: A systematic review was conducted in the first study, which is the first to determine the performance of existing MetS prediction models. The second study analysed the 1958 British birth cohort, which includes 18,558 individuals born in the first week of March 1958. Variables utilised were obtained prospectively at birth, 7, 16, 23 and 45 years. The outcome (MetS) was defined based on the National cholesterol education programme –adult treatment panel III (NCEP-ATP III) clinical criteria. In the first stage, a simple MetS prediction model was developed using multivariate logistic regression and evaluated using measures of discrimination and calibration. In the second stage, an existing model (Chinese MetS risk score), identified during the systematic review, was applied on the 1958 British birth cohort. Performance parameters were assessed, and comparison was made with the model developed in this thesis.Results: In the systematic review, twenty seven studies met the inclusion criteria. The overall performance of the included models was judged to be poor, mainly due to a general inadequacy both in terms of conduct and reporting as well as lack of external validation.ivSimilarly, the developed model has good accuracy (AUROC 0.91 (0.90, 0.92)), is well calibrated (Hosmer-Lemeshow 4.01 (0.856)) and has good internal validity. On the contrary, the Chinese MetS risk score showed poor performance on the host data. This suggests that in its current form, the Chinese MetS risk score may not be suitable to be applied directly on an external population data without being updated.Conclusions: As research in MetS prediction evolves, emphasis may be placed on improving the conduct and reporting of model studies, external validation including comparative studies, and updating the existing ones.

Date of Award	Nov 2020
Original language	English
Awarding Institution	University of Bedfordshire
Supervisor	Yannis Pappas (Supervisor), Gurch Randhawa (Second supervisor) & Dong Pang (Third supervisor)