Ethnic-Specific Threshold Analysis and BMI and Waist Circumference Cutoffs for Cardiovascular Disease and Subjective Wellbeing: Results using Data from the UK Biobank.

Abstract

Objectives: We aimed to identify ethnicity-specific BMI and waist circumference cutoffs for cardiovascular disease (CVD) and to define optimal thresholds for CVD risk and subjective wellbeing (SWB) through predictive modelling, to inform precise public health initiatives.

Methods: We used data from 296,767 UK Biobank participants and adjusted logistic and linear regression models for CVD and SWB, respectively, complemented by receiver operating characteristic analysis, to explore optimal risk thresholds of CVD in six different ethnic groups and to calculate ethnicity-specific cutoffs of BMI and waist circumference (WC) to further elucidate the relationships between demographic factors and cardiovascular risk among diverse populations.

Results: The logistic regression model of CVD revealed moderate discriminative ability (AUROC ~ 64-65%) across ethnicities for CVD status, with sensitivity and specificity values indicating the model's predictive accuracy. For SWB, the model demonstrated moderate performance with an AUROC of 63%, supported by significant variables that included age, BMI, WC, physical activity, and alcohol intake. Adjusted-incidence rates of CVD revealed the evidence ethnic-specific CVD risk profiles with Whites, South Asians and Blacks demonstrating higher predicted CVD events compared to East Asians, mixed and other ethnic groups.

Conclusion: Alterations of ethnicity-specific BMI and waist circumference are required to ensure ethnic minorities are provided with proper mitigation of cardiovascular risk, addressing the disparities observed in CVD prevalence and outcomes across diverse populations. This tailored approach to risk assessment can facilitate early detection, intervention and management of CVD, ultimately improving health outcomes and promoting health equity. The moderate accuracy of predictive models underscores the need for further research to identify additional variables that may enhance predictive accuracy and refine risk assessment strategies.