Study Title
Predictors of cardiovascular disease risk: findings from the National Survey of Health and Development Cohort
Principal Investigator
Sumantra Ray
Affiliation
NNEdPro Global Institute for Food, Nutrition and Health
Start Date
January 2022
End Date
November 2025
Study Objective
This study aimed to explore the incremental effect of risk factors and respective biomarkers on CVD events risk estimation models.
Short Abstract
Background: Cardiovascular disease (CVD) remains a significant cause of mortality and morbidity in the UK. CVD and all-cause mortality can be predicted with varying certainty using different models.
Purpose: To explore the incremental effect of risk factors and biomarkers on CVD risk estimation.
Methods: The National Survey for Health and Development (NSHD) birth cohort (2,547 women and 2,815 men; n=5,362) was analysed for relationships between conventional and emerging risk factors and cardiometabolic outcomes (myocardial infarction, stroke) from 1999–2009. Logistic regression and XGBoost models were used. Model fit was assessed using binary error matrices.
Results: For myocardial infarction, significant predictors included alcohol intake (β = –0.029g; p<0.05), BMI (β = 0.095; p<0.05), HbA1c (β = 0.033; p<0.01), TC/HDL ratio (β = 0.459; p<0.0001), waist-to-hip ratio (β = 5.632; p=0.05) and waist circumference (β = 0.031; p<0.01). For stroke, significant predictors were exercise 5+ times (β = –1.47; p<0.01), SBP (β = 0.040; p<0.001), DBP (β = 0.027; p<0.001), pulse pressure (β = 0.033; p<0.01), triglycerides (β = 0.388; p<0.05) and TC/HDL ratio (β = 0.419; p<0.01). Adding BMI, HbA1c and TC/HDL ratio to the baseline model improved F1 score to 0.123. XGBoost initially reduced F1 to 0.079 but improved to 0.103 with added alcohol and waist-hip data.
Conclusion: Clinical and laboratory markers substantially improve risk prediction models for CVD events. Using large cohorts enhances certainty and replicability across populations.
Study Design
Cohort study
Population
The NSHD birth cohort: 2,547 women and 2,815 men followed between 1999 and 2009.
Sample Size
5,362 participants
Inclusion Criteria
Participants in the NSHD birth cohort study.
Exclusion Criteria
Not specified
Intervention/Exposure
Risk factors and biomarkers (BMI, HbA1c, cholesterol, waist measures, blood pressure, triglycerides, physical activity, alcohol intake).
Outcome Measures
Myocardial infarction, stroke, cardiometabolic and vascular outcomes.
Funding Source
Swiss Re Institute
Collaborating Institutions
Not specified
Ethics Approval
Not specified
Publication Status
Not published
Keywords
Cardiovascular disease (CVD), Risk Prediction
Data Collection Methods
Logistic regression and machine learning (XGBoost) modelling using NSHD cohort data (clinical and laboratory measures, lifestyle questionnaires).
Primary Data Availability
Not available
Contact Information