Background Obesity, hypertension and diabetes are known conditions that increase the risk of cardiovascular disease (CVD). An accurate estimate of the risk of CVD requires blood tests. The aim of this work is to establish which non-invasive measures, without the blood tests, would be suitable for an early determination of the risk of CVD. Some ethnic groups such as south Asians have higher risk of CVD and may have a different risk profile.

Methods A purposive cohort of 1087 patients was recruited to establish non-invasive risk factors. The cohort was enriched for south Asians in order to explore ethnicity and in those with type 2 diabetes. As well as the usual measurements of body-mass-index (BMI) and systolic-brachial-blood pressure, systolic-ankle-blood pressures, waist and waist-to-height-ratio (WHtR) were also collected. Treated hypertension status, age, ethnicity, and sex were taken from patient records. Classification trees were employed to ascertain potential risk factors for CVD and also for diabetes which enable handling of collinear measures such as BMI and WHtR.

Results The classification tree for CVD showed that within our cohort, treated hypertension, age, diabetic status, and ankle-brachial-index (systolic-ankle-blood pressures/systolic-brachial-blood pressure) were strong predictors. For diabetes the strong predictors were treated hypertension, age, ethnicity, and WhtR. Although available, BMI did not feature as a risk factor since WHtR shows greater discrimination.

Conclusion When using non-invasive measures alone, waist-to-height ratio is shown as the more sensitive measure of obesity when predicting diabetes and ankle-brachial-index is more sensitive predictor than systolic-brachial-blood-pressure for CVD after treated hypertension-status.