Introduction and Objectives
One-fourth of the world’s population is in South Asia1 and South Asian migrants are found worldwide. Cardiovascular diseases (CVD) are the commonest cause of death globally and three-quarters of CVD deaths take place in low- and middle-income countries 2. Prevention of CVD is the most cost-effective especially for developing countries like South Asian countries. A total risk-based approach is recommended for CVD prevention3. Cardiovascular (CV) risk prediction is an essential entity in this approach. However, there are no CV risk prediction models developed in South Asians or South-East Asians. Therefore, different risk models developed in white Caucasians; e.g. Framingham risk scores (FRS)4 or developed by modelling approach; e.g. World health organization/International society of hypertension (WHO/ISH) risk charts 5; are being used to risk predict South Asians. Both the standard models (FRS with- with-cholesterol and WHO/ISH charts with-cholesterol)and the low-information models made up without cholesterol values in equation ( FRS with-BMI and WHO/ISH without-cholesterol) are being used depending on the availability of recourses. However, the best risk score in risk prediction of South Asians is not known. Only a very few studies have looked into this question and the literature is inconclusive and also the available studies are difficult to be compared 6.
South Asians have a high risk of CVD compared to other Asians and white Caucasians 7. They have a different CVD profile; high risk of CVD than Whites in the UK 8,9 and America10, a rising trend of CVDs despite CVDs having a declining trend in the west 11, more strokes than coronary heart diseases and CVDs at younger ages12,13. Furthermore, they have different genetics and have high prevalences of vascular risk factors like diabetes mellitus and metabolic syndrome than white Caucasians 14-17. Therefore, the risk prediction models developed of Western cohorts might not be accurately predicting CV-risk of South Asians.
Therefore, we compared 10-year general-cardiovascular risk predictions of four commonly used models in South Asians; Framingham BMI-based, Framingham cholesterol-based, WHO/ISH with-cholesterol and WHO/ISH without-cholesterol for agreement in a sample of Sri Lankans without prior CVDs.
Methods
All consecutive adults attending a non-communicable disease clinic at the Faculty of Medicine, University of Kelaniya, Sri Lanka were screened in 2019 over one year. Patients with vascular risk factors and having complete data to calculate CV-risk scores but without a past history of CVD were enrolled in this study. Data on vascular risk factors were collected using an interviewer-administered questionnaire and referring clinic records. Height and weight were measured at the clinic. Two blood pressure measurements were done in the left arm 5 minutes apart in seated position with a mercury sphygmomanometer. 10-year CV-risk predictions of all participants were calculated using four models; FRS BMI-based, FRS cholesterol-based, WHO/ISH charts without and with-cholesterol. 10-year risk predictions of developing a fatal or non-fatal CVD were calculated using the formulas; Framingham 10-year general CVD risk 18 and WHO/ISH charts meant for South-East Asian Region- B (SEAR- B) 19. FRS was calculated using age, systolic blood pressure (SBP), antihypertensive use, current smoking status, diabetes status, and body mass index (BMI) or total cholesterol (TC) and high-density lipoprotein (HDL) level. WHO/ISH risk predictions were calculated using age, SBP, current smoking status, diabetes status and additionally with total cholesterol for WHO/ISH with-cholesterol estimate. BMI was calculated with weight and height. The mean of the two blood pressure measurements made at the clinic was used as the SBP. The most recent recorded TC and HDL values within the previous year were used in risk calculations. All current smokers and those who quit smoking less than 1 year before the assessment were considered current smokers. Persons with self-reported diabetes mellitus cross-checked with medical records or taking insulin or oral hypoglycaemic drugs were considered as having diabetes mellitus according to the World Health Organization, criteria20. People with self-reported hypertension cross-checked with medical records, physician-diagnosed hypertension, or taking antihypertensive medications were defined as hypertension, according to the Joint National Committee (JNC) VII criteria21. Past history of hyperlipidaemia was defined as someone with physician-diagnosed hyperlipidaemia in medical records on the National Cholesterol Education Program III criteria22. Patients were categorised into two risk groups using risk estimates; low risk (<20% ) and high risk (≥ 20%) risk.
IBM SPSS statistics version 22.0 was used for analysis. Continuous variables were reported as means with standard deviation (SD) or 95% confidence intervals, and categorical variables were reported as percentages. The significance level was set at p <0.05. Mean Framingham risks of BMI-based and cholesterol-based models were compared using the paired sample Students t-test. Risk predictions of the models were compared for agreement across risk categories with Cohen’s kappa coefficient (κ ). The strength of agreement was interpreted as,κ : ≤ 0.20 = poor, 0.21–0.40 = fair, 0.41–0.60 = moderate, 0.61–0.80 = good and 0.81–1.00 = very good 23-25.
Ethics approval was obtained from the Ethics Review Committee, Faculty of Medicine, University of Kelaniya, Sri Lanka. Informed consent of all the patients was obtained.