Optimum cardiovascular risk prediction algorithm for South-Asians - Are WHO risk prediction charts really the right answer?

Dear Editor,

We read with interest the review article by Ofori et al. “Risk assessment in the prevention of cardiovascular disease (CVD) in low-resource settings” and the accompanying editorial by Hariram. Both these articles draw our attention toward the need for ethnic-specific CVD risk scores for South Asian populations and other low middle income countries (LMIC) – an extremely important yet largely ignored issue so far! We congratulate the authors for the same.

Estimation of the risk of future atherosclerotic CVD events is one of the most important initial steps in the management of the patients requiring primary prevention of CVD. Such an estimate allows objective assessment of the ‘seriousness’ of the illness, provides a means to communicate the same to the patient and the patient's family, and most importantly, forms the basis for a number of important therapeutic decisions. Several CVD risk assessment algorithms are currently available for this purpose. However, as these CVD risk scores are based on population-specific epidemiological data, each risk algorithm is applicable only to the population from which it has been derived. Unfortunately, no such CVD algorithm is currently available for Indians and most other LMIC and this remains a major limitation to the delivery of appropriate preventive cardiovascular care in these populations.

To overcome this limitation, the World Health Organization (WHO), in collaboration with the International Society for Hypertension (ISH), published a series of risk prediction charts for different ethnic-geographic regions.3, 4, 5, 6 These risk assessment charts were derived with the help of statistical modeling using extrapolated data about the prevalence of various CVD risk factors in the respective populations. They are simple to use and are available in both lab-based and non-lab based versions. These attributes make them particularly attractive for use in low-resource settings as emphasized by Ofori et al., as well as by the editorial expert. However, it is important to remember that the WHO risk prediction charts have not been validated in prospective studies. Therefore, despite their ease of use, it is still important to document the validity of these risk prediction charts in different population groups before incorporating them into widespread clinical use.

While a large cross-sectional study in South Africans demonstrated the accuracy of WHO risk prediction Charts, their accuracy in other population groups has been rather questionable.8, 9, 10 Selvarajah et al. compared WHO risk prediction charts with several other CVD risk algorithms in a large-scale prospective study among Malaysians and found that the WHO risk prediction charts grossly underestimated the risk. We too have performed two studies in North Indians and have found similar results. The first study included 149 subjects who had no previous CVD and had presented with first acute myocardial infarction (MI). Four risk algorithms were applied in them (WHO risk prediction charts, Framingham risk score, American College of Cardiology/American Heart Association pooled cohort equations, and the 3rd iteration of Joint British Societies’ (JBS3) risk calculator) to determine their predictive accuracy if these patients had presented in the clinic immediately prior to their index event. Of the four risk algorithms, the JBS3 risk score was the most accurate in identifying these acute MI patients as ‘high risk’ whereas the WHO risk prediction charts most underestimated the CVD risk. However, an important limitation of this study was that it had included the subjects who had already had a CVD event and retrospective risk profiling was done in them. Therefore, in the subsequent study, we included a mixed population of subjects who were undergoing primary or secondary prevention of CVD. The same four risk algorithms were applied and were correlated with coronary calcium score (CCS) and carotid intima–media thickness (CIMT). Both CCS and CIMT are established surrogate measures of atherosclerosis, and CCS in particular has been shown to have consistent, strong, and independent predictive value for future CVD risk.11, 12, 13, 14, 15 Once again, we found that JBS3 risk score had the best correlation with CCS and CIMT whereas WHO risk prediction charts had only an inconsistent relation.

Based on the above two studies and the Malaysian study, we believe it would be inappropriate to recommend widespread use of WHO risk prediction charts, at least in South-Asian populations. Although the simplicity of these charts promises to increase uptake of CVD risk assessment in the low-resource settings widely prevalent in these nations, their use is likely to result in gross underestimation of the CVD risk in South Asians. Such underestimation of CVD risk would result in false sense of complacency, which would be clearly undesirable in these populations in which CVD epidemic is burgeoning at the present moment.