Original Research

Risk Prediction of Cardiovascular Disease in the Asia‑Pacific Region: The SCORE2 AsiaPacific Model

Register or Login to View PDF Permissions
Permissions× For commercial reprint enquiries please contact Springer Healthcare: ReprintsWarehouse@springernature.com.

For permissions and non-commercial reprint enquiries, please visit Copyright.com to start a request.

For author reprints, please email rob.barclay@radcliffe-group.com.
Information image
Average (ratings)
No ratings
Your rating

Abstract

Background and aims: To improve upon the estimation of 10-year cardiovascular disease (CVD) event risk for individuals without prior CVD or diabetes mellitus in the Asia-Pacific region by systematic recalibration of the SCORE2 risk algorithm. Methods: The sex-specific and competing risk-adjusted SCORE2 algorithms were systematically recalibrated to reflect CVD incidence observed in four Asia-Pacific risk regions, defined according to country-level WHO age- and sex-standardised CVD mortality rates. Using the same approach as applied for the original SCORE2 models, recalibration to each risk region was completed using expected CVD incidence and risk factor distributions from each region. Results: Risk region-specific CVD incidence was estimated using CVD mortality and incidence data on 8,405,574 individuals (556,421 CVD events). For external validation, data from 9 560 266 individuals without previous CVD or diabetes were analysed in 13 prospective studies from 12 countries (350,550 incident CVD events). The pooled C-index of the SCORE2 Asia-Pacific algorithms in the external validation datasets was 0.710 (95% CI [0.677–0.744]). Cohort-specific C-indices ranged from 0.605 (95% CI 0.597–0.613) to 0.840 (95% CI 0.771–0.909). Estimated CVD risk varied several-fold across Asia-Pacific risk regions. For example, the estimated 10-year CVD risk for a 50-year-old non-smoker, with a systolic blood pressure of 140 mmHg, total cholesterol of 5.5 mmol/l, and high-density lipoprotein cholesterol of 1.3 mmol/l, ranged from 7% for men in low-risk countries to 14% for men in very-high-risk countries, and from 3% for women in low-risk countries to 13% for women in very-high-risk countries. Conclusion: The SCORE2 Asia-Pacific algorithms have been calibrated to estimate 10-year risk of CVD for apparently healthy people in Asia and Oceania, thereby enhancing the identification of individuals at higher risk of developing CVD across the Asia-Pacific region.

Disclosure:The authors included in the SCORE2 Asia-Pacific writing group have nothing to disclose relevant to the current project.

Received:

Accepted:

Published online:

Data Availability Statement:

Data used for the current study are available upon reasonable request and approval of the individual cohorts or collaborative groups.

Ethics Approval Statement:

Relevant ethical approval and participant consent were already obtained in all studies that contributed data to this work (Supplementary Material).

Funding:

No study-wide funding was used for the SCORE2-ASIA project, however, the individual researchers working on the SCORE2 Asia-Pacific model and data sources used for the project did have specific funding sources to mention: HL was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea Ministry of Science and ICT [grant number 2022R1F1A1066181]. SHJH is supported by a grant from the Dutch Heart Foundation [grant number 03-006-2023-0095]. DJM is supported by National Health and Medical Research Council Investigator Grants. ELMB is supported by a National Health and Medical Research Council project grant and Australian Stroke and Heart Research Accelerator Centre Targeted Translation Research Accelerator funding. Also, for funding or logistical support, we are grateful to: National Health and Medical Research Council (NHMRC grants 233200 and 1007544), Australian Government Department of Health and Ageing, Abbott Australasia Pty Ltd, Alphapharm Pty Ltd, Amgen Australia, AstraZeneca, Bristol-Myers Squibb, City Health Centre-Diabetes Service-Canberra, Department of Health and Community Services—Northern Territory, Department of Health and Human Services—Tasmania, Department of Health—New South Wales, Department of Health—Western Australia, Department of Health—South Australia, Department of Human Services—Victoria, Diabetes Australia, Diabetes Australia Northern Territory, Eli Lilly Australia, Estate of the Late Edward Wilson, GlaxoSmithKline, Jack Brockhoff Foundation, Janssen-Cilag, Kidney Health Australia, Marian & FH Flack Trust, Menzies Research Institute, Merck Sharp & Dohme, Novartis Pharmaceuticals, Novo Nordisk Pharmaceuticals, Pfizer Pty Ltd, Pratt Foundation, Queensland Health, Roche Diagnostics Australia, Royal Prince Alfred Hospital, Sydney, Sanofi Aventis, sanofisynthelabo, and the Victorian Government’s OIS Program. The Singapore Multi-Ethnic Cohort Phase 1 is supported by individual research and clinical scientist award schemes from the National Medical Research Council (NMRC) and the Biomedical Research Council (BMRC) of Singapore, and infrastructure funding from the Singapore Ministry of Health (Population Health Metrics Population Health Metrics and Analytics PHMA), National University of Singapore and National University Health System, Singapore. LIFECARE—this study was partially funded by Pfizer Inc. through an investigator-initiated grant. The Philippines subcohort was additionally funded by: the Department of Health; the Philippine Council for Health Research and Development; Diabetes Philippines; the Philippine Society of Hypertension; the Philippine Heart Association; the Philippine Lipid and Atherosclerosis Society; and LRI Therapharma. The Thailand subcohort was additionally funded by: the Faculty of Medicine Ramathibodi Hospital, Mahidol University; the Thailand Research Fund; the National Research Council of Thailand; the Electricity Generating Authority of Thailand; the Office of the Higher Education Commission; and the project for Higher Education Research Promotion and National Research University Development. The Indonesian subcohort received additional funding from PT. Kalbe Farma. There were no additional sources of funding for the Malaysian subcohort. The CHERRY study was supported by the National Key Research and Development Program of China (grant number 2020YFC2003503) and the National Natural Science Foundation of China (NSFC) (grant number 82373662).

Acknowledgements:The AusDiab study, initiated and coordinated by the International Diabetes Institute, and subsequently coordinated by the Baker Heart and Diabetes Institute, gratefully acknowledges the support and assistance given by: K Anstey, B Atkins, B Balkau, A Cameron, S Chadban, M de Courten, D Dunstan, A Kavanagh, S Murray, N Owen, K Polkinghorne, T Welborn, and all the study participants. All named authors contributed equally. Authors from the SCORE2 Asia-Pacific collaborators, ESC CRC, AFC, and APSC are listed at the end of the manuscript. This article has been co-published with permission in the European Heart Journal, the Journal of Asian Pacific Society of Cardiology and the ASEAN Heart Journal. The articles are identical except for minor stylistic and spelling differences in keeping with each journal’s style. Any of the citations can be used when citing this article.

Correspondence Details:Sofian Johar: sofian.johar@moh.gov.bn and Steven HJ Hageman: S.H.J.Hageman-4@umcutrecht.nl

Open Access:

This work is open access under the CC-BY-NC 4.0 License which allows users to copy, redistribute and make derivative works for non-commercial purposes, provided the original work is cited correctly.

Graphical Abstract

Graphical Abstract

Article image

Introduction

Cardiovascular diseases (CVDs), which include coronary heart disease and stroke, are the most common fatal non-communicable diseases globally, responsible for an estimated 18.6 million deaths in 2019.1 Guidelines recommend the use of risk prediction models to enhance healthcare and population-wide prevention. These models integrate information on several CVD risk factors and typically estimate individual risk over a 10-year period. The goal is to identify people at higher risk of CVD who benefit most from preventive action. In 2021, the Systematic COronary Risk Evaluation 2 (SCORE2) was published and implemented in the 2021 European Society of Cardiology (ESC) CVD prevention guidelines.2,3 Improvements of the SCORE2 algorithms in comparison to its predecessors include competing risk adjustment and systematic recalibration using aggregate data.

While age-adjusted CVD mortality rates are higher in many highly populated Asian countries in comparison to other parts of the world,4 most CVD risk prediction algorithms have been developed and validated solely in Western countries.5 Locally derived risk prediction models are not widely available and used, with a few exceptions, such as the CHINA-PAR risk model or the Japanese JALS risk score.6,7 Risk prediction models developed in Western populations can provide useful tools for risk stratification in Asian populations, but first need recalibration (statistical adjustment) to reflect important differences in risk factor distributions and CVD incidence patterns between Asian and Western populations. Adequate recalibration to the Asia-Pacific region’s clinical practice is scarce, with the exception of the World Health Organization (WHO) risk charts and Globorisk.8,9

The aim of the current project is to recalibrate the SCORE2 risk algorithms, tailoring them to age- and sex-specific CVD incidence and risk factor distributions observed across the Asia-Pacific region. The recalibrated SCORE2 Asia-Pacific algorithms will provide a more accurate means of 10-year CVD risk estimation for individuals without prior CVD or diabetes mellitus in these populations.

Figure 1: Study Design

Article image

Methods

Study Design

The SCORE2 Asia-Pacific project involved multiple data sources (Figure 1). First, to adapt risk prediction models to each Asian region, the model was systematically recalibrated to several Asia-Pacific risk regions, using the same methods as have been used for the original SCORE2 recalibration. These methods were based on aggregate data on contemporary age- and sex-specific incidence and risk factor distributions.2 Second, the external performance was assessed by performing external validation in individual-participant data using independent data sources from all Asia-Pacific risk regions. Third, to validate whether the original SCORE2 coefficients are appropriate for Asia-Pacific populations, the SCORE2 coefficients were compared to those of locally derived models. Fourth, the variation of CVD risk across Asia-Pacific regions was illustrated using data from contemporary populations. Last, to illustrate the variation of CVD risk in all Asian-Pacific countries, we applied the model to simulated data based on contemporary populations and local risk factor levels.

The SCORE2 Asia-Pacific algorithms use the coefficients as derived in the original SCORE2 algorithms.2 As described in the SCORE2 paper, these coefficients were derived using individual-participant data from the 44 cohorts included in the Emerging Risk Factor Collaboration, and the UK Biobank.10,11 The sex-specific, competing risk-adjusted algorithms included the following predictors: age, current smoking, history of diabetes mellitus, systolic blood pressure (SBP), and total and high-density lipoprotein (HDL)-cholesterol, as well as age-interactions for all included predictors to account for declining relative associations with CVD occurrence with increasing age.9 While the SCORE2 risk models are not intended for use in individuals with diabetes, participants with a history of diabetes were included at the model derivation stage (with appropriate adjustment for diabetes status), since it was not possible to exclude people with diabetes from population-level mortality statistics and risk factor data used in recalibration as these were only available on aggregate-level. Coefficients of predictors have been shown to be stable over time and geographic region.9

Data Sources and Procedures

For recalibration of the algorithms, we obtained country-specific CVD mortality rates reported by the WHO’s Global Health Estimates (GHE) 2019,12,13 and converted these to estimated fatal and non-fatal CVD incidence by using age- and sex-specific multipliers. Risk region-specific multipliers were obtained by combining the multipliers observed in the Singhealth dataset (Singapore), Korean National Health Insurance Service (NHIS) (South Korea)14 the CHinese Electronic health Records Research in Yinzhou (CHERRY) study (China),15 Brunei Healthcare Information Management System (BruHIMS) (Brunei Darussalam), and the Health Checks Ubon Ratchathani study (HCUR) (Thailand). Original SCORE2 multipliers were included in the modelling process to get more stable estimates of the multiplier’s age-slopes (Supplementary Methods).16 Details of these data sources and methods are provided in Supplementary Table 1 and Supplementary Appendix 1. Age-specific and sex-specific risk factor values (SBP, total and HDL cholesterol, diabetes prevalence, smoking status) were obtained from the Non-Communicable Disease Risk Factor Collaboration (NCD-RisC).17,18 A visual representation of the underlying data and steps of the recalibration process are presented in Supplementary Figures 1–5.

For external validation of the algorithms, we included 12 independent data sources that did not contribute to the model derivation, although some sources contributed to multiplier derivation as part of the recalibration process as well as to the external validation. Details of the cohorts are provided in Supplementary Appendix 1 and Supplementary Table 2.15,16,19–21

In alignment with the original SCORE2 models, the target population is individuals aged 40–69 years without prior CVD or diabetes mellitus and the primary outcome estimated by the SCORE2 Asia-Pacific algorithms was defined as a composite of cardiovascular mortality, non-fatal myocardial infarction, and non-fatal stroke.2 Cardiovascular disease mortality was defined as death due to coronary heart disease, heart failure, stroke, and sudden death.22 Follow-up was until the first non-fatal myocardial infarction, non-fatal stroke, death, or end of the registration period. Deaths from non-CVD were treated as competing events. Details of the different ICD-10 codes included in both the fatal and non-fatal components of the endpoint are provided in Supplementary Table 3.

Statistical Analysis

Details of statistical analysis and recalibration process are provided in Supplementary Methods. Risk models were recalibrated to risk regions using age- and sex-specific mean risk factor levels and CVD incidence rates.23 All countries in the Asia-Pacific region were grouped into four risk regions according to their most recently reported WHO’s GHE age- and sex-standardised overall CVD mortality rates per 100 000 population (ICD 10 chapters IX, I00–I99, Figure 2 and Supplementary Table 4).12 Using the same cut-offs as the European SCORE2, the four groupings were: low risk (<100 CVD deaths per 100,000), moderate risk (100 to <150 CVD deaths per 100,000), high risk (150 to <300 CVD deaths per 100,000), and very high risk (≥300 CVD deaths per 100,000). Incidence rates were estimated by rescaling region-specific CVD mortality rates, by applying age-, sex-, and region-specific multipliers, estimated in contemporary representative cohorts.

Figure 2: Risk regions Based on Age and Sexstandardised CVD Mortality Rates from the Global Health Estimates

Article image

We assessed discrimination using external validation cohorts by calculating Harrell’s C-index, adjusted for competing risks.24 Comparison of SCORE2 Asia-Pacific and WHO risk charts (laboratory-based model)9 was assessed using the respective regions for both algorithms. No direct comparison in terms of discrimination was made to the European SCORE2, as the recalibration of the model does not alter the model coefficients and has therefore little impact on the discrimination. The appropriateness of the original SCORE2 coefficients for use in the Asia-Pacific clinical practice was further validated by repeating the SCORE2 derivation process in several Asia-Pacific prospective cohort studies. The locally derived coefficients were visually compared to the original SCORE2 coefficients to identify any substantial geographical heterogeneity. This visual inspection aimed to detect clear patterns indicating potentially different predictor effects in Asian populations, which were not identified in prior risk scores, such as the WHO CVD risk charts.9 We did not assess calibration in most of our external validation cohorts, as the incidence in the cohorts is likely not nationally representative, due to healthy participant bias or the fact that these are non-contemporary cohorts.25 Therefore, model calibration was only assessed in cohorts deemed approximately nationally representative, which were the NHIS (South Korea), the NHG/NUHS health cluster (Singapore), the HCUR (Thailand), and BruHims (Brunei Darussalam). Calibration was assessed by plotting the predicted SCORE2 Asia-Pacific risk per 5-year age group versus the observed cumulative CVD incidence as this best reflects our methods of recalibration. No formal statistical testing was performed on the calibration because the modified Nam–D’Agostino has no extension to the competing risk setting and because these tests are inherently power dependent. The calibration of the SCORE2 Asia-Pacific model was also compared to the calibration of the WHO CVD risk charts (laboratory-based version) using visual comparison in 5-year age groups.

To compare the proportion of the population at different levels of CVD event risk according to the SCORE2 Asia-Pacific algorithms, predicted risk distributions were simulated using age- and sex-specific risk factor value means and prevalences from NCD-RisC and risk factor correlation structures observed in NHIS cohort.

Approaches used to handle missing data are described in the Supplementary Methods. We adopted analytical approaches and reporting standards recommended by the PROBAST guidelines26 and TRIPOD.27 Analyses were performed with R-statistic programming (version 4.3.2, R Foundation for Statistical Computing) and Stata (version 15.1, StataCorp). The study was designed and completed by the SCORE2 Asia-Pacific Working Group in collaboration with the ESC Cardiovascular Risk Collaboration.

Figure 3: SCORE2 AsiaPacific Risk Charts for the Prediction of 10year Risk in Four AsiaPacific Risk Regions

Article image

Figure 3: Continued

Article image

Figure 3: Continued

Article image

Figure 3: Continued

Article image

Results

Regional sex- and age-specific multipliers for conversion of CVD mortality rates to incidence rates including non-fatal events involved 8,405,574 individuals (556,421 CVD events, Supplementary Table 2). Multipliers were somewhat higher in women than in men and decreased with age in a similar pattern as was seen for European multipliers.2 Similarly, multipliers were lower in the higher risk regions compared to low/moderate-risk regions.

The SCORE2 Asia-Pacific charts for CVD risk estimation in four Asia-Pacific risk regions are presented in Figure 3. For practical and presentational purposes, the charts are displayed according to non-HDL cholesterol rather than total cholesterol and HDL cholesterol. The estimated absolute risk for a given age and combination of risk factors differed substantially across regions as a result of recalibration. For example, the estimated 10-year CVD risk for a 50-year-old non-smoker, with a SBP of 140 mmHg, total cholesterol of 5.5 mmol/l, and HDL cholesterol of 1.3 mmol/l, ranged from 7% for men in low-risk countries to 14% for men in very-high-risk countries, and from 3% for women in low-risk countries to 13% for women in very-high-risk countries (Supplementary Figure 6). Given the same risk factor profiles, risks predicted by SCORE2 Asia-Pacific were generally higher in comparison to the original SCORE2 (Supplementary Figure 7).

There was no substantial geographical heterogeneity between the European SCORE2 coefficients and risk factor effects in Asian populations (Supplementary Figure 8). External validation of risk algorithms was completed using data from 9,560,266 individuals without previous CVD or diabetes in 13 prospective studies from 12 Asia-Pacific countries (350,550 CVD events were observed). C-indices showed moderate-to-good discrimination in all regions, with an overall pooled C-index of 0.710 (95% CI [0.677–0.744]; Figure 4).

Figure 4: Cindex

Article image

Cohort-specific C-indices ranged from 0.605 (95% CI [0.597–0.613]) to 0.840 (95% CI [0.771–0.909]). The C-index for the SCORE2 Asia-Pacific algorithms was broadly similar for men and women and in each of the four risk regions (Supplementary Figures 9 and 10). In comparison to the WHO CVD risk charts, SCORE2 Asia-Pacific showed comparable risk discrimination (difference in C-index: −0.003, 95% CI [−0.034, 0.028]) (Supplementary Figure 11). The calibration of the SCORE2 Asia-Pacific algorithms is shown in Supplementary Figure 12. In NHG/NUHS from Singapore and HCUR from Thailand, the predicted risks matched the observed risks well, whereas in BruHims from Brunei Darussalam and NHIS from South Korea, predicted risks were higher than observed risks. Observed risks were best matched from the SCORE2 Asia-Pacific algorithms in the HCUR study, and in BruHims, the SCORE2 Asia-Pacific model overestimated and the WHO CVD risk charts underestimated the observed incidence. Both models had similar performance in the NHIS and NHG/NUHS studies (Supplementary Figure 13). Predicted risks from the SCORE2 Asia-Pacific model also generally better matched the observed incidence in comparison to the original SCORE2 model (Supplementary Figure 14).

When the recalibrated SCORE2 Asia-Pacific algorithms were applied to simulated data representing populations from each risk region, the proportion of individuals aged 40–69 years with an estimated risk >10% varied by region, from 38% in the low-risk region to 92% in the very-high-risk region in men and from 0.8% to 87%, respectively, in women, with these proportions increasing with age, as would be expected (Figure 5).

Figure 5: Distribution of 10year CVD Risk

Article image

Discussion

In the current study, we present the SCORE2 Asia-Pacific 10-year CVD risk estimation algorithms, an adaptation of the SCORE2 algorithms to the region, sex-, and age-specific CVD incidence and risk factor characteristics of Asia-Pacific populations (Graphical Abstract). The SCORE2 algorithms are the recommended CVD prediction model in the 2021 European guidelines on CVD prevention in clinical practice. By extending the SCORE2 algorithms to the Asia-Pacific population, we enhance the identification of individuals at higher risk of developing CVD across the Asia-Pacific region.

Several country-specific cardiovascular prediction models are being used in several Asia-Pacific countries, such as the China-PAR model or the Japanese JALS risk score.6,7 In addition, the WHO CVD risk charts were recalibrated to several Asia-Pacific regions, which were mostly selected based on geographical location rather than expected CVD incidence as was used for SCORE2 Asia-Pacific. The WHO CVD risk charts consisted of a separate stroke and coronary heart disease endpoint, recalibrated separately, whereas the SCORE2 Asia-Pacific model had a simpler design using a single composite outcome. A direct comparison to the WHO CVD risk charts showed similar discriminatory performance of the SCORE2 Asia-Pacific algorithms, reflecting the largely similar derivation data and predictors and indicating that separate recalibration of the endpoints does not make a substantial difference to discrimination. Neither of the models had separate predictors for different stroke aetiologies. Future models could explore whether accounting for differences in stroke aetiology may improve prediction accuracy.

In terms of calibration, the SCORE2 Asia-Pacific and WHO CVD risk charts had similar performance in two external validation data sources, whereas the SCORE2 Asia-Pacific model showed better calibration in the HCUR data. In the BruHims data, the SCORE2 Asia-Pacific model overestimated the predicted risks, whereas the WHO CVD risk charts underestimated the risks. A limitation of this data source is the follow-up duration of 7 years. Because people get older, risks are highest in the last part of the 10-year duration. Validation at 7 years may therefore have contributed to overestimation of CVD risks of the SCORE2 Asia-Pacific model. Another limitation in this dataset was the high number of missing data, which could be one explanation of diminished risk factors associations and low discrimination of both models. However, it is unclear in which direction this might affect model calibration.

The clinical performance of risk prediction models depends importantly on the differing ability to predict the correct level risk in the target population (i.e. extent of ‘calibration’).23 Previous studies validating the European SCORE2 model in the Asia-Pacific region found that it overestimated risk in a Korean population and variably under- or overestimated risk in subgroups of Malaysian individuals.28,29 Therefore, we ensured that the SCORE2 Asia-Pacific algorithms are now also well-calibrated in Asia-Pacific populations by adapting the SCORE2 algorithms to contemporary Asia-Pacific CVD incidence rates. With this, the SCORE2 Asia-Pacific model is the first model available that has been recalibrated to several regions that were grouped according to age standardised CVD mortality rates. On top of this, the SCORE2 Asia-Pacific has several other advantages in comparison to existing alternatives.

First, the SCORE2 Asia-Pacific model accounts for the impact of competing risks by non-CVD outcomes whereas most national Asia-Pacific risk scores as well as the WHO CVD risk charts did not do so. This statistical adjustment prevents overestimation of CVD risk and overestimation of the benefit of treatment in populations where the risk of competing non-CVD deaths is high.24,30 For example, this adjustment should predominantly benefit treatment decisions in older individuals, and those from high- or very-high-risk regions.30

Second, SCORE2 Asia-Pacific has been systematically recalibrated to the Asia-Pacific clinical practice, using the most contemporary, powerful, and representative CVD rates available. The recalibration methods have previously been effectively applied within Europe and have now been repeated to adapt the model for Asia-Pacific populations, which ensures that the SCORE2 Asia-Pacific predicted risks are in line with the alarmingly high and rapidly changing CVD incidence.1 Even though the same cut-offs in age- and sex-standardised CVD mortality rates were used to define the risk regions, SCORE2 predictions differed considerably between the respective risk regions in both continents. This further verifies that risk models may need to be adapted to the local situation, even if average levels of risk are similar as in places a model is currently recalibrated to.

Third, because the recalibration approach is based on registry data, the SCORE2 Asia-Pacific algorithms can be readily updated to reflect future CVD incidence and risk factor profiles of any target population of apparently healthy individuals to be screened.2,23 This means that if descriptive age- and sex-specific epidemiological data are available from individual countries, they can be readily incorporated to revise models at a country-level. The calibration results of the SCORE2-Asia model in the low-risk region illustrate that within-region differences still exist after the region-specific recalibration, implying that further improvement can be obtained from country-specific recalibration. Especially for large countries with substantial within-country variation, a country-level approach would still not capture the complete geographical variation of incident CVD.31 The current recalibration approach would be suitable to recalibrate the model to within-country areas, even up to the neighbourhood level. This requires, however, high-quality data on CVD risk factors and incidence for the intended regions to ensure adequate calibration of risk algorithms.

Similar to the European SCORE2, SCORE2 Asia-Pacific can be used in a simplified form via the two-dimensional risk charts as provided in Figure 3. However, to accommodate more accurate predictions that do not require rounding to broad categories of CVD risk factors, the SCORE2 Asia-Pacific algorithms will be integrated into online calculators, such as the ESC CVD risk prediction app or the CE-marked U-Prevent medical device, available from www.U-Prevent.com. Because of the time required for implementation into a CE-marked medical device, the SCORE2 Asia-Pacific algorithms are integrated in an R-shiny app for scientific purposes only (i.e. not for clinical use) from https://hagemanshj.shinyapps.io/SCORE2ASIAPACIFIC/.

The SCORE2 Asia-Pacific risk charts have been provided with different colours matching categories of predicted 10-year CVD event risk. Specific colours on the charts do not necessarily reflect ‘treatment thresholds’, in which individuals with higher risks would automatically qualify for treatment. This is because such thresholds can be highly dependent on the local CVD burden, CVD prevention guidelines, and socio-economic circumstances. To aid national and regional guideline makers, we have illustrated the performance of the SCORE2 Asia-Pacific algorithms with data estimated from all Asia-Pacific countries, showing the expected proportions of individuals in specific risk categories across countries. These analyses and SCORE2 Asia-Pacific risk charts may help to determine suitable risk thresholds, which can be age-specific such as the European charts, or be independent of age. Apart from such risk thresholds, treatment decisions will likely also depend on other factors, such as preferences of the patient and physician, the risk of side effects, and other comorbidities or personal factors that may play a role.

On top of these points, strengths of the SCORE2 Asia-Pacific model include the use of very powerful, contemporary datasets and the use of proven recalibration methods based on nationally representative aggregate data. However the potential limitations of the SCORE2 Asia-Pacific algorithms need to be considered. The original SCORE2 algorithms were derived using data from mostly European regions and populations. Ideally, the derivation of risk models would have involved large nationally representative, prospective cohorts also from Asia-Pacific countries. However, analyses from the current study have shown that the coefficients from the European SCORE2 algorithms apply well to Asia-Pacific populations. The SCORE2 Asia-Pacific algorithms showed adequate discrimination in all external validation cohorts, similar to the validation of the European SCORE2 algorithms in Europe. This was further verified by refitting the SCORE2 models in several external validation cohorts, showing very similar subdistribution HRs between Asia-Pacific and European data. These findings align with the WHO CVD risk charts, in which no evidence of geographical heterogeneity in the model coefficients was observed either.9

Another limitation of the study is that several of the external validation data sources were also used for derivation of the multipliers, as part of the recalibration procedure. This approach was chosen because powerful, approximately nationally representative data sources with both fatal and non-fatal events are scarce, whereas those are necessary for both the multiplier derivation as for external validation. As the quality of the recalibration procedure directly improves predicted risks in clinical practice, this was prioritised over keeping the processes completely separate. In addition, the recalibration process does not affect model discrimination, ensuring discrimination can be evaluated unbiased.22 Because the recalibration procedure involves all multipliers from different data sources as well as CVD mortality rates and risk factor from all Asian-Pacific countries, the effect of multipliers from a single dataset on the final predicted risks (and therefore calibration) is also rather limited.

In the current study, the SCORE2 Asia-Pacific algorithms were validated in 12 Asian-Pacific countries. In the low- and moderate-risk regions, we were able to access multiple high-quality data sources that provided a more comprehensive view of the population. These data sources were from different geographical areas within these regions, allowing us to demonstrate the model’s discrimination and calibration. However, in several countries, especially in high- and very-high-risk regions, suitable high-quality longitudinal data for external validation were not available. This limitation also extends to the recalibration, as the current approach relies on high-quality age-specific CVD mortality data to accurately adapt the SCORE2 model to the Asian-Pacific region. Similar to the external validation data, these data for recalibration were often of lower quality in high- and very-high-risk countries.

Moreover, while high-quality data were available for external validation in some countries, these datasets were not always nationally representative. This limitation is particularly significant in larger, more diverse countries where the need for further external validation remains essential to ensure the model’s applicability across various sub-populations. For all countries, however, CVD mortality and risk factor data were used in the recalibration process. Should high-quality data become available in these countries, this may contribute to the ongoing validation efforts of SCORE2 Asia-Pacific and other CVD prediction algorithms, ensuring ongoing accuracy and applicability.

Data on medication use, family history, socio-economic status, nutrition, physical activity, renal function, or ethnicity were not included in the original SCORE2 algorithms as these were unavailable in cohorts and registries.2 Hence, interpretation of SCORE2 estimates may require clinical judgement, especially for individuals in whom these factors may be relevant (e.g. those taking lipid or blood pressure lowering treatments, with a family history of CVD,32 with chronic kidney disease,33 or in at-risk socio-economic and ethnic groups).32,34 For individuals with several of these risk factors, solutions have been developed to accurately incorporate these additional risk factors on top of existing prediction models.33,35

In conclusion, the SCORE2 Asia-Pacific algorithms have been adapted to different Asia-Pacific risk regions for prediction of 10-year CVD risk for individuals without DM or CVD. The SCORE2 coefficients are the foundation of the algorithms, which have now been shown to apply well to Asian-Pacific populations. As the SCORE2 Asia-Pacific algorithms, they are adapted to reflect risk factor levels and CVD incidences across the Asian-Pacific region. SCORE2 Asia-Pacific is part of a family of high-quality prediction algorithms using contemporary data and methodology and provides an accurate tool for identification of individuals at high CVD risk in the Asia-Pacific region. With the recalibration approach used, the model can be readily updated to further refine predictions to even smaller geographical regions and to adapt to changing CVD incidences.

Click here to view Supplementary Material.

Appendix

ESC’s Cardiovascular Risk Collaboration (ESC CRC): Emanuele Di Angelantonio (Cardiovascular Epidemiology Unit, Department of Public Health and Primary Care, University of Cambridge, Cambridge, UK); Michael Papadakis (Cardiovascular clinical academic group, St George’s, University of London, UK); Adam Timmis (William Harvey Research Institute, Queen Mary University London, London, UK); Victor Aboyans (Department of Cardiology, Dupuytren University Hospital, and EpiMaCT, Inserm1094/IRD270, Limoges University, Limoges, France); Panos Vardas (Hygeia Hospitals Group, HHG, Athens, Greece Biomedical Research Foundation Academy of Athens (BRFAA), Athens, Greece); Frank LJ Visseren (Department of Vascular Medicine, University Medical Center Utrecht, The Netherlands); John William McEvoy (National University of Ireland Galway, Galway, Ireland); Maryam Kavousi (Department of Epidemiology, Erasmus MC, University Medical Center Rotterdam, Rotterdam, The Netherlands); Jean Ferrieres (Department of Cardiology, Toulouse Rangueil University Hospital, Institut National pour la Santé et la Recherche Médicale Unité Mixte de Recherche, Toulouse cedex, France Emergency Department, Rangueil Hospital, Toulouse, France); Radu Huculeci (European Society of Cardiology, Brussels, Belgium).

ASEAN Federation of Cardiology (AFC): Alex Junia (Cebu Institute of Medicine, Cebu, Philippines); Rungroj Krittayaphong (Division of Cardiology, Department of Medicine, Siriraj Hospital, Mahidol University, Bangkok, Thailand); Quang Ngoc Nguyen (Department of Cardiology, Hanoi Medical University); Abdul Halim Raynaldo (Department Cardiology and Vascular Medicine, University of Sumatera Utara); Alan Fong (Sarawak Heart Centre, Kota Samarahan, Malaysia).

Asia-Pacific Society of Cardiology (APSC): Hyo-Soo Kim (Seoul National University Hospital, Seoul, South Korea); Jack Tan (National Heart Centre Singapore, Singapore); Issei Komuro (International University of Health and Welfare/The University of Tokyo); Wael Almahmeed (Heart and Vascular and Thoracic Institute, Cleveland Clinic Abu Dhabi); Khung Keong Yeo (National Heart Centre Singapore, Singapore); Junya Ako (Kitasato University School of Medicine, Tokyo, Japan); Kyung Woo Park (Seoul National University College of Medicine, Seoul, South Korea).

SCORE2 Asia-Pacific

Authors listed alphabetically (all authors listed alphabetically contributed equally): Noraidatulakma Abdullah (UKM Medical Molecular Biology Institute (UMBI), Hospital Canselor Tuanku Muhriz, Universiti Kebangsaan Malaysia Jalan Yaacob Latif, Bandar Tun Razak, Cheras, 56000 Kuala Lumpur, Malaysia); Muhammad Irfan Abdul Jalal (UKM Medical Molecular Biology Institute (UMBI), Hospital Canselor Tuanku Muhriz, Universiti Kebangsaan Malaysia Jalan Yaacob Latif, Bandar Tun Razak, Cheras, 56000 Kuala Lumpur, Malaysia); Elizabeth L.M. Barr (Epidemiology and Clinical Diabetes, Baker Heart and Diabetes Institute); Parinya Chamnan (Sanpasitthiprasong Hospital, Ubon Ratchathani, Thailand); Chean Lin Chong (PAPRSB Institute of Health Sciences, Universiti Brunei Darussalam); Lucky Cuenza (Department of Adult Cardiology, Philippine Heart Center); Praveen Deorani (Singapore Ministry of Health); Pei Gao (Department of Epidemiology and Biostatistics, School of Public Health, Peking University Health Science Center; Key Laboratory of Epidemiology of Major Diseases, Peking University, Ministry of Education; Center for Real-world Evidence Evaluation, Peking University Clinical Research Institute); Ian Graham (School of Medicine, Trinity College Dublin, The University of Dublin, College Green, Dublin, Ireland); Saima Hilal (Saw Swee Hock School of Public Health, National University of Singapore and National University Health System, Singapore); Joris Holtrop (Department of Vascular Medicine, University Medical Center Utrecht); Rahman Jamal (UKM Medical Molecular Biology Institute (UMBI), Hospital Canselor Tuanku Muhriz, Universiti Kebangsaan Malaysia Jalan Yaacob Latif, Bandar Tun Razak, Cheras, 56000 Kuala Lumpur, Malaysia); Tosha Ashish Kalhan (Saw Swee Hock School of Public Health, National University of Singapore); Hidehiro Kaneko (The Department of Cardiovascular Medicine, The University of Tokyo, Tokyo, Japan; The Department of Advanced Cardiology, The University of Tokyo, Tokyo, Japan); Chi-Ho Lee (Department of Medicine, School of Clinical Medicine, University of Hong Kong); Charlie G.Y. Lim (Saw Swee Hock School of Public Health, National University of Singapore and National University Health System, Singapore); Xiaofei Liu (Department of Epidemiology and Biostatistics, School of Public Health, Peking University Health Science Center); Dianna J. Magliano (Clinical Diabetes and Epidemiology, Baker Heart and Diabetes Institute, Melbourne, Australia); Nima Motamed (Gastrointestinal and Liver Diseases Research Center, Iran University of Medical Sciences, Tehran, Iran); Maziar Moradi-Lakeh (Gastrointestinal and Liver Diseases Research Center, Iran University of Medical Sciences, Tehran, Iran); Sok King Ong (PAPRSB Institute of Health Sciences, Universiti Brunei Darussalam (UBD)); Ruwanthi Perera (Department of Rogavijnana, Faculty of Indigenous Medicine, Gampaha Wickramarachchi University of Indigenous Medicine, Sri Lanka.); Kameshwar Prasad (Department of Neurology, All India Institute of Medical Sciences & Fortis Hospital, Vasant Kunj, New Delhi); Jonathan E Shaw (Clinical Diabetes and Epidemiology, Baker Heart and Diabetes Institute, Melbourne, Australia); Janaka de Silva (Department of Medicine, Faculty of Medicine, University of Kelaniya, Sri Lanka); Xueling Sim (Saw Swee Hock School of Public Health, National University of Singapore and National University Health System, Singapore); Yuta Suzuki ((i) The Department of Cardiovascular Medicine, The University of Tokyo, Tokyo, Japan; (ii) Center for Outcomes Research and Economic Evaluation for Health, National Institute of Public Health, Saitama, Japan); Kathryn C.B. Tan (Department of Medicine, School of Clinical Medicine, University of Hong Kong); Xun Tang (Department of Epidemiology and Biostatistics, School of Public Health, Peking University Health Science Center; Key Laboratory of Epidemiology of Major Diseases, Peking University, Ministry of Education); Kavita Venkataraman (Saw Swee Hock School of Public Health, National University of Singapore); Rajitha Wickremasinghe (Department of Public Health, Faculty of Medicine, University of Kelaniya, Sri Lanka); Hideo Yasunaga (The Department of Clinical Epidemiology and Health Economics, School of Public Health, The University of Tokyo, Tokyo, Japan); Farhad Zamani (Gastrointestinal and Liver Diseases Research Center, Iran University of Medical Sciences, Tehran, Iran).

References

  1. Roth GA, Abate D, Abate KH, et al. Global, regional, and national age-sex-specific mortality for 282 causes of death in 195 countries and territories, 1980–2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet Rheumatol 2018;392:1736–88. 
    Crossref | PubMed
  2. Hageman S, Pennells L, Ojeda F, et al. SCORE2 risk prediction algorithms: new models to estimate 10-year risk of cardiovascular disease in Europe. Eur Heart J 2021;42:2439–54. 
    Crossref | PubMed
  3. Visseren FLJ, Mach F, Smulders YM, et al. 2021 ESC guidelines on cardiovascular disease prevention in clinical practice. Eur Heart J 2021;42:3227–337. 
    Crossref | PubMed
  4. Ohira T, Iso H. Cardiovascular disease epidemiology in Asia. Circ J 2013;77:1646–52. 
    Crossref | PubMed
  5. Zhang Y, Miao H, Chia Y, et al. Cardiovascular risk assessment tools in Asia. J Clin Hypertens 2022;24:369–77. 
    Crossref | PubMed
  6. Yang X, Li J, Hu D, et al. Predicting the 10-year risks of atherosclerotic cardiovascular disease in Chinese population. Circulation 2016;134:1430–40. 
    Crossref | PubMed
  7. Harada A, Ueshima H, Kinoshita Y, et al. Absolute risk score for stroke, myocardial infarction, and all cardiovascular disease: Japan Arteriosclerosis Longitudinal Study. Hypertens Res 2019;42:567–79. 
    Crossref | PubMed
  8. Hajifathalian K, Ueda P, Lu Y, et al. A novel risk score to predict cardiovascular disease risk in national populations (Globorisk): a pooled analysis of prospective cohorts and health examination surveys. Lancet Diabetes Endocrinol 2015;3:339–55. 
    Crossref | PubMed
  9. Kaptoge S, Pennells L, De Bacquer D, et al. World Health Organization cardiovascular disease risk charts: revised models to estimate risk in 21 global regions. Lancet Glob Health 2019;7:e1332–45. 
    Crossref | PubMed
  10. Danesh J, Erqou S, Walker M, et al. The Emerging Risk Factors Collaboration: analysis of individual data on lipid, inflammatory and other markers in over 1.1 million participants in 104 prospective studies of cardiovascular diseases. Eur J Epidemiol 2007;22:839–69. 
    Crossref | PubMed
  11. Hewitt J, Walters M, Padmanabhan S, Dawson J. Cohort profile of the UK Biobank: diagnosis and characteristics of cerebrovascular disease. BMJ Open 2016;6:e009161. 
    Crossref | PubMed
  12. World Health Organization. Global Health Estimates. 2022. https://www.who.int/data/global-health-estimates (accessed 5 May 2022).
  13. World Health Organization. WHO Mortality Database. 2022. https://apps.who.int/healthinfo/statistics/mortality/whodpms/ (accessed 20 April 2022).
  14. Cheol Seong S, Kim YY, Khang YH, et al. Data resource profile: the national health information database of the national health insurance service in South Korea. Int J Epidemiol 2017;46:799–800. 
    Crossref | PubMed
  15. Lin H, Tang X, Shen P, et al. Using big data to improve cardiovascular care and outcomes in China: a protocol for the CHinese Electronic health Records Research in Yinzhou (CHERRY) Study. BMJ Open 2018;8:e019698. 
    Crossref | PubMed
  16. Suebsamran P, Choenchoopon H, Rojanasaksothorn S, et al. Association between alcohol consumption and pre-diabetes among 383,442 Thai population aged 15 years and older in Ubon Ratchathani: analytical cross-sectional study. J Med Assoc Thail Chotmaihet Thangphaet 2016;99:S35–42.
  17. NCD Risk Factor Collaboration. Worldwide trends in blood pressure from 1975 to 2015: a pooled analysis of 1479 population-based measurement studies with 19.1 million participants. Lancet Rheumatol 2017;389:37–55. 
    Crossref | PubMed
  18. NCD Risk Factor Collaboration. Worldwide trends in diabetes since 1980: a pooled analysis of 751 population-based studies with 4.4 million participants. Lancet Rheumatol 2016;387:1513–30. 
    Crossref | PubMed
  19. Thulani UB, Mettananda KCD, Warnakulasuriya DTD, et al. Validation of the World Health Organization/International Society of Hypertension (WHO/ISH) cardiovascular risk predictions in Sri Lankans based on findings from a prospective cohort study. PLoS One 2021;16:e0252267. 
    Crossref | PubMed
  20. Kasturiratne A, Ediriweera DS, De Silva ST, et al. Patterns and predictors of mortality in a semi-urban population-based cohort in Sri Lanka: findings from the Ragama Health Study. BMJ Open 2020;10:e038772. 
    Crossref | PubMed
  21. Nagai K, Tanaka T, Kodaira N, et al. Data resource profile: JMDC claims database sourced from health insurance societies. J Gen Fam Med 2021;22:118–27. 
    Crossref | PubMed
  22. Conroy RM, Pyörälä K, Fitzgerald AP, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project. Eur Heart J 2003;24:987–1003. 
    Crossref | PubMed
  23. Pennells L, Kaptoge S, Wood A, et al. Equalization of four cardiovascular risk algorithms after systematic recalibration: individual-participant meta-analysis of 86 prospective studies. Eur Heart J 2019;40:621–31. 
    Crossref | PubMed
  24. Wolbers M, Koller MT, Witteman JCM, Steyerberg EW. Prognostic models with competing risks. Epidemiology 2009;20:555–61. 
    Crossref | PubMed
  25. Huang JY. Representativeness is not representative: addressing major inferential threats in the UK Biobank and other big data repositories. Epidemiology 2021;32:189. 
    Crossref | PubMed
  26. Wolff RF, Moons KGM, Riley RD, et al. PROBAST: a tool to assess the risk of bias and applicability of prediction model studies. Ann Intern Med 2019;170:51–51. 
    Crossref | PubMed
  27. Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMC Med 2015;13:1–10. 
    Crossref | PubMed
  28. Choi J, Sung S, Park SK, et al. SCORE and SCORE2 in East Asian population. JACC Asia 2024;4:265–74. 
    Crossref | PubMed
  29. Kasim SS, Ibrahim N, Malek S, et al. Validation of the general Framingham Risk Score (FRS), SCORE2, revised PCE and WHO CVD risk scores in an Asian population. Lancet Reg Health West Pac 2023;35:100742. 
    Crossref | PubMed
  30. Hageman SHJ, Dorresteijn JAN, Pennells L, et al. The relevance of competing risk adjustment in cardiovascular risk prediction models for clinical practice. Eur J Prev Cardiol 2023;30:1741–7. 
    Crossref | PubMed
  31. Rossello X, Dorresteijn JAN, Janssen A, et al. Risk prediction tools in cardiovascular disease prevention: a report from the ESC Prevention of CVD Programme led by the European Association of Preventive Cardiology (EAPC). Eur J Prev Cardiol 2019;26:1534–44. 
    Crossref | PubMed
  32. D’Agostino RB, Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation 2008;117:743–53. 
    Crossref | PubMed
  33. Matsushita K, Kaptoge S, Hageman SHJ, et al. Including measures of chronic kidney disease to improve cardiovascular risk prediction by SCORE2 and SCORE2-OP. Eur J Prev Cardiol 2023;30:8–16. 
    Crossref | PubMed
  34. Kist JM, Vos RC, Mairuhu ATA, et al. SCORE2 cardiovascular risk prediction models in an ethnic and socioeconomic diverse population in the Netherlands: an external validation study. EClinicalMedicine 2023;57:101862. 
    Crossref | PubMed
  35. Hageman SHJ, Petitjaen C, Pennells L, et al. Improving 10-year cardiovascular risk prediction in apparently healthy people: flexible addition of risk modifiers on top of SCORE2. Eur J Prev Cardiol 2023;30:1705–14. 
    Crossref | PubMed