Cardiovascular Disease Mortality Burden in Singapore From 1990 to 2050

Graphical Abstract

Singapore is a high-income city-state with an advanced public health system that serves a diverse ethnic population of 6 million people. Despite having the lowest age-standardized morbidity and mortality rates for cardiovascular disease (CVD) in Southeast Asia, a recent forecast based on the Singapore Myocardial Infarction Registry predicts a nearly 2-fold increase in the incidence of ischemic heart disease (IHD) between 2025 and 2050.^1,2 The annual direct and indirect costs of IHD in Singapore are estimated at US$2.2 billion and US$1.4 billion, respectively, accounting for almost 2% of the country's gross domestic product.³

Recognizing the urgent need for health policy interventions to address the growing CVD burden, this study provides a contextualized forecast of Singapore's CVD-related mortality through 2050. It highlights key shifts in cardiovascular risk factors over time, offering actionable insights to prioritize CVD prevention strategies and identifying the most impactful risk factors for primordial, primary, and secondary prevention in Singapore.⁴

Methods

Using data from the 2023 Global Burden of Diseases, Injuries, and Risk Factors Study (GBD 2023), we projected trends to 2050 using GBD Foresight modeling, which accounts for demographic changes and the continuation of recent epidemiological trends.⁵ Age-standardized rates (per 100,000) and crude estimates were reported with 95% uncertainty intervals (UIs). Age-standardized rates (per 100,000) were computed by direct standardization to the GBD World Standard Population (the fixed reference age structure used by GBD across all locations, sexes, and years). This standard population was held constant across the observed period (1990-2023) and the forecast period (2025-2050) so that rates are directly comparable across time. Age-standardized rates and crude estimates were reported with 95% uncertainty intervals (UIs).

For projections to 2050, we used GBD Foresight, which is a multistaged forecasting framework developed by the Institute for Health Metrics and Evaluation (IHME) and used to generate the future-health scenarios in GBD 2023.⁶ The framework forecasts 220 mutually exclusive causes of mortality and 68 risk-factor summary exposure values (SEVs).⁷ For each cause-location-age-sex stratum, the underlying (risk-deleted) mortality is modeled with a 3-component mixed-effects specification on the log scale: 1) a location-age–specific random intercept; 2) a global fixed slope on the sociodemographic index (SDI); and 3) an age-specific random slope on a secular time trend, with a risk-factor scalar entered as an offset to capture combined risk-factor effects and an ARIMA(0,1,0) process (with attenuated exponential drift for all-cause mortality and no drift for cause-specific mortality) capturing residual temporal variation. Risk-factor exposure (SEV) is forecast independently through a true ensemble of 12 submodels—6 annualized-rate-of-change models and 6 meta-regression–Bayesian regularized trimmed (MR-BRT) spline models driven by SDI—with recency-weighting schemes varied across submodels. A risk-factor mediation layer routes changes in upstream exposures (dietary factors, body mass index, fasting plasma glucose, smoking, and physical activity) through intermediate metabolic mediators (systolic blood pressure, low-density lipoprotein [LDL] cholesterol, and plasma glucose) to avoid double-counting of attributable mortality. The projected SEVs are then propagated through the GBD comparative risk-assessment framework to obtain risk-attributable mortality and feed the risk-factor scalar in the cause-of-death model. Age-specific rates are translated into all-age and age-standardized mortality counts using IHME's population forecasts, which combine UN World Population Prospects 2022 migration estimates with age- and sex-specific fertility and mortality projections. Uncertainty is propagated by drawing 500 posterior samples through the multistage pipeline, and incorporate uncertainty in the cause-of-death model parameters, the risk-factor SEVs ensemble, the GBD comparative risk-assessment relative risks, and the IHME demographic projections (age- and sex-specific fertility, mortality, and net migration).⁷ However, the forecast tools are unable to fully account for transformative changes in future health care access, treatment uptake, or policy implementation, or geopolitical and economic developments, that may alter disease trajectories in ways that cannot be accurately predicted. Accordingly, the forecasts estimates should be interpreted as broad indications of possible future trajectories to inform scenario planning and prevention efforts, rather than as precise point estimates.

The cause-of-death model uses SDI as the principal time-varying covariate, together with the time index and the risk-factor scalar; location-age random intercepts and age-specific random slopes on time provide hierarchical pooling across geographies. SEV forecasts similarly use SDI as the principal driver within the MR-BRT branch. Risk-attributable mortality is governed by the projected SEV trajectories for all risk factors mapped to each cause, with mediated effects handled through the mediation layer. GBD Foresight employs a single global model specification estimated jointly across all 204 GBD locations, thereby borrowing strength from cross-national patterns while allowing location-specific variation through the random-effects structure.⁷ However, there are several model assumptions. GBD Foresight assumes that historical relationships between SDI, secular time trends, and log-mortality will persist, smoothed by the hierarchical structure; demographic transition will proceed per IHME and UN population projections; risk-factor exposures will evolve smoothly within logical bounds, as captured by the SEV ensemble; and comparative-risk-assessment relative risks remain constant throughout the projection period. GBD Foresight evaluates predictive performance using a 2010 to 2019 hold-out design, with skill defined relative to a naive baseline holding the 2009 value constant and squared errors winsorized at the 95th percentile. A positive skill value indicates that GBD Foresight outperforms this baseline. Skill is reported for mortality and disability-adjusted life year (DALY) estimates at level 0 (all causes), level 1, and level 2 aggregates.⁷ However, we did not conduct an independent out-of-sample validation for Singapore, as GBD Foresight is a centrally maintained IHME framework not amenable to location-specific refitting. Although Singapore maintains high-quality registries for myocardial infarction and stroke,⁸ dedicated registries for the remaining CVD subtypes are lacking, precluding a comprehensive national-registry–only analysis across the full cardiovascular spectrum. GBD estimates were prioritized in this study given that Singapore's underlying data infrastructure is robust—the GBD estimation for Singapore draws primarily on vital registration data rated 4 stars or above on the GBD data-quality scale,⁹ supplemented by clinically adjudicated inputs from national disease registries, lending confidence to the modeled estimates.

Moreover, to estimate the potential impact of intensified national preventive policies on the CVD mortality forecast, we additionally report the GBD Foresight “Improved Behavioral and Metabolic Risks” scenario,⁶ which models a trajectory in which exposure to all modifiable behavioral and metabolic CVD risk factors reaches their theoretical-minimum-risk exposure level (TMREL) by 2050. The reference scenario reflects the continuation of current trends. The improved behavioral and metabolic risks scenario modifies SEV trajectories only—driving all modifiable behavioral and metabolic exposures to their theoretical-minimum-risk levels by 2050—while leaving cause-of-death and demographic components unchanged.⁶ The latter scenario therefore represents the upper bound of what comprehensive risk-factor control alone could deliver, and serves as a policy-intensification counterfactual against the reference projection.

Decomposition analysis was performed using the Das Gupta method to quantify the drivers of temporal changes in CVD and risk-attributable burden.¹⁰ The total change in burden between 1990 and 2023 was decomposed into 3 mutually exclusive and additive components: population growth, population aging, and epidemiological change. Age-specific burden estimates for each risk factor were obtained from GBD 2023, restricted to both sexes combined. Population data by age group were merged with burden estimates to calculate age-specific rates. Total percentage change estimates were derived from GBD aggregate data. For each tier-2 GBD risk factor, we repeated the decomposition on risk-attributable deaths to isolate the contribution of declining age-specific risk-attributable mortality from the demographic forces acting in the opposite direction.

The Das Gupta method decomposes the change in total burden between 2 time points into 3 additive components, following these formulas:

$P^t={\sum }_i{N}_i^t$: total population

$A_i^t=\frac{N_i^t}{P^t}$: proportion of the population in age group i

$B^t={\sum }_i{N}_i^t\times R_i^t$: total burden at time t

$$\Delta B=B^{2023}-B^{1990}$$

$$\Delta B={\Delta }_{pop}+{\Delta }_{age}+{\Delta }_{rate}$$

$${\Delta }_{pop}={\sum }_i\left(P^{2023}-P^{1990}\right)\times \overline{A_i}\times \overline{R_i}$$

$${\Delta }_{age}={\sum }_i\overline{P}\times \left(A_i^{2023}-A_i^{1990}\right)\times \overline{R_i}$$

$${\Delta }_{rate}={\sum }_i\overline{P}\times \overline{A_i}\times \left(R_i^{2023}-R_i^{1990}\right)$$

where t denotes the time point, i denotes the age group, and $R_i^t$ denotes the age-specific burden rate. Overbars denote arithmetic means across time points: $\overline{P}=\left(P^{1990}+P^{2023}\right)÷2$, $\overline{A}=\left(A^{1990}+A^{2023}\right)÷2$, $\overline{R}=\left(R^{1990}+R^{2023}\right)÷2$.¹⁰

This study was exempt from Institutional Review Board review by the National Healthcare Group Domain Specific Review Board C, as it used publicly available data that did not contain confidential or identifiable patient information.

Results

Between 1990 and 2023, the age-standardized CVD mortality rate in Singapore fell by 69.6% (from 270.1 [95% UI: 255.2-280.1] to 82.0 [95% UI: 70.6-92.0] per 100,000), while crude all-age CVD deaths rose by 44.8% (from 5,204 [95% UI: 4,991-5,362] to 7,533 [95% UI: 6,485-8,459] per 100,000). Between 2025 and 2050, crude all-age CVD deaths in Singapore are projected to rise by 76.5%, reaching 12,397 (95% UI: 9,583-15,301) deaths in 2050, while the age-standardized CVD mortality rate is projected to fall by 43.4% (68.0 [95% UI: 59.6-73.5] to 38.5 [95% UI: 31.2-47.7] per 100,000). Singapore's 2023 age-standardized CVD mortality rate is the fourth lowest across the Asia-Pacific, with its improvement in the age-standardized mortality rate between 1990 and 2023 being the third fastest in the region.

In 2023, the leading causes of CVD-related age-standardized mortality in Singapore were IHD (54.4 [95% UI: 47.4-61.0] per 100,000; 66.5% of CVD deaths), followed by stroke (12.3 [95% UI: 10.1-14.4] per 100,000; 15.0% of CVD deaths) and hypertensive heart disease (6.5 [95% UI: 5.3-7.5] per 100,000; 8.1% of CVD deaths), collectively accounting for 89.6% of all CVD deaths. From 1990 to 2023, the largest increases in crude all-age CVD deaths in Singapore were contributed by endocarditis (11-64 deaths; 467.9% increase; 5.4% annual increase; 0.2%-0.8% of CVD deaths), aortic aneurysm (51-223 deaths; 332.7% increase; 4.5% annual increase; 1.0%-3.0% of CVD deaths), and atrial fibrillation and flutter (31-126 deaths; 301.4% increase; 4.3% annual increase; 0.6%-1.7% of CVD deaths). Because of medical advancements, every CVD subtype showed an improvement in its age-standardized mortality rate between 1990 and 2023, with the exception of endocarditis (0.6-0.7 per 100,000; 24.5% increase; 0.5% annual increase) (Table 1). By 2050, the leading causes of CVD-related mortality in Singapore ranked by projected age-standardized rate will remain IHD (16.7 [95% UI: 12.7-22.4] per 100,000), hypertensive heart disease (7.1 [95% UI: 5.4-9.3] per 100,000), and stroke (6.5 [95% UI: 5.3-7.9] per 100,000) (Figure 1).

Table 1.

Mortality and Burden Contributions of Cardiovascular Disease Subtypes, Singapore (1990-2023)

Cause	ASMR (per 100,000), 1990	ASMR (per 100,000), 2023	ASMR, Total Change (%), 1990-2023	ASMR, Annualized Change (% per year), 1990-2023	Crude Deaths, 1990	Crude Deaths, 2023	Crude Deaths, Total Change (%), 1990-2023	Crude Deaths, Annualized Change (% per year), 1990-2023	Share of Total CVD Deaths (%), 1990	Share of Total CVD Deaths (%), 2023	Change in Share of Total CVD Deaths (percentage points), 1990-2023
Overall cardiovascular disease	270.1	82	−69.6	−3.55	5,204	7,533	44.8	1.13	100	100	NA
Ischemic heart disease	146.1	54.4	−62.8	−2.95	2,872	5,006	74.3	1.7	55.2	66.5	11.3
Stroke	88.5	12.3	−86.1	−5.8	1,660	1,132	−31.8	−1.15	31.9	15	−16.9
Hypertensive heart disease	16.1	6.5	−59.4	−2.71	280	610	117.7	2.39	5.4	8.1	2.7
Atrial fibrillation and flutter	2.0	1.4	−29.5	−1.08	31	126	301.4	4.3	0.6	1.7	1.1
Aortic aneurysm	2.6	2.4	−6.4	−0.24	51	223	332.7	4.54	1.0	3.0	2.0
Endocarditis	0.6	0.7	24.5	0.47	11	64	467.9	5.41	0.2	0.8	0.6

Each row represents a CVD subtype, with the first row showing overall CVD as the reference total. Columns present the age-standardized mortality rate per 100,000 population in 1990 and 2023, together with the total percentage change and the annualized percentage change (per year) over the 1990 to 2023 period. Crude death counts are shown for the same 2 years alongside their total and annualized percentage changes. The final 3 columns provide each subtype's share of total CVD deaths (%) in 1990 and 2023, as well as the absolute change in share from 1990 to 2023 expressed in percentage points.

ASMR = age-standardized mortality rate; CVD = cardiovascular disease; NA = not applicable.

Figure 1.

Age-Standardized Mortality and All-Age Deaths From Cardiovascular Subtypes, Singapore (1990-2050)

Age-standardized mortality rates (per 100,000 population) and all-age deaths from cardiovascular disease subtypes in Singapore, from 1990 to 2050. Solid lines represent observed estimates (1990-2023) and dashed lines represent projections (2024-2050) derived from Global Burden of Disease (GBD) Foresight. Ribbons represent the 95% uncertainty intervals (UIs). Data source: Global Burden of Disease Study 2023.

Sex-disaggregated analysis showed that the age-standardized CVD mortality rate in 2023 was 1.9-fold higher in males (108.9 [95% UI: 95.1-121.8] per 100,000) than in females (57.0 [95% UI: 44.2-65.8] per 100,000), with males bearing a higher burden across all CVD subtypes except peripheral artery disease, rheumatic heart disease, and pulmonary arterial hypertension. Age-specific CVD mortality peaked earlier in males (70-74 years) than in females (90-94 years). Between 2025 and 2050, the age-standardized CVD mortality rate is projected to decline by 43.8% in males and 43.0% in females, narrowing the absolute sex difference by 15.0%.

A metabolic-centric risk factor landscape underpins the growing CVD burden in Singapore. In 2023, the leading cardiovascular risk factors in Singapore ranked by age-standardized risk-attributable CVD mortality rate were high systolic blood pressure (37.1 [95% UI: 29.2-45.7] per 100,000), followed by dietary risks (25.2 [95% UI: 11.3-37.5] per 100,000) and high LDL cholesterol (19.4 [95% UI: 13.0-27.7] per 100,000); the smallest reductions in metabolic risk factor-related age-standardized mortality were seen with high body mass index (from 12.2 to 8.4 per 100,000; 31.4% decrease), kidney dysfunction (from 36.1 to 11.0 per 100,000; 69.6% decrease), and high fasting plasma glucose (from 31.8 to 9.6 per 100,000; 69.9% decrease). Specific to behavioral risk factors, the smallest decreases in age-standardized mortality were observed with dietary risks (from 91.5 to 25.2 per 100,000; 72.5% decrease), tobacco use (from 30.0 to 8.1 per 100,000; 73.0% decrease), and low physical activity (from 3.9 to 0.9 per 100,000; 78.1% decrease). Specific to dietary risk factors, diet high in red meat contributed to the largest increase in age-standardized mortality (from 1.2 to 1.4 per 100,000; 19.9% increase), and the smallest reductions were in diet high in sugar-sweetened beverages (from 0.7 to 0.5 per 100,000; 25.3% decrease) and diet high in processed meat (from 0.7 to 0.5 per 100,000; 30.7% decrease). In terms of environmental/occupational risk factors, age-standardized cardiovascular-related mortality attributed to nonoptimal temperature observed the least reduction (from 1.0 to 0.6 per 100,000; 39.6% decrease; largely driven by high temperature), while the reduction in air pollution contributed to the largest reduction (from 60.7 to 10.3 per 100,000; 83.0% decrease).

Decomposition analysis revealed that reductions in cardiovascular risk exposures averted an estimated 9,357 deaths from 1990 to 2023 (−179.8% decrease), but this gain was offset by population aging (7,677 deaths; 147.5% increase) and population growth (4,009 deaths; 77.0% increase). When the same decomposition was applied to risk-attributable CVD deaths, the largest reductions in risk exposure from 1990 to 2023 were related to high systolic blood pressure (contributing to 5,732 fewer deaths; 199.4% decrease), dietary risks (3,354 fewer deaths; 182.4% decrease), high LDL cholesterol (2,666 fewer deaths; 181.1% decrease), and air pollution (2,500 fewer deaths; −213.6% decrease), with smaller but meaningful contributions from reductions in tobacco use, kidney dysfunction, and high fasting plasma glucose exposures. The notable exception was high body mass index, recording much smaller risk exposure reduction between 1990 and 2023 (78.0% decrease), which was offset by demographic forces, yielding an overall 191% rise in body mass index–attributable CVD deaths. Under the improved behavioral and metabolic risks scenario, the age-standardized CVD mortality rate in Singapore would fall from 68.0 per 100,000 in 2025 to 16.1 per 100,000 (95% UI: 13.4-19.0) by 2050, representing a 76.4% decline—substantially greater than the 43.4% decline projected under the reference scenario. In absolute terms, this alternative scenario would avert an estimated 6,112 CVD deaths by 2050 compared with the reference projection.

Discussion

The favorable age-standardized CVD mortality trajectory in Singapore therefore reflects substantial improvements in risk-attributable mortality, driven primarily by improved prevention of hypertension, hyperlipidemia, dietary factors, and air pollution. This is largely contributed by Singapore's affluence, strong health care infrastructure and resources, early detection and prevention, and cardiac rehabilitation. However, the absolute death toll continues to grow because population aging and population growth together outpaced these gains. This demographic-epidemiological tension in Singapore demonstrates that effective prevention alone will not curb the absolute risk-factor burden in rapidly aging populations without concurrent health care infrastructure expansion and intensified primordial prevention.¹¹ CVD preventive policies must therefore preemptively scale cardiovascular services to meet the rising metabolic disease challenge, promote sustainable healthy aging through targeted surveillance, and contain health care costs through primordial and primary prevention. Sustaining progress in Singapore will require accelerated upscaling of current efforts to outpace the demographic forces driving crude CVD burden upward.

The 33.0 percentage-point gap between the reference (43.4%) and improved behavioral and metabolic risks (76.4%) scenarios quantifies the additional CVD mortality reduction achievable through comprehensive risk-factor control, and offers a sense of optimism that the greatest yield lies in targeting these modifiable metabolic exposures. In practical terms, intensifying population-level management of hypertension, hyperlipidemia, hyperglycemia, and obesity—the metabolic drivers that collectively account for the largest share of risk-attributable CVD deaths in Singapore—could nearly double the projected decline in age-standardized CVD mortality by 2050. Health policymakers in Singapore are presently targeting the top 3 most actionable cardiovascular risk factors in Singapore—hypertension, dietary risks, and hyperlipidemia—with “whole-of-country” health systems interventions. The Ministry of Health launched Healthier SG in 2022, alongside related initiatives such as the National Steps Challenge, the Eat Drink Shop Healthy Challenge, the Nutri-Grade scheme, Primary Tech-Enhanced Care Home Blood Pressure Monitoring Programme, and the national Familial Hypercholesterolemia genetic testing program, shifting the emphasis from reactive care for acute CVD to proactive primordial and primary CVD prevention.12, 13, 14 Healthier SG supports family doctors in providing holistic screening and care, empowering the engagement with community partners to promote patient education and healthy lifestyles.^3,15 These programs are consistent with the direction of the projected continued decline in age-specific CVD mortality. However, the present forecast analyses are trend-extrapolative and do not model specific policy interventions as covariate; the national programs discussed are presented as contextually aligned initiatives rather than causally attributed drivers of the projected trajectory.^2,3,16 This roadmap provides insights into achievable targets and CVD prevention strategies, shifting the focus towards whole-of-society approaches in enhancing the standards of care and cardiovascular care delivery in Singapore, all aimed at addressing the forecasted increase in cardiovascular mortality in Singapore over the next 25 years.¹⁷

Funding Support and Author Disclosures

Drs Hausenloy and Chan are supported by the CADENCE (CArdiovascular DiseasE National Collaborative Enterprise) National Clinical Translational Program (MOH-001277-01). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.