Introduction

Stroke is a major chronic non-communicable disease globally and a leading cause of death and disability in adults; it is characterised by high rates of incidence, disability, mortality, and recurrence, as well as a substantial economic burden1. According to the latest Global Burden of Disease (GBD) 2021 data, stroke ranks as the third leading cause of death (7.3 million deaths, accounting for 10.7% of all deaths) and the fourth leading cause of disability (160.5 million DALYs, 5.6% of global DALYs)2. Despite a decline in recent years in age-standardised incidence rates (ASIRs) and age-standardised mortality rates (ASMRs) of stroke worldwide, the absolute number of cases and deaths has continued to increase markedly due to population ageing, lifestyle changes, and other complex factors3. However, while ASMRs have decreased, the decline in ASIRs has been more modest, indicating that stroke prevention efforts may face additional challenge compared to treatment4. Therefore, despite advances in treatment and management, the global burden of stroke continues to increase, with prevention efforts facing numerous challenges. In-depth research on the trends in stroke and the development of effective prevention strategies are crucial for reducing future incidence rates and mortality risk.

The global geographical distribution and trends of stroke cases show notable differences, particularly in East Asia. Asia accounts for about two-thirds of the global population, and in 2002, East Asia accounted for 60% of global stroke deaths5. According to a 2022 report by the American Heart Association, East Asia has the highest ASIR of stroke globally6. China, Japan, and South Korea share common risk factors such as rapidly ageing populations and increasingly Westernised diets. However, differences in healthcare resource allocation, cultural lifestyle practices, and the implementation of prevention policies have led to variations in stroke epidemiology and disease burden across these countries. For example, although ASIRs and ASMRs have decreased in China in recent years, it still bears the highest global stroke burden and is the main contributor to the upward trend in East Asia7,8. In Japan, stroke incidence and mortality have declined steeply over the past half-century, but these reductions have plateaued in recent decades9,10. South Korea, meanwhile, maintains low ASIRs and experienced a downward trend in the ASMR11 but experiences a notable shift toward younger age at stroke onset, indicating an increasing economic burden12. Comparing the epidemiological characteristics among these three countries can help clarify the impact of health policies, prevention measures, and demographic shifts, and offer evidence for targeted, region-specific strategies. Additionally, the United States serves as an example of a high-resource, high-income country with comprehensive stroke prevention and control measures. Its successful reduction in stroke ASIRs and ASMRs3—mainly attributed to long-term management of known risk factors such as smoking and hypertension13,14—provides a valuable reference for evaluating policy effectiveness in other regions. A systematic comparison of stroke trends in China, Japan, and South Korea with those in the United States and globally facilitates a deeper understanding of how varying levels of economic development, health systems, and prevention strategies influence stroke epidemiology and burden, and highlights the unique challenges faced by East Asia.

Although previous studies have explored global trends and regional differences in stroke8,15 research on the disease burden of stroke in Asia has been limited to analyses of a specific stroke subtype without future trend projections. Long-term trends and future projections of stroke incidence and mortality in East Asian countries (particularly China, Japan, and South Korea) remain understudied. To fill this research gap, we used the Global Burden of Disease (GBD) 2021 data to analyse the trends in stroke incidence and mortality from 1990 to 2021 in China, Japan, and South Korea, compare the trends with those in the U.S. and globally, and project trends to 2035. This multi-country analysis not only comprehensively reveals the progress and existing challenges in stroke prevention and control in East Asia, but also provides international insights and robust evidence to inform the development of regionally tailored public health strategies.

Methods

Data sources

The data for this study were obtained from the Global Health Data Exchange (GHDx) tool (https://vizhub.healthdata.org/gbd-results). The GBD 2021 data generation methodology comprised the following steps: incidence data integration employed the DisMod-MR 2.1 model, synthesising multiple epidemiological data sources (including systematic reviews, disease registries, hospital records, and surveillance system microdata) through cascading modelling and healthcare access and quality (HAQ) index covariates to adjust for diagnostic discrepancies and geographical biases. Mortality data generation utilised the cause-of-death ensemble model (CODEm), which synthesised 22 data sources (e.g., cause-of-death registries, verbal autopsies, and excess mortality records) under an ensemble learning framework to weight submodel outputs, generating 95% uncertainty intervals (UIs) from 1,000 Markov chain Monte Carlo (MCMC) draws. Population data calibration applied the Bayesian hierarchical cohort component model (BCCMP), integrating census, migration, and fertility/mortality data from 1950. Age structures were unified via super-regional prior constraints to ensure demographic accuracy16,17. The GBD 2021 database provides comprehensive assessments of the health impacts of 371 diseases and injuries and 88 risk factors since 1970, covering 204 countries and territories worldwide. We extracted stroke data directly from the GBD 2021 database, including new cases, deaths, crude incidence rates (CIRs), crude mortality rates (CMRs), age-standardised incidence rates (ASIRs), and age-standardised mortality rates (ASMRs), for China, Japan, South Korea, the U.S., and globally, from 1990 to 2021. ASIRs and ASMRs were calculated in the GBD database using the direct standardisation method with the GBD world standard population as a reference, which ensures comparability across countries and over time. All extracted data were stratified by country/region, year, sex (male, female), and 17 age groups (15–19 years to 95 + years, in 5–year intervals). Individuals under 15 years were excluded because of low stroke occurrence18. In addition, to project the stroke burden from 2022 to 2035 using the BAPC model, we extracted age-specific population projections from the IHME database (https://ghdx.healthdata.org/record/ihme-data/global-population-forecasts-2017-2100/).

Statistical analysis

Joinpoint regression model

A joinpoint regression model was used to assess stroke trends in ASIRs and ASMRs from 1992 to 2021 in China, Japan, and South Korea, with U.S. and global trends included for comparison. The model analysed long-term trends by dividing the entire period into multiple segments and evaluating the trend within each segment. Optimal inflection points were identified using the Monte Carlo permutation test, which permitted up to five joinpoints19. The annual percent change (APC) and average annual percent change (AAPC) were calculated with 95% confidence intervals (CIs)20. If the APC and AAPC values were identical, no inflection points were identified. If APC or AAPC > 0, an increasing trend in the stroke ASIR or ASMR was indicated, meaning the entire period was a single segment. if APC or AAPC < 0, a decreasing trend was indicated; and if APC or AAPC = 0, a stable trend was indicated.

Age-period-cohort model

This study used the age-period-cohort model, based on Poisson regression, to analyse the independent effects of age, period, and birth cohort on stroke incidence and mortality within each country or region (China, Japan, South Korea, the U.S., and globally). For this analysis, stroke incidence and mortality rates were stratified into 17 age groups (15–19, 20–24, …, 90–94, \(\:\ge\:\) 95) and six periods (1992–1996, 1997–2001, …, 2017–2021). Data from 1990 to 1992 were excluded due to insufficient information. The model is specified as follows:

$$\:\text{ln}\left({R}_{ijk}\right)=\mu\:+{\alpha\:}_{i}+{\beta\:}_{j}+{\gamma\:}_{k}+{\epsilon\:}_{ijk}$$

where \(\:{R}_{ijk}\) represents the stroke incidence or mortality rate (per 100,000 population) for the \(\:i\)-th age group, \(\:j\)-th period, and \(\:k\)-th birth cohort; \(\:\mu\:\) is the intercept; and \(\:{\alpha\:}_{i}\), \(\:{\beta\:}_{j}\) and \(\:{\gamma\:}_{k}\) denote the regression coefficients for age, period, and birth cohort effects, respectively. \(\:{\epsilon\:}_{ijk}\) denotes the error term. Given the inherent collinearity among age, period, and cohort, we adopted the intrinsic estimator (IE) method to obtain unique and valid effect estimates21,22. The age, period, and cohort effects estimated by this method represent the stroke incidence or mortality for a specific age group, period, or birth cohort, with the reference group defined as the average level across all groups within each country and region23. For intuitive interpretation, regression coefficients were exponentiated to yield relative risks (RR = exp(coefficient)). Thus, RR represents the risk relative to the overall average: RR > 1 indicates a higher risk than average, RR < 1 indicates a lower risk, and RR = 1 indicates risk equal to the average.

Bayesian age-period-cohort (BAPC) model

The Bayesian age-period-cohort (BAPC) model was used to predict trends in stroke incidence and mortality, including ASIRs, ASMRs, new cases and deaths from 2022 to 2035 for China, Japan, South Korea, the U.S., and globally. As a statistical tool for analysing and predicting disease incidence and mortality data24 the BAPC model is based on integrated nested Laplace approximations (INLA) algorithms. The BAPC model incorporates age, period, and cohort effects, as well as baseline effects, enabling predictions of future stroke trends across different age groups, periods, and birth cohorts. In this study, we estimated and projected age-specific stroke incidence and mortality rates by country and sex. The projected annual new cases and deaths were calculated by multiplying the estimated incidence and mortality rates by the projected population in each age group, based on IHME Global Population Forecasts (2017–2100). Published studies have shown that compared with models such as general summation, smoothing spline, and Nordpred, the BAPC model achieves higher accuracy in short- and medium-term forecasts25,26 leading to more reliable disease burden projections.

The joinpoint regression model was conducted using Joinpoint Software (version 4.9.1.0, National Cancer Institute). All other analyses were performed using R (version 4.3.2): the “Epi” package was used for age-period-cohort decomposition, the “BAPC” and “INLA” packages for Bayesian predictions, and the “tidyverse” and “ggplot2” packages for data organisation and visualisation. All the statistical tests were two-sided, with a significance level of 0.05.

Results

Descriptive analysis of stroke incidence and mortality

During 1990–2021, the ASIRs decreased in China, Japan, and South Korea, consistent with the U.S. and global trends. South Korea exhibited the largest reduction (68% in both sexes), followed by Japan (36% in males, 42% in females), and then China (4% in males, 17% in females) (Table 1). Throughout the period, the ASIRs of all three Asian countries remained higher than those of the U.S. Specifically, China’s ASIRs consistently exceeded the global average; Japan’s were below the global average but above those of the U.S.; and South Korea’s ASIRs became lower than China’s after 2004 for males and after 2001 for females, yet remained higher than those in Japan and the U.S. (Fig. S1a).

Table 1 The cases and age-standardised rates (per 100,000 population) of stroke incidence and mortality by sex in China, Japan, South Korea, the United States, and globally from 1990 to 2021.
Full size table

The ASMR of stroke decreased across all countries and regions studied. China saw reductions of 34% in males and 52% in females; Japan experienced reductions of 64% in males and 72% in females; and South Korea had the largest reductions, with decreases of 82% in males and 84% in females, thereby narrowing the gap with Japan (Table 1). China’s ASMRs were above the global and U.S. rates, while Japan’s were below the global rate and, after 2002, fell below the U.S. rates for females. South Korea’s male and female ASMRs fell below global rates after 2002 and 2001, respectively, but remained higher than those of the U.S. (Fig. S1b).

Additionally, in China, Japan, and South Korea, stroke ASIRs and ASMRs were higher in males than in females. Notably, in China, the male–female gap for both rates not only persisted but also widened over time.

Joinpoint regression analysis of stroke incidence and mortality

A joinpoint regression analysis showed that during the study period, South Korea had the fastest decline in the ASIR (males AAPC = − 3.62%, females AAPC = − 3.59%), followed by Japan (males AAPC = − 1.48%, females AAPC = − 1.75%), both countries declining more rapidly than the U.S. (males AAPC = − 1.43%, females AAPC = − 1.18%) and the global average (males AAPC = − 0.70%, females AAPC = − 0.94%). In contrast, China’s ASIRs declined more slowly (males AAPC = − 0.19%, females AAPC = − 0.62%), with decreases smaller than those observed in the U.S. and globally. Specifically, China’s ASIR for females experienced the fastest decline between 2005 and 2012 (APC = − 2.62%), and China’s ASIR for males declined most rapidly between 2019 and 2021 (APC = − 1.96%). Notably, there was a brief increase in ASIRs in 2014 and 2015 for Chinese males and females (APC = 0.82% and 2.83%, respectively), which decreased again after 2019. In Japan, the steepest declines in ASIRs occurred between 1996 and 2000 for males (APC = − 4.03%) and between 1995 and 2000 for females (APC = − 4.32%). South Korea’s ASIRs for both males and females began to decrease rapidly in 2003 and 1996, respectively, but the decline slowed after 2010 (Fig. 1a and Table S1).

The trends in ASMRs paralleled those in ASIRs. During the study period, South Korea had the fastest decline in stroke ASMR (males AAPC = − 5.36%, females AAPC = − 5.88%), followed by Japan (males AAPC = − 3.20%, females AAPC = − 4.10%) and China (males AAPC = − 1.37%, females AAPC = − 2.31%), all of which exceeded the declines in the global average (males AAPC = − 1.34%, females AAPC = − 1.85%) and the U.S. (males AAPC = − 1.33%, females AAPC = − 1.14%). Specifically, the most pronounced decline in the stroke ASMR in China occurred between 2004 and 2007, with a change of − 5.09% for males and − 6.82% for females, followed by a slower decline after 2007. China’s stroke ASMR for females declined rapidly between 2010 and 2014 (APC = − 5.29%), with the ASMR for males also declining (APC = − 3.52%) during the same period. Notably, from 2015 to 2019, the stroke ASMRs for both Chinese males and females declined, although the stroke ASIRs increased during this period. In Japan, the most notable decline in stroke ASMR occurred in the early 1990s, with males experiencing a -4.40% change from 1990 to 1993 and females experiencing a -6.56% change from 1990 to 1992. After this time, the decline slowed, but from 2001 to 2017, the stroke ASMR for males in Japan resumed a steady decline (APC = − 3.54%), and the stroke ASMR for females also declined rapidly during 1995–2002 and 2002–2016 (APC = − 6.20% and − 3.86%, respectively). South Korea maintained a persistently high rate of decline for both sexes after 1990 (Fig. 1b and Table S2).

Fig. 1
figure 1

Using the joinpoint regression model, trends in age-standardised rates of stroke incidence (a) and mortality (b) by sex in China, Japan, South Korea, the United States (U.S.), and globally from 1990 to 2021. The observed values are indicated by points, and the solid lines represent model estimates.

Full size image

Age-period-cohort analysis of stroke incidence and mortality

The age-period-cohort model showed that the RR of stroke incidence increased with age in both sexes across China, Japan, South Korea, the U.S., and globally. For Chinese males, the incidence RR peaked at ages 80–84 before decreasing, whereas in females, it continued to rise at older ages. RRs > 1 were observed in Japan and South Korea for males aged \(\:\ge\:\) 45 years and females aged \(\:\ge\:\) 50 years, and in China for males aged \(\:\ge\:\) 55 years and females aged \(\:\ge\:\) 60 years (Fig. 2a and Table S3).

Stroke mortality RR also increased with age in the aforementioned countries and regions. In China, Japan, and globally, the stroke mortality RR for males rose with age, peaking at ages 90–94 before declining, while for females, the mortality RR continued to rise in the oldest age groups. In South Korea and the U.S., stroke mortality RR increased with age, peaking at ages 95–99. Higher mortality RRs (RRs > 1) were faced by Chinese males and females, by Japanese and South Korean males aged ≥ 55 years, and by Japanese and South Korean females aged ≥ 60 years (Fig. 2b and Table S4).

Fig. 2
figure 2

Age-specific relative risks of stroke incidence (a) and mortality (b) by sex in China, Japan, South Korea, the United States (U.S.), and globally, based on the age-period-cohort model.

Full size image

After controlling for age and cohort effects, stroke incidence RR increased over time in China, Japan, the U.S., and globally, but decreased in South Korea. From 1992 to 1996 to 2017–2021, RR increased from 0.77 to 1.31 (a 70% increase) for Chinese males, and from 0.85 to 1.23 (a 45% increase) for Chinese females. In Japan, the RR for females declined until 2007–2011, before rising again from 2017 to 2021 (RR = 1.15). In South Korea, the incidence RR decreased by 30% for males and by 34% for females from 1992 to 1996 to 2017–2021 (Fig. 3a and Table S3).

Period effects on mortality revealed increased RRs in China, the U.S., and globally for both sexes and in Japan for males. In contrast, decreased RRs in Japan for females and in both sexes in South Korea. In China, the mortality RR for males rose above 1.0 from 2002 to 2006, after which the RR for males continued to increase, whereas that for females declined from 2007 to 2011 to 2012–2016 and then rose again from 2017 to 2021. The mortality RR in Japanese females declined overall, although it increased from 2007 to 2011 to 2017–2021. From 1992 to 1996 to 2017–2021, South Korea’s stroke mortality declined by 47% for males and 53% for females (Fig. 3b and Table S4).

Fig. 3
figure 3

Period-specific relative risks of stroke incidence (a) and mortality (b) by sex in China, Japan, South Korea, the United States (U.S.), and globally, based on the age-period-cohort model.

Full size image

After controlling for age and period effects, stroke incidence and mortality RRs in China, Japan, South Korea, the U.S., and globally exhibited a pattern of rapid decline followed by slower decline across successive birth cohorts during the study period. Notably, China experienced a substantial reduction in incidence RR, with the RR decreasing by 94% in males and 93% in females, reaching below 1.0 in males born between 1952 and 1956 and females born between 1957 and 1961. Similar declines were observed among the 1940–1950 s birth cohorts in Japan and South Korea (Fig. 4a and Table S3).

Mortality RRs declined by 98% for males and 97% for females in Japan, 97% for males and 98% for females in China, and also decreased in South Korea, the U.S., and globally. Among Chinese males, Japanese females, and South Korean females, the RR of stroke mortality in the earliest birth cohorts increased slightly, but this trend reversed in later cohorts. All countries demonstrated a marked decrease in mortality RR among cohorts born after 1950 (Fig. 4b and Table S4).

Fig. 4
figure 4

Cohort-specific relative risks of stroke incidence (a) and mortality (b) by sex in China, Japan, South Korea, the United States (U.S.), and globally, based on the age-period-cohort model.

Full size image

Projected trends of stroke incidence and mortality from 2022 to 2035

The predicted ASIRs for males and females in China and South Korea are expected to decrease, consistent with global predictions, while those in Japan are projected to increase, similar to the trend observed in the U.S. The predicted number of new cases is expected to increase for both males and females in China and Japan, and for females in South Korea. The highest predicted increase is in Chinese females, from 1,790.81 thousand in 2022 to 2,760.95 thousand in 2035, a 1.54-fold increase (Fig. 5a and Table S5).

The predicted ASMRs for China, Japan, and South Korea are expected to decrease, consistent with global projections, but in contrast to trends in the U.S. The predicted number of deaths is projected to increase for males and females in China, for males and females in South Korea, and for males in Japan, but is expected to decrease for Japanese females. South Korea is projected to show the largest relative increase in predicted deaths, increasing 1.56-fold in males and 1.58-fold in females. Additionally, the predicted ASIR and ASMR levels are higher in males than in females (Fig. 5b and Table S5).

Fig. 5
figure 5

Using the BAPC model, trends of the predicted number of incidence cases and age-standardised incidence rate (ASIR) (a), and the predicted number of deaths and age- standardised mortality rate (ASMR) (b) of stroke by sex in China, Japan, South Korea, the United States, and the global level from 2022 to 2035. The solid lines indicate the observed ASIR or ASMR, the dashed lines indicate predicted ASIR or ASMR by the BAPC model, and the vertical dashed grey lines indicate year of 2021.

Full size image

Discussion

This study used GBD 2021 data to analyse stroke trends from 1990 to 2021 in China, Japan, and South Korea in comparison with those in the U.S. and at the global level, and predicted trends from 2022 to 2035. By providing a direct comparison with both U.S. and global trends, this analysis highlights both commonalities and regional disparities in the evolution of the stroke burden across major East Asian countries, high-income nations, and the world as a whole. From 1990 to 2021, stroke ASIRs and ASMRs in China, Japan, and South Korea decreased overall, consistent with U.S. and global patterns, with South Korea showing the largest decrease. The age-period-cohort model indicated that age and birth cohort were the main factors influencing stroke incidence and mortality in these regions. Stroke incidence and mortality increased with age in all three countries. After controlling for age and cohort effects, stroke incidence and mortality RRs decreased over time in South Korea, whereas those in China (both sexes) and in Japan (males) gradually increased. In Japan, the stroke incidence RR of females increased over time, whereas mortality RR decreased. After controlling for age and period effects, stroke incidence and mortality RRs generally decreased with later birth cohorts, although earlier cohorts of Chinese males and South Korean and Japanese females experienced rising mortality rates.

Analyses of sex differences revealed that both ASIRs and ASMRs were higher in males than females, aligning with the findings of Qiang He et al.8. Potential reasons include physiological factors, such as the protective effect of oestrogen in females27; and lifestyle disparities: males exhibit higher smoking, alcohol consumption, and obesity rates—all stroke risk factors. In China, male smoking rate (52%) vastly exceeds females (3%)28 and globally, the rate of heavy drinking is 6.7% in males compared with 0.6% in females29. Higher hypertension rates in males and a greater proportion of men performing manual labour increase stroke risk30. In contrast, females generally have higher health literacy and are more likely to focus on health and adopt healthier lifestyles31. Furthermore, cross-country comparisons show that the stroke burden among males—in terms of incidence and mortality—is particularly pronounced in China, and the gender gap widened further by 2021. This highlights the different risk profiles and health awareness patterns between Eastern and Western countries.

This research, like previous studies, found a declining trend in global stroke ASIR and ASMR3,8. Globally, the decline, especially in high-income countries, is due to improved risk factor management, better hypertension control, enhanced acute stroke care, and raised public health awareness3. However, many low- and middle-income countries are still experiencing higher rates and slower progress than high-income countries like the U.S32. These differences highlight the influence of health system capacity, preventive services accessibility, public health infrastructure, and sociocultural factors on stroke outcomes. Although the gaps in ASIR and ASMR between Japan and the U.S. have narrowed over time, China and South Korea still have higher rates than the U.S. This is mainly due to higher hypertension prevalence and poorer awareness, treatment, and control rates in these three East Asian countries14,33 as well as high-sodium diets34. The U.S. established advanced stroke care infrastructure earlier35 enabling more comprehensive care. Sociocultural factors differ: the U.S. ‘FAST’ education promotes public awareness36 while health consciousness and stroke symptom recognition are comparatively lower in East Asia37,38. Thus, China, Japan, and South Korea need to strengthen risk factor management, optimise healthcare resources, and improve public health awareness to reduce stroke incidence and mortality.

Between 1990 and 2021, the ASIR and ASMR trends in China, Japan, and South Korea exhibited distinct patterns, likely influenced by health control policies implemented in each country. China’s ‘Stroke Screening and Prevention Program’ (2009) improved early diagnosis and outcomes, potentially explaining the temporary rise in ASIRs in 2014 and 2015, whereas mortality declined. The establishment of the China Stroke Prevention Project Committee (2011) strengthened nationwide prevention39 potentially serving as a key factor in the ongoing mortality decline. The ‘Healthy China 2030’ initiative (2016) and the Stroke Emergency Map improved public awareness and the emergency response system40 which may have contributed to the decrease in incidence and mortality after 2019. Since 2002, the Korean Stroke Society’s stroke registry may have promoted epidemiological research and treatment evaluation, contributing to reductions in incidence and mortality41. In 2011, the Korean Ministry of Health and Welfare launched the Regional Comprehensive Stroke Centre programme, which was strengthened from 2014 to 201841, these developments may explain why South Korea has effectively controlled the incidence and mortality of stroke. The Japanese stroke database, established in 1999 42, may have promoted research and contributed to reducing stroke incidence and mortality. The 2016 five-year plan by the Stroke Society and Circulation Society, followed by the Stroke and Cardiovascular Disease Control Act (2018) and the National Plan for Stroke and Cardiovascular Disease Prevention (2020)42,43,44 may have supported these reductions. Additionally, in the mid-1990s, the U.S. approved intravenous recombinant tissue plasminogen activator (rt-PA) for acute stroke treatment. The Brain Attack Coalition (BAC), established in 1996, initiated stroke centre infrastructure development. It put forward the stroke centre concept in 2000, which led to the American Stroke Association’s certification programme (2003, based on BAC’s recommendations), implemented from 200435.

The age effect refers to the impact of physiological ageing or changes in life stages on stroke incidence and mortality. The age-period-cohort model in this study showed that stroke incidence and mortality RRs increased with age in both sexes in China, Japan, and South Korea, possibly related to ageing. Research indicates that ageing affects cerebrovascular diseases through various physiological mechanisms45. As people age, blood vessels undergo degenerative changes, reducing elasticity and increasing fragility, thereby elevating the risk of stroke46. Moreover, stroke is closely related to chronic conditions, including hypertension, diabetes, and hyperlipidaemia47,48. During ageing, physiological functions decline, leading to a higher risk of diseases that indirectly affect vascular health, and thus contribute to stroke occurrence49. Once a stroke occurs, the reduced recovery ability in older adults further increases the mortality RR50. Therefore, enhancing health education for middle-aged and elderly populations and raising stroke prevention awareness are crucial for reducing incidence and mortality RRs.

The period effect reflects the impact of population structure changes, lifestyle shifts, or medical technological advancements during specific periods on stroke incidence and mortality. In our study, period effects showed that stroke incidence and mortality RRs increased over time in China and Japan, whereas South Korea experienced a decline. Our period effect trends in China differ from those reported in previous studies applying alternative age-period-cohort modelling strategies18,51. This inconsistency is mainly attributable to the different methods used to address the ‘identifiability’ problem inherent in age-period-cohort models. Different statistical methods, even when applied to identical datasets, may yield divergent period effect trends52. While constrained generalised linear model (CGLIM) method (as used in previous studies) and IE method (employed in our analysis) may allocate observed trends differently among age, period, and cohort components, our findings closely align with the national crude trends of stroke incidence and mortality in China. In addition to methodological differences, updates in GBD data sources and underlying statistical frameworks may also contribute to variations in findings between studies53. In contrast to our findings in Japan, Okui (2020) reported that stroke mortality RR showed a declining period effect54possibly due to differences in population and data sources. Okui examined only two communities in Osaka and Akita, focusing on residents aged 40 and above, whereas our study included all Japanese aged 15 and over. Okui used Vital Statistics for precise, local data; we used GBD 2021 data, which, though broader, may lack local specificity. These trends in China and Japan are likely driven by population ageing and lifestyle changes55,56. As the older population increases, stroke incidence and mortality may rise, aligning with global ageing trends57. In China, unhealthy lifestyles, including high intake of animal foods, lack of exercise, and the high prevalence of smoking and alcohol consumption may increase stroke risk58,59. Improvements in diagnostic technology and health registries may also have increased stroke reporting60. China’s low hypertension awareness, treatment, and control rates, and a high hypertension prevalence in stroke survivors likely worsen stroke risk61. In Japan, hypertension management has improved14and smoking rates have steadily declined. However, smoking rates are still high, and the prevalence of stroke risk factors, such as obesity and metabolic disorders, has increased sharply over time62. In contrast, South Korea’s declining stroke incidence and mortality RRs have been attributed to successful risk factor management. Smoking rates among men have decreased11 and hypertension awareness, treatment, and control rates have markedly increased, even surpassing those in the U.S. in certain age and sex groups60. South Korea’s stroke mortality RR, especially from haemorrhagic stroke, has steadily decreased63. Improving lifestyles, particularly through better hypertension management, dietary changes, and physical activity, is likely crucial for stroke prevention. Notably, the increasing period effect observed in the age-period-cohort model appears to contradict the continuous decline in age-standardised rate (ASR); in fact, this reflects the complementary nature of these methods—ASR shows overall trends after adjusting for age, while the period effect may capture offsetting influences such as population ageing and advances in diagnostics.

The cohort effect reflects changes in stroke incidence and mortality among males and females born in different years, due to differences in social, economic and other factors. After controlling for age and period effects, the analysis showed a declining trend in stroke incidence and mortality RRs across China, Japan, and South Korea, likely related to improvements in education, healthcare, and health awareness. China had the most pronounced decrease in incidence RR among the three countries, with a notable reduction in mortality RR. In recent years, China’s economic growth and strengthened healthcare system have enhanced health security. The increase in medical resources and technological advancements was especially notable from 2010 to 2020. During this period, substantial growth in intravenous thrombolysis, carotid endarterectomy, and carotid artery stenting likely played key roles in improving early diagnosis and treatment, thereby reducing incidence and mortality RRs64. Additionally, widespread health education and environmental management have contributed to reducing adverse health factors, further lowering incidence and mortality RRs in later-born cohorts. In South Korea, the decreasing stroke incidence and mortality RRs are likely due to progress in stroke prevention, early detection, and medical interventions, especially the control of risk factors in hypertension and diabetes. Establishing efficient emergency systems, stroke units, and public education has further reduced stroke risk41. Similarly, Japan’s success can be attributed to its stroke screening, health education, and universal health insurance system, with advances in medical technology enhancing the public’s ability to manage stroke risk factors10,65.

The BAPC predictive model showed that by 2035, both new cases and deaths in Japan, China, and South Korea will rise, with the exception of deaths among females in Japan, likely due to population growth and ageing. The ageing trend is unlikely to reverse in the short term66,67 and as life expectancy increases and the proportion of elderly grows57,68 declining immunity and rising comorbidities will worsen the health burden of stroke. Except for a slight increase in the ASIRs for both sexes in Japan, the ASMRs for both sexes in Japan and the ASIRs and ASMRs in China and South Korea are expected to continue declining, providing a positive signal for stroke prevention in China and South Korea. The increased ASIR in Japan may be related to genetic susceptibility and high sodium intake, both key stroke risk factors48,69. The predictions suggest that in countries with high levels of ageing, despite strong prevention measures, stroke risk management will need strengthening.

By situating these long-term trends within a cross-national framework, this study provides important epidemiological evidence for global stroke surveillance and prevention. Although this study offers novel multinational insights into stroke burden and its determinants, several limitations should be noted: First, the analysis relies on the GBD database, which, despite rigorous quality control, may be affected by data collection, reporting, and coding issues, so model estimates may contain errors. Second, the study focuses on overall trends and sex differences, without exploring other potential factors, such as stroke subtypes, limiting the comprehensive understanding of stroke. Future research should explore subtype analysis. Third, the joinpoint regression and age-period-cohort models are descriptive and cannot establish causal links, limiting their interpretability. Additionally, life expectancy in East Asian countries typically does not exceed 85 years, so the results for the oldest age groups may be less accurate and should be treated with caution. Finally, the predictive models rely on assumptions, and future demographic, technological, and policy changes are not fully predictable, requiring caution when interpreting the results.

Conclusion

This study examined trends in stroke incidence and mortality from 1990 to 2021 in China, Japan, and South Korea, in comparison with U.S. and global data, and projected trends for 2022–2035. Overall, the ASIRs and ASMRs declined in the three countries, with South Korea experiencing the greatest decline, reflecting the success of its interventions. An age-period-cohort model analysis indicated that age and cohort effects were key determinants of stroke trends: older age groups had the highest incidence and mortality RRs, and incidence and mortality RRs decreased in later birth cohorts in China, Japan, and South Korea. The effectiveness of stroke prevention and control is closely linked to risk factor management, health policies, and public health awareness. Projections for the next 14 years suggest that, despite existing interventions, stroke risk management must be strengthened in countries with ageing populations. This study indicates that international comparison and experience can inform more flexible and targeted prevention strategies. Ongoing collaboration, policy refinement, and tailored interventions are needed to continue reducing the stroke burden and contribute to global stroke prevention efforts.