Study data on the disease burden of stroke attributable to HSBP between 1990 and 2019, including annual cases of deaths/DALYs and their age-standardized rates of mortality (ASMR) and DALY (ASDR) by sex, age, and location, were extracted from the GBD 2019 [a publicly available resource (http://ghdx.healthdata.org/gbd-results-tool) recorded 369 diseases and injuries in 21 regions and 204 countries and territories]. This study complied with the Guidelines for Accurate and Transparent Health Estimates Reporting (GATHER) guidelines.

A SBP of at least 110–115 mmHg was defined as HSBP according to a theoretical minimum exposure aimed at capturing the maximum-caused burden, which was related to a higher risk of cardiovascular diseases (particularly stroke). The detailed data selection and input have been described previously (6). Each cause and related state of stroke, covering the years 1990–2019, was defined according to the World Health Organization (WHO) criteria and the International Statistical Classification of Diseases (ICD, the 10th Related Health Problems) (8), and the Cause List Mapped to the ICD codes of I60–I62, I62.9–I64, I64.1, I65–I69.998, Z82.3, and G45–46.8 (9, 10). Stroke attributable to HSBP was defined as a stroke combining exposure to the risk-related HSBP (11).

Disease burden was quantified by the numbers of deaths, DALYs, ASMR, and ASDR, with 95% uncertainty intervals (UI). The socio-demographic index (SDI), a composite index of per capita income, educational attainment, and total fertility rate of all areas, can be derived from the Global Health Data Exchange (GHDx, http://ghdx.healthdata.org), wherein the socio-demographic development status in different countries or regions are identified, and the geometric mean on a scale of 0–1 is calculated, by which five groups are low, low-middle, middle, middle-high, and high countries (12). Among 204 countries and territories, 5 SDIs and 21 GBD regions were available to assess the disease burdens by location, according to epidemiological and geographical conditions. More details on the SDI calculation have been described previously (4, 13), wherein ages were categorized into 15 groups from 25 to 94 years and older than 95 years.

The GBD 2019 comparative risk assessment approach was used to estimate population fractions. A no-weighted mean of the GBD year's age-related proportional distributions for national locations, with more than 5 million populations in the GBD year to update the global population age standard (4), was used to calculate ASR. In brief, several parameters were summed up to calculate the following: the age-standardized rate products (a_i, wherein i is the ith age class), a number (or the weight) of persons (w_i) in the same age subgroup I (a standard population reference), and a dividend of the standard population weights (14, 15); in the formula of ASR=∑i=1A⁡aiwi/∑i=1A⁡wi×100,000, ASMR (per 100,000 person-years) and ASDR (per 100,000 people) estimates are presented according to WHO's standard population reference (16), with 95% UIs for every metric.

The estimated annual percentage change (EAPC) can quantify the secular trend of ASRs according to a previously well-established formula (17): ln(y) = α + βx + ε, wherein y = ln(ASR), x = calendar year, and ε = the error term. Its 95% confidence interval (CI) was from a linear regression model; the absolute EAPC value represents the rate changes over time; the smoothing splines model fitted the association between EAPC and SDI; ASMR or ASDR was recognized to be on the rise if the EAPC estimation and 95% CI > 0, a decreasing trend if the EAPC estimation and 95% CI < 0, and stable if 95% CI = 0 (4). The Pearson correlation analysis was used to measure the degree of correlation between two variables; for the influential factors for EAPC, the associations between EAPC and SDI, ASR, and human development index (HDI) at the national level were assessed (18). Based on the social development and health outcomes in 2019 and the ASR in 1990, the HDI in 2019 was referred to as a national healthy condition. A hierarchical cluster analysis was conducted in four categories of countries and territories ((1) minor increase; (2) remain stable; (3) minor decrease; (4) significant decrease) in accordance with the temporal trends in ASMR and ASDR by a join point regression model (version 4.9.1.0), wherein the Z Test was used, with a two-sided P value of 0.05 as a significant level. A traditional linear regression was used when H0 (a segmentation point as 0) was first assumed, wherein H1 was a segmentation with at least 1 point; once H0 was rejected, the significance of the segment points of 1 − n was estimated. According to the estimate of annual percentage change (APC) (19), the time series and the ASRs were set up and taken as independent and dependent variables, respectively. R (version 4.2.1; R Core Team, Vienna, Austria) was used in the visualization of indicators and linear correlations and coefficients between the ASRs and SDI; P < 0.05 was defined as significance; draw maps were conducted in a visualization tool [GBD Compare, IHME Viz Hub (healthdata.org)].

Globally, 52.57% of deaths and 55.54% of DALYs from stroke were due to HSBP, with approximately 3.44 million in 2019 and 2.36 million in 1990 for deaths, and 79.6 million in 2019 and 58.8 million in 1990 for DALYs; furthermore, men bore more deaths and higher DALYs (Figure 1). The corresponding ASMRs (66.96 per 100,000, 95% UI: 55.11–78.88 in 1990 vs. 43.6 per 100,000, 95% UI: 36.19–51.06 in 2019) and ASDRs (1,419.59 per 100,000, 95% UI: 1,197.05–1,629.35 in 1990 vs. 969.4 per 100,000, 95% UI: 823.77–1,110.2 in 2019) experienced a decline by 34.89% and 31.71%, respectively, with the highest levels of the ASMR-related EAPC (−1.99, 95% CI: −2.12 to −1.85) and ASDR-related EAPC (−1.81, 95% CI: −1.93 to −1.69) in women (Table 1 and Figure 2).

Significant geographical variations in the GBD region-related deaths and DALYs showed disparities over the past three decades. Middle SDI regions had the most number of deaths (12.64 × 10⁵, 95% UI: 10.51–14.7 in 2019 vs. 6.61 × 10⁵, 95% UI: 5.36–7.81 in 1990) and DALYs (300.62 × 10⁵, 95% UI: 254.31–345.66 in 2019 vs. 168.37 × 10⁵, 95% UI: 138.96–198.53 in 1990), with both accounting levels over 35% worldwide; middle and low-middle SDI regions had the highest levels of ASMR (57.07 per 100,000, 95% UI: 47.02–67.25) and ASDR (1,215.12 per 100,000, 95% UI: 1,028.78–1,396.49). Notably, the ASMR and ASDR between 1990 and 2019 experienced a decline across the five SDI regions, with the largest decline in the ASMR-related EAPC (−3.86, 95% CI: −4.08 to −3.65) and ASDR-related EAPC (−3.38, 95% CI: −3.58 to −3.18) in high SDI regions (Table 1).

Across the GBD regions in 2019, East Asia had over 33% of global deaths (11.58 × 10⁵, 95% UI: 9.16–14.08) and DALYs (259.59 × 10⁵, 95% UI: 210.4–310); Central Asia had the highest levels of ASMR (89.6 per 100,000, 95% UI: 74.61–104.23) and ASDR (1,866.16 per 100,000, 95% UI: 1,575.09–2,145.87). The ASMR-related EAPCs between 1990 and 2019 were <0, with the largest decline in high-income Asia Pacific (EAPC = −5.17, 95% CI: −5.38 to −4.96); the ASDR-related EAPCs in most of the GBD regions between 1990 and 2019 were <0, with only one level >0 in Oceania (EAPC = 0.16, 95% CI: 0.01–0.3) and the largest decline in high-income Asia Pacific (EAPC = −4.34, 95% CI: −4.53 to −4.16) (Table 1).

The percentages of deaths and DALYs in 1990 showed an approximately twofold difference, with their highest levels in Eastern and Central Europe, Central Asia, and southern sub-Saharan Africa but their lowest levels in Andean Latin America, Oceania, and southern Latin America. These data in 2019 accounted for over 40% of all GBD regions, with the highest percentages of deaths in Central Asia and DALYs in Southeast Asia (Figure 1).

China ranked the highest in the number of deaths (1,126,038, 95% UI: 889,923–1,370,837) and DALYs (25,176,256, 95% UI: 20,271,928–30,148,436) (Table 2). North Macedonia (142.76, 95% UI: 105.37–187.27), Mongolia (134.57, 95% UI: 101.96–174.56), and Indonesia (128.32, 95% UI: 105.53–149.57) shared the top three ASMRs; Nauru (3,223.45, 95% UI: 2,461.86–4,130.88), Mongolia (3,111.56, 95% UI: 2,378.13–4,060.13), and Vanuatu (3,056.07, 95% UI: 2,281.04–4,064.91) shared the top three ASDRs in 2019 (Figures 3A,B and Table 2). Azerbaijan had the highest rise in ASMR (ASMR-related EAPC = 1.79, 95% CI: 1.39–2.19) and the Philippines had the highest rise in ASDR (ASDR-related EAPC = 1.94, 95% CI: 1.34–2.55) (Figures 3C,D and Table 2).

According to the cluster analyses, 32, 110, 59, and 3 countries or territories were grouped into “minor increase” (Indonesia, the Philippines, Omen, Kuwait, etc.), “remained stable” (Togo, Iraq, Cuba, Sudan, etc.), “minor decrease” (Japan, Italy, Brazil, etc.), and “significant decrease” (Singapore, Estonia, Korea, etc.), respectively (Supplementary Table 1).

The number of deaths by age experienced an upward trend but decreased since then in both men and women; those aged 70–84 years accounted for most of the deaths, with the peak in those aged 75–79 years. The regions with middle to high-middle SDI had more ASMR. Men and women had a similar ASMR, wherein those aged ≥95 years had an increased ASMR and shared the peak rate (Figure 4A). The number of ASDR had a similar pattern to ASMR, except for the peak point in those aged 65–69 years and for a higher DALY in those aged 60–74 years; ASDR increased in those aged 85–89 years and experienced a decline since then in men aged ≥90 years, while it experienced an upward trend and reached a maximum in women aged ≥95 years (Figure 4B).

Globally, between 1990 and 2019, ASMR experienced a decline in all age groups, with the most declines in those aged 90–94 years; it also experienced a decline in both women and men (Figure 5A), had a decline across the five SDI regions, with the biggest decline in those aged 70–74 years in high SDI regions (Figure 5B), which was similar to the pattern of ASDR-related EAPCs.

Overall, SDI was negatively related to ASMR (R = −0.33, P < 0.001), with a sharp decline in SDI >0.7 (Figure 6A); the SDI in 2019 was negatively related to the ASMR-related EAPC (ρ = −0.63, P < 0.001), particularly the SDI >0.5 (Figure 6B), while the ASMR-related EAPC was not related to the ASMR in 1990 (R = 0.072, P = 0.31) (Figure 6C). Similar patterns were also observed between SDI and ASDR and the ASDR-related EAPC (Figure 7).

The HDI in 2019 was negatively related to both ASMR-related EAPC (ρ = −0.65, P < 0.001) and ASDR-related EAPC (ρ = −0.66, P < 0.001), particularly the HDI >0.8 (Figure 8).