Skip to content

Epidemiology and burden

TL;DR — GBD 2021 estimated 11.9 million incident strokes, 93.8 million prevalent cases, 7.3 million deaths, and 160.5 million DALYs worldwide; estimates are modelled and carry uncertainty intervals, not registry counts (GBD 2021 Stroke Risk Factor Collaborators 2024, PMID 39304265). Ischemic stroke comprised 65.3% of incident events, ICH 28.8%, and SAH 5.8%. The apparent paradox is declining age-standardized mortality alongside rising absolute burden: population growth, aging, unequal prevention, and stalled incidence reductions since 2015 keep total events high (GBD 2019 Stroke Collaborators 2021, PMID 34487721; GBD 2021 Stroke Risk Factor Collaborators 2024, PMID 39304265). Low-income, low-socioeconomic-status, Indigenous, and younger populations expose different layers of inequity, while blood pressure remains the largest modifiable population risk. Stroke burden is therefore simultaneously a prevention, acute-care, rehabilitation, and long-term social-care problem.

Global counts

Metric GBD 2019 estimate (95% UI) GBD 2021 estimate (95% UI) Interpretation
Incident strokes 12.2 M (11.0–13.6) 11.9 M (10.7–13.2) Model revisions and pandemic-era inputs prevent naive trend subtraction
Prevalent stroke 101 M (93.2–111) 93.8 M (89.0–99.3) Survivors living with prior stroke, not necessarily disabled
Deaths 6.55 M (6.00–7.02) 7.3 M (6.6–7.8) 2021: 10.7% of all deaths
DALYs 143 M (133–153) 160.5 M (147.8–171.6) Years of life lost plus years lived with disability
Global rank 2nd cause of death; 3rd cause of death/disability combined 3rd cause of death; 4th cause of DALYs COVID-19 changes 2021 rank

Sources: GBD 2019 Stroke Collaborators 2021, PMID 34487721; GBD 2021 Stroke Risk Factor Collaborators 2024, PMID 39304265.

GBD combines vital registration, verbal autopsy, registries, administrative data, surveys, and Bayesian modelling. Its uncertainty intervals quantify sampling and model uncertainty, but incomplete primary surveillance remains most consequential where burden is highest (GBD 2016 Stroke Collaborators 2019, PMID 30871944; GBD 2019 Diseases and Injuries Collaborators 2020, PMID 33069326).

Pathological types

Stroke type 2019 incident count/share 2021 share Epidemiologic consequence
Ischemic 7.63 M (6.57–8.96); 62.4% 65.3% (62.4–67.7) Largest event count; survival creates long prevalence tail
Intracerebral hemorrhage 3.41 M (2.97–3.91); 27.9% 28.8% Disproportionate early mortality and DALYs
Subarachnoid hemorrhage 1.18 M (1.01–1.39); 9.7% 5.8% Younger mean age amplifies years of life lost

The change in SAH share between GBD rounds is a methodological warning: burden estimates should be cited with study round, geography, age, and definition rather than treated as timeless constants (GBD 2019 Stroke Collaborators 2021, PMID 34487721; GBD 2021 Stroke Risk Factor Collaborators 2024, PMID 39304265).

From 1990 to 2019, absolute incident strokes rose 70.0% (95% UI 67.0–73.0), prevalence 85.0% (83.0–88.0), deaths 43.0% (31.0–55.0), and DALYs 32.0% (22.0–42.0). Over the same period, age-standardized incidence fell 17.0%, mortality 36.0%, prevalence 6.0%, and DALYs 36.0% (GBD 2019 Stroke Collaborators 2021, PMID 34487721).

Trend lens Observed direction What drives it
Absolute events Up Population growth and aging
Age-standardized incidence Down through much of 1990–2019 Prevention and competing method effects
Incidence after 2015 Reduction stagnated globally Uneven risk-factor control
People under 70 Several rates increasing Risk-factor exposure and demographic change
Lower-SDI regions Burden rates remain highest Prevention, care access, surveillance, and social determinants

GBD 2021 found increases in incidence, death, prevalence, or DALY rates in parts of southeast Asia, east Asia, Oceania, lower-SDI settings, and people younger than 70 (GBD 2021 Stroke Risk Factor Collaborators 2024, PMID 39304265). These are heterogeneous regional signals, not evidence of one universal reversal.

Age and premature stroke

Stroke incidence rises steeply with age, but premature stroke creates more lifetime disability, work loss, and caregiving. The causes in younger adults are more heterogeneous, including conventional risks, dissection, cardiac sources, inflammatory and genetic arteriopathies, and pregnancy-associated mechanisms (Boot 2020, PMID 32015089; Potter 2022, PMID 36374365).

In GBD 2019, people younger than 70 had a 22.0% rise in prevalence rates and 15.0% rise in incidence rates from 1990 to 2019, contrasting with declining global all-age age-standardized rates (GBD 2019 Stroke Collaborators 2021, PMID 34487721). A Northern Manhattan population study estimated first-stroke incidence at ages 20–44 as 23 per 100,000 person-years: 10 ischemic, 7 ICH, and 6 SAH; estimates were based on only 74 events and therefore imprecise (Jacobs 2002, PMID 12468771).

Sex and gender

Women have more lifetime events in many populations because they live longer; age-specific risks vary across the life course. Pregnancy, hypertensive pregnancy disorders, oral hormonal exposures, migraine with aura, and AF interact with conventional risks, while men often have higher rates at younger ages (Potter 2022, PMID 36374365; Boot 2020, PMID 32015089).

Sex-stratified counts must not be mistaken for biological causation. Access, symptom recognition, social support, treatment intensity, prestroke disability, and survival all contribute to outcome differences. The GBD 2016 prevalence estimate was 41.1 million women and 39.0 million men, within a total 80.1 million prevalent cases (GBD 2016 Stroke Collaborators 2019, PMID 30871944).

Geographic and socioeconomic inequality

In 2019, age-standardized mortality was 3.6-fold higher and DALY rate 3.7-fold higher in low-income than high-income countries (GBD 2019 Stroke Collaborators 2021, PMID 34487721). An earlier systematic review similarly found more incidence, risk exposure, mortality, severe presentation, and—inconsistently—less evidence-based care at lower socioeconomic status (Addo 2012, PMID 22363052).

Inequality level Measured signal Evidence limitation
Between countries 3.6× mortality and 3.7× DALY rate, low- vs high-income, 2019 GBD model dependence
Within countries Lower SES generally associates with higher incidence and mortality SES measures differ across studies
Indigenous populations in very-high-HDI countries Incidence rate ratios 1.08–3.2 in studied populations Only 24 studies from seven countries; heterogeneous quality
LMIC stroke services Sparse stroke units and very low thrombolysis in reviewed reports Service data often local and dated

The Indigenous-incidence review found rate ratios of 1.7–3.2 for Aboriginal and Torres Strait Islander Australians, 1.2 for American Indians, 1.08–2.14 for Sámi populations, and 1.7–1.9 for Singaporean Malay populations versus respective comparators. Few studies reported Indigenous governance, limiting both validity and research justice (Balabanski 2024, PMID 38354325).

Risk-attributable burden

Risk factor 2019 stroke DALYs attributable Share of stroke DALYs (95% UI)
High systolic blood pressure 79.6 M 55.5% (48.2–62.0)
High body-mass index 34.9 M 24.3% (15.7–33.2)
High fasting plasma glucose 28.9 M 20.2% (13.8–29.1)
Ambient particulate matter 28.7 M 20.1% (16.6–23.0)
Smoking 25.3 M 17.6% (16.4–19.0)

Source: GBD 2019 Stroke Collaborators 2021, PMID 34487721. Attributable fractions overlap and must not be summed: one stroke can be jointly attributable to several correlated exposures.

Between 1990 and 2021, global DALYs attributable to high BMI increased 88.2% (95% UI 53.4–117.7), high ambient temperature 72.4% (51.1–179.5), and high fasting glucose 32.1% (26.7–38.1). High systolic blood pressure remained central despite a smaller 6.7% increase in attributable DALYs (GBD 2021 Stroke Risk Factor Collaborators 2024, PMID 39304265).

Case-control triangulation of risk

INTERSTROKE recruited 13,447 first-stroke cases and 13,472 controls across 32 countries. Ten modifiable factors together accounted for a population-attributable risk (PAR) of 90.7% for all stroke, 91.5% for ischemic stroke, and 87.1% for ICH; hypertension had OR 2.98 (99% CI 2.72–3.28) and PAR 47.9% (45.1–50.6) (O'Donnell 2016, PMID 27431356). The earlier 22-country phase had estimated a combined PAR of 88.1% (82.3–92.2), providing internal historical triangulation rather than an independent design (O'Donnell 2010, PMID 20561675).

PAR depends on exposure prevalence and causal assumptions; it is not the percentage of individual strokes that a clinic can certainly prevent. The consistency across regions coexists with marked regional variation in the magnitude and prevalence of particular risks (O'Donnell 2016, PMID 27431356).

Population-based studies versus global models

Population-based incidence studies directly enumerate events in a defined denominator but cover few locations. A 56-study review found incidence falling 42% in high-income countries while more than doubling in low- and middle-income countries across four decades; by 2000–2008 LMIC incidence was 20% higher (Feigin 2009, PMID 19233729). In 13 high-income population studies, incidence declined 28% over a mean 16.5 years (pooled rate ratio 0.72, 95% CI 0.66–0.79), yet UK first-ever event counts were still projected to rise 13% by 2045 with population aging (Li 2020, PMID 32208842).

Evidence source Strength Principal limitation
Population registry Direct events and denominator Geographically sparse and expensive
Administrative data Large scale and longitudinal Coding and access behavior
Case-control risk study Efficient multivariable risk estimation Recall/selection and PAR assumptions
GBD model Comparable global coverage and UIs Dependent on input density and modelling choices

Additional heterogeneity hidden by global totals

  • A 98-report review found male incidence 33% and prevalence 41% higher, but women were older at first stroke (72.9 vs 68.6 years) and had higher one-month case fatality (24.7% vs 19.7%); these historical pooled differences mix populations and eras (Appelros 2009, PMID 19211488).
  • Aneurysmal-SAH incidence across 75 studies and 67.7 million person-years was 7.9 per 100,000 person-years (95% CI 6.9–9.0) and fell an estimated 1.7% annually from 1955–2014, with strong regional variation (Etminan 2019, PMID 30659573).
  • A South African review found only 12 eligible national prevalence reports, all based on self-report, and only one incidence study—evidence scarcity that modelled regional estimates cannot erase (Abdelatif 2021, PMID 33769427).
  • Asia's population size and epidemiologic diversity imply roughly 9.5–10.6 million strokes annually in a recent review, alongside large differences in resources and mechanisms such as intracranial atherosclerosis, ICH, moyamoya, and infection (Tan 2024, PMID 38657577).
  • Broad burden reviews consistently find ischemic stroke more frequent but hemorrhagic stroke responsible for disproportionate death and DALYs (Katan 2018, PMID 29791947).

GBD cause-of-death modelling placed stroke second globally in 2019 and third in 2021 after COVID-19, illustrating that rank changes can reflect a new competing cause without a reduction in stroke's absolute importance (GBD 2021 Causes of Death Collaborators 2024, PMID 38582094).

Fatality, disability, and recurrence

Population burden is shaped by three linked probabilities: having a first stroke, surviving the acute phase, and living with disability or recurrence. ICH and SAH cause fewer events than ischemic stroke but greater early case fatality; ischemic stroke contributes a larger survivor pool requiring prevention and rehabilitation (GBD 2019 Stroke Collaborators 2021, PMID 34487721; Claassen 2022, PMID 35985353).

Functional disability is commonly represented by the modified Rankin Scale, but DALYs incorporate both fatal and non-fatal loss through modelled disability weights. Neither captures communication participation, fatigue, cognition, caregiver effects, or unequal access to rehabilitation in full (Bernhardt 2017, PMID 28697708; Winstein 2016, PMID 27145936).

Economic and household burden

A US systematic review found that acute hospitalization dominates short-term costs, while nursing-home care, ambulatory care, lost earnings, and informal caregiving dominate lifetime cost (Demaerschalk 2010, PMID 20645668). Direct cost comparisons age rapidly because prices, treatment mix, payer structure, and survival change; cost year and perspective are essential fields.

Cost perspective Components that must be counted Common omission
Health system Ambulance, imaging, thrombolysis/thrombectomy, admission, rehabilitation, readmission Community and long-term care
Patient/household Travel, copayments, home modification, equipment, paid support Informal time and foregone income
Caregiver Time, health loss, reduced employment Treated as cost-free substitute for services
Society Direct medical plus productivity and informal care Double counting transfers and benefits

Caregiving carries measurable health utility loss across chronic illnesses, but stroke-specific estimates vary by patient dependency and instrument (Wittenberg 2013, PMID 23572441). Economic evaluations that stop at discharge systematically understate the long tail.

Surveillance quality

The ideal incidence study has a defined population, overlapping case-ascertainment sources, rapid clinical review, standardized imaging and definitions, first-ever/recurrent separation, and complete follow-up. Administrative codes miss non-hospitalized events and inherit coding changes; hospital series cannot estimate population incidence; GBD models fill gaps but do not replace registries (GBD 2016 Stroke Collaborators 2019, PMID 30871944).

LMIC service research identified feasible ambulance prenotification, telemedicine, training, stroke-unit, and subsidized thrombolysis approaches, but outcome and sustainability evidence was limited (Pandian 2017, PMID 28848165). Better measurement and better care are inseparable: countries without surveillance cannot reliably locate access failures.

Denominators that should not be mixed

  • Incidence counts new events in a defined population and period; first-ever and recurrent events must be distinguished.
  • Prevalence counts people living after stroke and therefore depends on both incidence and survival.
  • Mortality rate uses a population denominator; case fatality uses people who had stroke and cannot substitute for it.
  • DALYs add modelled fatal and non-fatal loss; they are neither event counts nor a direct clinical outcome measure (GBD 2019 Stroke Collaborators 2021, PMID 34487721).

Open questions

  • Why did global incidence reduction stagnate after 2015, and which risk or surveillance changes explain regional reversals? (GBD 2021 Stroke Risk Factor Collaborators 2024, PMID 39304265)
  • Which interventions reduce the 3.6-fold low- versus high-income mortality-rate gap rather than merely increasing measured treatment volume? (GBD 2019 Stroke Collaborators 2021, PMID 34487721; Pandian 2017, PMID 28848165)
  • How much of rising premature stroke reflects true incidence versus improved ascertainment and shifting definitions? (Boot 2020, PMID 32015089; Potter 2022, PMID 36374365)
  • What surveillance designs can produce Indigenous-governed, comparable incidence estimates without collapsing distinct peoples into one category? (Balabanski 2024, PMID 38354325)
  • Which household costs are prevented by acute reperfusion, secondary prevention, and rehabilitation over a lifetime horizon? (Demaerschalk 2010, PMID 20645668)

References

  1. GBD 2021 Stroke Risk Factor Collaborators. Global burden of stroke and its risk factors, 1990–2021. Lancet Neurol. 2024. PMID 39304265
  2. GBD 2019 Stroke Collaborators. Global burden of stroke and its risk factors, 1990–2019. Lancet Neurol. 2021. PMID 34487721
  3. GBD 2016 Stroke Collaborators. Global burden of stroke, 1990–2016. Lancet Neurol. 2019. PMID 30871944
  4. GBD 2019 Diseases and Injuries Collaborators. Global burden of 369 diseases and injuries, 1990–2019. Lancet. 2020. PMID 33069326
  5. Addo J, et al. Socioeconomic status and stroke: an updated review. Stroke. 2012. PMID 22363052
  6. Balabanski AH, et al. Incidence of stroke in Indigenous populations of countries with very high HDI. Neurology. 2024. PMID 38354325
  7. Boot E, et al. Ischaemic stroke in young adults: a global perspective. J Neurol Neurosurg Psychiatry. 2020. PMID 32015089
  8. Potter TBH, et al. Epidemiology, risk factors, etiology, and outcomes of premature stroke. Curr Atheroscler Rep. 2022. PMID 36374365
  9. Jacobs BS, et al. Stroke in the young in the Northern Manhattan Stroke Study. Stroke. 2002. PMID 12468771
  10. Demaerschalk BM, et al. US cost burden of ischemic stroke: a systematic review. Am J Manag Care. 2010. PMID 20645668
  11. Wittenberg E, et al. Disutility of illness for caregivers and families: a systematic review. Pharmacoeconomics. 2013. PMID 23572441
  12. Pandian JD, et al. Strategies to improve stroke care services in low- and middle-income countries. Neuroepidemiology. 2017. PMID 28848165
  13. Claassen J, et al. Spontaneous subarachnoid haemorrhage. Lancet. 2022. PMID 35985353
  14. Bernhardt J, et al. Standards in stroke recovery research. Int J Stroke. 2017. PMID 28697708
  15. Winstein CJ, et al. Guidelines for adult stroke rehabilitation and recovery. Stroke. 2016. PMID 27145936
  16. O'Donnell MJ, et al. Global and regional effects of modifiable risk factors in 32 countries: INTERSTROKE. Lancet. 2016. PMID 27431356
  17. O'Donnell MJ, et al. Risk factors for ischemic and intracerebral hemorrhagic stroke in 22 countries. Lancet. 2010. PMID 20561675
  18. Feigin VL, et al. Worldwide stroke incidence and early case fatality in 56 population studies. Lancet Neurol. 2009. PMID 19233729
  19. Li L, et al. Trends in stroke incidence in high-income countries. Stroke. 2020. PMID 32208842
  20. Appelros P, et al. Sex differences in stroke epidemiology: a systematic review. Stroke. 2009. PMID 19211488
  21. Etminan N, et al. Worldwide incidence of aneurysmal subarachnoid hemorrhage. JAMA Neurol. 2019. PMID 30659573
  22. Abdelatif N, et al. National prevalence of coronary heart disease and stroke in South Africa. Cardiovasc J Afr. 2021. PMID 33769427
  23. Tan KS, et al. Stroke in Asia. Cerebrovasc Dis Extra. 2024. PMID 38657577
  24. Katan M, Luft A. Global burden of stroke. Semin Neurol. 2018. PMID 29791947
  25. GBD 2021 Causes of Death Collaborators. Global burden of 288 causes of death, 1990–2021. Lancet. 2024. PMID 38582094