Association of Younger vs Older Ages With Changes in Incidence of Stroke and Other Vascular Events, 2002-2018

Key Points

Question

Did stroke incidence rates diverge in younger vs older people from 2002 to 2018?

Findings

In this cohort study that included 94 567 participants in Oxfordshire, England, and compared temporal changes in stroke rates between 2002-2010 and 2010-2018, there was a 67% increase in stroke incidence among participants younger than 55 years and a 15% decrease in stroke incidence among participants aged 55 years or older. No similar divergence by age was seen for other vascular events.

Meaning

Between 2002-2010 and 2010-2018, there was an increase in stroke incidence in younger persons and a decrease in older persons, without similar divergence for other vascular events.

Abstract

Importance

Some studies have reported increasing stroke incidence at younger ages (<55 years) but have often relied only on administrative data, and more population-based studies of adjudicated stroke are required. An understanding of the drivers of any increase in incidence of young stroke also requires comparisons with stroke trends at older ages and with trends in incidence of other vascular events at younger ages.

Objective

To determine temporal changes in incidence of stroke and other major vascular events at younger vs older ages.

Design, Setting, and Participants

Prospective population-based incidence study conducted from April 2002 to March 2018 with a mean catchment population of 94 567 in Oxfordshire, England.

Exposures

Calendar time, premorbid vascular risk factors, and occupation.

Main Outcomes and Measures

Changes in incidence of stroke, transient ischemic attack (TIA), and other major vascular events (myocardial infarction, sudden cardiac death, and peripheral vascular events) stratified by age, sex, diagnostic workup, etiology, and severity.

Results

A total of 2429 incident strokes were ascertained (mean age, 73.6 [SD, 14.4] years; 51.3% female). From 2002-2010 to 2010-2018, stroke incidence increased significantly among participants younger than 55 years (incidence rate ratio [IRR], 1.67; 95% CI, 1.31-2.14) but fell significantly among participants aged 55 years or older (IRR, 0.85; 95% CI, 0.78-0.92; P < .001 for difference). The significant increase in incidence at younger than 55 years was independent of sex, stroke severity, pathological subtype, and changes in investigation and was also seen for TIA (IRR, 1.87; 95% CI, 1.36-2.57) but not for myocardial infarction and other major vascular events (IRR, 0.73; 95% CI, 0.58-0.93). Although TIA and stroke at younger than 55 years were significantly associated with diabetes (risk ratio [RR], 3.47; 95% CI, 2.54-4.74), hypertension (RR, 2.52; 95% CI, 2.04-3.12), current smoking (RR, 2.38; 95% CI, 1.92-2.94), and obesity (RR, 1.36; 95% CI, 1.07-1.72), the significant increase in incidence from 2002-2010 to 2010-2018 was still seen in individuals without these risk factors. The increase was greatest in professional/managerial occupations (IRR, 2.52; 95% CI, 1.75-3.62) and least in partially skilled/unskilled occupations (IRR, 1.17; 95% CI, 0.79-1.74). The proportion of TIAs and strokes among those younger than 55 years without known vascular risk factors increased significantly over time (45 [30.4%] vs 115 [42.4%]; absolute difference, 12.0%; 95% CI, 2.6-21.5), especially in patients with cryptogenic events (10 [18.5%] vs 63 [49.2%]; absolute difference, 30.7%; 95% CI, 17.2-44.2; P < .001; P = .002 for heterogeneity).

Conclusions and Relevance

Comparing persons living in Oxfordshire, England, in 2002-2010 vs 2010-2018, there was a significant increase in stroke incidence in those younger than 55 years, but a decrease in those aged 55 years or older. Given the absence of this divergence for other vascular events, further research is needed to understand the causes of this difference.

This study assesses time trends in incidence of stroke, transient ischemic attack, and other major vascular events at younger vs older ages in a population living in Oxfordshire, England, in 2002-2018.

Introduction

Increasing life expectancy in high-income countries has slowed or even reversed since 2010, mostly due to increased mortality between ages 35 to 50 years, particularly in the UK and US.^1,2 A similar age-specific divergence was reported in incidence of colorectal cancer, with increasing rates at younger ages,³ possibly due to increases in obesity, lack of exercise, and poor diet. Given the overlap in risk factors, similar age-specific divergence in incidence of stroke and other vascular events might also be expected.

In contrast to the continued overall decline of stroke incidence in high-income countries in the 21st century,⁴ incidence at younger ages (usually defined as <55 years) appeared to be increasing in the US and some other countries.^5,6,7 However, many studies relied on routinely collected administrative data on deaths or hospitalization, which are prone to bias, as diagnostic coding practices and admission policies change over time.⁸ Furthermore, increased use of diffusion-weighted brain imaging, and changes in public knowledge and behaviors following public health campaigns may have improved the ascertainment of young stroke.

In light of these uncertainties, this population-based study aimed to determine the incidence of transient ischemic attack (TIA) and stroke and other vascular events in Oxfordshire, England, from 2002 to 2018, taking into account changes in diagnostic practices, control of traditional vascular risk factors, other exposures, lifestyle, and sex-specific causes of stroke.^9,10 In light of conflicting findings in previous studies, with stable or decreasing incidence of young stroke in some countries,^11,12,13 in a linked systematic review¹⁴ the divergence in stroke incidence at younger vs older ages was quantified to determine whether a less favorable change in incidence at younger ages was a more consistent finding.

Methods

Participants

The Oxford Vascular Study (OXVASC) is an ongoing prospective population-based study of the incidence and outcome of all acute vascular events in a mixed rural/urban population.¹⁵ The study population comprised all individuals (2002-2018 mean = 94 567), irrespective of age, registered with about 100 general practitioners in 9 general practices in Oxfordshire, England. At the mid–study point (2011), the population was 91% White for all age groups and 88% White for individuals younger than 50 years (eAppendix in the Supplement). The detailed study methodology and case definitions have been published.¹⁵ Briefly, multiple overlapping methods of prospective and retrospective ascertainment were used to achieve near-complete identification of all individuals with acute vascular events (eAppendix in the Supplement).

Written informed consent or assent from relatives was obtained for all participants. OXVASC was approved by the local research ethics committee (OREC A: 05/Q1604/70).

Exposures

The main exposure of the study is the calendar year of the incident vascular event. All first-ever incident events ascertained between April 1, 2002, and March 31, 2018, were included.

Other exposures included were premorbid vascular risk factors and occupation. To stratify TIA and stroke incidence in the cohort by the presence of risk factors and by occupation in the underlying population, age- and sex-specific data for the background population were obtained from the Health Survey for England (using the South East England region as a proxy) and the Annual Population Survey (using Oxfordshire as a proxy) (eAppendix and eFigure 1 in the Supplement).

Race and ethnicity were ascertained to determine if disease incidence differed by these characteristics and was based on predefined categories of Bangladeshi, Black African, Black Caribbean, Chinese, Indian, Pakistani, White, and other. Judgment was made by researchers and confirmed with patients or relatives during face-to-face interviews. When interview was not possible due to death before assessment, researchers extracted relevant information recorded in medical notes.

Both for TIA and stroke cases and the underlying population, deprivation was measured by postcode of residence using the English Indices of Multiple Deprivation (eAppendix in the Supplement).

Outcomes

The outcomes of the study included changes in incidence of stroke, TIA, and other major vascular events (myocardial infarction, sudden cardiac death, and peripheral vascular events).

Patients with TIA or stroke were ascertained and assessed face-to-face by study physicians as soon as possible after the initial presentation in the hospital, in an emergency clinic, or at home. In the UK, more than 70% of patients with TIA and minor stroke are assessed in emergency clinics, with all diagnostic workup performed in an outpatient setting.¹⁶ If a patient died before assessment, an eyewitness account of the clinical event was obtained and any relevant clinical records reviewed. Baseline demographic data, premorbid use of preventive medication, premorbid blood pressure measurements, traditional vascular risk factors, other potential risk factors, and family history of stroke were collected from face-to-face interviews and cross-referenced with primary care records (eAppendix in the Supplement). Detailed clinical history was recorded for all patients and assessments were made for stroke severity using the National Institutes of Health Stroke Scale (NIHSS) on admission to the hospital or at the time of the first assessment, either at the emergency department or in the acute TIA/minor stroke clinic, for those who were not admitted. Patients routinely had brain imaging, vascular imaging, cardiac investigation and standard blood tests (eAppendix in the Supplement). All cases were reviewed by the same senior stroke neurologist (P.M.R.) throughout the study to ensure consistency of diagnosis, and ischemic stroke etiology was classified according to the modified TOAST criteria.¹⁷

All patients with TIA or stroke were followed up face-to-face at 1, 6, 12, 60, and 120 months by a study nurse or physician to determine their functional status (modified Rankin Scale [mRS]; score range, 0-6, with 0 being the best outcome).¹⁸ Major stroke is defined as new disability or death (mRS score >2) using the 1-month mRS score, or as progression of disability (an increase of 1 point in mRS score) in those with premorbid disability (premorbid mRS score >2). For patients who had moved out of the study area, follow-up was performed by telephone. All patients were flagged for Office for National Statistics mortality data and all deaths during follow-up were recorded with causes.

Patients with any other incident major acute vascular event (myocardial infarction, peripheral vascular disease and sudden cardiac death) were also ascertained over the same period (eAppendix in the Supplement) as reported previously.^15,19

Three other sources of data on stroke incidence were also studied. First, stroke incidence between 2002 and 2018 was compared with that in the Oxfordshire Community Stroke Project, a previous population-based study of all first-ever strokes over 4 years (1981-1984 and 1986) covering the same general practice population. The methods have also been published, and the case diagnosis, assessment, and follow-up were similar to those in the later OXVASC cohort (2002-2018).²⁰ To try to ensure consistency in the application of definitions of stroke between the 2 studies, all potential cases in the first 2 years of the 2002-2018 cohort were reviewed by the principal investigators of both studies.¹⁵ Second, data on hospitalization for strokes at younger than 55 years in Oxfordshire in 1975-1976 were extracted from an earlier study.²¹ Third, data on annual stroke admissions at younger than 60 years in England between 1998 and 2018 were extracted from the Hospital Episode Statistics database,²² based on International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes, including I60 (nontraumatic subarachnoid hemorrhage), I61 (nontraumatic intracerebral hemorrhage), I63 (cerebral infarction), and I64 (unspecified stroke), and stratified by reported subtype (unknown, ischemic, intracerebral hemorrhage, or subarachnoid hemorrhage).

Statistical Analyses

Crude incidence rates per population per year were calculated by age groups (<55 years vs ≥55 years), sex, stroke severity (major vs minor), stroke subtypes (ischemic stroke, intracerebral hemorrhage, or subarachnoid hemorrhage), and common stroke risk factor status (smoking, obesity, hypertension, diabetes, and deprivation) with 95% confidence intervals estimated assuming a Poisson distribution. Incidence rates using the England national data on stroke admissions at younger than 60 years were also calculated. All incidences were standardized to the 2011 population of England and Wales. Poisson regression models, adjusted for the population age and sex structures, were used to calculate incidence rate ratios (IRRs) for 2002-2018 vs 1981-1986 and for 2010-2018 vs 2002-2010. The IRRs for the changes in incidence of TIA, TIA and stroke, and other acute vascular events at younger than 55 years comparing 2010-2018 vs 2002-2010 were also calculated, and sensitivity analyses stratified by the participating general practices and by rural vs urban catchment area were performed. To measure the age-specific divergence in IRR, the relative temporal difference was calculated by dividing the IRR at younger than 55 years by the IRR at age 55 years or older (eAppendix in the Supplement).¹⁴

For all patients younger than 55 years with TIA and stroke, changes in the following factors between the two 8-year periods were also compared using a χ² test for categorical variables and a t test for continuous variables: demographics (age, sex, and occupation), traditional vascular risk factors; other potential risk factors; family history of stroke; premorbid blood pressure; event territory (anterior vs posterior circulation); stroke etiology (TOAST subtypes for ischemic stroke, hematoma location for intracerebral hemorrhage, and aneurysmal vs nonaneurysmal for subarachnoid hemorrhage), mode of diagnostic workup and patient behavior, and the nonlaboratory version of the Framingham General Cardiovascular Risk Score.²³ Sensitivity analyses excluding TIAs were also performed. Given the multiple nature of these comparisons and the potential for type I error, these analyses should be interpreted as exploratory.

The prevalence of traditional vascular risk factors in patients with TIA and stroke was also compared with the prevalence reported in the background population for each 4-year time period respectively, and the risk ratio (RR) was then pooled using random-effects meta-analysis. For deprivation, patients were divided into 4 groups according to the English quartiles of the English Indices of Multiple Deprivation score, with quartile 1 being the most affluent and quartile 4 being the most deprived. The same approach was undertaken for the income subcomponent score. Age-specific incident stroke rates (per 1000 population per year) were calculated and stratified by quartile of deprivation.

Complete analyses without imputation were performed and missing data were reported where applicable. P values were 2-sided with statistical significance set at P < .05.

All analyses were done using SPSS version 25 (IBM) and Stata version 17.0 (StataCorp).

Results

Overall, 2429 incident strokes (51.3% among women) were ascertained (1166 in 2002-2010 and 1263 in 2010-2018). There was a significant decline in stroke incidence at age 55 years or older in 2010-2018 vs 2002-2010 (IRR, 0.85; 95% CI, 0.78-0.92; P < .001) (Figure 1; eTable 1 in the Supplement) but a significant increase at younger than 55 years (IRR, 1.67; 95% CI, 1.31-2.14; P < .001) (Table 1 and Figure 1; eTable 1 in the Supplement), with an annual percentage change of 5.5% (95% CI, 3.4%-7.7%; P < .001 for trend). These divergent changes (relative temporal difference, 1.96; 95% CI, 1.52-2.55; P < .001) (eTable 1 in the Supplement) were consistent for non–atrial fibrillation–related events (eFigure 2 in the Supplement) and also by sex, stroke type, and stroke severity (eTable 1 in the Supplement). The decrease of stroke incidence at age 55 to 64 years was less prominent than at older ages and was not statistically significant (eTable 1 in the Supplement).

Figure 1. Temporal Changes in Standardized Stroke Incidence at Younger Than 55 Years vs 55 Years or Older and at Younger Than 55 Years for Stroke vs All Other Incident Major Vascular Events.

Other major vascular events included acute myocardial infarction, peripheral vascular events, and sudden cardiac death. The study year began on April 1 and ended on March 31 of the next year. Whiskers indicate 95% CIs.

Table 1. Standardized Incidence per 10 000 per Year for All Acute Vascular Events at Younger Than 55 Years in Oxfordshire From 1981 to 2018, Stratified by Sex, Stroke Subtype, and Stroke Severity.

Events	Standardized incidence per 10 000 per year (95% CI)		Difference (95% CI)		P valuea	Standardized incidence per 10 000 per year (95% CI)		Difference (95% CI)		P valuea
	OCSP, 1981-1986	OXVASC, 2002-2006	Absolute	Relative		OXVASC, 2002-2010	OXVASC, 2010-2018	Absolute	Relative
Stroke	1.96 (1.37-2.55)	1.68 (1.18-2.18)	−0.29 (−1.41 to 0.57)	0.85 (0.56-1.31)	.47	1.88 (1.51-2.25)	3.14 (2.67-3.61)	1.27 (0.67 to 1.86)	1.67 (1.31-2.14)	<.001
Transient ischemic attack	0.74 (0.40-1.07)	0.69 (0.37-1.01)	−0.04 (−0.51 to 0.42)	0.94 (0.49-1.80)	.85	1.04 (0.76-1.32)	1.95 (1.58-2.32)	0.91 (0.44 to 1.37)	1.87 (1.36-2.57)	<.001
Other acute vascular eventsb						3.01 (2.54-3.48)	2.21 (1.81-2.60)	−0.80 (−1.42 to −0.18)	0.73 (0.58-0.93)	.01
Myocardial infarction and sudden cardiac death						2.60 (2.16-3.04)	1.94 (1.57-2.31)	−0.67 (−1.24 to −0.09)	0.74 (0.58-0.96)	.02
Peripheral vascular events						0.43 (0.25-0.61)	0.33 (0.18-0.48)	−0.10 (−0.34 to 0.13)	0.77 (0.41-1.43)	.40
For strokes only
Sex
Men	2.25 (1.36-3.15)	1.86 (1.13-2.59)	−0.39 (−1.55 to 0.76)	0.83 (0.47-1.45)		2.07 (1.53-2.62)	3.71 (3.00-4.42)	1.64 (0.74-2.53)	1.79 (1.30-2.46)
Women	1.66 (0.90-2.42)	1.48 (0.79-2.16)	−0.18 (−1.20 to 0.84)	0.89 (0.46-1.71)		1.67 (1.17-2.18)	2.53 (1.92-3.13)	0.86 (0.07 to 1.65)	1.51 (1.03-2.22)
Stroke subtype
Ischemic stroke	1.33 (0.85-1.82)	1.00 (0.62-1.39)	−0.33 (−0.95 to 0.29)	0.75 (0.44-1.29)		1.26 (0.96-1.57)	2.43 (2.02-2.85)	1.17 (0.65 to 1.68)	1.92 (1.44-2.57)
ICH	0.25 (0.03-0.48)	0.12 (0.00-0.26)	−0.13 (−0.39 to 0.13)	0.49 (0.11-2.08)		0.23 (0.10-0.36)	0.45 (0.28-0.63)	0.23 (0.01 to 0.45)	2.00 (1.02-3.91)
SAH	0.38 (0.13-0.63)	0.55 (0.26-0.84)	0.17 (−0.21 to 0.55)	1.46 (0.63-3.35)		0.39 (0.22-0.56)	0.26 (0.12-0.39)	−0.13 (−0.35 to 0.09)	0.66 (0.33-1.31)
Stroke severity (NIHSS score at presentation) c , d
Minor stroke (NIHSS score <3)						0.94 (0.68-1.20)	1.79 (1.44-2.14)	0.85 (0.41 to 1.29)	1.90 (1.36-2.66)	<.001
Ischemic stroke						0.88 (0.62-1.14)	1.63 (1.29-1.97)	0.75 (0.32 to 1.17)	1.85 (1.30-2.61)
ICH						0.06 (0.00-0.12)	0.16 (0.06-0.27)	0.10 (−0.02 to 0.23)	2.80 (0.82-9.59)
Major stroke (NIHSS score ≥3)						0.55 (0.35-0.75)	1.10 (0.82-1.38)	0.54 (0.20 to 0.89)	2.00 (1.30-3.08)	.002
Ischemic stroke						0.38 (0.21-0.55)	0.80 (0.56-1.04)	0.42 (0.13 to 0.71)	2.11 (1.26-3.52)
ICH						0.17 (0.06-0.28)	0.29 (0.15-0.44)	0.12 (−0.06 to 0.31)	1.73 (0.77-3.87)
Stroke outcome (mRS score change at 1 mo) c , e
Nondisabling stroke	0.98 (0.56-1.40)	0.93 (0.56-1.30)	−0.05 (−0.61 to 0.51)	0.95 (0.53-1.70)		1.19 (0.89-1.49)	2.34 (1.94-2.75)	1.15 (0.65 to 1.65)	1.97 (1.46-2.64)
Ischemic stroke	0.89 (0.49-1.29)	0.89 (0.52-1.25)	0.00 (−0.54 to 0.54)	1.00 (0.55-1.83)		1.13 (0.84-1.42)	2.14 (1.75-2.53)	1.01 (0.52 to 1.49)	1.89 (1.39-2.57)
ICH	0.09 (0.00-0.23)	0.04 (0.00-0.12)	−0.05 (−0.21 to 0.10)	0.43 (0.04-4.83)		0.06 (0.00-0.12)	0.20 (0.08-0.32)	0.14 (0.01 to 0.28)	3.45 (1.06-11.22)
Disabling or fatal stroke	0.60 (0.27-0.94)	0.20 (0.02-0.37)	−0.40 (−0.78 to −0.03)	0.33 (0.12-0.93)		0.30 (0.15-0.45)	0.55 (0.35-0.74)	0.24 (0.00 to 0.49)	1.81 (1.00-3.29)
Ischemic stroke	0.44 (0.16-0.72)	0.12 (0.00-0.25)	−0.33 (−0.64 to −0.02)	0.26 (0.07-0.93)		0.13 (0.03-0.23)	0.29 (0.15-0.43)	0.16 (−0.01 to 0.33)	2.21 (0.94-5.24)
ICH	0.16 (0.00-0.34)	0.08 (0.00-0.20)	−0.08 (−0.29 to 0.14)	0.52 (0.08-3.22)		0.17 (0.06-0.28)	0.25 (0.12-0.39)	0.09 (−0.09 to 0.26)	1.50 (0.66-3.45)

Abbreviations: ICH, intracerebral hemorrhage; mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; OCSP, Oxfordshire Community Stroke Project; OXVASC, Oxford Vascular Study; SAH, subarachnoid hemorrhage.

^a P values refer to relevant difference in stroke incidence.

^b Including acute myocardial infarction, peripheral vascular event, and sudden cardiac death.

^c Analysis excluding SAH.

^d The NIHSS quantifies stroke severity. Scores range from 0 to 42, with higher scores indicating greater severity.

^e The mRS ranges from 0 to 6, with higher scores indicating greater disability. Disabling or fatal stroke is defined as new disability or death (mRS score >2) using the 1-month mRS score or as progression of disability (mRS score increase of 1 point) in individuals with premorbid disability (premorbid mRS score >2).

Incidence of TIA at younger than 55 years also increased significantly from 2010-2018 to 2002-2010 (Table 1), whereas incidence of myocardial infarction and of all major nonstroke vascular events at younger than 55 years declined significantly (IRR, 0.73; 95% CI, 0.58-0.93; P = .01) (Table 1 and Figure 1).

In contrast to the significant increase in incidence of TIA and stroke at younger than 55 years during 2002-2018, there was no significant difference in incidence rates between 2002-2006 and the earlier cohort (1981-1986) or a hospital-based study of young stroke from the same county in 1975-1976 (Figure 2).²¹ However, incidence of stroke at age 55 years or older decreased significantly between 1981-1986 and 2002-2006 (IRR, 0.77; 95% CI, 0.69-0.87; P < .001). After 2002-2006, incidence of both stroke and TIA at younger than 55 years increased approximately linearly in the cohort (Figure 2; eFigure 3 and eTable 2 in the Supplement), consistent with national stroke admission rates (available for 1998 onward) at younger than 60 years (Figure 2), and with a significant increase in incidence of both ischemic stroke and intracerebral hemorrhage (Table 1), again consistent with national stroke admission data (eFigure 4 in the Supplement).

Figure 2. Temporal Changes of Standardized Stroke Incidence in Individuals Younger Than 60 Years in England and in Those Younger Than 55 Years in Oxfordshire.

TIA indicates transient ischemic attack. Data on stroke incidence in individuals younger than 60 years in England are based on hospital admission data on any stroke, 1998-2018; data on stroke incidence in individuals younger than 55 years in Oxfordshire are based on stroke incidence studies, 1975-1976,²¹ 1981-1986, and 2002-2018. Number of events and population at risk for each study are reported in eTable 12 in the Supplement. For the population-based study between 2002-2018, the study year began on April 1 and ended on March 31 of the next year. Whiskers indicate 95% CIs.

Of 350 incident TIAs or ischemic strokes at younger than 55 years between 2002 and 2018, 245 (70.0%) were of cryptogenic or small vessel disease subtypes. For each of the main TOAST subtypes, the absolute number of events was greater in 2010-2018 than in 2002-2010 (Table 2; eTables 3 and 4 in the Supplement), but not for events caused by arterial dissection (9 in 2002-2010 vs 7 in 2010-2018). The absolute increase in numbers was greatest for cryptogenic and small vessel events (Table 2).

Table 2. Demographics, Risk Factors, Medication at Baseline, and Premorbid Blood Pressure in Patients Younger Than 55 Years With Incident Transient Ischemic Attack or Stroke^a.

Event type	No. (%)b		Absolute difference (95% CI)
	2002-2010 (n = 148)	2010-2018 (n = 273)
Ischemic events	n = 116	n = 234
Cardioembolic	18 (15.5)	25 (10.7)	−4.8 (−12.5 to 2.9)
Large artery disease	8 (6.9)	14 (6.0)	−0.9 (−6.4 to 4.6)
Small vessel disease	20 (17.2)	43 (18.4)	1.2 (−7.3 to 9.7)
Cryptogenic	54 (46.6)	128 (54.7)	8.1 (−3.0 to 19.2)
Unknown/multiple/other	16 (13.8)	24 (10.3)	−3.5 (−10.9 to 3.9)
Intracerebral hemorrhage	n = 12	n = 25
Lobar	5 (41.7)	11 (44.0)	2.3 (−31.7 to 36.3)
Deep/posterior or intraventricular	7 (58.3)	14 (56.0)	−2.3 (−36.3 to 31.7)
Subarachnoid hemorrhage, No.	20	14
Demographics
Age, mean (SD), y	45.8 (7.5)	45.2 (9.1)	−0.6 (−2.3 to 1.1)
Female	67 (45.3)	108 (39.6)	−5.7 (−15.6 to 4.2)
Male	81 (54.7)	165 (60.4)	5.7 (−4.2 to 15.6)
Occupationc	n = 135	n = 249
Professional/managerial	33 (24.4)	103 (41.4)	17.0 (7.5 to 26.5)
Skilled	47 (34.8)	87 (34.9)	0.1 (−9.9 to 10.1)
Partially skilled/unskilled	50 (37.0)	51 (20.5)	−16.5 (−26.1 to −6.9)
Unemployed	5 (3.7)	8 (3.2)	−0.5 (−4.4 to 3.4)
Traditional risk factors
Current smoker	63 (42.6)	95/270 (35.2)	−7.4 (−17.2 to 2.4)
Ex-smoker	27 (18.2)	65/270 (24.1)	5.8 (−2.2 to 13.9)
Hypertension	43 (29.1)	61 (22.3)	−6.7 (−15.5 to 2.1)
Hyperlipidemia	37 (25.0)	36 (13.2)	−11.8 (−19.9 to −3.8)
Diabetes	12 (8.1)	25 (9.2)	1.0 (−4.5 to 6.6)
Atrial fibrillation	3 (2.0)	5 (1.8)	−0.2 (−3.0 to 2.6)
None of the above risk factors	45 (30.4)	115/271 (42.4)	12.0 (2.6 to 21.5)
Previous atherosclerotic disease	11 (7.4)	11 (4.0)	−3.4 (−8.2 to 1.4)
Overweight or obesity	83/144 (57.6)	158/262 (60.3)	2.7 (−7.3 to 12.7)
Obesity	37 (25.7)	74 (28.4)	2.7 (−6.3 to 11.6)
Regular alcohol use	36 (24.3)	62/266 (23.3)	−1.0 (−9.6 to 7.6)
10-y risk of cardiovascular event, %d	n = 146	n = 257
<10	69 (47.3)	121 (47.1)	−0.2 (−10.3 to 9.9)
10-29	66 (45.2)	124 (48.2)	3.0 (−7.1 to 13.1)
≥30	11 (7.5)	12 (4.7)	−2.8 (−7.8 to 2.2)
Other potential risk factors
Exercise (none or below average)	40/124 (32.3)	77/233 (33.0)	0.8 (−9.4 to 11.0)
Previous migraine	45 (30.4)	71/272 (26.1)	−4.3 (−13.4 to 4.8)
Depression	31 (20.9)	82/272 (30.1)	9.2 (0.7 to 17.7)
Previous autoimmune disease	16 (10.8)	47 (17.2)	6.4 (−0.3 to 13.1)
Previous cancer	8 (5.4)	11 (4.0)	−1.4 (−5.7 to 2.9)
Previous venous thrombosis	4 (2.7)	6 (2.2)	−0.5 (−3.6 to 2.6)
Family history of young stroke
Parent(s)	8/118 (6.8)	14/229 (6.1)	−0.7 (−6.2 to 4.8)
Sibling(s)	1/111 (0.9)	10/216 (4.6)	3.7 (0.4 to 7.0)
Premorbid medication
Current oral contraceptives or hormonal therapye	16 (23.9)	22 (20.4)	−3.5 (−16.2 to 9.2)
Antihypertensives	34 (23.0)	47 (17.2)	−5.8 (−13.9 to 2.4)
Antidepressants	16 (10.8)	32 (11.7)	0.9 (−5.3 to 7.2)
Statins	14 (9.5)	31 (11.4)	1.9 (−4.1 to 7.9)
Antiplatelets	9 (6.1)	17 (6.2)	0.1 (−4.7 to 4.9)
Anticoagulants	4 (2.7)	9 (3.3)	0.6 (−2.8 to 4.0)
Premorbid blood pressure, mm Hgf	n = 146	n = 249
Systolic, mean (SD)	131.7 (17.7)	127.9 (13.9)	−3.8 (−7.0 to −0.6)
Diastolic, mean (SD)	80.8 (10.7)	79.0 (9.9)	−1.8 (−3.9 to 0.3)
Mean ≥140/90	46 (31.5)	54 (21.7)	−9.8 (−18.9 to −0.7)
Any ≥140/90	86 (58.9)	144 (57.8)	−1.1 (−11.2 to 9.0)

^a Comparisons of additional factors are shown in eTable 5 in the Supplement.

^b Data are No. (%) of participants unless otherwise indicated.

^c Examples of the occupation categories are described in the eAppendix in the Supplement.

^d Based on the nonlaboratory version of the Framingham General Cardiovascular Risk Score.

^e Among female participants.

^f Median time from most recent blood pressure measurement to baseline was 322 days.

Of 270 incident strokes at younger than 55 years, most were nondisabling (n = 213 [78.9%]), but a significant increase in incidence was seen for disabling or fatal strokes (Table 1; eTable 2 in the Supplement), with consistent results also using the initial NIHSS score as a measure of stroke severity (Table 1; eFigure 5 in the Supplement). The 1-year risk of recurrent stroke at younger than 55 years was also similar to that at age 55 years or older (7.4% vs 9.6%; P = .33) (eFigure 6 in the Supplement).

Among demographic characteristics of patients with stroke or TIA at younger than 55 years (Table 2; eTables 5-8 in the Supplement), there was a significant increase in the proportion who were in more skilled occupations (Table 2), particularly for professional/managerial occupations, which remained after adjustment for age, sex, history of hypertension, diabetes, hyperlipidemia, myocardial infarction, peripheral vascular disease, and current smoking (adjusted odds ratio for the step change toward a more skilled occupation, 1.51; 95% CI, 1.17-1.96), and after adjustment for occupational status in the underlying study population (professional or managerial occupations: IRR, 2.52 [95% CI, 1.75-3.62]; skilled occupations: IRR, 2.10 [95% CI, 1.47-3.00]; partially skilled/unskilled occupations: 1.17 [95% CI, 0.79-1.74]). The increase in the proportion in professional/managerial occupations was present for analyses of young stroke alone and for events in men and women separately (eTables 5-8 in the Supplement). However, there was no significant increase among younger patients with myocardial infarction (n = 39 [33.1%] vs n = 28 [33.3%]) (eTable 9 in the Supplement).

The absolute prevalence of traditional vascular risk factors in patients with TIA/stroke at younger than 55 years was high (61.8%) (Table 2), especially for current smoking (38% vs regional prevalence of 20% in 2011 and 15% in 2018²⁴) and obesity (Table 2; eTable 10 in the Supplement). Prevalences were broadly similar in men vs women (eTable 10 in the Supplement), but the proportion of women having a predicted 10-year risk of cardiovascular events of 10% or greater was significantly lower than that of men (26.2% vs 71.1%, respectively) (eTables 7, 8, and 10 in the Supplement).

Prevalence of traditional vascular risk factors in patients with TIA/stroke at younger than 55 years was significantly higher than in the underlying population (eFigure 7 in the Supplement) for diabetes (RR, 3.47; 95% CI, 2.54-4.74), hypertension (RR, 2.52; 95% CI, 2.04-3.12), current smoking (RR, 2.38; 95% CI, 1.92-2.94), and obesity (RR, 1.36; 95% CI, 1.07-1.72). Only 19.2% (41/214) of patients with TIA/stroke at younger than 55 years had an ideal premorbid blood pressure (mean blood pressure <120/80 mm Hg²⁵) during the year prior to the index stroke/TIA, compared with 39.3% in the underlying population in 2002-2017. The prevalence of vascular risk factors in patients younger than 55 years was similar for TIA/stroke vs myocardial infarction, except for current smoking (37.8% vs 54.7%, respectively; P < .001).

The proportion of patients with TIA/stroke at younger than 55 years without any traditional vascular risk factors was significantly greater in 2010-2018 vs 2002-2010 (Table 2), especially among men (16 [19.8%] vs 65 [39.9%]; absolute difference, 20.1%; 95% CI, 8.6%-31.6%; P = .002; P = .05 for heterogeneity) (eTable 7 in the Supplement) and among patients with cryptogenic events (10 [18.5%] vs 63 [49.2%]; absolute difference, 30.7%; 95% CI, 17.2%-44.2%; P < .001; P = .002 for heterogeneity) (eTable 11 in the Supplement). This difference was reversed in patients younger than 55 years with myocardial infarction (39 [29.1%] in 2002-2010 vs 15 [14.9%] in 2010-2018; P = .01) (eTable 9 in the Supplement).

Increased incidence of TIA and stroke at younger than 55 years during 2002-2018 was present irrespective of any individual traditional vascular risk factor or deprivation (Figure 3) and also irrespective of the number of vascular risk factors (eFigure 8 in the Supplement).

Figure 3. Temporal Changes in TIA and Stroke Incidence at Younger Than 55 Years Stratified by Presence of Individual Risk Factors.

IMD indicates Indices of Multiple Deprivation; IRR, incidence rate ratio; TIA, transient ischemic attack. Obesity is defined as body mass index ≥30; hypertension, blood pressure ≥140/90 mm Hg; diabetes, fasting glucose ≥126 mg/dL (7 mmol/L) on 2 separate tests. The study year began on April 1 and ended on March 31 of the next year. Populations at risk are based on prevalence of risk factors in the underlying population. Risk factor prevalence data are from the Health Survey for England; sex and deprivation data for the underlying population are by data extraction from study general practices. Whiskers indicate 95% CIs.

The frequency of other potential risk factors in patients with TIA/stroke at younger than 55 years was also high, including for low exercise levels and history of depression or other mental health disorder (Table 2; eTables 5-8 in the Supplement) but did not change significantly from 2002-2010 to 2010-2018, with the possible exception of depression (Table 2).

In terms of patients presenting to medical attention after TIA or stroke at younger than 55 years, proportions calling for medical attention or arriving to the hospital within 3 hours after symptom onset did not change significantly from 2002-2010 to 2010-2018 (eTable 3 in the Supplement). Of the 270 individuals with stroke at younger than 55 years, the proportion admitted to the hospital was also stable over time (60 [61.8%] vs 103 [59.9%]; P = .83). Although the use of brain magnetic resonance imaging (MRI) for initial investigation increased significantly during the study period for TIA, ischemic stroke, and stroke mimics at younger than 55 years, the same pattern was also seen at older ages (eFigure 9 in the Supplement).

Discussion

In this population-based study in Oxfordshire, England, there was a significant increase in stroke incidence from 2002-2010 to 2010-2018 in individuals younger than 55 years, but a decrease in those aged 55 years or older.

This finding of an absolute increase in incidence of young stroke validated UK hospital admission trends and was consistent with findings in the US.^{26,27,28,29,30} The relative divergence in incidence in younger vs older patients was also consistent with a linked systematic review in which analysis of all stroke incidence studies in the 21st century in high-income countries revealed less favorable trends in stroke incidence at younger ages vs older ages.¹⁴

The increase in stroke incidence in younger patients from 2002-2010 to 2010-2018 could not be explained by changes in diagnostic workup, stroke definition and adjudication, patient behavior, hospital admission policy, or health care coverage (universal in the UK since 1948), lending support to similar changes in other studies in high-income countries that did not address these methodological issues. Moreover, the incidence of TIA also increased significantly at younger than 55 years in Oxfordshire as well as in the US and Denmark,^29,31 suggesting that the increase in stroke incidence is unlikely to be due to a shift in diagnosis from TIA to stroke. Although MRI rates increased at younger ages over time, the same magnitude of increase was also seen at older ages, in keeping with findings in the US.^6,30

The prevalence of traditional vascular risk factors in patients with stroke at younger than 55 years was high in this study, but changes in prevalence in the underlying population did not explain the increase in stroke incidence at younger ages, consistent with the few previous studies that have reported data.^5,6,32 In contrast to patients with myocardial infarction at younger than 55 years, there was an increase in the proportion of patients younger than 55 years with TIA/stroke without known traditional vascular risk factors, particularly in cryptogenic events. Taken along with the decrease in incidence of myocardial infarction and peripheral vascular events at younger than 55 years, it seems unlikely that the overall increase in incidence of young stroke is driven by atherosclerosis. Nevertheless, traditional vascular risk factors were still more prevalent among young patients with stroke than among the underlying population, and so their identification and management are still important.

Vascular risk factors tend to be undertreated at younger ages, due at least partly to the widespread use of model-based predictions of vascular risk to identify individuals with an apparently sufficient 10-year risk of events.²³ This approach is particularly limiting in terms of primary prevention in women, with 75% of women with TIA/stroke at younger than 55 years in this study having a predicted premorbid risk of vascular events below the current treatment threshold, highlighting the need for better prognostication in younger women or consideration in the meantime of the possibility of a lower treatment threshold in younger women. It is possible that incorporation of early disease markers, such as arterial stiffness, and sex-specific risk factors, such as migraine or use of hormonal treatments, could improve prognostication.

In terms of other risk factors for stroke, the similar increase of incidence of young stroke in men and women suggests that endogenous estrogen or hormone use are unlikely to be the driver, although there was no measure of any exposure to environmental estrogens. The increase in young stroke was most marked in people doing professional or managerial jobs, which again contrasted with myocardial infarction, possibly suggesting a role for work-related stress, low physical activity, and long working hours, each of which are more strongly associated with risk of stroke than myocardial infarction.³³ A role of atmospheric pollution is also possible,³⁴ and further analyses in OXVASC are ongoing.

The increase of stroke incidence at younger ages might possibly have slowed during 2014-2018 in this study, although interpretation is very difficult given limited statistical power. Similar plateauing after 2010 may also be evident in some studies in North America and in Europe^35,36,37 but is not universal.³⁰ However, there is no evidence of any reversal of the increase in incidence, and no reduction in incidence at ages 55 to 64 years was found in this study. The Global Burden of Disease Study also reported increasing stroke incidence rates in people younger than 70 years in low- to middle-income countries.³⁸

Although a large proportion of strokes at younger ages in this study were minor as measured by NIHSS scores at baseline, the long-term risks of future vascular events and cognitive decline might be substantial. The focus over recent decades on prevention of vascular events at older ages in light of the aging population has been successful, but there is an urgent need to better understand the causes and routes to prevention of stroke at younger ages, as well as potential barriers, such as risk-based thresholds for preventive treatment with imperfect models. There is also a need for more research to determine how best to prevent recurrent strokes in young patients, very few of whom have been included in secondary prevention trials, and there are particular issues relating to the potential durations of treatments required and the time course of risk of recurrent events.

Limitations

This study has several limitations. First, as stroke is uncommon at younger than 55 years, there is a trade-off between the smaller sample size in prospective population-based studies of adjudicated strokes and the greater statistical power of large administrative data sets. The numbers in this study were too small to allow reliable determination of trends in incidence in narrower age bands, or of changes in uncommon risk factors over time. Second, the possibility that some unmeasured changes in recognition of stroke symptoms or in care-seeking behavior accounted for the increase of young stroke in this study cannot be excluded, but the consistency of the age-specific divergence in stroke incidence across all studies in the linked systematic review would indicate that any such bias would have to be very widespread. The finding of similar changes in incidence of minor vs major strokes also suggests that such recognition/behavior bias is unlikely to be a major factor. Third, although the possibility that more minor strokes were identified at younger ages due to the increased use of MRI during the study period cannot be excluded, use of MRI should not bias the time trends in more major stroke. Time trends in investigation are more problematic for studies based only on administrative data, and probably explain the apparent decline in strokes coded as unknown subtype in hospital admission data in England. Fourth, the current study results are based on a predominantly White population and might not be generalizable to other ethnic groups in whom stroke incidence might be higher.³⁹ However, a similar magnitude of increase in incidence of young stroke was also found in the other population in this study, and no differences between ethnic groups were seen in the linked systematic review.¹⁴ Fifth, the diagnosis and management of acute myocardial infarction have changed over the last 20 years, including more sensitive assays for troponin over time and the introduction of primary percutaneous coronary intervention. However, the same definition for acute myocardial infarction based on symptoms, electrocardiogram changes, and a consistent threshold for troponin were kept throughout the study.

Conclusions

Comparing persons living in Oxfordshire, England, in 2002-2010 vs 2010-2018, there was a significant increase in stroke incidence in those younger than 55 years, but a decrease in those aged 55 years and older. Given the absence of this divergence for other vascular events, further research is needed to understand the causes of this difference.

Supplement.

eAppendix. Additional Methods for the Oxford Vascular Study

eTable 1. Standardised Stroke Incidence per 1000 per Year (95%CI) in Oxfordshire From 2002 to 2018 Stratified by Age Groups, Sex, Stroke Subtype and Stroke Severity

eTable 2. Numbers of Incident Stroke at Age <55 Years in OXVASC (2002-2018) Stratified by Participating GP Practices

eTable 3. Changes in Stroke Aetiology and Patient Behaviour in Young Patients (Age<55) With Incident TIA or Stroke During OXVASC

eTable 4. Vascular Territory and Aetiology in Young Male vs. Female Patients With First-Ever Ischaemic Stroke in OXVASC (2002-2018)

eTable 5. Additional Demographics, Risk Factors, Medication and Blood Pressure at Baseline in Young Patients (Age<55) With Incident TIA or Stroke in OXVASC

eTable 6. Demographics, Traditional Risk Factors, Young Stroke Specific Risk Factors and Medication Prior to the Index Event in Young Patients (Age<55) With Incident Stroke in OXVASC (2002-2018)

eTable 7. Demographics, Risk Factors, Medication and Blood Pressure Prior to the Index Event in Young Male Patients (Age<55) With Incident TIA or Stroke in OXVASC (2002-2018)

eTable 8. Demographics, Risk Factors, Medication and Blood Pressure Prior to the Index Event in Young Female Patients (Age<55) With Incident TIA or Stroke in OXVASC (2002-2018)

eTable 9. Demographics, Risk Factors and Premorbid Medication in Young Patients (Age<55) With Incident Myocardial Infarction or Sudden Cardiac Death in OXVASC (2002-2018)

eTable 10. Demographics, Risk Factors, Medication and Blood Pressure Prior to the Index Event in Young Male vs. Female Patients (age<55) With Incident TIA or Stroke in OXVASC (2002-2018)

eTable 11. Number of Incident Transient Ischaemic Attack (TIA), Ischaemic Stroke and Myocardial Infarction Stratified by Number of Risk Factors and TOAST Subtypes in OXVASC (2002-2018)

eTable 12. Number of Events and Population at Risk Included in the Calculation of Temporal Trends of Standardised Stroke Incidence in Those Aged <60 Years in England (Hospital Admission Data of Any Stroke; 1998-2018) and in Those Aged<55 Years in Oxfordshire (Stroke Incidence Studies; 1975-1976;²¹ 1981-1986; 2002-2018)

eFigure 1. Creating the Health Survey for England Time Series Dataset for South East England (1994-2017)

eFigure 2. Time Trends of Standardised Stroke Incidence at Age<55 Years vs. Age≥55 Years Within OXVASC (2002-2018) Excluding Those With Known Atrial Fibrillation

eFigure 3. Temporal Trends in Standardised Stroke Incidence in Those Aged<55 Years in Oxfordshire

eFigure 4. Temporal Trends of Standardised Stroke Incidence in Those Aged <60 Years in England Using the Hospital Episode Statistics Record (1998-2018)

eFigure 5. Number of Incident Stroke Cases <55 Years Stratified by the NIHSS Score in OXVASC (2002-2018)

eFigure 6. Risks of Recurrent Stroke at One Year in Those Without Atrial Fibrillation Stratified by Age

eFigure 7. Time Trends in the Prevalence of Risk Factors in South East England Age 16-54 Years (Health Survey for England)

eFigure 8. Time Trends of Stroke Incidence at Age<55 Years Within OXVASC (2002-2018) Stratified by Number of Vascular Risk Factors

eFigure 9. Time Trends in MRI Brain Imaging Rates in OXVASC Stratified by Age