Abstract
Introduction
Patients who suffer a stroke at a young age, remain at a substantial risk of developing recurrent vascular events and information on very long-term prognosis and its risk factors is indispensable. Our aim is to investigate this very long-term risk and associated risk factors up to 35 years after stroke.
Patients and methods
Prospective cohort study among 656 patients with a first-ever ischaemic stroke or transient ischaemic stroke (TIA), aged 18–50, who visited our hospital (1980–2010). Outcomes assessed at follow-up (2014–2015) included TIA or ischaemic stroke and other arterial events, whichever occurred first. Kaplan–Meier analysis quantified cumulative risks. A prediction model was constructed to assess risk factors independently associated with any ischaemic event using Cox proportional hazard analyses followed by bootstrap validation procedure to avoid overestimation.
Results
Mean follow-up was 12.4 (SD 8.2) years (8105 person-years). Twenty-five years cumulative risk was 45.4% (95%CI: 39.4–51.5) for any ischaemic event, 30.1% (95%CI: 24.8–35.4) for cerebral ischaemia and 27.0% (95%CI: 21.1–33.0) for other arterial events. Risk factors retained in the prediction model were smoking (HR 1.35, 95%CI: 1.04–1.74), poor kidney function (HR 2.10, 95%CI: 1.32–3.35), history of peripheral arterial disease (HR 2.10, 95%CI: 1.08–3.76) and cardiac disease (HR 1.84, 95%CI: 1.06–3.18) (C-statistic 0.59 (95%CI: 0.55–0.64)).
Discussion and conclusion
Young stroke patients remain at a substantial risk for recurrent events; almost 1 of 2 develops a recurrent ischaemic event and 1 of 3 develops a recurrent stroke or TIA during 25 years of follow-up. Risk factors independently associated with recurrent events were poor kidney function, smoking, history of peripheral arterial disease and cardiac disease.
Keywords: Young stroke, prognosis, risk factors, cardiovascular disease
Introduction
Long-term prognosis after a stroke at a young age is significantly affected by occurrence of incident vascular events. It is already known that this risk is not only high in the first months or years after stroke, but patients remain at a substantial risk even 10 years after their initial stroke.1
Reliable information thereafter is largely lacking, though important to know as after 10 years these patients are still young and can expect to live for many more decades. Providing young adults with reliable information on the course of their disease, including the risk factors associated with this prognosis, also after the first 10 years is therefore of utmost importance.
Only a few large studies have tried to identify which particular risk factors were associated with recurrent vascular events. However, in those studies mean follow-up did not exceed 5 years,2,3 or only the association between the etiologic subgroups based on TOAST classification and recurrent events was taken into account rather than subgroups based on (potentially modifiable) risk factors.4 Also, prognosis has only been examined for the overall stroke population. Though this is important information from health care perspective, this does not always answer the question of individual patients who have a particular interest in their own, personalised prognosis. One study did develop a score for individual risk estimation of recurrent thrombotic events, with a moderate predictive value and follow-up duration of more than 3 years.2 Especially for younger patients, an individualised long-term risk assessment, after this first years after stroke is also of great importance.
Therefore, the aims of this study were to investigate the very long-term risk of recurrent vascular events up to 35 years after a stroke at a young age. In addition, we aimed to identify risk factors independently associated with the risk of recurrent vascular events on the long-term. Subsequently, predictive performance of the resulting model was assessed.
Patients and methods
Study population
This study is part of the “Follow-Up of Transient ischaemic attack and stroke patients and Unelucidated Risk factor Evaluation” study (FUTURE study), a prospective cohort study on causes and consequences of a stroke in young adults.5 Detailed methods have been described previously.1 The Medical Review Ethics Committee region Arnhem-Nijmegen approved the study. All patients signed informed consent.
In the present study, we included all consecutive patients with a first-ever transient ischaemic attack (TIA) or ischaemic stroke, aged 18–50 years, admitted to the Radboud University Nijmegen Medical Centre from 1980 till 2010. The World Health Organization definition for TIA and stroke was used.6
Patients were identified through a prospective registry with standardised collection of baseline characteristics. Cardiovascular risk factors were defined as a history of a risk factor or detected during analysis of the initial event: hypertension as mean systolic blood pressure ≥ 135 mmHg or a diastolic blood pressure of ≥ 85 mmHg, measured after the first week of the index event7 or the use of antihypertensive medication; diabetes mellitus as random blood glucose level > 200 mg/dL (11.1 mmol/L) or two consecutive fasting venous plasma glucose levels ≥ 6.1 mg/dL (7.0 mmol/L)8 or the use of antidiabetics; dyslipidemia as total cholesterol ≥ 5.0 mmol/L, LDL ≥ 2.5 mmol/L or triglycerides ≥ 2.0 mmol/L1 or the use of statins; smoking as at least 1 cigarette per day in the year prior to the event; atrial fibrillation when identified on electrocardiogram or during continuous electrocardiographic recording; excess alcohol consumption as consuming more than 200 g of pure alcohol per week. Kidney function was divided in estimated glomerular filtration rate (eGFR) < 60, eGFR 60–120 and eGFR > 120 mL/min/1.73m2.9 In addition, a history of cardiovascular disease (myocardial infarction, cardiac intervention procedures and peripheral artery revascularisation procedures) was collected.
The assessment of stroke etiology (Trial of Org 10172 in Acute Stroke Treatment (TOAST))10 and severity (National Institutes of Health Stroke Scale11) was done for all cases retrospectively by a validated approach.12
Follow-up
Follow-up assessment took place from 2009 to 2012 and subsequently from August 2014 to January 2015. In case information from this last follow-up assessment was missing, follow-up data from the previous follow-up (2009–2012) were used (n = 28, 4.3%). Patients underwent structured questionnaires on the occurrence of recurrent events. In case a patient had died, information was retrieved by the general practitioner. All confirmed events were verified by a neurologist or cardiologist. TIA and ischaemic stroke were defined similar to the index event, myocardial infarction was defined according to the universal definition of myocardial infarction.13
Outcome
Primary outcome was the risk of any recurrent ischaemic event, defined as the composite of cerebral ischaemia (ischaemic stroke or TIA) and other arterial events (myocardial infarction, coronary artery bypass grafting, percutaneous coronary intervention, carotid endarterectomy or peripheral artery revascularisation procedures). Secondary outcomes were the risk of recurrent cerebral ischaemia and the risk of other arterial events.
Statistical analysis
Kaplan–Meier analysis was used to estimate cumulative risks of any ischaemic event, cerebral ischaemia and other arterial event, whichever occurred first. Person-years at risk were calculated for each patient from date of the index stroke until recurrent event, death or end of follow-up. Patients who died and/or did not reach an endpoint were censored. Survival plots were curtailed at 25 years to ensure that the provided survival plots were reliable for all subgroups,14 all events were retained in subsequent analyses. Analyses were stratified by stroke subtype, age categories (18–29, 30–39 and 40–50 years) and TOAST subtype with undetermined cause as reference.
Annual risks were calculated from survival estimates obtained by Kaplan–Meier analysis. Mean annual risks were subsequently calculated for the periods 2–5 years, 6–10 years, 11–15 years, 16–20 years, and 21–25 years.
Missing data were handled using multiple imputation by the method of chained equations (eGFR 29.9% and dyslipidemia 20%),15 under the assumption that missing data were missing at random. Thirty imputed datasets were created using predictive mean matching for eGFR and logistic regression for dyslipidemia. All variables considered in the subsequent regression analyses were included in the imputation model. Regression analyses were performed on each of the imputed datasets individually and subsequently the coefficients were pooled using Rubin’s rules.16 Restricting the analyses to only complete patients yielded similar point estimates as obtained in the imputed datasets. Cox proportional hazard analysis was used to calculate hazard ratios with their 95% confidence intervals (CIs) for baseline characteristics adjusted for age and sex for the risk of any ischaemic event, cerebral ischaemia and other arterial events.
A composite variable with number of traditional risk factors was used (diabetes mellitus, hypertension, smoking, dyslipidemia or atrial fibrillation).
Finally we constructed a multivariable Cox regression model to predict recurrent ischaemic events. All predictors were considered for entrance in the models irrespective of their univariable association with the outcome. Subsequently, the model was simplified using a backward selection procedure (exclusion if p < 0.20).
A bootstrap validation procedure (b = 100) was performed to avoid overestimation in regression coefficients.17 A shrinkage factor was derived from this procedure and subsequently applied to the regression coefficients. Discriminative performance of the model was assessed by calculating the concordance (c) statics. Also the C-statistic was corrected for overoptimism by applying the bootstrap validation procedure. All p-values of less than 0.05 were considered as significant. IBM SPSS 20.0 and R version 3.2.1 software package (packages MICE version 2.25 and RMS version 4.4-0) were used.
Results
Study population
Six hundred fifty-six patients entered present study, 209 (31.9%) patients with TIA and 447 (61.7%) patients with ischaemic stroke. Mean follow-up was 12.4 (SD8.6) years and 21.3% had a follow-up of more than 20 years. At the end of the follow-up, 167 (25.5%) patients were deceased. Baseline characteristics are shown in Table 1.
Table 1.
Baseline characteristics.
| Total 656 | TIA 209 | Ischaemic stroke 447 | |
|---|---|---|---|
| Mean age at time of event, years (SD) | 40.7 (7.7) | 40.6 (8.0) | 40.8 (7.6) |
| Men, n (%) | 309 (47.1) | 94 (45.0) | 215 (48.1) |
| Mean follow-up duration, years (SD) | 12.4 (8.6) | 11.9 (8.1) | 12.6 (8.8) |
| >15 years | 217 (33.1) | 56 (26.8) | 171 (38.3) |
| >20 years | 140 (21.3) | 41 (19.6) | 99 (22.1) |
| >25 years | 73 (11.1) | 26 (12.4) | 47 (10.5) |
| Median NIHSS at admission (IQR)a | 3 (1–7) | 0 (0–1) | 5 (2–10) |
| TOAST, n (%) | |||
| Large artery disease | 166 (25.3) | 47 (22.5) | 119 (26.6) |
| Cardio-embolic stroke | 86 (13.1) | 27 (12.9) | 59 (13.2) |
| Small vessel disease | 65 (13.1) | 11 (5.3) | 54 (12.1) |
| Other defined | 96 (14.6) | 23 (11.0) | 73 (16.3) |
| Multiple causes | 17 (2.6) | 2 (1.0) | 15 (3.4) |
| Unknown | 226 (34.5) | 99 (47.4) | 127 (28.4) |
| History of vascular risk factors, n (%) | |||
| Hypertension | 231 (35.2) | 71 (34.0) | 160 (35.8) |
| Diabetes | 54 (8.2) | 13 (6.2) | 41 (9.2) |
| Dyslipidemiaa | 424 (80.8) | 147 (83.5) | 277 (79.4) |
| Atrial fibrillation | 15 (2.3) | 5 (2.4) | 10 (2.2) |
| Smokinga | 324 (51.8) | 83 (40.7) | 241 (57.2) |
| Excess drinking | 48 (7.3) | 15 (7.2) | 33 (7.4) |
| Number of traditional risk factorsb | |||
| 0 | 72 (11.0) | 26 (12.4) | 46 (10.3) |
| 1 | 243 (37.0) | 82 (39.2) | 161 (36.0) |
| 2 | 238 (36.3) | 71 (34.0) | 167 (37.4) |
| ≥3 | 103 (15.7) | 30 (14.4) | 73 (16.3) |
| History of cardiovascular disease, n (%)c | 38 (5.8) | 11 (5.3) | 27 (6.0) |
| Cardiac disease | 29 (4.4) | 9 (4.3) | 20 (4.5) |
| Peripheral arterial disease | 21 (3.2) | 7 (3.3) | 14 (3.1) |
| Baseline kidney function, n (%) | |||
| eGFR 60–120 | 392 (85.2) | 122 (88.4) | 270 (83.9) |
| eGFR < 60 | 29 (6.3) | 7 (5.1) | 22 (6.8) |
| eGFR > 120 | 39 (8.5) | 9 (6.5) | 30 (9.3) |
TIA: transient ischaemic attack; NIHSS: national institutes of health stroke scale; IQR; interquartile range.
Missing data: NIHSS 0.5%; dyslipidemia 20%; smoking 4.7%; eGFR 29.9%.
Risk factors included diabetes mellitus, hypertension, dyslipidemia, atrial fibrillation and smoking.
Some patients had a history of both PAD and cardiac disease.
There were no differences in baseline characteristics between participants and non-participants (lost to follow-up n = 73 and refusers n = 139) except for age (39.1 years (SD8.2) versus 40.7 years (SD7.7) respectively).
Cumulative risk of ischaemic events
During 8105 person-years of follow-up, 197 (30%) patients had at least one ischaemic event of whom 67 (34.0%) had more than one event.
Twenty-five years cumulative risk of any ischaemic event was 45.4% (95%CI: 39.4–51.5), 30.1% (95%CI: 24.8–35.4) for cerebral ischaemia and 27.0% (95% CI: 21.1–33.0) for other arterial events (Figure 1(a) to (c)). There was no sex difference. Annual risks are shown in Figure 2.
Figure 1.
Cumulative risks of ischaemic events: (a) any ischaemic event; (b) cerebral ischaemia; (c) other arterial events.
Figure 2.
Annual risk of ischaemic events. Data points represent mean annual risks for the periods 2 to 5 years, 6 to 10 years, 11 to 15 years, 16 to 20 years and 21 to 25 years.
There were no differences between patients with a TIA or ischaemic stroke with regards to their 25-years cumulative risk of any ischaemic event (TIA 40.7% (95%CI: 29.7–46.4) and ischaemic stroke 47.1% (95%CI: 40.0–50.8), cerebral ischaemia (TIA 27.8% (95%CI: 23.0–37.0) and ischaemic stroke 30.8% (95%CI: 27.6–37.2), or other arterial events (TIA 23.7% (95%CI: 18.3–34.2) and ischaemic stroke 28.5% (95%CI: 24.9–35.8).
There were no differences in 25-years cumulative risk of any ischaemic event or cerebral ischaemia by age category. Patients aged 40–50 years at the time of their qualifying stroke at a young age had a higher 25-years cumulative risk of other arterial events (38.2%, 95%CI: 28.9–47.6) compared to patients aged 30–39 years (15.5%, 95%CI: 7.2–23.7) (log rank p = 0.004) and patients aged 18–29 years (7.8%, 95%CI: 0.3–15.4) (log rank p = 0.002).
Figure 3 shows the cumulative risk of any ischaemic event stratified by TOAST subtype. For cerebral ischaemia, patients with a cardio-embolic stroke (50.8%, 95%CI: 25.8–75.7) or small vessel disease (37.5%, 95%CI: 22.7–52.4) had a higher 25-years cumulative risk than patients with undetermined cause (22.9%, 95%CI: 15.2–30.6) (log rank p = 0.02 and p = 0.02 respectively). For other arterial events patients with a large artery stroke (48.5, 95%CI: 37.8–59.2) or cardio-embolic stroke (60.9, 95%CI: 6.0–115.8) had a higher 25-years cumulative risk than those with undetermined cause (17.3, 95%CI: 8.6–26.1) (log rank p < 0.001 and p = 0.05 respectively).
Figure 3.
Cumulative risk of any ischaemic event stratified by TOAST subtype. Patients with large artery (61.4%, 95%CI: 51.1–71.8%) or cardio-embolic stroke (62.7%, 95%CI: 38.6–86.7%) had a higher cumulative risk of any ischaemic event compared to undetermined cause (33.8%, 95%CI: 24.3–43.2%) (log rank p < 0.001 and p = 0.001 respectively).
Risk factors
In age- and sex-adjusted univariable analyses, age (HR 1.03 per year increase, 95%CI: 1.01–1.05), diabetes mellitus (HR 1.62, 95%CI: 1.03–2.57), hypertension (HR 1.36, 95%CI: 1.02–1.82), eGFR < 60 (2.51, 95%CI: 1.47–4.26, history of peripheral arterial disease (PAD) (HR 3.44, 95%CI: 21.95–6.08) and history of cardiac disease (HR 3.33, 95%CI: 2.01–5.52) were associated with any ischaemic event. In addition, number of traditional risk factors was associated with any ischaemic event (Table 2).
Table 2.
Univariate age- and sex-adjusted associations with recurrent vascular events.
| Any ischaemic event |
Cerebral ischaemia |
Other arterial events |
||||
|---|---|---|---|---|---|---|
| HR (95%CI) | p-Value | HR (95%CI) | p-Value | HR (95%CI) | p-Value | |
| Sex, men | 0.91 (0.68–1.21) | 0.50 | 0.94 (0.67–1.33) | 0.73 | 0.77 (0.51–1.17) | 0.22 |
| Age | 1.03 (1.01–1.05) | 0.01 | 1.01 (0.99–1.03) | 0.35 | 1.08 (1.04–1.11) | <0.001 |
| Diabetes mellitus | 1.62 (1.03–2.57) | 0.04 | 1.24 (0.68–2.26) | 0.48 | 2.54 (1.42–4.52) | 0.002 |
| Hypertension | 1.36 (1.02–1.82) | 0.04 | 1.08 (0.75–1.54) | 0.68 | 1.61 (1.06–2.43) | 0.02 |
| Dyslipidemia | 1.38 (0.90–2.12) | 0.14 | 1.12 (0.69–1.81) | 0.65 | 1.72 (0.88–3.36) | 0.11 |
| Excess alcohol consumption | 0.89 (0.53–1.51) | 0.68 | 0.96 (0.51–1.80) | 0.89 | 1.15 (0.59–2.26) | 0.68 |
| Smoking | 1.34 (1.00–1.80) | 0.05 | 1.18 (0.83–1.68) | 0.35 | 2.12 (1.32–3.42) | 0.002 |
| Atrial fibrillation | 1.34 (0.55–3.27) | 0.52 | 1.51 (0.55–4.10) | 0.42 | 0.43 (0.06–3.09) | 0.40 |
| Number of traditional risk factors | ||||||
| 0 | 1 | 1 | ||||
| 1 | 1.27 (0.57–2.81) | 0.55 | 1.03 (0.46–2.30) | 0.94 | NA | |
| 2 | 2.07 (0.95–4.49) | 0.07 | 1.19 (0.54–2.63) | 0.67 | NA | |
| ≥3 | 2.54 (1.14–5.66) | 0.02 | 1.46 (0.64–3.35) | 0.37 | NA | |
| Baseline kidney function | ||||||
| eGFR 60–120 | 1 | 1 | ||||
| eGFR < 60 | 2.51 (1.47–4.26) | 0.001 | 3.18 (1.74–5.80) | <0.001 | 2.03 (0.92–4.49) | 0.08 |
| eGFR > 120 | 0.86 (0.40–1.82) | 0.69 | 1.10 (0.49–2.46) | 0.82 | 0.29 (0.04–2.14) | 0.23 |
| History of cardiovascular disease | ||||||
| Peripheral arterial disease | 3.44 (1.95–6.08) | <0.001 | 1.38 (0.56–3.39) | 0.48 | 6.87 (3.60–13.12) | <0.001 |
| Cardiac disease | 3.33 (2.01–5.52) | <0.001 | 1.80 (0.87–3.74) | 0.11 | 4.99 (2.70–9.21) | <0.001 |
Age- and sex-adjusted hazard ratios were calculated using Cox proportional hazards analyses. Imputed data were used for missing data. It was not possible to determine hazard ratios for ‘number of traditional risk factors’ for the outcome event ‘other arterial events’, due to total separation of the dataset (none of the patients without risk factors reached the endpoint ‘other arterial event’).
Kidney function (eGFR < 60) was associated with recurrent cerebral ischaemia (HR 3.18, 95%CI: 1.74–5.80). Factors associated with other arterial event were age (HR 1.08 per year increase, 95%CI: 1.04–1.11), diabetes mellitus (HR 2.54, 95%CI: 1.42–4.52), hypertension (HR 1.61, 95%CI: 1.06–2.43), smoking (HR 2.12, 95%CI: 1.32–3.42), history of PAD (HR 6.87, 95%CI: 3.60–13.12) and cardiac disease (HR 4.99, 95%CI: 2.70–9.21).
Table 3 shows the multivariable Cox proportional hazard model. Smoking (HR 1.35, 95%CI: 1.04–1.74), eGFR < 60 (HR 2.10, 95%CI: 1.32–3.35), history of PAD (HR 2.10, 95%CI: 1.08–3.76) and cardiac disease (HR 1.84, 95%CI: 1.06–3.18) were independently associated with any ischaemic event. Reported hazard ratios have been shrunk by applying a shrinkage factor of 0.82 derived from the bootstrap validation procedure. C-statistic of the model was 0.59 (95%CI: 0.55–0.64) after correction for overoptimism.
Table 3.
Predictors of any ischaemic event.
| HR (95%CI) | p-Value | |
|---|---|---|
| Age | 1.02 (1.00–1.04) | 0.06 |
| Diabetes mellitus | 1.39 (0.93–2.07) | 0.11 |
| Alcohol | 0.73 (0.47–1.15) | 0.18 |
| Smoking | 1.35 (1.04–1.74) | 0.02 |
| Baseline kidney function | ||
| eGFR 60–120 | 1 | |
| eGFR < 60 | 2.10 (1.32–3.35) | 0.002 |
| eGFR > 120 | 0.80 (0.42–1.52) | 0.50 |
| History of peripheral arterial disease | 2.01 (1.08–3.76) | 0.03 |
| History of cardiac disease | 1.84 (1.06–3.18) | 0.03 |
Predictors of any ischaemic events selected by backward LR (exclusion if p < 0.200). C-statistic 0.59 (95%CI: 0.55–0.64). The regression coefficients were corrected for overfitting with a shrinkage factor of 0.82.
Discussion
We showed that patients who have suffered a stroke at a young age remain at a high risk of developing recurrent ischaemic events, even decades after their stroke. Traditional vascular risk factors are common in these patients and are associated with this life-long risk.
A unique and essential strength of our study is the large study population in combination with a very long-term follow-up duration up to 35 years. For patients who suffer a stroke at a young age, a long-term risk assessment is crucial, as patients are still young after only a few years of follow-up. Moreover, the single centre, prospective design, made it possible to uniformly collect data, making information bias less likely.
However, some possible limitations need to be addressed. First of all selection bias might have occurred as patients who did not participate in the study might have been the ones with a more severe outcome. The number of patients who refused to participate and those lost to follow-up is rather high, which is probably due to the long follow-up duration, however there were no differences in baseline characteristics, making selection bias less likely.
Second, since our study features a long inclusion period, recall bias may have influenced our findings, especially when patients experienced their event a long time ago or died, however, this would only have resulted in an underestimation of cumulative incidence. In addition, we have tried to overcome this by collecting information in an identical way in patients who died and those still alive. Along the course of the inclusion period, specific young stroke investigations might have changed and secondary prevention strategies changed over time, which might have influenced risk of recurrent vascular events. However, these are unavoidable features of a long-term follow-up study.
We have previously shown that patients who suffer a stroke at a young age remain at a high risk of developing recurrent vascular events.1 Due to the extension of our follow up we were now able to show that even up to 25 years after a stroke at a young age the risk of recurrent events remains high. Moreover, we found an even higher risk compared to other large cohorts,2–4,18–20 most likely due to the far longer follow-up duration and inclusion of TIA as an outcome event in our cohort. In addition, most other studies were small and/or retrospective.3,18–20
To determine secondary prevention strategies and provide information on the course of the disease for individual patients, it is necessary to assess risk of recurrent events at an individual level and treatment decisions should ideally be based on clinical trials. As younger patients were usually underrepresented in trials on secondary prevention,21 it remains uncertain how to treat these patients and information should be obtained from large cohorts.22 We therefore aimed to identify risk factors associated with recurrent vascular events. Poor kidney function, smoking and a history of PAD and cardiac disease were independently associated with recurrent ischaemic events. However, the prediction model only had a moderate discriminative ability. Possible reasons for this finding might include residual confounding which might have arisen from inevitable differences in measurement of risk factors during the course of an inclusion period spanning almost 30 years. Furthermore, due to the long inclusion period in our cohort, secondary prevention strategies changed, which may have influenced the prediction model as well.
Two previous studies also tried to identify risk factors associated with recurrent vascular events. One study found age, heart failure, previous TIA and diabetes mellitus to be important predictors3 while the other prospective study found especially discontinuation of medication, antiphospholipid antibodies, family history, migraine with aura and any increase in traditional vascular risk factors to be associated with recurrent events2. However, follow-up in those studies was shorter and they had a smaller number of events to include in the analyses. Differences in study design and inclusion criteria between the studies could have contributed to the variation in associated risk factors. Patients in our study and the Finnish study were older (41.5 years) than in the Italian stroke project (36.8 years), which may have resulted in a higher incidence of traditional risk factors. In addition, it may be possible that due to the longer follow-up duration, traditional vascular risk factors become more important as patients are longer exposed to the risk factors. In addition, the Italian study developed a score for individual risk estimation, with a better predictive value than the prediction model in our study (C-statistic 0.66). Unfortunately, none of the models had a good discriminative ability, which may also indicate that there is considerable potential to be gained. Other unknown factors may contribute to the risk of recurrent events. For example, increasing evidence suggests that stroke at a young age has a stronger genetic basis than stroke occurring at older ages.23,24 Genetic factors might also interact with vascular risk factors, possibly making patients who have suffered a stroke at a young age more vulnerable to traditional vascular risk factors.
Furthermore, we found that patients with a cardio-embolic stroke or large artery disease had a higher risk of any ischaemic event or other arterial event. Another large study on the association between etiologic subgroups and recurrent events, also found large artery atherosclerosis and cardioembolism to be important predictors of recurrent events.4 Moreover, in our study patients with small vessel disease or large artery disease had a higher risk of recurrent ischaemia. This is probably explained by the fact that patients with these etiologies of their stroke are more likely to have traditional risk factors than patients with an undetermined cause. Surprisingly, age was not associated with the risk of recurrent events, which indicates that even very young patients remain at a substantial risk of developing recurrent vascular events.
Our findings may indicate that risk factors for vascular diseases and a history of vascular diseases such as peripheral artery or cardiac disease should create awareness among the treating physician for risk of future recurrent events. Especially smoking is an important modifiable risk factor, which can be addressed more straightforward than pre-existing atherosclerosis in daily practice. Although prediction models are not perfect, traditional vascular (partly) modifiable risk factors do seem to play an important role in developing recurrent events. This might suggest that strict long-term secondary prevention strategies should be encouraged, even in very young patients, however it should be noted that the use and adherence of medication is not taken into account in this study. Therefore, future research should especially aim at secondary prevention strategies in this specific age group.
In conclusion, we have shown that patients with a stroke at young age remain at substantial risk for developing recurrent vascular events, even decades after their initial event. Particularly young patients, with vascular risk factors that are usually prevalent at higher ages should raise the level of suspicion of their treating physicians as they are at the highest risk for future vascular events.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Frank-Erik de Leeuw received research support from the “Dutch Epilepsy Fund” (grant number 2010-18), ‘Dutch Heart Foundation’ (clinical established investigator grant, grant number 2014-T060) and “The Dutch Organisation for Health Research and Development” (VIDI innovational grant, ZonMw, grant number 016-126-351). Loes Rutten-Jacobs was supported by a British Heart Foundation Immediate Research Fellowship (FS/15/61/31626) (www.bhf.org.uk).
Ethical approval
The Medical Review Ethics Committee region Arnhem-Nijmegen approved the study.
Informed consent
Written informed consent was obtained from the patients for their anonymised information to be published in this article.
Guarantor
FEdL.
Contributorship
RA, MvA, NS, JvP, NM, HS, MvdV, LR-J were involved in patient recruitment and data collection. RA and LR-J were involved in data analysis. RA and MvA wrote the first draft of the manuscript and contributed equally. FEdL gained ethical approval. LR-J and FEdL were involved in the protocol development and conceived the study. All authors reviewed and edited the manuscript and approved the final version of the manuscript.
References
- 1.Rutten-Jacobs LC, Maaijwee NA, Arntz RM, et al. Long-term risk of recurrent vascular events after young stroke: The FUTURE study. Ann Neurol 2013; 74: 592–601. [DOI] [PubMed] [Google Scholar]
- 2.Pezzini A, Grassi M, Lodigiani C, et al. Predictors of long-term recurrent vascular events after ischemic stroke at young age: The Italian Project on Stroke in Young Adults. Circulation 2014; 129: 1668–1676. [DOI] [PubMed] [Google Scholar]
- 3.Putaala J, Haapaniemi E, Metso AJ, et al. Recurrent ischemic events in young adults after first-ever ischemic stroke. Ann Neurol 2010; 68: 661–671. [DOI] [PubMed] [Google Scholar]
- 4.Aarnio K, Siegerink B, Pirinen J, et al. Cardiovascular events after ischemic stroke in young adults: A prospective follow-up study. Neurology 2016; 86: 1872–1879. [DOI] [PubMed] [Google Scholar]
- 5.Rutten-Jacobs LC, Maaijwee NA, Arntz RM, et al. Risk factors and prognosis of young stroke. The FUTURE study: A prospective cohort study. Study rationale and protocol. BMC Neurol 2011; 11: 109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Hatano S. Experience from a multicentre stroke register: A preliminary report. Bull World Health Organ 1976; 54: 541–553. [PMC free article] [PubMed] [Google Scholar]
- 7.Rutten-Jacobs LC, Arntz RM, Maaijwee NA, et al. Long-term mortality after stroke among adults aged 18 to 50 years. JAMA 2013; 309: 1136–1144. [DOI] [PubMed] [Google Scholar]
- 8.Alberti KG, Zimmet PZ. Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1: Diagnosis and classification of diabetes mellitus provisional report of a WHO consultation. Diabet Med 1998; 15: 539–553. [DOI] [PubMed] [Google Scholar]
- 9.Synhaeve N, van Alebeek ME, Arntz R, et al. Kidney dysfunction increases mortality and incident events after young stroke: The FUTURE study. Cerebrovasc diseases 2016; 42: 224–231. [DOI] [PubMed]
- 10.Bousser MG, Amarenco P, Chamorro A, et al. Rationale and design of a randomized, double-blind, parallel-group study of terutroban 30 mg/day versus aspirin 100 mg/day in stroke patients: The prevention of cerebrovascular and cardiovascular events of ischemic origin with terutroban in patients with a history of ischemic stroke or transient ischemic attack (PERFORM) study. Cerebrovasc Dis 2009; 27: 509–518. [DOI] [PubMed] [Google Scholar]
- 11.Brott T, Adams HP, Jr, Olinger CP, et al. Measurements of acute cerebral infarction: A clinical examination scale. Stroke 1989; 20: 864–870. [DOI] [PubMed] [Google Scholar]
- 12.Williams LS, Yilmaz EY, Lopez-Yunez AM. Retrospective assessment of initial stroke severity with the NIH Stroke Scale. Stroke 2000; 31: 858–862. [DOI] [PubMed] [Google Scholar]
- 13.Thygesen K, Alpert JS, White HD, et al. Universal definition of myocardial infarction. Circulation 2007; 116: 2634–2653. [DOI] [PubMed] [Google Scholar]
- 14.Pocock SJ, Clayton TC, Altman DG. Survival plots of time-to-event outcomes in clinical trials: Good practice and pitfalls. Lancet 2002; 359: 1686–1689. [DOI] [PubMed] [Google Scholar]
- 15.van Buuren S, Groothuis-Oudshoorn K. Mice: Multivariate imputation by chained equations in R. J Stat Softw 2011; 45: 1–67. [Google Scholar]
- 16.Rubin DB. Multiple imputation for nonresponse in surveys, New York: John Wiley & Sons, Inc., 1987. [Google Scholar]
- 17.Harrell FE, Jr, Lee KL, Mark DB. Multivariable prognostic models: Issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996; 15: 361–387. [DOI] [PubMed] [Google Scholar]
- 18.Naess H, Nyland HI, Thomassen L, et al. Long-term outcome of cerebral infarction in young adults. Acta Neurol Scand 2004; 110: 107–112. [DOI] [PubMed] [Google Scholar]
- 19.Nedeltchev K, der Maur TA, Georgiadis D, et al. Ischaemic stroke in young adults: Predictors of outcome and recurrence. J Neurol Neurosurg Psychiatry 2005; 76: 191–195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Varona JF, Bermejo F, Guerra JM, et al. Long-term prognosis of ischemic stroke in young adults. Study of 272 cases. J Neurol 2004; 251: 1507–1514. [DOI] [PubMed] [Google Scholar]
- 21.Hong KS, Yegiaian S, Lee M, et al. Declining stroke and vascular event recurrence rates in secondary prevention trials over the past 50 years and consequences for current trial design. Circulation 2011; 123: 2111–2119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Naess H, Waje-Andreassen U, Thomassen L, et al. Do all young ischemic stroke patients need long-term secondary preventive medication? Neurology 2005; 65: 609–611. [DOI] [PubMed] [Google Scholar]
- 23.Traylor M, Bevan S, Rothwell PM, et al. Using phenotypic heterogeneity to increase the power of genome-wide association studies: Application to age at onset of ischaemic stroke subphenotypes. Genet Epidemiol 2013; 37: 495–503. [DOI] [PubMed] [Google Scholar]
- 24.Cheng YC, Cole JW, Kittner SJ, et al. Genetics of ischemic stroke in young adults. Circ Cardiovasc Genet 2014; 7: 383–392. [DOI] [PMC free article] [PubMed] [Google Scholar]