Long-Term Mortality After First-Ever and Recurrent Stroke in Young Adults
Background and Purpose—Mortality after first-ever stroke, and particularly after recurrent stroke, and predictors of long-term mortality among young and middle-aged stroke patients are not well-known. We assessed 17-year risk of mortality with focus on the effect of recurrence on the risk of death of young and middle-aged patients with stroke.
Methods—Mortality and recurrent stroke rate of 970 consecutive 30-day survivors of first-ever ischemic stroke aged 15 to 49 years (1994–2007) were studied. Prospective follow-up data came from the Finnish Care Register for Health Care and Statistics Finland. Mean follow-up was 10.2±4.3 years. We compared survival between clinical subgroups and identified factors associated with mortality. Standardized mortality ratio was calculated for demographic and pathogenetic subgroups using mortality data of the general population matched with age, sex, calendar year, and geographical area.
Results—At the end of follow-up, 152 (15.7%) patients had died (cumulative risk, 23.0%; 95% confidence interval, 19.1%–26.9%) and 132 (13.6%) had experienced a recurrent stroke. After adjusting for baseline characteristics, recurrent stroke was statistically the most important risk factor for mortality after first-ever ischemic stroke (hazard ratio, 16.68; 95% confidence interval, 2.33–119.56; P=0.005). Observed mortality was 7-fold higher than the expected mortality (standardized mortality ratio, 6.94; 95% confidence interval, 5.84–8.04) and particularly high among patients who experienced a recurrent stroke (standardized mortality ratio, 14.43; 95% confidence interval, 10.11–18.74).
Conclusions—The high mortality rates and the striking impact of recurrent stroke on the risk of death should lead to development of more robust primary and secondary prevention strategies for young patients with stroke.
Several lines of evidence suggest a rising incidence of stroke among younger people but a declining trend of stroke in the elderly.1–3 Stroke in the young and middle-aged patients has serious consequences. Those <50 years of age are typically vocationally active, and they have families to take care of. Dying or being left disabled cause enormous social and economical burden to the society, as well as patients themselves and their next-of-kin. Recent studies have revealed that mortality among young patients with stroke is lower than in the elderly but significantly higher than in the age-adjusted general population.4–8
Previous studies show that the cumulative mortality among young and middle-aged ischemic patients with stroke is 10.7% at 5 years, and this figure rises ≤27% after a mean follow-up of 18.3 years.9,10 When transient ischemic attack and hemorrhagic stroke are added, the cumulative mortality after mean 11.1 year follow-up was 20%.6 Accordingly, the recurrence rates of stroke in the young have varied from cumulative 5-year rate of 9.4% to cumulative 10-year rate of 24%.11–14 Several factors associated with mortality after ischemic stroke in the young have been identified, including increasing age, active malignancy, heart failure, heavy drinking, large-artery atherosclerosis, living alone, preceding infection, type 1 diabetes mellitus, seizures, and stroke pathogenesis.5,9 However, to the best of our knowledge, no mortality data comparison between patients with ischemic stroke with single or recurrent episodes of stroke exists.
We aimed at investigating the effect of recurrent stroke on long-term mortality, at developing a prediction model of mortality based on stroke characteristics and comorbidities of patients and at comparing mortality rates of 970 young 30-day survivors of first-ever ischemic stroke with general population matched for age, sex, calendar year, and geographical area.
Patients and Methods
The study was conducted at the Department of Neurology, Helsinki University Central Hospital. Our hospital has the only neurological emergency unit for a defined population of 1.5 million. In Finland, virtually all patients with stroke are treated in a hospital regardless of symptom severity. All 1008 consecutive patients aged 15 to 49 years with first-ever ischemic stroke from January 1994 to May 2007 were identified from a prospective computerized hospital discharge database and entered into the Helsinki Young Stroke Registry.15 Ischemic stroke was defined according to the World Health Organization definition as focal neurological deficit persisting for 24 hours with no apparent other than vascular cause.16 Of the 1008 patients, we excluded here patients with a history of hemorrhagic stroke (n=8) and those who died within the first 30 days after the index stroke (n=24). Furthermore, we excluded 4 patients who were judged to be stroke mimics during their later follow-up and 2 patients who were lost to follow-up (eg, moved abroad without registry follow-up data).
Baseline data consisted of diagnostic test results, risk factors, stroke subtype, localization, and pathogenesis, described in detail elsewhere.15 Family history of any stroke was defined as a history of ischemic or hemorrhagic stroke or transient ischemic attack in a first-degree relative. Diabetes mellitus was defined according to the 1999 World Health Organization criteria as fasting glucose ≥7.0 mmol/L (126 mg/dL). Type 1 diabetes mellitus was distinguished from type 2 diabetes mellitus based on insulin dependency within 1 year from diabetes mellitus diagnosis.17 History of transient ischemic attack, coronary heart disease, atrial fibrillation, congestive heart failure, and active malignancy was recorded. Cigarette smoking was defined as a patient smoking ≥1 cigarette per day within 1 year before stroke. Heavy drinking was defined as a patient drinking regularly an estimated amount of >200 g of pure alcohol per week. Patients were considered having hypertension if they were treated with antihypertensive medication or had a history or present diagnosis of hypertension according to the 2003 World Health Organization criteria (systolic blood pressure ≥140 mm Hg and diastolic blood pressure ≥90 mm Hg). Obesity was defined as body mass index of ≥30 kg/m2, or patient clearly stated as heavily obese if body mass index data were not available. Dyslipidemia was defined as on lipid-lowering medication or total cholesterol level ≥5.0 mmol/L (193 mg/dL), low-density lipoprotein level ≥3.0 mmol/L (116 mg/dL), or high-density lipoprotein level <1.0 mmol/L (39 mg/dL). Obstructive sleep apnea was defined as apnea–hypopnea index ≥5 with clinical symptoms.
All patients had undergone brain imaging on admission either with computed tomography or MRI. A neuroradiologist evaluated all images while study neurologists rated the findings. Localization was divided into 1 or multiple territories. Laterality was either right, left, or both. The size of the infarct was determined as the size of the largest ischemic lesion on the basis of documented criteria with the following modifications: small, <1.5 cm lesion in the anterior or posterior circulation; medium, lesion in a cortical superficial branch of the anterior cerebral artery, middle cerebral artery, posterior cerebral artery, or in a deep branch of middle cerebral artery or posterior cerebral artery, or lesion in internal borderzone territories; large anterior, lesion involving complete territory of anterior cerebral artery or middle cerebral artery or lesion involving >1 arterial territory; large posterior, >1.5 cm lesion involving brain stem or cerebellum or involving complete territory of posterior cerebral artery together with borderzone territories. If no visible ischemic lesion was present but symptoms lasted for 24 hours, lesion size was considered small.18
Stroke severity was assessed using the National Institutes of Health Stroke Scale (NIHSS) both as a total score at admission and as divided into mild (0–6), moderate (7–14), and severe (≥15).19 The classification of stroke pathogenesis was done independently by pairs of investigators, reaching consensus when necessary. We used the Trial of Org 10172 in Acute Stroke Treatment criteria, with modifications to better itemize the relationship of specific underlying causes with outcomes in young adults: sources of cardioembolism (Trial of Org 10172 in Acute Stroke Treatment 2) were divided into low-risk and high-risk sources, the former comprising patent foramen ovale with or without atrial septal aneurysm, or sole atrial septal aneurysm, and the latter all other sources.20,21 Other determined pathogenesis (Trial of Org 10172 in Acute Stroke Treatment 4) was divided into 3 large groups as follows: internal carotid artery dissection, vertebral artery dissection, and rare causes other than dissection.
Data on all hospitalizations in inpatient units from January 1994 until December 31, 2011 for all 970 patients included in the study were retrieved from the Care Register for Healthcare maintained by the National Institute for Health and Welfare. This electronic register is obligatory, regulated by law and contains the data of all hospital stays in Finland beginning from the early 1970s.22 The discharge codes used in this study were according to the Finnish version of the International Classification of Diseases (ICD), 9th Revision (ICD-9), 1994 to 1995, and according to the 10th Revision (ICD-10) from 1996 onwards.23–25 We looked for the first hospitalization after the last hospitalization related to the index stroke with any of the discharge diagnoses being ischemic or hemorrhagic stroke or cerebral venous thrombosis. Stroke codes taken into account according to ICD-9 were 430 to 431, 4330A, 4331A, 4339A, 4340A, 4341A, 4349A, 436, and 437 and according to ICD-10 were I60-I61, I63-I64, I67-I68, and G46. All except 2 recurrent stroke hospital stays were verified from the original patient records by 2 investigators (K.A. and E.H.) reaching a consensus with a third investigator (J.P.) when necessary. The 2 nonverified patients were treated in other university hospitals of Finland and were regarded as valid end point events. Mortality data were obtained from Statistics Finland.26,27 Fatal strokes that did not lead to hospitalization were identified based on the mortality registry data (n=5).
A recurrent stroke was defined as a rapid onset of a new persistent neurological deficit attributed to an obstruction in cerebral blood flow and intracerebral hemorrhage with no apparent nonvascular cause. Ischemic stroke with hemorrhagic transformation was classified as ischemic stroke. Traumatic intracranial hemorrhages were not considered strokes. Subarachnoid hemorrhage with late ischemia in the brain parenchyma was classified as subarachnoid hemorrhage.
All personal identifiers were removed and replaced with a code in the final compiled data set. The protocol was approved by the local ethics committee, Dnro 73/13/03/00/11 April 12, 2012, and by the National Institute of Health and Welfare, Dnro THL/956/5.05.00/2012 November 12, 2012.
Statistical analyses used SPSS 21.0 for Macintosh (SPSS Inc, IBM, Armonk, USA). Two-sided values of P<0.05 were considered significant. We compared the cumulative mortality risks between clinical subgroups with Kaplan–Meier analysis. Actual annual risks with 95% confidence intervals (CI) were calculated with the Life Tables function using the formula 1−[(1−Ic)1/n], with Ic as cumulative incidence rate and n as the number of years. We used the Cox proportional hazards model for univariate and multivariable analyses to identify factors associated with mortality. Proportionality assumptions for the analysis were assessed and met. A multivariable model using backward stepwise method was constructed by selecting significant (P<0.05) variables from the univariate analysis. Recurrent stroke was entered as a time-dependent covariate in the Cox proportional analysis, meaning adjusted for time from the index stroke until recurrent stroke.
To compare the mortality rates of first-ever patients with ischemic stroke with those in the general population, standardized mortality ratios (SMRs) were calculated as the ratio of observed number of deaths to the expected number of deaths, and CI was calculated according to the Poisson distribution. The expected number of deaths was calculated with data of the general Finnish population matched with calendar year, age, sex, and geographical area. We calculated the absolute excess number of deaths as the difference between the observed and expected deaths, divided by the number of person-years at risk and expressed per 1000 person-years. Person-years at risk were calculated from the date of the index stroke until death or end of follow-up.
Table 1 shows the baseline characteristics of 970 consecutive patients aged 15 to 49 years who survived ≥30 days after the index stroke. The median NIHSS count on admission for the 24 patients who died within the first 30 days after the index stroke was 17.5 (interquartile range, 12.3–20.0) and 17 (70.8%) of these patients experienced severe strokes (NIHSS >14). Of the survivors, 152 died (15.7%) and 132 (13.6%) had a recurrent ischemic (n=117) or hemorrhagic (n=13 intracerebral hemorrhages, n=2 subarachnoid hemorrhages) stroke during our mean follow-up time of 10.2 (±4.3) years from the index stroke until death or the end of follow-up. Eleven of all recurrent patients with stroke died within 30 days after recurrency. No cerebral venous thrombosis occurred.
The median time from the first ischemic stroke to the first recurrent stroke was 3.7 years (interquartile range, 1.2–7.7). Of the patients who died, 15 (9.9%) died of ischemic stroke, 8 (5.3%) of hemorrhagic stroke, 45 (29.6%) of cardioaortic cause, 29 (19.1%) of malignancy, 10 (6.6%) of infection, and 45 (29.6%) of other causes.
Among the 30-day survivors, women and those <40 years of age had markedly increased survival rate compared with men and patients aged ≥40 years (Figure 1). Figure 2 shows the best prognosis in patients having internal carotid artery dissection and the worst prognosis in patients with large-artery atherosclerosis. Figure 3 shows the excess percentage of people surviving after first-ever ischemic stroke if no recurrency occurred during the follow-up.
The average annual rate of mortality for all study subjects at 16 years from the index stroke was 1.6%, being highest for the patients with large-artery atherosclerosis (5.6%). Accordingly, the cumulative risk of death in all patients was 23.0% (95% CI, 19.1%–26.9%) and that of patients with large-artery atherosclerosis 60.0% (95% CI, 42.4%–77.6%; Table I in the online-only Data Supplement).
In the univariate analysis, male sex, age >40 years, active malignancy, cigarette smoking, coronary heart disease, type 1 and type 2 diabetes mellitus, heart failure, heavy drinking, hypertension, stroke located anteriorly or in both anterior and posterior region, stroke in multiple territories, and increasing NIHSS score were associated with higher mortality. About stroke pathogenesis, large-artery atherosclerosis, high-risk source of cardioembolism, small-vessel occlusion, rare causes other than dissection, and undetermined pathogenesis were associated with significantly higher mortality compared with the ones with the lowest cumulative mortality (ie, the internal carotid artery dissection group). Furthermore, recurrent stroke was associated with higher risk of death. After adjusting for sex and age, patients with malignancy, as also those with high-risk source of cardioembolism, rare causes other than dissection, and large-artery atherosclerosis as the pathogenesis of the index stroke exhibited all hazard ratios nearly or >10 with respect to long-term risk of death. However, the statistically strongest association with long-term risk of death appeared with recurrent stroke (hazard ratio, 13.90; 95% CI, 1.94–99.56; P=0.009). Of note, also those with stroke because of small-vessel occlusion and undetermined pathogenesis had higher age- and sex-adjusted mortality risks compared with internal carotid artery dissection patients (Table II in the online-only Data Supplement).
In the multivariable Cox regression model, increasing age, active malignancy, type 1 diabetes mellitus, heart failure, heavy drinking, increasing NIHSS score, large-artery atherosclerosis, high-risk sources of cardioembolism, small-vessel occlusion, rare causes other than dissection, undetermined pathogenesis, and recurrent stroke all were independently associated with higher risk of death during the 17-year follow-up. Recurrent stroke was statistically the most important factor associated with this risk (hazard ratio, 16.68; 95% CI, 2.33–119.56; P=0.005; Table 2).
Compared with the general population, young patients with first-ever ischemic stroke had SMR of 6.94 (95% CI, 5.84–8.04). About demographic groups, the absolute excess risk of mortality was higher among patients aged 40 to 49 years in both sexes and higher in men across both age groups (Table 3). Both men and women having recurrent stroke had higher SMR and absolute excess risk compared with those without recurrent events (Table III in the online-only Data Supplement). With respect to pathogenetic subgroups, patients with large-artery atherosclerosis exhibited the highest SMR and excess risk, whereas those with internal carotid artery dissection or vertebral artery dissection depicted the lowest figures compared with the general population (Table IV in the online-only Data Supplement).
Our results show unexpectedly high cumulative mortality rates during the 17-year follow-up in young patients with ischemic stroke who survived 30 days after the index stroke compared with the general population. Having a recurrent stroke more than doubled the cumulative mortality rate and raised the hazard of death in such patients to ≈17× higher than in those with only a single episode of stroke after adjusting for demographics, risk factors, and stroke characteristics. Even more importantly, patients who had a stroke recurrence were imposed to ≈30× absolute excess risk of death per 1000 person-years compared with the age- and sex-matched general population.
In our population, SMR for all patients compared with the sex- and age-matched Finnish general population was ≈7-fold. This number is higher than the SMR of 3.9 reported for ischemic stroke survivors by a recent Dutch study involving patients aged 18 to 50 years.6 The reasons for this difference are unclear but may be explained by differences among demographic features, follow-up methodology, or some selection in the patient cohort. However, the study design in that study closely resembles that of ours, and the slight differences cannot solely explain the ≈2-fold difference. In another recent registry-based study from Sweden, the SMR for all patients with ischemic stroke aged 18 to 54 years was 8.15, which is in line with our figures considering the different age cuf off.7 The average annual mortality rate in our study (1.6%) was in accordance with previous studies (1.4%–1.6%).4,28
In our study, factors associated with mortality during the follow-up are similar to previous ones because our analysis showed that increasing age, active malignancy, type 1 diabetes mellitus, heart failure, heavy drinking, stroke in multiple territories, and increasing NIHSS score are associated with long-term mortality at young age.9,10,29 About stroke pathogenesis, those with stroke because of large-artery atherosclerosis experienced a strikingly high cumulative 17-year mortality rate (60%) and the highest SMR (17.08) and absolute excess risk (38.93 per 1000 person-years). Also their multivariable hazard (13.58 compared with internal carotid artery dissection patients) was among the highest. These figures are in accordance with prior reports9,11,30,31 and highlight the importance of aggressive preventive measures in this subgroup. However, rare causes other than dissection were attributable to nearly comparably high-risk estimates. Those with small-vessel occlusion and undetermined pathogenesis also exhibited relatively high mortality risk figures, groups traditionally considered having more benign prognosis.30 Dissection is typically related to good prognosis,32 a scenario complemented by the long-term follow-up of our young patients with stroke. The risk for death among young patients with atrial fibrillation has been considered low,33 but our study shows that the risk for death is much higher among stroke patients with atrial fibrillation than among the normal population matched with age, sex, and geographical area. Our estimates of the observed and expected mortality among smaller Trial of Org 10172 in Acute Stroke Treatment subgroups may be instable, but our findings suggest that in all pathogenetic subgroups except low-risk source of cardioembolism and internal carotid artery dissection groups, the observed mortality exceeds the expected mortality in the general population. This holds true, in particular, for large-artery atherosclerosis group, on contrary to recent observations.6
The strengths of our study include a nearly population-based design, large consecutive exceptionally well-characterized patient series, long follow-up period, the use of data from the Finnish Care Register for Healthcare, 100% complete mortality data, and estimates of the excess risk of death compared with general population. The role of age on death in our study is low because of the young age group of our patients.
Limitations of our study include that recurrent stroke data relies only on hospital admissions in Finland. This is a rather theoretical bias because patients with stroke in general, and particularly young and middle-aged patients with stroke, are almost exclusively treated as inpatients and those who are initially treated abroad are invariably transported to their own hospitals. According to a study by Tolonen et al,34 only 0.6% of first ischemic stroke diagnoses, and 1.8% of all first either ischemic or hemorrhagic stroke diagnoses were made in outpatient clinics in Finland.34 Furthermore, false registry information may occur because data used in this study to catch up recurrent events have been assessed to be accurate in 75% to 99% of cases.22,27 However, discharge registry data are more accurate in a long-term follow-up.35 This accuracy problem was largely overcome by verifying the outcome events from the actual medical records in 125 of 132 (94.7%) events. In addition, because only 18% of patients were followed for >15 years, their results should be interpreted with caution compared with patients with shorter follow-ups. Including only those patients who survived >30 days might have an impact on the final results because this underestimates the proportion of severe strokes, but this results in a better understanding of the long-term survival rather than the in-hospital mortality or case fatality.
Because more and more strokes occur in children and young and middle-aged adults in low-income and middle-income countries around the world, it is important to surmount the burden of this disease.36 In conclusion, the unexpectedly high long-term mortality rates compared with the general population and the striking impact of recurrent stroke should lead to improving risk factor control and both primary and secondary prevention strategies of stroke occurring at young age.
We thank the National Institute for Health and Welfare and Statistics Finland, and their collaborators for guidance and computing.
Sources of Funding
The study was funded by the Helsinki University Central Hospital Research Fund (TYH2011212). The investigators are supported by the Sigrid Juselius Foundation (Dr Tatlisumak), Academy of Finland (Dr Tatlisumak), Helsinki University Central Hospital Research Fund (Drs Tatlisumak, Putaala, and Kaste), the Finnish Medical Foundation (Dr Putaala), and Maire Taponen Foundation (Dr Tatlisumak).
Guest Editor for this article was Bruce Ovbiagele, MD, MSc, MAS.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.005648/-/DC1.
- Received March 31, 2014.
- Revision received June 24, 2014.
- Accepted July 7, 2014.
- © 2014 American Heart Association, Inc.
- Sultan S,
- Elkind MSV
- Giang KW,
- Björck L,
- Nielsen S,
- Novak M,
- Sandström TZ,
- Jern C,
- et al
- Marini C,
- Totaro R,
- Carolei A
- Putaala J,
- Curtze S,
- Hiltunen S,
- Tolppanen H,
- Kaste M,
- Tatlisumak T
- Kappelle LJ,
- Adams HP Jr.,
- Heffner ML,
- Torner JC,
- Gomez F,
- Biller J
- Putaala J,
- Metso AJ,
- Metso TM,
- Konkola N,
- Kraemer Y,
- Haapaniemi E,
- et al
- Paciaroni M,
- Agnelli G,
- Corea F,
- Ageno W,
- Alberti A,
- Lanari A,
- et al
- Brott T,
- Adams HP Jr.,
- Olinger CP,
- Marler JR,
- Barsan WG,
- Biller J,
- et al
- Adams HP Jr.,
- Bendixen BH,
- Kappelle LJ,
- Biller J,
- Love BB,
- Gordon DL,
- et al
- Sund R
- 23.↵Lääkintöhallitus. Tautiluokitus 1987, Part 1–2. Helsinki, Finland: Valtion Painatuskeskus; 1986.
- 24.↵World Health Organization. Manual of the International Statistical Classification of Diseases, Injuries, and Causes of Death, Ninth Revision. Volume 1. Geneva, Switzerland: World Health Organization; 1977.
- 25.↵World Health Organization. International Statistical Classification of Diseases and Related Health Problems. Geneva, Switzerland: World Health Organization; 1992.
- 26.↵Official Statistics of Finland (OSF): Deaths [e-publication]. ISSN=1798–2545. Helsinki, Finland: Statistics Finland. http://www.tilastokeskus.fi/til/kuol/tau_en.html. Accessed November 19, 2013.
- Heikinheimo T,
- Broman J,
- Haapaniemi E,
- Kaste M,
- Tatlisumak T,
- Putaala J
- Melgaard L,
- Rasmussen LH,
- Skjoth F,
- Lip G,
- Larsen T
- Feigin VL,
- Forouzanfar MH,
- Krishnamurthi R,
- Mensah GA,
- Connor M,
- Bennett DA,
- et al