Predictive Variables for Mortality After Acute Ischemic Stroke
Background and Purpose— Stroke is a major healthcare issue worldwide with an incidence comparable to coronary events, highlighting the importance of understanding risk factors for stroke and subsequent mortality.
Methods— In the present study, we determined long-term (all-cause) mortality in 545 patients with ischemic stroke compared with a cohort of 330 age-matched healthy control subjects followed up for a median of 7.4 years. We assessed the effect of selected demographic, clinical, biochemical, hematologic, and hemostatic factors on mortality in patients with ischemic stroke. Stroke subtype was classified according to the Oxfordshire Community Stroke Project criteria. Patients who died 30 days or less after the acute event (n=32) were excluded from analyses because this outcome is considered to be directly attributable to the acute event.
Results— Patients with ischemic stroke were at more than 3-fold increased risk of death compared with the age-matched control cohort. In multivariate analyses, age, stroke subtype, atrial fibrillation, and previous stroke/transient ischemic attack were predictive of mortality in patients with ischemic stroke. Albumin and creatinine and the hemostatic factors von Willebrand factor and β-thromboglobulin were also predictive of mortality in patients with ischemic stroke after accounting for demographic and clinical variables.
Conclusions— The results indicate that subjects with acute ischemic stroke are at increased risk of all-cause mortality. Advancing age, large-vessel stroke, atrial fibrillation, and previous stroke/transient ischemic attack predict mortality; and analysis of albumin, creatinine, von Willebrand factor, and β-thromboglobulin will aid in the identification of patients at increased risk of death after stroke.
Stoke is a major healthcare issue worldwide representing the third most common cause of death in the United Kingdom.1 The Oxford Vascular Study reported the incidence of cerebrovascular events was 1.2-fold higher than coronary events.2 Approximately 50% of cerebrovascular events were in those aged under 75 years despite previous indications that stroke was more a disease of the elderly population.2 Although stroke incidence and subsequent mortality have declined in recent years attributable to improved risk factor management,3 stroke can be considered to represent a greater healthcare burden than acute coronary disease resulting from residual disability. A greater understanding of the underlying causes of stroke and subsequent mortality will be required to establish appropriate prevention and treatment strategies.
Prospective studies have identified risk factors for stroke and stroke mortality, including advancing age, hypertension, diabetes, smoking, and atrial fibrillation (AF).2,4–7 Acute thrombosis is central to the pathogenesis of ischemic stroke and elevated levels of hemostatic factors, including fibrinogen and tissue plasminogen activator (tPA), predict acute stroke.8,9 Case–control studies, including the present cohort, also identified elevated levels of hemostatic factors, including fibrinogen, von Willebrand Factor (vWF), and β-thromboglobulin (βTG), in subjects with acute stroke.10–12 The relationship between biochemical, hematologic, and hemostatic factors and poststroke mortality, however, has not been extensively studied.
The aims of this study were to (1) determine all-cause mortality in a hospital-based cohort with ischemic stroke compared with an age-matched healthy cohort, (2) identify demographic and clinical variables predictive for mortality after ischemic stroke, (3) identify routine biochemical and hematologic factors that add to the predictive value of demographic and clinical variables, and (4) identify hemostatic factors predictive for mortality.
Materials and Methods
Consecutive white European patients (n=609) with a clinical diagnosis of acute stroke, defined by World Health Organization clinical criteria, were recruited from 4 hospitals in Leeds as previously described.10,13 Hemorrhagic or ischemic stroke was confirmed by noncontrast cranial CT scan. Ischemic stroke was subclassified according to the Oxfordshire Community Stroke Project criteria into total and partial anterior circulation infarction (TACI and PACI), posterior circulation syndrome, and lacunar infarction (LACI) as previously described10,13; subjects with equivocal Oxfordshire Community Stroke Project classification (n=7) were excluded from analysis. A total of 545 patients with ischemic stroke (274 males, 271 females) were included in the present analysis.
Subjects were classified according to smoking history as current, ex-, or nonsmokers. Medical history was obtained (by A.J.C.), including history of previous stroke or transient ischemic attack (TIA), ischemic heart disease (IHD), and peripheral vascular disease (PVD).10,13 AF was established on clinical examination at the time of acute stroke and confirmed on 12-lead electrocardiogram.14 Presence of hypertension and diabetes were determined from case notes and by current use of antihypertensive and antihyperglycemic agents.
Age-matched healthy white European control subjects (139 males, 191 females) with no personal or family history of atherothrombotic disease, recruited from local Family Health Services Authority general practice registers as previously described,10 were followed up as a reference cohort. Patients and control subjects were flagged with the Office for National Statistics for notification of death as previously described.13 All subjects gave informed consent according to a protocol approved by the Leeds Teaching Hospitals Research Ethics Committee as previously described.10,13
A venous blood sample was taken for analysis of biochemical (albumin, creatinine, cholesterol, triglycerides, glucose, bilirubin, uric acid, total protein), hematologic (hemoglobin, platelet count, white cell count), and hemostatic factors (fibrinogen, factor [F]VII, FVIII, FXIII A and B subunit and activity, vWF, tPA, plasminogen activator inhibitor-1 [PAI-1], and βTG). Biochemical and hematologic analyses were carried out using standard hospital laboratory protocols, and hemostatic factors were analyzed by clotting assays (fibrinogen, FVII, FVIII) or enzyme-linked immunosorbent assay (FXIII, vWF, tPA, PAI-1, and βTG) as previously described.10–13,15–17 The first 115 patients recruited did not have a sample taken for analysis of hemostatic factors and in a further 15 patients, insufficient blood was obtained for these analyses. There were no differences in demographic, biochemical, or hematologic factors or mortality between those subjects with and without hemostasis samples except for age (no sample: 75 [range, 66 to 81] years; with sample: 72 [range, 65 to 79] years, P=0.03) and platelet count (no sample: 237 [range, 227 to 251]×109L−1; with sample: 256 [range, 247 to 264]×109L−1, P=0.04). Samples for analysis of βTG were only taken in 275 patients and 278 control subjects in whom clean venipuncture was performed on the first attempt with minimal venous stasis as previously described.12
The outcome measure was all-cause long-term mortality after acute ischemic stroke. Mortality in control subjects was monitored as a reference. Surviving patients and control subjects were censored on January 19, 2002. Univariate associations between demographic, clinical, biochemical, hematologic, and hemostatic factors and mortality were assessed using Kaplan–Meier survival analysis and significance determined using the log rank test with continuous variables analyzed as quartiles. Hazard ratios for mortality were determined by univariate and multivariate Cox proportional hazards regression analyses with data presented as hazard ratio (compared with the lowest quartile) with 95% CIs. Log minus log plots were evaluated to test the validity of the proportionality of hazards assumption over time; all variables met this assumption. Initial analyses identified the demographic and clinical variables that independently predicted poststroke mortality. Significant demographic and clinical variables were entered into subsequent multivariate Cox regression analyses with biochemical, hematologic, and hemostatic factors included in the models using forward stepwise selection with a probability value of 0.05 required for entry into the model. A probability value of 0.05 or less was considered significant. All statistical analyses were performed using SPSS for Windows v12.
Subjects were followed up for a median of 7.4 years; minimum and maximum follow-up periods for survivors were 5.3 years and 8.5 years, respectively. During follow up, 343 patients (63%) and 73 control subjects (22%) died with median times to death of 1.6 and 3.0 years, respectively.
Baseline Patient and Control Subject Characteristics and Mortality Rates
The characteristics of patients with ischemic stroke who survived compared with those who died 30 days or less and those dying more than 30 days are presented in Table 1. Survivors were younger than those who died. The proportion of exsmokers was higher in survivors than in those who died; there was no difference in time since cessation of smoking or pack-years in exsmokers who survived compared with those who died (data not shown). The proportion of subjects with LACI was highest in survivors, whereas the proportion of subjects with TACI was highest in those who died. A greater proportion of subjects with AF, previous stroke, IHD, PVD, and aspirin use at the time of stroke died during follow up. Compared with patients dying after more than 30 days, a greater proportion of patients dying 30 days or less had experienced a TACI and were in AF (Table 1). Subjects dying 30 days or less after the acute event were excluded from subsequent analyses.
The characteristics of the control subjects who died during follow up and survivors are presented in supplemental Table I (available online at http://stroke.ahajournals.org). Those who died were significantly older than survivors. There was no significant difference in gender, smoking, or other clinical variables between those who subsequently died and survivors.
Kaplan–Meier survival curves comparing age-matched patients and control subjects analyzed by gender are presented in Figure 1. Average annual mortality rates in male and female patients were 8.0% and 8.4% compared with 2.6% and 3.2% in male and female controls, respectively. The hazard ratios (95% CI) for mortality in male patients compared with male controls was 3.59 (2.38 to 5.42) and for female patients compared with female controls was 3.14 (2.26 to 4.38).
Demographic and Clinical Variables Predictive for Long-Term Poststroke Mortality
In univariate analyses, in patients more than 30-day all-cause mortality was significantly associated with age and stroke subtype (Figure 2A, B). Previous stroke/TIA, AF, IHD, PVD, aspirin therapy at the time of stroke, and being a nonsmoker were also significantly associated with mortality (Table 2). There were no associations between mortality after acute stoke and gender, diabetes, warfarin treatment at the time of acute stroke, hypertension, or admission blood pressure (data not shown). In a multivariate Cox regression analysis, including age, smoking, stroke subtype, previous stroke/TIA, IHD, PVD, aspirin therapy, and AF, independent predictors of poststroke mortality were age, stroke subtype, previous stroke/TIA, and AF (Table 2).
Biochemical and Hematologic Factors Predictive for Poststroke Mortality
In univariate analyses, lower concentrations of albumin and hemoglobin and higher concentrations of creatinine at the time of stroke were associated with higher mortality (Table 3). There were no significant associations between plasma glucose, cholesterol, triglycerides, platelet count, white cell count, bilirubin, plasma uric acid, or total protein and poststroke mortality (data not shown). In multivariate Cox regression analyses, after accounting for age, stroke subtype, previous stroke/TIA, and AF, only albumin and creatinine remained associated with mortality (Table 3). Kaplan–Meier survival curves for albumin and creatinine are presented in Figure 2C and D. Inclusion of current and former smoking in the model did not influence the association between albumin and stroke mortality, and no significant interaction between smoking and albumin was identified (data not shown).
Hemostatic Factors Predictive for Poststroke Mortality
In univariate analyses, elevated levels of fibrinogen, FVIII, βTG, vWF, and tPA and lower levels of FXIII A subunit were each associated with higher poststroke mortality (Table 4). There were no significant associations between PAI-1, FVII, FXIII B subunit or FXIII activity and mortality (data not shown). In multivariate Cox regression analyses, after accounting for age, stroke subtype, previous stroke/TIA, and AF, only βTG and vWF were associated with mortality (Table 4). Kaplan–Meier survival curves for βTG and vWF are presented in Figure 2E and F.
Patients with ischemic stroke who died more than 30 days after the acute event were at more than 3-fold increased risk of death after the acute event compared with an age-matched healthy cohort, which is in keeping with the results of large population-based studies.5,19 We excluded patients who died 30 days or less after the acute event because death is largely attributable to the acute event itself reflected by the large proportion of patients with TACI in those dying within 30 days in the present study. In contrast, longer-term mortality is generally attributable to vascular disease, in particular cardiovascular disease.5,18
Demographic and Clinical Variables Predictive for Mortality After Ischemic Stroke
In keeping with results of population studies,5,19,20 mortality increased with age and was highest (80%) in subjects aged 75 years or older. Notably, half of those subjects aged less than 75 years also died during follow up. The high mortality rate, irrespective of age, highlights the importance of research into the underlying mechanisms predisposing to stroke and stroke mortality.
We used the Oxfordshire Community Stroke Project classification of stroke21 to determine the association of stroke subtype with mortality and found an approximately 2-fold overall increased risk of mortality for those with larger infarcts compared with those with LACI (1.6-fold for PACI and 2.6-fold for TACI). These results are in keeping with previous studies, summarized in a recent meta-analysis,22 indicating that our hospital-based cohort is representative of larger population-based studies. In keeping with other studies,18 we also identified history of stroke/TIA as predicting mortality, independent of stroke subtype, indicating the need for improved management of vascular risk to prevent secondary stroke and subsequent cardiovascular mortality. Univariate analysis indicated a paradoxic association of aspirin use at the time of stroke with subsequent mortality. Aspirin use is likely to reflect a history of atherothrombotic disease supported by the fact that aspirin use was not independently predictive of mortality in multivariate analyses, including history of atherothrombotic disease.
Diabetes, smoking, and hypertension were not independently associated with long-term mortality in contrast to results of population-based studies.6,20,23 A possible explanation for this discrepancy may be that population-based studies include those who died before hospitalization, a cohort that may be overrepresented by diabetes, hypertension, and smoking. Furthermore, we cannot exclude that diabetes and hypertension were undiagnosed at the time of recruitment, because presence of these risk factors was determined with reference to case notes and current use of antihypertensive and antihyperglycemic agents. Nor can we exclude misreporting of smoking status, which may be a source of error in our analyses. It must also be recognized that the present hospital-based cohort is relatively small in comparison with population-based studies, which may contribute to the observed discrepancies.
AF is associated with an approximately 5-fold increased risk for the development of stroke.24,25 Stroke in subjects in AF is generally associated with embolization of thrombus formed within the left atrium.26 Previous studies showed that AF predicted poor outcome at 1 month and 1 year in comparison with subjects in sinus rhythm,24,27 with mortality rates of 50% at 1 year. In keeping with previous studies,20,25 AF was a significant predictor of mortality in the present study, independent of stroke subtype and other measured demographic and clinical factors. Only 16% of patients in AF survived during follow up, and only 34 subjects in AF (33%) were taking antithrombotic treatments at the time of acute stroke.
Biochemical and Hematologic Variables Predictive for Mortality After Ischemic Stroke
Of the routine factors analyzed, only albumin and creatinine were associated with poststroke mortality after accounting for demographic and clinical variables. Low albumin concentrations predicted mortality, which is consistent with the findings of previous studies.28,29 It has been reported that low albumin concentrations only predict mortality in smokers,28 and it has been suggested that smokers with decreased albumin are particularly sensitive to smoking-induced inflammation and presumably, therefore, also its consequences.28,30 In contrast to these studies, we found that low albumin (<38 g/L) was predictive for mortality irrespective of smoking history. Decreased albumin levels may be associated with poor nutritional status29; however, total protein was not associated with subsequent mortality in the present study. Albumin levels could be decreased attributable to an acute inflammatory response to stroke or ongoing chronic inflammation; however, the association between decreased albumin and mortality was independent of stroke subtype and white blood cell count was not associated with stroke mortality, suggesting that these are unlikely to explain the observed association. Elevated creatinine at the time of acute stroke also predicted long-term mortality. These data suggest that patients with increased creatinine and/or decreased albumin may have impaired renal function, possibly attributable to advancing age and hypertension. Such patients should be investigated further to determine underlying causes of these biochemical disturbances to facilitate appropriate management.
Hemostatic Factors Predictive for Mortality After Ischemic Stroke
Fibrin formation and platelet activation occur in the acute phase of stroke.31 We previously reported elevated levels of fibrinogen, FVIII, vWF, PAI-1, and βTG in subjects with acute stroke.10–13 Of the hemostatic factors analyzed, only vWF and βTG were independently associated with poststroke mortality during follow up. vWF is stored in endothelial cell Weibel-Pallade bodies and platelet α-granules and is released in response to endothelial cell and platelet activation. vWF plays an essential role in platelet adhesion to subendothelial matrix, exposed at sites of endothelial disruption, promoting thrombus formation. βTG is released on platelet activation and is considered a specific marker of ongoing platelet activation. Results from the present study indicate that vWF and βTG are predictive for long-term mortality after the acute event independent of stroke subtype and other predictors of mortality and are likely to reflect both endothelial disruption and platelet activation.
Limitations of the Study
We analyzed only data collected at the time of presentation with stroke in this study to determine the predictive value of variables that could readily be obtained after stroke. The information on treatment at discharge was not obtained, and consequently, we may have underestimated the influence of variables on stroke mortality. We only analyzed total mortality in this study because of the recognized inconsistencies in cause of death reporting by death certificate.
The results of this study indicate that, in addition to well-recognized risk factors for poststroke mortality, including advancing age, more severe strokes, AF, and previous stroke/TIA, low albumin and elevated creatinine predict mortality after stroke after accounting for other factors, which may indicate impaired renal function in these patients. Furthermore, we have shown that the hemostatic factors vWF and βTG predict poststroke mortality after accounting for other factors, likely indicating ongoing vascular damage and platelet activation. We acknowledge that βTG requires particularly stringent sample handling and is therefore unlikely to be adopted as a routine hospital measurement. However, vWF antigen measurement is straightforward and could be considered as a routine assessment when evaluating patients with acute stroke to identify patients in whom ongoing vascular damage and platelet activation is likely to be occurring. These subjects may reflect a subgroup of patients at particular risk for subsequent thrombotic events in whom combined antiplatelet treatment may be most beneficial. The results of the present study indicate that in subjects with acute ischemic stroke, analysis of albumin, creatinine, vWF, and βTG will aid in the identification of patients who are at increased risk of death after stroke.
Sources of Funding
This study was funded by the Stroke Association, UK.
- Received October 20, 2006.
- Revision received December 21, 2006.
- Accepted January 9, 2007.
Wolfe CD. The impact of stroke. Br Med Bull. 2000; 56: 275–286.
Rothwell PM, Coull AJ, Silver LE, Fairhead JF, Giles MF, Lovelock CE, Redgrave JN, Bull LM, Welch SJ, Cuthbertson FC, Binney LE, Gutnikov SA, Anslow P, Banning AP, Mant D, Mehta Z. Population-based study of event-rate, incidence, case fatality, and mortality for all acute vascular events in all arterial territories (Oxford Vascular Study). Lancet. 2005; 366: 1773–1783.
Rothwell PM, Coull AJ, Giles MF, Howard SC, Silver LE, Bull LM, Gutnikov SA, Edwards P, Mant D, Sackley CM, Farmer A, Sandercock PA, Dennis MS, Warlow CP, Bamford JM, Anslow P. Change in stroke incidence, mortality, case-fatality, severity, and risk factors in Oxfordshire, UK from 1981 to 2004 (Oxford Vascular Study). Lancet. 2004; 363: 1925–1933.
Wolf PA, D’Agostino RB, Belanger AJ, Kannel WB. Probability of stroke: a risk profile from the Framingham Study. Stroke. 1991; 22: 312–318.
Dennis MS, Burn JP, Sandercock PA, Bamford JM, Wade DT, Warlow CP. Long-term survival after first-ever stroke: the Oxfordshire Community Stroke Project. Stroke. 1993; 24: 796–800.
Hu G, Sarti C, Jousilahti P, Peltonen M, Qiao Q, Antikainen R, Tuomilehto J. The impact of history of hypertension and type 2 diabetes at baseline on the incidence of stroke and stroke mortality. Stroke. 2005; 36: 2538–2543.
Menotti A, Jacobs DR Jr, Blackburn H, Kromhout D, Nissinen A, Nedeljkovic S, Buzina R, Mohacek I, Seccareccia F, Giampaoli S, Dontas A, Aravanis C, Toshima H. Twenty-five-year prediction of stroke deaths in the seven countries study: the role of blood pressure and its changes. Stroke. 1996; 27: 381–387.
Carter AM, Catto AJ, Bamford JM, Grant PJ. Gender specific associations of the fibrinogen Bβ 448 polymorphism, fibrinogen levels and acute cerebrovascular disease. Arterioscler Thromb Vasc Biol. 1997; 17: 589–594.
Carter AM, Catto AJ, Bamford JM, Grant PJ. Platelet GPIIIa PlA and GPIb variable number tandem repeat (VNTR) polymorphisms and markers of platelet activation in acute stroke. Arterioscler Thromb Vasc Biol. 1998; 18: 1124–1131.
Carter AM, Catto AJ, Grant PJ. Association of the alpha-fibrinogen Thr312Ala polymorphism with poststroke mortality in subjects with atrial fibrillation. Circulation. 1999; 99: 2423–2426.
Heywood DM, Carter AM, Catto AJ, Bamford JM, Grant PJ. Polymorphisms of the factor VII gene and circulating FVII:C levels in relation to acute cerebrovascular disease and poststroke mortality. Stroke. 1997; 28: 816–821.
Hankey GJ, Jamrozik K, Broadhurst RJ, Forbes S, Burvill PW, Anderson CS, Stewart-Wynne EG. Long-term risk of first recurrent stroke in the Perth Community Stroke Study. Stroke. 1998; 29: 2491–2500.
Hardie K, Hankey GJ, Jamrozik K, Broadhurst RJ, Anderson C. Ten-year survival after first-ever stroke in the Perth Community Stroke Study. Stroke. 2003; 34: 1842–1846.
Vernino S, Brown RD Jr, Sejvar JJ, Sicks JD, Petty GW, O’Fallon WM. Cause-specific mortality after first cerebral infarction: a population-based study. Stroke. 2003; 34: 1828–1832.
Jackson C, Sudlow C. Comparing risks of death and recurrent vascular events between lacunar and non-lacunar infarction. Brain. 2005; 128: 2507–2517.
Hart CL, Hole DJ, Smith GD. Risk factors and 20-year stroke mortality in men and women in the Renfrew/Paisley study in Scotland. Stroke. 1999; 30: 1999–2007.
Feinberg WM, Pearce LA, Hart RG, Cushman M, Cornell ES, Lip GY, Bovill EG. Markers of thrombin and platelet activity in patients with atrial fibrillation: correlation with stroke among 1531 participants in the stroke prevention in atrial fibrillation III study. Stroke. 1999; 30: 2547–2553.
Marini C, De Santis F, Sacco S, Russo T, Olivieri L, Totaro R, Carolei A. Contribution of atrial fibrillation to incidence and outcome of ischemic stroke: results from a population-based study. Stroke. 2005; 36: 1115–1119.
Heppell RM, Berkin KE, McLenachan JM, Davies JA. Haemostatic and haemodynamic abnormalities associated with left atrial thrombosis in non-rheumatic atrial fibrillation. Heart. 1997; 77: 407–411.
Lin HJ, Wolf PA, Kelly-Hayes M, Beiser AS, Kase CS, Benjamin EJ, D’Agostino RB. Stroke severity in atrial fibrillation. The Framingham Study. Stroke. 1996; 27: 1760–1764.
Davis JP, Wong AA, Schluter PJ, Henderson RD, O’Sullivan JD, Read SJ. Impact of premorbid undernutrition on outcome in stroke patients. Stroke. 2004; 35: 1930–1934.
Nelson JJ, Liao D, Sharrett AR, Folsom AR, Chambless LE, Shahar E, Szklo M, Eckfeldt J, Heiss G. Serum albumin level as a predictor of incident coronary heart disease: the Atherosclerosis Risk in Communities (ARIC) study. Am J Epidemiol. 2000; 151: 468–477.
Feinberg WM, Erickson LP, Bruck D, Kittelson J. Hemostatic markers in acute ischemic stroke. Association with stroke type, severity, and outcome. Stroke. 1996; 27: 1296–1300.