Cause of Stroke Recurrence Is Multifactorial
Patterns, Risk Factors, and Outcomes of Stroke Recurrence in the South London Stroke Register
Background and Purpose— This article examines stroke recurrence and whether the subtype of the initial stroke influences the risk and subtypes of further strokes. The proportion of recurrences attributable to conventional risk factors is quantified.
Methods— From January 1995 to August 2000, all first-in-a-lifetime strokes (n=1626) were identified and prospectively followed up in a defined multiethnic inner city population of 234 533. Twelve overlapping referral sources and face-to-face follow-up at 3 months and 1 and 3 years were used to attain complete registration of stroke recurrence. Index and recurrent stroke were classified according to the Oxford Community Stroke Project classification.
Results— In 2744 person-years of follow-up, 153 recurrences were observed. At 5 years, the cumulative risk of first stroke recurrence was 16.6% (95% CI, 13.5 to 20.4), and the combined risk of death or stroke recurrence was 65.3% (95% CI, 61.9 to 68.6). Ethnicity and subtype of index stroke were not associated with stroke recurrence. A change in subtype between index and recurrent stroke occurred in 45.5% (95% CI, 35.8 to 55.2) of cases and was most frequent among index lacunar strokes and primary intracerebral hemorrhages. In multivariable analyses, diabetes mellitus and atrial fibrillation were associated with both stroke recurrence and recurrence-free survival. In the stroke population, 9.1% (95% CI, −2.0 to 20.2) of recurrences were attributable to diabetes and 4.9% (95% CI, −7.3 to 17.2) to atrial fibrillation during the first year after the index stroke.
Conclusions— The cause of stroke recurrence is multifactorial, and the subtypes of index and recurrent strokes are often not identical. Most recurrences remain unexplained by conventional risk factors.
Prognostic studies in stroke medicine focus on the risks of stroke recurrence, disability, and death. Statements on the risk of stroke recurrence are often based on studies that were retrospective or hospital-based and did not consider both ischemic and hemorrhagic strokes together.1–8 Studies without these methodological limitations have given point estimates of 3-year stroke recurrence rates of between 6% and 25% but have been inconclusive as to whether recurrence rates differ between ischemic and hemorrhagic strokes.5,6,8
There has been a debate about whether the subtype of the index stroke predicts the type of subsequent strokes. Two population-based studies found that recurrences were of the same subtype in almost 90% of cases.5,8 However, this high level of agreement could have been due to the crude classification used, which differentiated only between ischemic and hemorrhagic strokes. A retrospective hospital-based study with a more detailed categorization of stroke subtypes suggested that stroke recurrences in lacunar and hemorrhagic index strokes are often of a different type9—hence, the hypothesis of the multifactorial origin of stroke recurrence.10
The relationship between risk factors such as ethnicity, diabetes mellitus, hypertension, and atrial fibrillation (AF) and stroke recurrence has been examined in several studies.1–4,6,8,11 However, few had sufficient sample sizes or used multivariable analysis to control for confounding factors.1,3,6 Furthermore, no estimates have been made of the proportion of stroke recurrences that are attributable to these risk factors.
This article prospectively examines stroke recurrence in a defined multiethnic population to answer the following questions: What is the overall risk of stroke recurrence, and does the risk differ between subtypes of index stroke? Do recurrent strokes tend to be of the same or of a different subtype than the index stroke? What are the risk factors associated with stroke recurrence, and what proportion of stroke recurrences is attributable to modifiable risk factors?
Initial Case Ascertainment
The South London Community Stroke Register was established in 1995 to study the incidence, management, and outcome of stroke in a multiethnic population. This population-based register prospectively collected data on first-in-a-lifetime stroke patients. Twelve overlapping referral sources were used to attain complete notification of strokes. The study area compromised 22 wards of the Lambeth, Southwark and Lewisham Health Authority with a population of 234 533. The methodology has been described in detail elsewhere and is summarized here.12
Stroke was defined according to the World Health Organization (WHO) criteria.13 Stroke diagnosis and initial assessment were made by a study doctor within the first week after the event. Classification of stroke subtypes was based on clinical and radiological (CT/MRI scan) findings. According to the Oxford Community Stroke Project (OCSP), ischemic strokes were differentiated clinically into total anterior circulation infarcts (TACIs), partial anterior circulation infarcts (PACIs), posterior circulation infarcts (POCIs) and lacunar infarcts (LACIs).14 Other subtypes included primary intracerebral hemorrhage (PICH), subarachnoid hemorrhage (SAH), and unclassified (UNC) strokes. The final subtype was decided in conference with the lead stroke physician and additional information from brain imaging.
The subtype-specific rates of stroke recurrence and recurrence-free survival reported in the present study were based on the 1626 cases registered until August 31, 2000.
Recording of Stroke Recurrence
Notification of death was confirmed by the Office for National Statistics. The definition of stroke recurrence was the same as for the index stroke with an additional criterion5: There had to be either a new neurological deficit or a deterioration of the previous deficit not considered to be due to edema, hemorrhagic transformation, or intercurrent illness. Only recurrences 21 days after the index stroke or, if earlier, clearly in another part of the brain were included. Referral sources were also used for notification of stroke recurrences. In addition, face-to-face follow-up visits were carried out at 3 months (n=901), 1 year (n=731), 2 years (n=213; 1995 cohort only), and 3 years (n=320) after the index stroke, and patients were asked whether they had experienced another stroke. Positive responses for cases fulfilling the above definition of stroke recurrence were included, and when patients were rehospitalized after the index stroke, we checked whether a stroke recurrence was the cause. Registration of stroke recurrence was performed in the same way as the index stroke. The subtype of stroke recurrence was determined clinically according to the OCSP classification with consensus among study clinicians.
Detailed information on risk factors has been reported previously.15 Risk factor details were obtained from hospital records, general practitioners, and patient self-report. Demographic factors included age, sex, and ethnicity. Ethnicity was based on census categories.12 Patients of African-Caribbean (n=196) and African (n=86) origin were classified as black; Caucasian (n=1258) and other (n=86) origin was classified as white or other. Potentially modifiable risk factors included history of previous transient ischemic attack (TIA), ischemic heart disease, AF, hypertension, diabetes mellitus, alcohol abuse, and smoking. TIA was defined as a focal brain deficit caused by vascular disease that cleared completely in <24 hours.16 Diabetes mellitus was diagnosed in keeping with the WHO classification.17 Alcohol consumption was classified as heavy if the intake was ≥14 U/wk for women and ≥21 U/wk for men.18 Smoking status was recorded as never or current/ever. Hypertension was defined as a blood pressure reading >160/95 mm Hg (WHO)19 and from patients recalling the prescription of antihypertensive treatment. Ischemic heart disease was defined as a history of angina or myocardial infarction. Diagnosis of AF was made when reported before the index stroke or by the study physician on the basis of the ECG taken during the acute stroke period.20 Because data on risk status were available only for patients registered in 1995 to 1998 (n=1254), patients with a later index stroke were not included in the analyses of risk factors.
The life-table cumulative risks of stroke recurrence and of recurrence-free survival were obtained by the use of unadjusted Kaplan-Meier survival analyses. The censoring date was August 31, 2000. Changes in subtypes between the index and first recurrent stroke were analyzed by cross-tabulation for all strokes. The univariate associations of risk factors with stroke recurrence and with death/stroke recurrence (or non–recurrence-free survival, based on death or stroke recurrence, whichever occurred first) were examined by proportional-hazards analysis. For stroke recurrence, patients were censored if they died of another cause. Underlying hazard functions were examined, and from the change in the log likelihood statistic (Δ−2LL), which is a measure of goodness of fit, the best-fitting model of stroke recurrence was obtained from the Gompertz distribution and that of recurrence-free survival from the Weibull distribution.21 Multivariable modeling used the stepwise method with criteria for entry of Pe=0.05 and for removal of Pr=0.10. Cross-tabulations of outcomes and risk factors 1 year after index stroke were used to compute relative risk ratios and attributable risks. The attributable fraction in the exposed group gives the proportion by which the incidence rate of the outcome among those exposed to a risk factor would be reduced if the exposure were absent; the attributable fraction in the population is the equivalent but refers to the entire stroke population, not only exposed subjects. Hence, the latter gives the proportion by which the incidence rate of the outcome in the entire stroke population would be reduced if exposure were absent.
Rates of Stroke Recurrence and Recurrence-Free Survival
The study comprised 1626 first-ever stroke patients registered over 68 months between January 1, 1995, and August 31, 2000. Of the index strokes, 765 (47.1%) were non-LACI (TACI, PACI, and POCI), 401 (24.7%) were LACI, 222 (13.7%) were PICH, 88 (5.4%) were SAH, and 150 (9.2%) were UNC strokes. The mean±SD age was 71.4±14.2 years; 834 (51.3%) were female.
The median follow-up time was 432 days (interquartile range, 32 to 1090 days). The total analysis time for all patients taken together was 2744 years. During this period, 153 stroke recurrences were observed: 127 patients had 1 stroke recurrence, 24 patients had a second recurrence, and 2 patients had a third. The estimates of the life-table cumulative risks of first stroke recurrence and recurrence-free survival are given in Table 1, and the recurrence-free survival is presented in the Figure. Incidence rates of first stroke recurrence were highest in the first year after index stroke, with a risk of 8.0% (95% CI, 6.5 to 9.8). During the second, third, fourth and fifth years, incidence rates were 3.3% (95% CI, 2.2 to 4.9), 3.5% (95% CI, 2.1 to 5.8), 1.2% (95% CI, 0.4 to 3.7), and 1.8% (95% CI, 0.4 to 7.4).
Annual incidence recurrence rates for infarct index strokes were 7.8% (95% CI, 6.2 to 9.9), 3.1% (95% CI, 1.2 to 4.9), 4.4% (95% CI, 2.7 to 7.3), 1.1% (95% CI, 0.3 to 4.6), and 2.5% (95% CI, 0.6 to 10.0), respectively.
Patterns of First Stroke Recurrence
In 206 (12.7%) index strokes, no CT/MRI scan was done; this rate increased to 35.4% (n=45) in stroke recurrences (P<0.01). Table 2 shows the relationship between the index and first recurrence subtype. Change in subtype between index and first recurrent stroke occurred in 46 patients (45.5%), excluding those with UNC strokes (95% CI, 35.8 to 55.2).
Association Between Subtypes of Index Stroke and Risk of Stroke Recurrence
The univariate and multivariable associations of first stroke recurrence and of death/recurrence with subtype of index stroke are given in Tables 3 and 4⇓. Stroke subtype was differentiated into 5 categories (LACI, non-LACI, PICH, SAH, and UNC) and was significantly related to stroke recurrence in univariate analyses only (P=0.04). Subdivision of nonlacunar infarcts did not lead to a better-fitting model for stroke recurrence (df=2, Δ−2LL=0.13, P=0.94). Death/recurrence was significantly associated with the 5 categories of stroke subtype. Further differentiation of nonlacunar strokes improved the model only up to 3 months after the index stroke; the model including 3-month survivors only was improved by further differentiation (df=2, Δ−2LL=23.75, P<0.0001), whereas that including all patients was not (df=2, Δ−2LL=0.59, P=0.74).
Association Between Risk Factors and Stroke Recurrence
Of the potentially modifiable risk factors, diabetes mellitus and AF reached significance in the multivariable models of both stroke recurrence and death/recurrence (Tables 3 and 4⇑), and smoking was significantly associated with death/recurrence. We found that 197 cases (16.4%) were exposed to diabetes, and 249 (21.1%) were exposed to AF. The impact of exposure to diabetes and AF on stroke recurrence and death/recurrence is shown in Table 5. The fraction of stroke recurrences over the first year attributable to exposure to diabetes was greater than that of AF in both the exposed group and the entire stroke population. The reverse was seen when considering the fraction of death/recurrence over the first year.
This article presents the rates, patterns, and determinants of stroke recurrence in a multiethnic population. The strengths of this study include its defined multiethnic population, large sample size, inclusion of first-ever strokes only, prospective case ascertainment, complete notification of index and recurrent strokes, registration of both ischemic and hemorrhagic strokes, and comparisons of subtypes between index and recurrent strokes.
Risks of Stroke Recurrence
This study confirms the important previous finding that the risk of stroke recurrence is highest in the immediate period after index stroke.5,8 Consequently, initiation and monitoring of secondary preventative treatment without delay appear to be paramount.20 There has been considerable variation in the stroke recurrence rates reported by stroke registries that have included both ischemic and hemorrhagic strokes.5,6,8 They have reported 3-year cumulative risks of stroke recurrence of 6% (Malmö), 18% (Perth), and 25% (Oxfordshire). The 3-year cumulative risk of stroke recurrence of 14%, which was observed in the present study, fell within this range. The annual risks of first stroke recurrences in the second to fifth year after index stroke were similar in the present and all 3 previous studies. Differences between studies looking at stroke recurrences seem to occur particularly in the first year after index stroke and may be explained by differences in definition of stroke recurrence, acute and preventative management, case mix, and research methodology.
One problem in the study of stroke recurrence is the difficulty in identifying recurrences, particularly when they occur shortly after the index stroke. Reliable detection of new neurological signs in a patient who has depressed consciousness or is paralyzed or bedridden is more difficult than in a recovered patient. Hence, the definition of stroke recurrence often specifies additional criteria for the acute phase. The present study used the same criteria as the Perth and Oxfordshire studies and excluded vascular events within the first 21 days after index stroke unless they were clearly in another vascular territory.5,8 The stroke registry in Malmö, which reported the lowest recurrence rates, excluded all events within the first 4 weeks after the index stroke.6 Studies in which the definition of recurrence had a shorter time qualification between index and recurrent stroke report a higher recurrence rate.7 Estimates of the risks of stroke recurrence become less precise with follow-up periods over several years because of death and dropout. Data on 5-year cumulative risks in the present study were based on ≈350 patients. This is the sample size with which most other cohort studies started their follow-up.1,2,4,8,22 When stroke recurrence rates were analyzed, some patients may have been erroneously censored because they died of a stroke recurrence that remained undetected because autopsy was not performed on all patients. Because of this and the potential overlap between risk factors of cardiovascular disease and death and stroke recurrence, this study also analyzed death and stroke recurrence as a combined outcome.
Patterns of Stroke Recurrence
Clinicians may assume that strokes recur in the same form as on index presentation. This explains why the rate of brain scans was considerably lower for recurrent than for index strokes in the present and previous studies.5,8 However, the present study found changes in the subtype between index and first recurrent strokes in 45.5% (46 of 101) of the cases. It was unlikely that the higher number of missing brain scans resulted in a bias. When only stroke recurrences with scans at index and recurrent strokes were considered, results were similar 44.0% (33 of 75; 95% CI, 32.8 to 55.2). The frequent changes in subtype between index and recurrent strokes suggested a multifactorial cause of stroke recurrence.10 Presence of only small-vessel disease could not be assumed in a patient with index LACI stroke because 62.5% of LACI strokes (20 cases) recurred in a different form. Likewise, 50% of hemorrhagic strokes (6 cases) that recurred were ischemic. The pathophysiology of non-LACI infarcts would have been of particular interest because the present and previous studies found these to recur often in the same form.9,23 Unfortunately, the present study did not further differentiate non-LACI strokes according to pathophysiology and relied solely on the clinical presentation. Nevertheless, it became clear that the index stroke poorly predicts the subtype of recurrent strokes. This has clinical implications for diagnostic and therapeutic procedures.
Association Between Subtype of Index Stroke and Risk of Stroke Recurrence
It has been suggested that certain pathological subtypes of stroke are associated with higher risks of recurrence than others during the first months after index stroke.7,14 However, long-term risk of stroke recurrence has been found to be similar across subtypes of stroke.1,5,7 This is in keeping with the present study in which differences in stroke recurrence rates between subtypes reached significance only in the first 3 months and the final multivariable model for stroke recurrence did not include subtype of index stroke. Further differentiation of non-LACI strokes according to the classification of Bamford et al14 did not help to predict stroke recurrence rates and improved the model of recurrence-free survival only over the first 3 months. That study, as well as the present study, found low rates of stroke recurrence in TACIs during the first few months after index stroke. This may be an artifact resulting from the difficulty in diagnosing new neurological deficits in severely affected patients.
Association Between Risk Factors and Stroke Recurrence
Several authors have investigated the effect of potentially modifiable risk factors on stroke recurrence to help to target secondary preventative interventions. Their results have often been contradictory, which may be accounted for by small sample sizes and differences in the definitions of risk factors. The best established risk factors associated with stroke recurrence in previous studies have been diabetes mellitus1,4,6,8 and AF.6,11 This is in keeping with the present study in which diabetes mellitus and AF reached significance. Previous findings regarding hypertension have been more heterogeneous, and almost as many studies have found a significant association with stroke recurrence1,4,11 as have not.3,5,6,8 This may be related to different criteria and methods for measuring hypertension and issues of antihypertensive management.20 According to our multivariable analysis, smoking seems to affect survival rather than stroke recurrence. It should be acknowledged that not all risk factors of index stroke were investigated in this study.24 The present study estimated that among affected patients, 19.7% of recurrences and 36.9% of deaths/stroke recurrences in the first year were attributable to AF, and 7.4% and 36.9%, respectively, were attributable to diabetes. Therefore, effective antidiabetic and antithrombotic management should be used rapidly in this group.20 This does not necessarily mean that all of these events could have been prevented by risk factor treatments. The ascertainment of risk status was undertaken at the time of the index stroke, and treatment of risk factors was not taken into account. Furthermore, only 4.9% and 9.1% of stroke recurrences in the entire stroke population were attributable to AF and diabetes mellitus, respectively, and these estimates did not reach statistical significance. More research is needed in this area.
Of all first-in-a-lifetime stroke patients, 17% have a second stroke within the next 5 years, and stroke recurrence rates are similar across subtypes. Except for non-LACI strokes, more than half of recurrent strokes are of a different subtype than the index stroke, suggesting a multifactorial origin of stroke recurrence. Although AF and diabetes mellitus put affected individuals at significantly increased risk, they account only for a small proportion of stroke recurrences in the entire stroke population, and most stroke recurrences remain unexplained by conventional risk factors.
This study was funded by the Northern and Yorkshire Region Research and Development Programme (United Kingdom), the Charitable Foundation of Guy’s and St Thomas’ Hospitals (United Kingdom), and the Stanley Thomas Johnson Foundation, Switzerland.
- Received October 9, 2002.
- Revision received December 3, 2002.
- Accepted January 13, 2003.
Hier DB, Foulkes MA, Swiontoniowski M, Sacco RL, Gorelick PB, Mohr JP, et al. Stroke recurrence within 2 years after ischemic infarction. Stroke. 1991; 22: 155–161.
Prencipe M, Culasso F, Rasura M, Anzini A, Beccia M, Cao M, et al. Long-term prognosis after a minor stroke: 10-year mortality and major stroke recurrence rates in a hospital-based cohort. Stroke. 1998; 29: 126–132.
Petty GW, Brown RD Jr, Whisnant JP, Sicks JD, O’Fallon WM, Wiebers DO. Survival and recurrence after first cerebral infarction: a population-based study in Rochester, Minnesota, 1975 through 1989. Neurology. 1998; 50: 208–216.
Sacco RL, Shi T, Zamanillo MC, Kargman DE. Predictors of mortality and recurrence after hospitalized cerebral infarction in an urban community: the Northern Manhattan Stroke Study. Neurology. 1994; 44: 626–634.
Burn J, Dennis M, Bamford J, Sandercock P, Wade D, Warlow C. Long-term risk of recurrent stroke after a first-ever stroke: the Oxfordshire Community Stroke Project. Stroke. 1994; 25: 333–337.
Elneihoum AM, Goransson M, Falke P, Janzon L. Three-year survival and recurrence after stroke in Malmö, Sweden: an analysis of stroke registry data. Stroke. 1998; 29: 2114–2117.
Petty GW, Brown RD, Whisnant JP, Sicks JD, O’Fallon WM, Wiebers DO. Ischemic stroke subtypes: a population-based study of functional outcome, survival, and recurrence. Stroke. 2000; 31: 1062–1068.
Hankey GJ, Jamrozik K, Broadhurst RJ, Forbes S, Burvill PW, Anderson CS, et al. Long-term risk of first recurrent stroke in the Perth Community Stroke Study. Stroke. 1998; 29: 2491–2500.
Yamamoto H, Bogousslavsky J. Mechanisms of second and further strokes. J Neurol Neurosurg Psychiatry. 1998; 64: 771–776.
Lai SM, Alter M, Friday G, Sobel E. A multifactorial analysis of risk factors for recurrence of ischemic stroke. Stroke. 1994; 25: 958–962.
Stewart JA, Dundas R, Howard RS, Rudd AG, Wolfe CDA. Ethnic differences in incidence of stroke: prospective study with stroke register. BMJ. 1999; 318: 967–971.
WHO Task Force on Stroke and other Cerebrovascular Disorders. Recommendations on stroke prevention, diagnosis, and therapy: report of the WHO Task Force on Stroke and Other Cerebrovascular Disorders. Stroke. 1989; 20: 1407–1431.
Hajat C, Dundas R, Stewart JA, Lawrence E, Rudd AG, Howard R, et al. Cerebrovascular risk factors and stroke subtypes: differences between ethnic groups. Stroke. 2001; 32: 37–42.
Dennis M, Bamford J, Sandercock P, Warlow C. Prognosis of transient ischemic attacks in the Oxfordshire Community Stroke Project. Stroke. 1990; 21: 848–853.
Barker J, Cryer C, Maryon-Davis A. Health Quest South East: Regional Report. London, UK: South Thames Regional Health Authority; 1993.
Hillen T, Dundas R, Lawrence E, Stewart JA, Rudd AG, Wolfe CD. Antithrombotic and antihypertensive management 3 months after ischemic stroke: a prospective study in an inner city population. Stroke. 2000; 31: 469–475.
Sacco RL, Benjamin EJ, Broderick JP, Dyken M, Easton JD, Feinberg WM. AHA Conference Proceedings: risk factors. Stroke. 1997; 28: 1507–1517.