Abstract
Objective: To characterize the rates of recurrent intracranial hemorrhage (ICH), ischemic stroke, and death in survivors of primary ICH.
Methods: Systematic review of studies reporting recurrent stroke in survivors of primary ICH, identified at index ICH and followed forward. Studies were identified by computerized search of the literature and review of reference lists.
Results: Ten studies published between 1982 and 2000 reporting 1,880 survivors of ICH, followed for a total of 6,326 patient-years (mean follow-up, 3.4 patient-years), were included. The aggregate rate of all stroke from five studies was 4.3% per patient-year (95% CI, 3.5% to 5.4%). The rate in the three population-based studies was higher than in the two hospital-based studies, 6.2% versus 4.0% per patient-year (p = 0.04). About three fourths of recurrent strokes were ICH. Considering all 10 studies, a total of 147 patients had a recurrent ICH, an aggregate rate of 2.3% per patient-year (95% CI, 1.9% to 2.7%). Based on data from four studies, patients with a primary lobar ICH had a higher rate of recurrent ICH than those with a deep, hemispheric ICH (4.4% versus 2.1% per patient-year; p = 0.002). The aggregate rates of subsequent ischemic stroke and mortality were 1.1% per patient-year (95% CI, 0.8% to 1.7%) and 8.8% per patient-year (95% CI, 5.2% to 11.0%).
Conclusions: Recurrent stroke among survivors of primary ICH occurs at a rate of about 4% per patient-year, and most are recurrent ICH. Survivors of ICH have a higher risk of recurrent ICH than of ischemic stroke, and this has implications for the use of antithrombotic agents in these patients.
Intracranial hemorrhage (ICH) accounts for 10% to 15% of all strokes and carries substantial associated mortality and often permanent disability for those affected.1 For the estimated 50,000 Americans who survive ICH each year, the risk of recurrent stroke and death are unclear. We undertook a systematic review of this issue, seeking to characterize the occurrence of recurrent ICH, ischemic stroke and death among survivors of primary ICH, as well as to identify predictors of recurrent stroke. We hypothesized a priori that recurrence of ICH would be infrequent and that subsequent stroke more often would be ischemic than hemorrhagic.
Methods.
Types of studies.
Studies of patients identified at the time of a primary ICH, followed longitudinally, and evaluated for recurrent stroke were considered for inclusion. Minimum average follow-up required for a study to be included was 3 months.
Types of participants.
We included studies of patients of any age or gender surviving a primary ICH for at least 30 days. When possible, the studies confirmed ICH by CT or MRI and excluded those due to vascular malformation, coagulopathy, trauma, or neoplasm.
Types of interventions.
There were no restrictions as to the type of intervention.
Types of outcome measures.
The primary outcomes evaluated were recurrent intracranial hemorrhage (any hemorrhage confirmed by CT, MRI, or autopsy and excluding those related to vascular malformation, coagulopathy, trauma, or neoplasm), ischemic stroke (evaluated by CT, MRI, or autopsy), all stroke (ICH, ischemic, plus those of unknown type), and death. Outcomes (except death) in patients with ICH who survived 30 days were counted from the time of the index hemorrhage whenever possible.
Search strategy for identification of studies.
Studies following the natural history after primary ICH and published in any language were identified using an electronic search of MEDLINE from 1 January 1966 to 31 January 2000. Intracranial hemorrhage and hypertension were the primary keywords and textwords employed. Following the electronic searches, the reference lists of the articles collected were reviewed. Contact with other authors in the field was made in order to identify additional studies.
Methods of the review.
Selection of studies and data collection.
The results of the searches were evaluated and data abstracted by two reviewers (R.D.B. and R.G.H.), and then crosschecked; disagreements were resolved by consensus. Extracted information included age, gender, coexistent vascular disease and vascular risk factors, mean follow-up, use of antithrombotic therapy, ethnicity, outcomes of hemorrhagic and ischemic stroke, and death.
Assessment of methodological quality.
Two reviewers evaluated the methodological quality of each selected study, including type of population enrolled, definition of outcomes, duration and frequency of follow-up, and number of patients lost to follow-up. The design and execution of each study were independently rated according to five criteria below by two of the authors. If information was not provided to allow a criterion to be assessed, it was assumed not met. The criteria were a cohort of unselected survivors of ICH, identified at the time of their index ICH and followed forward (a); exclusion of patients with secondary causes of ICH, such as arteriovenous malformations, aneurysms, trauma, tumor, anticoagulant use, or bleeding disorders (b); confirmation of the index ICH by neuroimaging in >80% of study patients (c); follow-up of >90% of study patients until death or end of the study (d); and neuroimaging or autopsy in >80% of study patients with subsequent strokes (e) (table 1).
For the purposes of this review, studies were considered “high quality” if they met at least four of the five criteria.
Statistical methods.
Total patient follow-up for each individual study was calculated from the mean patient exposure or, if not provided, from the event rates and number of events reported. Comparison of the relative frequency of hemorrhagic versus ischemic stroke was restricted to studies reporting both outcomes.
Annualized event rates (events per person-years of observation) for individual trials were computed using a Poisson regression model, with a value of 0.5 events assumed when no relevant events were reported. An aggregate event rate across trials was then computed by taking the anti-ln of the average model coefficient (weighted by relative number of patient-years of observation). The estimated variance for the aggregate event rate coefficient estimate was calculated as the inverse of the total number of events observed. Event rates were compared between groups using a Poisson regression model. Significance was accepted at the 0.05 level and all tests were two-sided.
Results.
Of 109 studies reviewed, 10 published between 1982 and 2000 met our criteria for inclusion in these analyses (see table 1). An additional study that included only patients with lobar ICH is considered separately below.12 The most common reasons for exclusion were that the study involved retrospective chart review, did not report ICH recurrences, or did not follow patients for longer than 30 days. The studies involved follow-up of 1,880 survivors of ICH. The weighted mean quality score for the studies was 3.0 (range, 1 to 5). Three studies were population-based (146 participants), while the remainder were hospital-based (1,734 participants). Five studies were from Europe (601 participants, 32%), with one each from Korea (n = 670, 36%), Australia (n = 36, 2%), Mexico (n = 359, 19%), Canada (n = 172, 9%), and the United States (n = 42, 2%).
Description of included studies
Patient features of survivors of ICH often were not provided. Based on four studies, the mean age was about 65 years, about half of the patients were men, and about two-thirds had hypertension.3,5,9,10⇓⇓⇓ No studies reported the frequency of antiplatelet therapy or its relationship to stroke recurrence. In two studies, the cohort was restricted to those with first-ever strokes.2,3⇓
These 10 studies included 1,880 survivors of ICH (range in individual studies, 32 to 670) followed for about 6,326 patient-years (range, 77 to 2,092), for a mean follow-up of 3.4 years per patient (range, 1.0 to 7.0) ( table 2).
Outcomes of included studies*
All strokes.
Five studies involving a total of 561 patients followed their participants for the endpoint of all strokes.2-4,8,9⇓⇓⇓⇓ A total of 82 patients had a recurrent stroke during a mean follow-up period of 3.2 patient-years, an aggregate rate of 4.3% per patient-year (95% CI, 3.5% to 5.4%). The aggregate rate from the three population-based studies (6.2% per patient-year) was higher than from the two hospital-based studies (4.0% per patient-year; p = 0.04; rate ratio = 1.6).
Ratio of recurrent ICH to ischemic stroke.
Of the 82 first recurrent strokes observed in these five studies, 69 were confirmed as hemorrhagic or ischemic by CT or autopsy, and 13 were categorized as unknown. Excluding strokes of unknown type and weighting studies by relative amount of exposure-years, the aggregate ratio of recurrent ICH to ischemic stroke during follow-up was 2.9:1. The ratio was lower in population-based studies compared with hospital-based studies (1.1:1 versus 3.4:1; χ2 test, p = 0.04).
Recurrent intracranial hemorrhage.
Based on results from all 10 studies, 147 patients had a confirmed recurrent ICH during a mean follow-up of 3.4 years. Excluding results from three studies in which the total 13 strokes of an unknown type occurred, the aggregate rate of recurrent ICH was 2.3% per patient-year (95% CI, 1.9% to 2.7%).2,3,8⇓⇓ Including all studies and assuming that three fourths of 13 unknown strokes were ICH (n = 10), the aggregate rate of recurrent ICH was similar at 2.4% per patient-year (95% CI, 2.0% to 2.8%).
Continuing to consider three fourths of unknown strokes as ICH, the aggregate rate of recurrent ICH in the three population-based studies was 3.7% per patient-year, (95% CI, 2.3% to 5.9%) compared with 2.3% (95% CI, 2.0% to 2.8%) in the hospital-based studies (p = 0.1; rate ratio 1.6). Recurrent ICH was 2.3% per patient-year (95% CI, 1.7% to 3.3%) for the four highest-rated studies, compared with 2.4% per patient-year in the remainder (95% CI, 2.0% to 2.8%). The aggregate rate of recurrent ICH from the five European studies was 2.9% per patient-year (95% CI, 2.2% to 3.9%).
Three studies provided information about the rate of recurrent stroke over time, but the results were difficult to interpret.3,6,9⇓⇓ The rate of recurrent ICH was higher during the initial year following the index ICH in one study.9
Ischemic stroke.
Again considering only the five studies that followed patients for all strokes, a total of 21 confirmed ischemic strokes occurred. Counting these plus one fourth of the unknown strokes in these studies as ischemic (n = 3), the aggregate rate was 1.1% per patient-year (95% CI, 0.8% to 1.7%) during a mean follow-up of 3.2 years per patient.
Deaths.
Four studies involving 229 survivors of ICH reported late mortality rates (30 days or more after the index ICH).3,4,10,11⇓⇓⇓ Aggregate mortality rate was 8.8% per patient-year (95% CI, 6.6% to 11.8%) during a mean follow-up of 2.1 years per patient. In each study and in aggregate, deaths (n = 45) exceeded recurrent stroke (n = 19).
Predictors of recurrent ICH.
Patient features associated with recurrent ICH were assessed in two studies.3,7⇓ No trends relating recurrence to age or gender were apparent. Diabetes was associated with recurrent ICH by univariate analysis in one study.2
Lobar location of the index ICH appeared to confer a higher risk of recurrent ICH compared with deep, hemispheric ICH (table 3; rate ratio 2.2; p = 0.002) based on data from four studies.6,8,9,11⇓⇓⇓ The aggregate rate of recurrent ICH in the 192 patients with initial lobar ICH was 4.4% per patient-year (95% CI, 3.1% to 6.3%) versus 2.1% per patient-year (95% CI, 1.6% to 2.7%) among the 823 patients whose index event was a deep, hemispheric ICH. When data from an additional trial that was restricted to survivors of primary lobar ICH were added to the analysis, the rate of ICH recurrence in patients surviving primary lobar ICH was 5.4% per patient-year (95% CI, 4.1% to 7.1%).12 Apolipoprotein ε genotype was independently associated with the risk of recurrent ICH among those with lobar hemorrhage in one study.
Site of initial bleed versus recurrent ICH
Discussion.
From systematic review of the available data, particularly the studies with additional data on ischemic stroke and death, recurrent ICH is not rare and occurs at twice the rate of ischemic stroke in survivors of primary ICH ( figure). This result challenges the paradigm that small artery disease, likely related to hypertension, is the common precursor for both ischemic and hemorrhagic stroke. The prominently accelerated rate of recurrent ICH and the relatively low rate of subsequent ischemic stroke suggest that vascular mechanisms are different between these stroke subtypes.
The aggregate rates of all stroke was higher in the population-based studies than in the hospital-based studies, likely due to a higher rate of recurrent ICH rather than ischemic stroke. Possible explanations include different patient characteristics or possibly more complete follow-up in the population-based studies. More information about the study patients than is available from the publications is needed to investigate these possibilities.
Several studies, including case-control studies ineligible for inclusion in this analysis, suggest that lobar hemorrhages have a higher risk for recurrent hemorrhage, which is also usually lobar.13,14⇓ In patients in whom the site of primary ICH was lobar, the rate of recurrent ICH was twice as great as for those with a primary deep, hemispheric ICH. The ratio of recurrent ICH versus recurrent ischemic stroke may be different for survivors of lobar ICH versus deep, hemispheric ICH, but available data were inadequate to assess this directly.
Some studies attempted to restrict analysis to hypertension-associated ICH. One study reported a cohort restricted to hypertensive ICH, but inclusion criteria included blood pressure measured after hospital admission for ICH. This study also reported 10% lobar hemorrhages, but did not provide data about additional, excluded patients with lobar ICH who did not meet criteria for hypertension. This series did not appear to represent purely hypertension-related ICH, but rather primary ICH, much like the other studies.6 Another study attempted to exclude ICH not associated with hypertension, but this exclusion was not clearly defined, as the authors did not provide data on the number of patients with primary ICH who failed to meet criteria for hypertension and were excluded from the study. In addition, relatively short follow-up and a small patient population make difficult any determination as to whether the lack of observed recurrence reflected a real difference between hypertension-related ICH and the play of chance.11 An additional study did not report the relationship between hypertension and lobar or nonlobar hemorrhage, neither at index event nor with recurrence. However, these authors also conclude that lobar ICH is more likely to recur.9 Overall, these studies do seem to suggest an association between index lobar ICH and recurrent lobar ICH.
This systematic review demonstrates the limitations of currently available data and the need for further research, pertinent to the management of hundreds of thousands of survivors of ICH. Among the studies included in this analysis, the data are based on clinically diagnosed, symptomatic strokes during follow-up. Lacunar infarcts occurring during follow-up are less likely to be identified than recurrent ICH. For this reason minor ischemic strokes and ICH may have been underdetected, and hence the rates of occurrence may be underestimated. Better data about the timing and severity of recurrent vascular events, influence of hypertension control, use of antiplatelet agents, and identification of high-risk subgroups are needed. Available studies of recurrent stroke in survivors of primary ICH suggest that recurrent ICH occurs more frequently than ischemic stroke, that ICH recurs at a rate of about 2.4% per year, and that those with lobar hemorrhage may be at higher risk.
- Received September 8, 2000.
- Accepted November 21, 2000.
References
Letters: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Alert Me
Recommended articles
Long-term antithrombotic treatment in intracranial hemorrhage survivors with atrial fibrillation
White matter hyperintensity load on stroke recurrence and mortality at 1 year after ischemic stroke
Recurrent thromboembolic events after ischemic stroke in patients with cancer
Anticoagulation for atrial fibrillation after intracranial hemorrhage