The iScore Predicts Effectiveness of Thrombolytic Therapy for Acute Ischemic Stroke
Background and Purpose—Tools to predict the clinical response after intravenous thrombolytic therapy (tPA) are scarce. The iScore is an existing validated tool to estimate outcomes after an acute ischemic stroke. The purpose of this study was to determine the ability of the iScore to predict clinical response and risk of hemorrhagic transformation after tPA.
Methods—We applied the iScore (www.sorcan.ca/iscore) to patients presenting with an acute ischemic stroke at 11 stroke centers in Ontario, Canada, between 2003 and 2009 identified from the Registry of the Canadian Stroke Network. A cohort of patients with stroke treated at 154 centers in Ontario was used for external validation. We compared outcomes between patients receiving and not receiving tPA after adjusting for differences in baseline characteristics using propensity-score matching. Patients were stratified into 3 a priori defined groups according to stroke severity using the iScore.
Results—Among 12 686 patients with an acute ischemic stroke, 1696 (13.4%) received intravenous thrombolysis. Higher iScores were associated with poor outcomes in both the tPA and non-tPA groups (P<0.001). Among those at low and medium risk based on their iScores, tPA use was associated with a benefit in the primary outcome (relative risk, 0.74 for those with low-risk iScores; 95% CI, 0.67–0.84; relative risk, 0.88 for those with medium risk iScores; 95% CI, 0.84–0.93). There was no difference in clinical outcomes between matched patients receiving and not receiving tPA in the highest iScore group (relative risk, 0.97; 95% CI, 0.94–1.01). Similar results were observed for disability at discharge and length of stay. The incident risk of neurological deterioration and hemorrhagic transformation (any or symptomatic) with tPA increased with the iScore risk. Results were similar in the validation cohort for risk of poor outcome with tPA by iScore level.
Conclusion—The iScore may be used to predict clinical response and risk of hemorrhagic complications after tPA for an acute ischemic stroke. Patients with high iScores may not have a clinically meaningful benefit from intravenous tPA at the time of carrying a higher risk of hemorrhagic complications.
Stroke is a leading cause of adult disability that can be devastating for patients and their families. Approximately two thirds of stroke survivors continue to experience functional deficits that are associated with diminished quality of life.1,2 Intravenous thrombolysis has been shown to improve clinical outcomes after an acute ischemic stroke in randomized trials and observational studies.3,4
Some factors that have been associated with improved outcome after thrombolysis include: shorter door-to-needle time, younger age, normoglycemia, the absence of comorbidities, and milder stroke.5,6 In contrast, advanced age, National Institutes of Health Stroke Scale >20, hyperglycemia on admission, and congestive heart failure among other comorbid conditions are associated with lower clinical response to thrombolysis.7–9
The decision to administer intravenous thrombolysis may be challenging, especially in patients with a higher prevalence of comorbid conditions, preadmission dependency, and dementia. Patients and families wonder about the likelihood of a good outcome if tissue plasminogen activator (tPA) is given, especially if the risk of developing hemorrhagic complications is high. Although a variety of risk prediction tools exist to estimate death or disability after stroke, it is not known whether any of these can be used to predict the likelihood of a good outcome after thrombolytic therapy.
The iScore is a recently developed and validated score that can be used to estimate the risk of short- and long-term mortality and clinical outcomes after an acute ischemic stroke.10,11 The iScore categorizes patients with ischemic stroke into risk categories, from very low to very high average risk, using clinical parameters and comorbid conditions.10 The prediction of response to intravenous thrombolysis may be useful for patients, their families, and clinicians in counseling or discussions related to the decision of giving intravenous thrombolysis. Our objectives were (1) to evaluate the ability of the iScore to predict clinical response after thrombolysis in patients admitted with an acute ischemic stroke; and (2) to determine whether a differential response to thrombolysis exists among low-, medium-, and high-risk groups predicted by the iScore.
We used the Registry of the Canadian Stroke Network (RCSN) to identify patients admitted to 11 acute stroke centers in the province of Ontario, Canada. Since its inception, the RCSN is a clinical database of >40 000 patients seen in the emergency department or admitted to the hospital with an acute stroke or transient ischemic attack. Further details on the RCSN can be obtained from the RCSN Report at www.rcsn.org and published elsewhere.10,12 Information on poststroke all-cause mortality was obtained through linkages to the Ontario Registered Persons Database at the Institute for Clinical Evaluative Sciences. The Registered Persons Database is a population-based administrative database including basic demographic data and date of death, which provides complete follow-up for all residents in the province. For the validation study (to determine the consistency of the results), we used the RCSN Ontario Stroke Audit database. Data collection for the Ontario Stroke Audit occurs every 2 to 4 years through the abstraction of charts of a random sample of eligible patients seen in the emergency department or admitted to all types of hospitals with a diagnosis of stroke or transient ischemic attack. All hospitals (n=154) in the province with >10 stroke admissions per year are included and include teaching hospitals and community-based institutions from rural and urban areas throughout Ontario. In the present study, we used Ontario Stroke Audit data collected in 3 periods: 2002 to 2003, 2004 to 2005, and 2008 to 2009. This external validation cohort consisted of 4908 patients with ischemic stroke admitted at 154 Ontario Stroke Audit hospitals.
The cohort used in the present study included patients from Phase 3 of the RCSN that were ≥18 years of age with a primary diagnosis of acute ischemic stroke and having presented to any of the 11 participating institutions in Ontario between July 1, 2003, and June 30, 2008. Patients with missing baseline characteristics (Congress of Neurological Surgeons score, glucose on admission, unique health identifier; n=1005; 7.3%) were excluded (Figure). Patients with transient ischemic attack were not eligible for this study.
For each patient, we calculated the iScore. The iScore is a recently developed and validated risk score that can be used to estimate the risk of death or disability after an acute ischemic stroke. The risk scoring system is represented in Table 1. Details of the selection of variables for the iScore, data sources, and the creation and conceptualization of the iScore have been published elsewhere.10,11 An online web-based tool (www.sorcan.ca/iscore) and iPhone version are currently available for practical use. Because our intention was to determine the ability of the iScore to predict a good clinical response after thrombolysis, patients were categorized a priori into 3 risk groups based on their probability of death or disability after stroke based on their iScore: low risk (iScore ≤139 with a >50% probability of a good outcome), medium risk (iScore 140–179 with a 10%–50% probability of a good outcome), and high risk (iScore ≥180 with a <10% probability of a good outcome).
Poor outcome, the primary outcome measure, was defined as death that occurred within 30 days of the stroke admission or severe disability at discharge according to on the modified Rankin Scale (mRS) score of 3 to 5. Secondary outcomes included: (1) disability at discharge (mRS 3–5); (2) death at 30 days; (3) length of hospital stay; and (4) intracerebral hemorrhage (any type or symptomatic) and neurological deterioration in the tPA cohort.
Chi-square tests were used to compare categorical variables; analysis of variance or Kruskal-Wallis tests were used to compare mean and median differences for continuous variables. Due to systematic differences in baseline characteristics between patients who did and did not receive tPA, propensity-score matching methods were used to estimate the effectiveness of tPA within each stratum of stroke severity as defined using the iScore. Propensity-score matching is an analytic technique that allows one to eliminate or minimize systematic differences between treatment groups so that any observed differences in outcomes can be attributed to the exposure. Propensity-score matching was conducted separately within each of the 3 stratum of the iScore. The propensity score (PS) was estimated using a logistic regression model in which receipt of tPA was regressed on the following baseline covariates that were thought to be associated with the outcomes of interest: age, sex, stroke severity, stroke subtype (lacunar versus other), hypertension, diabetes, hyperlipidemia, atrial fibrillation, previous stroke or transient ischemic attack, renal failure on dialysis, level of consciousness on arrival, dysphasia, glucose on admission, independence, and arrival by ambulance. Prior research has shown that including prognostically important variables in the PS model is a good analytic strategy.13 tPA and non-tPA patients were matched on the logit of the PS with a caliper of 0.2 SDs of the logit of the PS with a matching ratio of 1:1.14,15
In the final PS-matched sample, we compared the main (all-cause mortality at 30 days or disability at discharge) and secondary (disability at discharge, death at 30 days) outcomes between those receiving and not receiving intravenous tPA using paired t tests for continuous variables and McNemar tests for the binary variables. We calculated the relative risk and the corresponding 95% CIs in the PS-matched cohort using the method proposed by Agresti and Min.16
Statistical analysis was performed using SAS statistical software Version 9.2.2 (SAS Institute Inc, Cary, NC). All tests were 2-tailed, and probability values <0.05 were considered significant. Approvals from the St Michael's Hospital review board and the RCSN Publications Committee were obtained.
Among 12 686 patients with ischemic stroke in the RCSN sample, 1696 (13.4%) received intravenous tPA. In this sample, patients receiving tPA had more severe strokes, defined as a Congress of Neurological Surgeons score <4 (37.9% versus 16.1%; standardized difference 0.56), were more likely to have presented with aphasia (51.0% versus 28.6%; standardized difference 0.49), to have had a nonlacunar stroke subtype (93.10% versus 81.5%; standardized difference 0.31), and were more likely to have arrived by ambulance (91% versus 67.3%; standardized difference 0.60) compared with those who did not receive tPA (Table 2). Overall, the mean time from symptoms onset to tPA was 160.5 (SD ±87) minutes, and 18.8% received thrombolysis after 3 hours.
After PS matching in the low-risk group (iScore ≤139), 589 patients who received tPA were matched with 589 patients who did not receive tPA. Similarly, for the medium-risk (iScore 140–179) and high-risk (risk score ≥180) groups, 682 and 419 patients who received tPA were matched with 682 and 419 patients not receiving tPA, respectively. We were not able to find a match pair in only 6 patients (1.0%).
After matching on the PS, systematic differences between tPA and non-tPA patients in all groups were substantially reduced with all standardized differences being <0.1 (Table 3).
Clinical Outcomes by Risk Score
Overall, the absolute reduction in death or disability (mRS ≥3) at discharge was 10.2% favoring tPA (low iScore 16.1%; medium iScore 9.7%; high iScore 3.1%). In the matched sample, higher iScores were associated with poor functional outcomes in both the tPA and non-tPA groups (P<0.001; Table 4). In both the low- and medium-risk iScore groups, thrombolysis use was associated with a lower risk of a poor outcome (relative risk [RR], 0.74 in the low-risk group; 95% CI, 0.67–0.84; RR, 0.88 in the medium-risk group; 95% CI, 0.84–0.93). In the high-risk iScore group, tPA showed no significant benefit (RR, 0.97; 95% CI, 0.94–1.01). Similar results were observed for disability at discharge and length of stay (Table 4).
Death at 30 days occurred in 5.1% of non-tPA and 4.9% of tPA patients with low-risk iScores, 14.5% of non-tPA and 17.7% of tPA patients with medium-risk iScores, and 38.2% of non-tPA and 36.3% of tPA patients with high-risk iScores. tPA administration was not associated with a significant reduction in death at 30 days either in the overall sample nor in the matched risk groups (for the low iScore RR, 0.97; 95%CI 0.59–1.60; for the medium iScore RR, 1.22; 95% CI, 0.96–1.56; and for the high iScore RR, 0.95, 95% CI, 0.79–1.14).
The Figure represents the RR for a favorable outcome at each level of the iScore as determined from the multivariable model fit in the original cohort (n=12 868). These results complement those from the PS-matched analyses to suggest the lack of benefit with tPA (and probability of being harmful) for patients with an iScore >200 (note the upper CI crossed the line of 1; probability value for the iScore by treatment interaction <0.001).
Hemorrhagic Complications and Neurological Deterioration
Hemorrhagic transformation of any type occurred in 12.4% (211 of 1696), and symptomatic hemorrhage occurred in 6.9% (117 of 1696) of patients receiving tPA (Table 5). Patients receiving tPA in the high-risk iScore group had a higher incidence of intracerebral hemorrhage (any type and symptomatic hemorrhagic transformation) and neurological deterioration compared with lower risk iScore groups (Table 5; Supplemental Figure I; http://stroke.ahajournals.org).
Overall, 30-day mortality for patients with any hemorrhagic transformation after tPA was 38.4% (81 of 211), and a poor outcome was observed in the vast majority (92.9%) of these patients. Among those with hemorrhagic complications, outcomes were worse in the higher compared with the lower risk iScore groups (Table 5). The number needed to harm for death at 30 days or disability at discharge was 5 for any hemorrhagic complication and 17 for symptomatic hemorrhagic transformation.
Similar to the observation in the original cohort, tPA administration was associated with lower risk of poor outcomes in the low and medium iScore groups (low iScore RR, 0.78; 95% CI, 0.63–0.98; medium iScore RR, 0.83; 95% CI, 0.71–0.97). A nonsignificant benefit was observed between matched patients receiving and not receiving tPA in the high iScore group (RR, 0.91; 95% CI, 0.82–1.04). Similar results were observed for disability at discharge. Further details and other outcomes are shown in the supplemental material.
The prediction of a clinical response to a specific treatment or intervention is difficult. Particularly challenging is the prediction of a clinical response to thrombolysis considering the diverse interaction of coexisting comorbidities in patients with stroke and the potential risk of serious complications (ie, intracerebral hemorrhage).17,18 Physicians' perceptions on tPA risk and benefit assessment may not be accurate. For example, in a survey including emergency physicians and neurologists, only 11% (95% CI, 0%–22%) could correctly identify the magnitude of the benefit with tPA.19 Similarly, only 39% were able to estimate the risk of symptomatic and fatal intracerebral hemorrhage.19 As a result, misperceptions and misconceptions could affect clinical decision-making, especially in emergency situations. More important, decision support using clinical tools (eg, iScore) and electronic resources for diagnosis, screening, or prevention improve clinicians' performance.20 In previous studies, we found the iScore to be a useful tool to predict death at 30 days and 1 year as well as functional clinical outcomes after ischemic stroke.10,11 We have also demonstrated that the iScore improves on the accuracy of other simple models (only including age and stroke severity), which may over- or underestimate the risk of poor functional outcomes.10
In the present propensity-matched study, patients in the low (≤139) and medium (140–179) risk iScore categories who received tPA were significantly more likely than those who did not receive tPA to have a favorable outcome. The low and medium iScore group receiving tPA had a significant 26% and 12% reductions, respectively, in the risk of the combined outcome of death at 30 days or disability at discharge. In contrast, intravenous tPA resulted in a nonsignificant benefit among patients in the highest risk ≥180 iScore category (compared with those who did not receive tPA). Moreover, hemorrhagic complications after tPA increased with the iScore risk, being 3 times higher in the high iScore stratum (19.8%) compared with the lower risk category (6.4%). Results were consistent in the validation cohort despite a smaller sample size. The chance of achieving a favorable outcome with tPA sharply declined with higher iScore risk. Patients with iScore >200 had no apparent benefit after tPA with a potential harmful effect (Figure).
Together these findings suggest that the iScore can be used to predict the clinical response and major complications after tPA. These findings may be useful to clinicians in identifying groups of patients less likely to respond to tPA.
Previous studies have identified factors associated with good outcome and risk of hemorrhagic transformation after thrombolysis.3,5,21,22 In a pooled analysis of randomized trials including 2131 patients receiving tPA or placebo, the authors developed a model to predict good (mRS ≤1) and poor (mRS ≥5) outcome.23 Common variables associated with good outcome included the combination of age and National Institutes of Health Stroke Scale, hypertension and tPA, sex, and diabetes. Moreover, >60% of patients included in the study were treated between 3 and 6 hours after symptom onset. Despite this valuable information, no risk score has been validated in a large cohort including “real-world” patients to assist clinicians to determine the likely clinical response to thrombolysis. The iScore includes some of the well-established predictors (eg, age, stroke severity, hyperglycemia) and also adds other relevant concomitant conditions influencing stroke outcomes.10,11 Moreover, the observed absolute reductions in death or disability (mRS ≥3) in the matched cohort for tPA (10.2%) were similar to those reported in the National Institute of Neurological Disorders and Stroke (11.7%),24 European–Australian Acute Stroke Study (ECASS) III (6.8%),25 and Safe Implementation of Thrombolysis International Stroke Thrombolysis Registry and Virtual International Stroke Trials Archive (SITS/VISTA; 14.1%).26
The risk, benefits, and quality of life after tPA in selected high-risk groups has been under debate. In our study, the lack of benefit of tPA in the high-risk iScore group was in part driven by a 20% incident risk of intracerebral hemorrhage (any type and symptomatic), which was associated with a 48% mortality and 99% risk of death or major disability at discharge (Table 5). Although our functional outcome effectively integrates the benefits of tPA with any detrimental effect of hemorrhagic conversion, clinicians may have to balance options in the high iScore group for whom the chance of a good outcome is substantially lower than the risk of hemorrhagic complications. Many clinicians will still elect to administer intravenous tPA in this group given the poor outcomes in the absence of tPA (favorable outcome: 4.8% non-tPA versus 7.4% tPA group; RR, 2.8%; −0.8% to 6.2%) and the unlikely possibility of spontaneous recovery. Others may consider different therapeutic alternatives (eg, endovascular approach) after a discussion with the patient and/or their family. Regardless of the ultimate decision, the iScore facilitates useful information when counseling patients and their families.
Our study has several limitations and strengths. First, it is possible that some potentially relevant variables (eg, infarct size, imaging information) not included in the initial iScore may be associated with variations in the response to thrombolysis with the potential of confounding by indication. Because the iScore was not created for this primary purpose, these factors were not included in the original iScore and are not available in our clinical database. Second, although several ethnic groups were included in the present study, the majority of patients were non-Hispanic whites. Third, we cannot rule out a Type II error. We would argue, however, the power to detect a difference may be less relevant in the light of only 2.6% (nonsignificant) benefit in the primary outcome between matched tPA and non-tPA patients in the high iScore group. Fourth, due to sparse data in some iScore groups, the regression models to generate the Figure may be unstable and should be viewed as hypothesis-generating. Moreover, our study does not rule out a potential clinically meaningful benefit using other therapeutic approaches (eg, endovascular revascularization, intra-arterial thrombolysis, hypothermia) in the high iScore group. Fifth, the determination of the stroke subtype early after hospitalization may be a limitation for nonstroke experts. Finally, although we used PS matching to minimize baseline differences, this is not a randomized clinical trial, and the results from this observational study should not substitute for expert clinical assessment when making decisions on the administration of thrombolysis.
Strengths of our study include a large sample size comprising “real-world” patients (with 19% receiving tPA beyond 3 hours), the use of a previously validated score with a nearly complete ascertainment of stroke severity and follow-up, good matching for all patients receiving tPA, and consistency of the results in 2 independent cohorts.
This study suggests that the iScore (www.sorcan.ca/iscore) can be used to estimate the response and risk of complications after thrombolytic therapy for ischemic stroke. These results may provide important information to clinicians when discussing therapeutic options and prognosis with patients and their families and support rational decision-making for thrombolysis administration. Prospective studies may help determine the best approach in the high iScore group. From the organizational and resource allocation perspective, the iScore may also facilitate discussions about thrombolytic therapy, especially in high-risk individuals who may require more careful monitoring and expend more resources.
Dr Saposnik was supported in part by Clinician–Scientist Award from the Heart and Stroke Foundation of Ontario (HSFO). Drs Moira K. Kapral, and Peter Austin are supported in part by the Career Scientist Award from the HSFO. Dr Tu is supported by a Canada Research Chair in health services research and a Career Investigator Award from the HSFO.
Continuing medical education (CME) credit is available for this article. Go to http://cme.ahajournals.org to take the quiz.
The opinions, results, and conclusions reported in this article are those of the authors and are independent from the funding sources. No endorsement by the Institute for Clinical Evaluative Sciences or the Ontario Ministry of Health and Long Term Care (MOHLTC) is intended or should be inferred.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.111.646265/-/DC1.
- Received November 25, 2011.
- Revision received January 19, 2012.
- Accepted January 20, 2012.
- © 2012 American Heart Association, Inc.
- Nichols-Larsen DS,
- Clark PC,
- Zeringue A,
- Greenspan A,
- Blanton S
- Adams HP Jr..,
- del Zoppo G,
- Alberts MJ,
- Bhatt DL,
- Brass L,
- Furlan A,
- et al
- Konig IR,
- Ziegler A,
- Bluhmki E,
- Hacke W,
- Bath PM,
- Sacco RL,
- et al
- Weimar C,
- Konig IR,
- Kraywinkel K,
- Ziegler A,
- Diener HC
German Stroke Study Collaboration. Predicting outcome after acute ischemic stroke: an external validation of prognostic models. Neurology. 2004;62:581–585.
- Saposnik G,
- Kapral MK,
- Liu Y,
- Hall R,
- O'Donnell M,
- Raptis S,
- et al
- Saposnik G,
- Kapral MK,
- Liu Y,
- Raptis S,
- Tu JV,
- Mamdani M,
- et al
- Kapral MK,
- Silver FL,
- Richards JA,
- Lindsay MP,
- Fang J,
- Shi S,
- et al
- Austin PC
- Merino JG,
- Silver B,
- Wong E,
- Demaerschalk B,
- Tamayo A,
- Hachinski V
- Tanne D,
- Kasner SE,
- Demchuk AM,
- Koren-Morag N,
- Hanson S,
- Grond M,
- et al
- Kent DM,
- Selker HP,
- Ruthazer R,
- Bluhmki E,
- Hacke W
- Mishra NK,
- Ahmed N,
- Andersen G,
- Egido JA,
- Lindsberg PJ,
- Ringleb PA,
- et al