NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Sharma M, Clark H, Armour T, et al. Acute Stroke: Evaluation and Treatment. Rockville (MD): Agency for Healthcare Research and Quality (US); 2005 Jul. (Evidence Reports/Technology Assessments, No. 127.)

  • This publication is provided for historical reference only and the information may be out of date.

This publication is provided for historical reference only and the information may be out of date.

Cover of Acute Stroke: Evaluation and Treatment

Acute Stroke: Evaluation and Treatment.

Show details


Results of Literature Search

A total of 9,994 bibliographic records were retrieved through database searches (QUOROM flow chart, Appendix D). After 2,511 duplicate records were removed, 7,483 unique items remained. The reviewers nominated an additional four potentially relevant studies. A total of 7,487 reports were evaluated against the eligibility criteria and after the initial screening for relevance, 6,098 records were excluded. The remaining 1389 reports were then retrieved and subjected to a more detailed relevance assessment. After further relevance assessment, 1,253 of the 1,389 reports failed to meet the inclusion criteria. An additional eligibility criterion of level of evidence was then applied to the 136 remaining studies. One hundred bibliographic records that were examined and categorized to one of the 12 specific interventions examined, were deemed not to provide sufficient level of evidence for their related question. The reasons for exclusion are listed in the QUORUM flow chart (Appendix D). In total, 36 records (reporting on 37 studies) were deemed relevant and provided sufficient level of evidence for the systematic review. The Evidence Tables are presented in Appendix E and provide descriptive characteristics and results from the interventions. Experimental studies are presented first followed by observational studies in the following study design order: randomized clinical controlled trials; clinical controlled trials; prospective cohorts; retrospective cohorts; case-control studies; and before-after studies.

Report and Study Design Characteristics of Included Studies

The 36 studies included 6,960 individuals, ranging in age from 19 years to 95 years. Twenty-three studies reported on percentage of male participants, which ranged from 38.2% to 76%.67, 87108

Stroke Type

As expected, participants with acute ischemic stroke were drawn from admitting hospitals and EDs in 26 of 34 studies (72.7%)55, 67, 9294, 97, 99102, 104118 and nine of 34 studies (27.3%) admitted patients with ICH.8791, 96, 119121 Two studies (6%) recruited patients with both ischemic and hemorrhagic stroke.95, 103


Twenty-two studies specified the baseline severity of stroke using the NIHSS. Eighteen studies used only the NIHSS,55, 67, 93, 94, 97, 99102, 104106, 108, 109, 109118 whereas, four studies used other scales in addition to NIHSS.92, 97, 107, 116 Four studies included patients with all severity types assessed by NIHSS,67, 99, 102, 113 whereas, the remainder of the studies included subjects with moderate to severe stroke excluding mild strokes as defined by NIHSS of less than 4.55, 9294, 97, 100, 101, 103, 106, 109112, 114117 One study used the mRS alone,118 seven studies used other various scales such as the Glasgow Coma Scale (GCS),8789, 121 level of consciousness,119 New York Heart Association grade system,96 or clinical and neurological measures;120 five studies did not report on the baseline severity of stroke of the subjects recruited.90, 91, 95, 104, 105


The quality of included RCTs (n=24) was scored using the Jadad scale (scores range from 0 to 5). Only one RCT study reported on all Jadad items.113 Nine RCTs received only one point.55, 87, 93, 95, 97, 115, 119121 Allocation concealment was assessed as adequate in three studies,92, 111, 113 inadequate in one,93 and unclear for the remaining 20 studies.55, 67, 93, 94, 97, 99102, 104119 55, 67, 8789, 9497, 99, 109, 110, 112, 114, 115, 120, 121 One included controlled clinical trial103 was scored using a modified version of the Jadad scale (range 1 to 3) and received a score of 1. Only one of the nine included cohort studies reported on all the NOS items,117 for a maximum of 9 points. The remaining reports scored between 5 and 8 points.100102, 104106, 118 One included case-control study received 8 points.116 Three pre-post study designs were included in our review, however, the quality could not be determined.90, 91, 105

Intervention A: Does Surgery Impact the Outcome in Patients with Acute Intracerebral Hematoma?

Twenty-three studies investigating the effectiveness of surgery for ICH were identified by our searches. One relevant study was identified by an expert122and was published beyond our search dates. This study, along with five unique parallel RCTs met our eligibility criteria and was included in our final analyses (Summary Table 1).8789, 119121 Studies were published between the years 1989 and 2003. Eighteen studies were excluded for level of evidence and included one non-RCT,123 three single prospective cohorts,124126 five case-control studies,127131 eight non-comparative case series,132139 and one study whose design could not be determined.140

Summary Table 1. Intervention A.


Summary Table 1. Intervention A.

Summary Table 2. Intervention B.


Summary Table 2. Intervention B.

Primary intracerebral hematoma has a poor outcome, with case series approaching a 50% fatality rate.141 Hematoma growth is reported in 38% of patients in the first 24 hours, with consequent deterioration due to local pressure effects, perilesional ischemia, and toxic effects of thrombin and blood degradation products.142 Functional outcomes substantially impaired in survivors, with only 16% of patients randomized into the ISTICH Trial having either good recovery or moderate disability, as measured by the Glasgow Outcome Scale at 6 months.143

Morgenstern and colleagues reported on patients with ICH from a Prospective Registry and a randomized trial in Houston between 1993 and 1996.87 Sequential patients were initially added to the registry provided that they had hematomas greater than 9 mL in size and could be operated on within 12 hours. The registry contained a total of 41 patients, seven of whom had surgery (open craniotomy) and 34 of whom were treated medically. Surgical patients were more likely to have shorter median time from symptom onset to arrival at the ED (2.9 hours versus 5.4 hours). The median hematoma volumes were larger in patients receiving craniotomy (96.2 mL vs. 32.8 mL). In the nonrandomized group, there was a trend towards a better 6-month outcome as measured by the Barthel Index for medically treated patients (Barthel Index score of 85 versus 65 for surgically-treated patients). This difference was not, however, significant.

The prospectively randomized group contained 34 patients. All of the patients had a lobar or deep hemispheric hematoma greater than 9 mL but less than 20 mL in size. Patients with cerebellar and brain stem hematomas were excluded. The intervention group received craniotomy with hematoma extraction. The control group was admitted to a neurological intensive care unit with ICP monitoring. Treatment including mechanical ventilation, osmotic diuretics, and ventricular drainage, was carried out as needed to maintain an ICP of less than 21 cm of water. The primary outcome measure in this randomized group was 6-month survival, which slightly favored the surgical group (81% versus 76%). This difference, however, was not significant.

The surgical patients had somewhat poorer entry Glasgow Coma Outcome Scale (GOS) scores and, therefore, there was a bias against good outcome in these patients. The trial was also significantly hampered by low power. The limited number of subjects resulted in a 50% power to detect a 4-fold decrease in 6-month mortality of 25%. Of potential importance, the medically treated group did not have a standard regimen for blood pressure, and steroid treatment was not permitted.

Teernstra et al.88 reported on a multicenter RCT carried out in the Netherlands between 1996 and 1999 that examined the stereotactic treatment of intracerebral hematoma by means of a plasminogen activator (SICHPA). Seventy-one patients over the age of 45 with non-traumatic supratentorial hematomas with volumes of 10 mL or greater, and Glasgow eye motor scores between 2 and 10, were randomized to either the surgical group (n=36) or non-surgical (control) group (n=35). The surgical group had the intervention performed within 72 hours of onset. Treatment consisted of placement of a stereotactic catheter with 5000 IU of urokinase injected every 6 hours for eight cycles of treatment. At the end of each cycle, gentle aspiration was used to remove any liquefied hematoma. The control group was described as receiving “standard supportive care.” There was no requirement for ICP monitoring. The primary endpoint was death at 6 months. Mortality at this time point was 56% in the surgical group and 59% in the medical group with no significant difference noted in a Cox Regression Analysis. The stereotactic surgery group did demonstrate lower hematoma volumes. Supportive care and baseline characteristics were similar between the two groups, with the exception that surgical patients more often received low molecular weight heparin and mechanical ventilation, and generally had a slightly lower Glasgow Coma Score on admission.

Auer et al.119 compared endoscopic surgery within 24 hours with conservative management. One hundred patients, between 30 to 80 years of age, with hematomas over 10 cm3and a neurologic deficit, were randomized. All patients had supratentorial bleeds and angiograms to exclude aneurysmal hemorrhage or arteriovenous malformation. Patients randomized to surgery had a bur hole performed through which a 6 mm neuroendoscope was inserted. The scope was guided by intraoperative ultrasound. Once the probe was inserted, alternating irrigation and suction was performed under video guidance with bleeding cauterized by means of a laser within the instrument. The conservatively managed group was treated with hyperosmolar agents, cortisone, and antifibrinolytic agents. Blood pressure was kept between 140 and 160 mmHg irrespective of the presenting blood pressure. Outcome was analyzed at 6 months by means of the mortality rate and quality of survival measured on an ad hoc 6-point scale. At 6 months, mortality was 42% in the surgical group and 70% in the medical group, which was felt to be significant. There was no overall difference in quality of survival.

Batjer et al.120 studied 21 patients between 1983 and 1989 at a single center in Texas using a RCT design. Patients aged 30 to 75 years with putamenal hemorrhages greater than 3 cm in diameter by CT scanning, were included. All patients were required to be hypertensive, which was arbitrarily defined as a minimal recorded blood pressure within the medical record of 160/95. Patients with minimal hemiparesis, decorticate or decerebrate posturing were excluded. Three arms were studied: best medical management, best medical management plus intracranial pressure monitoring, and surgical evacuation. Medical management was rigorously defined to include Decadron, antihypertensive therapy to decrease blood pressure by 25% within 24 hours, intermittent Lasix and Mannitol with specified criteria for intubation and mechanical ventilation. The second intervention was best medical management, as defined above, with intracranial pressure monitoring. The monitor was used to modify medical intervention such that the pressure was maintained at 20 mmHg. The surgical group was treated with craniotomy, with control of blood pressure intraoperatively, and a standardized surgical approach. The trial had a pre-planned sample size of 60, but was terminated after 21 patients had been randomized, since no difference between the three treatment groups was observed, and the outcomes were felt to be poor. At 6 months, 15 of 21 patients were dead or vegetative. The numbers were felt to be too small for meaningful statistical comparison; regardless, no differences were noted between the three groups.

Juvela et al.121 reported on the experience of 52 patients with supratentorial spontaneous intracranial hemorrhage at a single center in Finland. Patients were enrolled between 1982 and 1986. Patients with aneurysmal hemorrhage along with hemorrhage from arteriovenous malformations were excluded. Twenty-six patients were randomized to either an intervention group or non-intervention group. The intervention group had surgical evacuation within 48 hours (median 14.5 hours), whereas, the non-intervention group was treated with conservative management. No details for perioperative care or conservative management were provided. Patients in the conservative group were more likely to be basal ganglion hemorrhage. Six-month mortality was 38% in the conservative group and 46% in the surgical group with no significant difference noted. There was likewise no significant difference in a dichotomized Glasgow Outcome Scale at 6 months.

Zucarrello et al.89 investigated early surgical treatment for supratentorial ICH in 20 patients randomized over 24 months (surgical intervention n=9, medical treatment n=11). Patients were recruited from one university and two community hospitals. Principal eligibility criteria were ICH volume >10cm3on baseline CT scan, GOS >4, randomization within 24 hours of symptom onset (median 3 hours 17 minutes), and <3 hours to time of surgery (1 hour 20 minutes), with no evidence of ruptured aneurysm or arterovenous malformation. No significant differences were noted for mortality rates. The likelihood of good outcome (<3 GOS) was 56% for the surgically treated group and 36% in the medically treated group, which did not differ significantly. A nonsignificant trend for good outcome for the surgically treated group for median GOS, Barthel Index, and Rankin Scale was observed at 3-month follow up. A significant difference in favor of surgical intervention for the NIHSS score was also observed (p=0.04).

Intervention B: Does Antihypertensive Treatment Reduce Stroke Related Mortality and Disability in Patients with Acute ICH?

Six studies were identified that investigated antihypertensive therapy for ICH.90, 91, 144147 Four studies were non-comparative case series designs and were excluded from our review.144147 Two unique studies met eligibility criteria for investigating the effectiveness of antihypertensive therapy for ICH and were a pre-post design (Summary Table 3).90, 91 The year of publication for each study was 199390 and 2000.91 Neither of these studies prospectively answers the clinical question posed but they were included to provide insight into relevant surrogate and safety measures.

Summary Table 3. Intervention C.


Summary Table 3. Intervention C.

Kay et al.90 reported a non-randomized uncontrolled clinical trial evaluating the efficacy of the serotonin antagonist, ketanserin, and its antihypertensive properties to lower mean arterial pressure in patients with ICH without a subsequent rise in ICP.90 Ten patients, five men and five women aged 49 to 64 years, were recruited from the Chinese University of Hong Kong. All patients had a spontaneous ICH confirmed by CT scan 8 to 48 hours prior to recruitment, a systolic BP >180 mm Hg and/or a diastolic BP >100 mm Hg, no previous history of hypertension, and required ICP monitoring in the intensive care unit. The patients were sedated, paralyzed, endotracheally intubated and mechanically ventilated to a target PaCO2 of 30 to 34 mm Hg. An intravenous bolus of 5 to 10 mg ketanserin was given and radial arterial pressure and ICP were measured continuously. BP readings were performed at multiple time points and compared with pre-injection pressures using analysis of variance (ANOVA) with a correction for multiple comparisons.

After the intravenous bolus of ketanserin the BP decreased on average by 40/21 mm Hg within 5 minutes and increased gradually over the next 2 hours remaining below pre-treatment levels. ICP remained stable throughout the observation period and thus the calculated CPP decreased by a mean of 27 mm Hg after 5 minutes, and on completion of the study, steadily increased to be 13mmHg below the pre-treatment value. There was no evaluation of clinical outcomes reported in this study.

A more recently published study by Nishiyama et al.91 explored the safety of calcium antagonist nicardipine and its effect on mean arterial pressure, ICP, and CPP. Twenty-two patients with an acute hypertensive putaminal hemorrhage requiring surgical drainage were recruited for a non-randomized uncontrolled clinical trial. There were 14 men and eight women ranging in age from 47 to 79 years. Mechanical ventilation was continued post-operatively to a target PaCO2of 30 to 35 mm Hg. Post-operatively nicardipine infusion was started at 1 μg/kg/min with rate adjustments to target systolic blood pressure between 120 to 160 mm Hg (a 20%-30% reduction from pre-infusion levels) for 72 hours. In addition all patients received: a hyperosmolar solution (glycerin fructose), anti-seizure medication (phenytoin) and antibiotics. BP was measured directly via a radial artery catheter and ICP monitored continuously via by an intraventricular catheter. Middle cerebral artery blood flow velocity (Vmca) and pulsatility index (PI) were measured and calculated using Transcranial Doppler Ultrasound. Platelet counts were also monitored as an anti-platelet effect of calcium antagonists has been previously reported and may increase the risk of bleeding.148 Clinical outcomes included level consciousness using the Japan Coma Scale (included in the appendix of the article, not referenced elsewhere) and repeat CT imaging to evaluate extension of hemorrhage.

Patients' BP decreased during and up to 24 hours after the end of the Nicardipine infusion compared with pre-infusion. There was no difference in platelet counts. Vmca and PI were unchanged and ICP decreased 24 hours after the end of the infusion. CPP was decreased at 24 and 72 hours of the infusion but was greater than 50 mm Hg at all times. Consciousness levels were unchanged and CT findings did not show any exacerbation of bleeding or edema.

Intervention C: Does IA thrombolysis reduce stroke-related mortality and disability in adults with acute ischemic stroke?

A total of 37 studies investigating IA thrombolytic therapy for ischemic stroke were identified by our searches. Nine single prospective cohort studies,149157 one controlled clinical trial,158 13 non-comparative case-series,159171 two case studies,172, 173 four abstracts174177 and three studies in which designs could not be determined178180 were excluded from our final analyses. Five unique studies met our eligibility criteria for inclusion (Summary Table 3).9294, 109, 110 All five studies were parallel RCTs and were published between 1999 and 2001.

Del Zoppo and colleagues110 described the results of the PROACT I study, a North American multi-centre study of IA pro-urokinase compared with placebo, carried out between 1994 and 1995. Patients aged 18 to 85 with new onset focal neurologic signs, and NIHSS scores between 4 and 30, were randomized on a 2:1 basis to either 6 mg pro-urokinase plus heparin (100 IU/kg bolus plus 1000 IU/hour infusion) or saline placebo plus matching heparin. After reviewing the results on the first 16 patients, the External Safety Committee changed the infusion to 2000 IU bolus followed by 500 IU/hour for 4 hours. Recanalization rates in the M1 or M2 vessels at 120 minutes, defined as TIMI 2 or 3 or better, were 57.7% in the treated group versus 14.3% in the control group. There was, however, no significant difference between mortality or mRS scores (either 0 or 1) at 90 days.

The results of the PROACT I trial were used to design PROACT II, which was reported by Furlan and colleagues.92 This trial was conducted between 1996 and 1998 in North America. A total of 12,323 patients were screened, resulting in 474 patients subjected to angiogram, and 180 patients randomized after meeting angiographic criteria. Eligible patients had clinical signs of less than 6 hours in duration in the middle cerebral artery (MCA). The NIH Stroke Scale Score was 4 to 30. Patients with isolated aphasia or hemianopsia were excluded. The CT scan excluded bleed or tumor and demonstrated early infarct signs in less than one third of the MCA territory. A diagnostic cerebral angiogram had to show complete occlusion or minimal perfusion (TIMI grade 0 or 1) in the M1 or M2 branches of the MCA. Dissection or severe carotid stenosis were exclusionary criteria. Randomization was on a 2:1 schedule favoring the intervention. One hundred and twenty-one patients received 9 mg of IA pro-urokinase over 2 hours. Heparin was delivered at a 2000 IU unit bolus followed by a 500 IU/hour infusion in all patients. The pro-urokinase was injected intra-thrombus or in the proximal face of the thrombus. Clinical assessments were blinded and the primary outcome was a mRS score of less than 2 at 90 days. This outcome was achieved by 40% of the treatment group compared with 25% of the control group (p=0.04). The calculated number needed-to-treat, on the basis of the absolute difference, was seven. Intracranial hemorrhage within 24 hours occurred in 35% of patients in the treated group compared with 13% in the control group. It should be noted that the control group contained more patients with diabetes than the treatment group (20% versus 8%).

Kase and colleagues109 presented the subgroup analysis of PROACT II regarding bleeding. The group who received treatment (n=110) versus the group that received no treatment (n=64) was compared. Symptomatic intracranial hemorrhage occurred in 12 of the 110 patients (10.9%) treated with urokinase with a mean onset time of 10 hours after initiation of treatment, compared with two of the 64 patients (3.1%) in the control group. Mortality with symptomatic ICH was 83%. Elevated blood glucose was associated with symptomatic intracranial hemorrhage, particularly if the baseline glucose was greater than 200 mg/dL (11.1 mmol/L). This resulted in a relative risk of bleeding of 4.2 (95% CI 1.04–11.7).

The EMS Bridging Trial was reported by Lewandowski and colleagues.94 This trial was a Phase I trial conducted between 1995 and 1996 in several centers in the United States. Patients with acute stroke (within 3 hours of symptom onset) and NIHSS scores greater than 5, and were CT-negative for hemorrhagic lesion, were enrolled. The interventional group received IV tPA 0.6 mg/kg plus IA tPA; the latter was delivered if a thrombus was seen on angiogram. The control group received IV placebo plus IA tPA (1 mg) injected beyond the thrombus with subsequent retraction into the thrombus followed by 10 mg/hour tPA infusion. Blood pressure was maintained less than 180/105. The calculated sample size was 30 patients per arm; however, only 35 were recruited prior to the study being halted. The primary outcome was a decrease of 7 or more points in the NIH Stroke Scale from baseline to 7 to 10 days, or a NIH Stroke Scale Score of 0 to 1 at 7 to 10 days. This outcome was achieved by 24% of the population in both groups. There was also no difference in the 90-day Glasgow Outcome Score, Barthel Index Score (95–100), or mRS Score (0 or 1).

Keris et al.93 combined IV and IA treatment in a single-center study performed in Latvia between 1997 and 1998. Patients with ischemic stroke of less than a 6-hour duration in the internal carotid distribution were included. Edema and effacement were described as exclusion arms; however, explicit exclusion criteria were not given. Intervention cases received 25 mg of tPA IA at the proximal surface of the clot followed by 25 mg IV tPA plus 5000 IU heparin initially and twice a day. Analysis was performed on those who had received the intervention (n=12) versus those who had received no intervention (n=33). A good outcome was defined as a mRS score of between 0 and 3. This outcome was obtained at 12 months by 10 of the 12 patients in the treatment arm and 11 of the 33 patients in the control arm. Fatal bleeding was noted in two patients, both of whom were in the combined treatment arm. Symptomatic ICH occurred in two patients in the combined treatment arm and 1 in the placebo/IA treatment arm. These rates were not significantly different between the two arms. Other systemic bleeding complications did not differ between the two arms. Hemorrhage was noted in two of 12 (17%) patients in the intervention group. The authors concluded that, in spite of baseline differences in the two groups, the outcome suggested a benefit for IA treatment.

Intervention D: Does Treatment to Normalize Blood Glucose Levels Reduce Stroke Related Mortality and Disability in Adults with Acute Stroke?

Our searches identified two unique studies investigating the effects of normalization of blood glucose levels in patients with ischemic stroke (Summary Table 4).95, 96 Both studies were parallel RCTs and were published in 2004 and 1999, respectively.

Summary Table 4. Intervention D.


Summary Table 4. Intervention D.

Post-stroke hyperglycemia has been identified in previous studies as being associated with poor stroke outcome. It is unclear to what extent this is a “normal” physiological response or whether it may lead to increased cerebral damage in the acute phase. The Glucose Insulin in Stroke Trial (GIST) has resulted in two published studies in this area. First, a Pilot study96 evaluating the feasibility of a large multicenter RCT of glucose, potassium, and insulin (GKI) in patients with acute stroke. Second, a small RCT that evaluated the natural history of post-stroke hyperglycemia and the immediate response to GKI.95

The pilot study96 was a non-blinded RCT from a single center. All patients who presented with an acute stroke within 24 hours of symptom onset and had a plasma glucose of 7.0 to 17.0 mmol/L were eligible. Exclusion criteria included: New York Heart Association (NYHA) grade 3 or 4 heart failure; renal failure (creatinine level >200 μmol/L); anemia (Hb <9 g/dL); radiologically documented pneumonia; coma (Glasgow Coma Scale motor subscore <4); previous disabling stroke (mRS Score >3); dementia; isolated posterior circulation stroke without physical disability; pure language disorder; previously diagnosed insulin-treated type 1 or 2 diabetes; or, subarachnoid hemorrhage.

Two hundred and forty-five consecutively admitted patients were screened over a 7-month period. Of these, 53 patients were randomized—28 to active treatment (with three patients withdrawn as stroke was not confirmed by post-randomization CT) and 25 to control treatment. Active treatment consisted of a combined infusate of 500 mL 10% dextrose, 16 U Human soluble insulin (Actrapid; Novo nordisk) and 20 mmol potassium chloride (KCl) administered through a peripheral vein at a fixed rate of 100 mL/h to a maximum volume of 2400 mL. Glucose testing was performed hourly by glycaemic strip aiming for a target glucose of 4 to 7 mmol/L. Above the target range, 4 U insulin was added to the infusate. Below the target range, the infusate was stopped and glucose repeated in 15 minutes interval 50% dextrose was given intravenously if glucose ≥4 mmol/L was not achieved spontaneously in 30 minutes. When target range was achieved, the infusate was restarted with 4 fewer units of insulin. Control treatment consisted of 154 mmol/L saline at 100 mL/h to a maximum volume of 2400 mL. Glucose values were not treated unless they exceeded 17 mmol/L.

Baseline blood work, BP and pulse were measured every 4 hours, and plasma glucose samples were measured every 8 hours during the 24 hours of infusate. All blood work was repeated at 48 hours. Clinical assessments of neurological impairment (European Stroke Scale (ESS)) and activities of daily living (Nottingham Extended Activities of Daily living for function) were performed at baseline, 24 hours, 48 hours, 7 days and 4 weeks by trained observers. Clinical assessments were not blinded to treatment allocation

The average glucose level in the group as a whole at randomization was 9.1 mmol/L. The glucose values for the GKI group were 6.4, 6.5, and 6.9 mmol/L at 8, 16, and 24 hours, respectively. For the control group, the values were 7.6, 7.2and 7.6 mmol/L, respectively. These differences did not achieve statistical significance. In the active treatment group the insulin concentration had to be adjusted at least once in 23 of 25 patients. Five patients required single doses of 10% dextrose for low glucose values although only one patient was symptomatic. There was no difference in clinical outcomes between the two groups.

The GIST-UK trial95 was a multicenter RCT. The eligibility and exclusion criteria were similar to the pilot study with the following exception that entry was based on plasma glucose levels of 6.0 to 17.0 mmol/L. Clinical outcomes were not measured in this study, as the focus was natural history of acute hyperglycemia in managed stroke care and efficacy, safety and practicability of routine intervention.

The first 452 patients recruited had a mean age of 74.8 years and 53.3% were women. Overall mean admission plasma glucose was 8.37 mmol/L (SD 2.13); of note, 28.3% of the patients had a glucose level of between 6.0 and 6.9 mmol/L at admission. Baseline demographics were the same between groups. Of the recruited patients, 221 were randomized to receive GKI and 231 received saline solution. Plasma glucose values were significantly lower in the GKI group at 8, 16 and 24 hours of infusion. In both groups, the glucose values were significantly lower during the active or control infusion as compared to baseline. Adjustment of the GKI regimen was required a median of two times per patient. Twenty cases of hypoglycemia occurred that required treatment with 10% dextrose. Diabetic patients within the GKI group required significantly more insulin to reach target, as well as more changes overall compared with non-diabetic patients in the GKI group.

Intervention E: Does Mechanical Thrombus Disruption Reduce Stroke-Related Mortality and Disability in Adults with Acute Ischemic Stroke?

Ten studies investigating the effectiveness of mechanical thrombus disruption for ischemic stroke were identified by our searches. One study was identified by an expert and screened for inclusion.111 This study was published beyond our literature search dates. One case-control,181 five non-comparative case series,182186 and three studies in which the design could not be determined,187189 were subsequently excluded from our review. Two unique parallel RCTs, which met our eligibility criteria, were included in our final analyses (Summary Table 5).97, 111 These studies were published in 2003 and 2004, respectively.

Summary Table 5. Intervention E.


Summary Table 5. Intervention E.

Eggers et al. reported on a single-center, prospective RCT, of ultrasound-enhanced thrombolysis in MCA occlusion.97 The study was carried out between 2000 and 2002, and involved individuals with ischemic stroke of less than 3 hours duration in the MCA territory. All participants met the criteria for the NINDS Thrombolysis Protocol and had a M1 occlusion diagnosed by transcranial ultrasound. tPA was given as per the NINDS Protocol.49 The intervention group received continuous transcranial color-coded sonography. The monitoring was in the pulse wave Doppler mode with an instrument developing an acoustic power of 179 mW/cm2. The control group had patency of the vessel established by transcranial sonography at baseline, 20, 40 and 60 minutes. Each assessment lasted less than 2 minutes.

Of the 1,177 candidates screened, 25 met the inclusion criteria. Of these, 11 were randomized to the treatment group and 14 to the control group. Exclusions were due to failure to meet the NINDS treatment criteria or an absence of an MI occlusion. The primary efficacy parameters at 90 days included the Barthel Index, mRS Score and mortality. Six of the 11 patients in the treatment group (54.5%) and 1 of 14 patients in the control group (7.7%) met the pre-specified criteria of Barthel greater than or equal to 95 (p=0.037). There was, however, no difference between the two groups in the 90-day mRS score or in mortality. The treated group demonstrated a higher median peak systolic blood flow velocity at the end of 1 hour of treatment. However, there was no significant difference in recanalization between the two groups. Four patients in the treated group experienced intracranial hemorrhage or hemorrhagic transformation of the infarct, compared with one patient in the control group (p=0.14).

Alexandrov and colleagues reported on the results of the CLOTBUST Study.111 This phase II multicenter trial was carried out in North America on patients treated with IV tPA within a 3-hour window. One hundred and twenty-six patients were randomized to receive either placebo or continuous ultrasonography. Head frames were placed on all patients. The patients in the treatment group began ultrasonographic monitoring prior to the administration of the tPA bolus and for the subsequent 2 hours. Emitted power output was set at the maximal achievable level with selected insonation depths under the FDA allowed threshold of 750 mW. In both groups, follow-up measurements were taken 30, 60, 90 and 120 minutes after the tPA bolus with arterial recanalization defined by Doppler criteria. The groups were comparable at baseline and equal in number (n=63 for each arm). Symptomatic intracranial hemorrhage occurred in three patients in each group. The pre-specified endpoints were complete recanalization or early or dramatic recovery from stroke. The latter was defined as a reduction of 10 or more points in the NIHSS or a total NIHSS of 3 or less within 2 hours after administration of the tPA bolus. Thirty-one patients (49% of the treated group) and 19 patients (30% of the control group) reached the pre-specified endpoints (p=0.03). Re-occlusion within 2 hours occurred in 11 patients in the treated group (18%) and 14 patients in the control group (22%, p=0.7). The 3-month mortality rates were 15% and 18% in the treated and controlled groups, respectively (p=0.4). Follow-up at 3 months was incomplete by four patients who were excluded from the outcome analysis. mRS scores of 0 or 1 were present in 22 of 53 treated patients (42%) and 14 of 49 control patients (29%) (relative risk 1.45; 95% CI 0.84–2.51; p=0.2).

Intervention F: Is the Effectiveness and Safety of Thrombolytic Therapy for Adults with Acute Ischemic Stroke Affected by Time From Onset to Treatment?

Six parallel RCTs (in seven publications) relevant to the timing of thrombolytic therapy for ischemic stroke published between 1993 and 2002 met our eligibility criteria (Summary Table 6).112 While these studies did not directly compare outcomes stratified by timing of therapy, they cover relevant time windows and examination of the results is instructive to the question posed.55, 67, 99, 107, 108, 112, 113 Albers et al.99 is considered to be a re-analysis of Clark et al.112 and Clark et al.67 Two abstracts,190, 191 one multiple prospective cohort study,192 one single prospective cohort study,193 and one case series,194 were excluded from our synthesis for level of evidence. No one publication investigated the link between onset to treatment time (OTT) and outcome. The included studies recruited patients across relevant time windows and in concert inform decision making on the relationship between time and outcome.

Summary Table 6. Intervention F.


Summary Table 6. Intervention F.

The use of tPA was well-established for the 0 to 3 hour time window by the NINDS Trial.49. This publication is well described elsewhere in this document. The initial trial publication did not describe an effect of time-to-treatment on treatment outcomes. A subsequent reanalysis55, along with the results of trials using tPA outside the 3-hour window, have been reported.

Hacke and colleagues report on a trial of Intravenous tPA for acute hemispheric stroke.108 A total of 620 patients were randomized to receive rtPA at a dose of 1.1 mg/kg versus placebo. This dose of tPA is higher than in the NINDS Trial.49 Adults with hemispheric stroke syndromes presenting within six hours were included. Those with rapidly improving symptoms or minor deficits, as defined by a Scandinavian Stroke Scale Score of greater than 50, were excluded as were those with increased risk for bleeding. Major early infarct changes including swelling of the affected hemisphere or changes in greater than 33% of the MCA territory were also excluded. All CT scans were read in a blinded fashion independently subsequent to randomization. The primary endpoint was clinical outcome as defined by the Barthel Index and Modified Rankin Score (mRS) at 90 days. The intention to treat analysis showed no difference in either the Barthel Index or Modified Rankin Score at 90 days. A secondary endpoint of the combined Barthel Index and mRS showed a significant difference in favor of the rtPA group. One hundred and nine patients were identified as having major protocol violations, 66 of these being violations of the CT inclusion criteria. More protocol violators were in the rtPA group. Neither mortality nor intracranial hemorrhage differed significantly between the two groups on the Intention to treat analysis.

The same group reported a second trial of 800 patients from centers in Europe, Australia and New Zealand randomized to receive intravenous tPA at 0.9 mg/kg or matching placebo within six hours of stroke onset.107 The inclusion/exclusion criteria were similar to those reported above. Four hundred and nine patients were randomized to the tPA group with 391 in the placebo group. The primary endpoint was proportion of patients with a mRS of 0 or 1 at 90 days. Subjects were recruited between 1996 and 1998. While an absolute difference of 3.7% in the proportion having an mRS score of 0 or 1 was noted in favor of tPA treatment, this difference was not statistically significant. Thirty and 90 day mortality did not differ between these two groups. During the first seven days there were more deaths in the treatment group from intracranial hemorrhage, and the combination of cerebral edema and intracranial hemorrhage. Parenchymal hemorrhage within the first seven days was more common in the tPA-treated group than in the placebo group (11.8 versus 3.1%). Symptomatic intracranial hemorrhage also occurred more often in the tPA-treated group (8.8% versus 3.4%).

Clark et al. reported the results of the rtPA 0- to 6-hour acute Thrombolytic Therapy in Acute Ischemic Stroke Study, Part A.112 This was a multi-center North American Trial that ran from 1991 to 1993. It included patients between the ages of 18 to 79 with acute ischemic stroke of less than 6 hours' duration. A CT scan was required to exclude hemorrhage. Patients with minor stroke (score of less than 4 on the NIH Stroke Scale) along with rapidly improving symptoms were excluded. Patients with CT evidence of mass effect with midline shift were also excluded. This was a phase II trial with three planned safety and futility analyses at 75, 150 and 225 patients. The trial was stopped on the basis of an interim safety analysis in October 1993 due to safety concerns in the 5 to 6 hour window. This paper reports the results of the 142 patients enrolled until that point. The mean time-to-treatment in this group was 4 hours 17 minutes in the placebo arm and 4 hours 24 minutes in the tPA arm. Only 17% of the placebo group and 14% of the tPA group were treated in less than 3 hours, whereas, 34% of the placebo group and 31% of the tPA group were treated between 5 and 6 hours. Patients in the placebo group were more likely to be diabetic. A dose of tPA of 0.9 mg/kg following the dosage schedule of the initial NINDS trial49 was used.

The primary planned efficacy endpoints of 30 or 90 days showed no difference in the percent of patients who achieved a greater than 4-point decrease on the NIH Stroke Scale at 30 days, though there was a significant difference at 24 hours with 40% of rtPA patients achieving this response versus 21% of placebo patients (p=0.02). Likewise, there was no significant difference in the infarct volume. Symptomatic bleeding was more likely in the tPA group by day 10 (11.3% versus 0% in the placebo group, p=0.003). There was likewise an increased death rate in the treated group at 90 days (22.5% versus 7.0% placebo, p=0.009). In the 5 to 6 hour population, by day 10 symptomatic intracranial hemorrhage was found in 18.2% of the treated group versus 0% of the placebo group (p=0.03). Death by 90 days was present in 36% of the tPA-treated group versus 4.2% of the placebo group (p=0.01). The trial continued to enroll patients in the 3 to 5 hour window and results are reported in a separate publication.

Subsequently, Clark et al.67 published data on the use of rtPA (Alteplase) for ischemic stroke administered within 3 to 5 hours after symptom onset. This phase 3, placebo-controlled, double-blind RCT was conducted between December 1993 and July 1998, with 90 days of follow-up measurement. Patients were recruited from 40 community and university hospitals in North America. Six hundred and thirteen (intent-to-treat) subjects with ischemic stroke were enrolled. Of these, 547 subjects were treated, as assigned, within 3 to 5 hours of symptom onset. A total of 39 other subjects were treated within 3 hours of symptom onset, 24 were treated more than 5 hours from symptom onset, and three subjects never received study medication. Subjects were administered 0.9 mg/kg of rtPA (n=272) or placebo (n=275), administered intravenously over a 1-hour period. Neurologic (NIHSS score ≤1) and functional (Barthel Index, mRS, Glasgow Coma Scale) outcomes were assessed up to 90 days follow-up. Thirty-two percent of placebo patients and 34% of the rtPA patients had excellent recovery at 90 days follow-up. Within the first 10 days of treatment with rtPA, there was a significantly increased rate of symptomatic ICH (7.0% vs. 1.1% for placebo, p<0.001). Mortality at 90 days was not significant between groups (11.0% in the treatment group and 6.9% in the placebo group). Results in the intent-to-treat population were similar.

A reanalysis of subjects (n=61) enrolled in the Alteplase Thrombolysis for Acute Non-interventional Therapy in Ischemic Stroke (ATLANTIS) study was conducted by Albers and colleagues in 2001.112 Patients had been randomized to receive either IV tPA or placebo within 3 hours of symptom onset. The pre-specified primary and secondary hypotheses of the ATLANTIS part B trial were used to evaluate clinical outcomes in these patients. The authors noted that although there was a significant increase in symptomatic intracranial hemorrhage, patients receiving IV tPA were more likely to have favorable outcome measured by NIHSS (≤1) at 90 day follow-up compared with placebo (p=0.01).

Haley and colleagues reported results of a pilot RCT of tPA for acute ischemic stroke conducted at three centers in the U.S. between 1990 and 1991.113 This was a feasibility trial prior to NINDS and was stratified into the 0 to 90 minute and 90 to 180 minute windows. Patients with a diagnosis of ischemic stroke verified by CT scan with a measurable neurologic deficit on the NIH Stroke Scale were included. Patients with minor stroke consisting of only sensory loss or ataxia were excluded. TPA was delivered in a dose of 0.85 mg/kg over 60 minutes, while a matching placebo was delivered to the control group.

Twenty-seven patients were randomized, 20 (10 rtPA, 10 placebo) to the time stratum from 0 to 90 minutes and 7 (4 rtPA, 3 placebo) to the time stratum of 90 to 180 minutes. While the median stroke scale scores in the early group were comparable, the median stroke scale score in the 90 to 180 minute stratum was 14 in the placebo arm and 6 in the tPA arm. No significant difference was found in change of NIH Stroke Scale score from baseline. No intracranial hemorrhage was noted in either the tPA or placebo group in the less than 90 minute time stratum. Small numbers in the later time stratum (i.e., 90 to 180 minutes) included one of three patients who died of ICH in the placebo group; none of the four patients in the tPA group died.

Marler and colleagues reported on a reanalysis of the relationship between onset to treatment time and outcome at 3 months, early improvement in 24 hours, and intracranial hemorrhage within 36 hours.55 The initial NINDS Report49 suggested that there was no difference between the 0 to 90 minute and the 90 to 180 minute stratum. This subsequent reanalysis was prompted by the concern that other variables may have masked this association. The NINDS Study was a multi-center RCT that enrolled patients between 0 and 180 minutes of stroke onset to rtPA treatment (0.9 mg/kg delivered as a 10% bolus followed by 1-hour infusion) or to matching placebo. Patients with ischemic stroke scale scores greater than 4 were included. The trial was performed in two parts with identical protocols, with the exception that the primary outcome for Part A was at 24 hours, whereas, that for Part B was at 3 months. A favorable outcome was defined as minimal or no deaths that measured on a composite scale at six months while a 4-point improvement in the NIH Stroke Scale Score was considered favorable at 24 hours. The analysis of time from onset to treatment demonstrated that within the 0 to 90 minute stratum, there was a tendency to cluster between 80 and 90 minutes prior to receiving treatment. There was a similar trend, though less marked, in the 90 to 180 minute stratum. The delay from ED arrival to treatment, however, was longer in the latter time window. No association was noted between onset to treatment time and baseline NIHSS. Suspected small vessel strokes were treated somewhat later than those due to large vessel occlusion; however, neither the 24 hour or three month outcomes varied by stroke subtype. The NIH Stroke Scale Score was a confounder for the relationship between onset to treatment time and outcome as the score was higher for tPA-treated patients in the earlier time stratum and higher for placebo patients in the later time stratum. After correction, an odds ratio for a good outcome at 24 hours was 1.71 in the 0 to 90 minute stratum (95% CI 1.09–2.70) and 1.12 in the 90 to 180 minute stratum (95% CI 0.71–1.76). The odds ratios for good outcome at 3 months adjusted for the NIH Stroke Scale Score shows a similar relationship with a value of 2.11 for the 0 to 90 minute stratum (95% CI 1.33–3.35) and an odds ratio of 1.69 in the 90 to 180 minute time stratum (95% CI 1.09–2.62). There was, however, no interaction between onset to treatment time and hemorrhage at 36 hours for symptomatic hemorrhage or all hemorrhages.

We did not conduct a meta-analysis of the effectiveness and safety of thorombolytic therapy for adults with acute ischemic stroke by time from onset to treatment since we identified a meta-analysis published in 2004 that used patient-level data from six RCTs to investigate the interval from onset to treatment using tPA.195 Findings of this meta-analysis indicated that the sooner the treatment is administered (<90 min) the more beneficial the outcome. Administration beyond 3 hours was found to have some benefit, although with some associated risks. Details pertaining to this patient-level meta analysis are described in the discussion section.

Intervention G: Do Pretreatment CT Scoring Systems Affect the Safety and Efficacy of Thrombolytic Therapy for Acute Ischemic Stroke?

Our searches for interventions examining the effectiveness of pretreatment CT scoring systems for ischemic stroke identified 11 studies. One case-control study,196 two single prospective cohort studies,197, 198 and six studies whose designs could not be determined,199204 were excluded for level of evidence. Two unique studies were included in our final analyses (Summary Table 7).114, 115 Both studies were parallel RCTs and were published in 2001 and 2002, respectively.

Summary Table 7. Intervention G.


Summary Table 7. Intervention G.

Patel et al.114 examined the frequency and significance of early infarct changes (EICs) on CT scans from the NINDS database. The NINDS trial described previously was a multicenter RCT carried out from 1991 to 1994 in American centers.49 The two parts of the trial (A and B) carried identical protocols differing only in the timing of the primary outcome collection with primary outcome being at 24 hours for Part A and 3 months for Part B. Patients with ischemic stroke that could be treated within 3 hours of onset of stroke symptoms were included. Of note, half of these were treated between 0 and 90 minutes of onset. Treatment consisted of tPA 0.9 mg/kg or matching placebo. The CT scan in this trial was used to exclude hemorrhage at the time of the trial. Changes other than hemorrhage were not used to exclude patients. The analysis reported in this paper was carried out in 1994 after conclusion and publication of primary trial results. All CT scans were obtained on third or fourth generation CT scanners with 10 mm thick slices. The coordinating center neuro-radiologist reviewed hard copies of the scans centrally. The site investigator supplied clinical information at the time of the initial treatment. The information included demographics, time since stroke onset, localization, and presumed stroke mechanism along with the component scores of the baseline NIHSS. The EIC's were classified into three groups: 1) loss of gray-white distinction 2) hypodensity, and 3) compression of CSF spaces. Visual inspection was used to classify the changes as being either less than one-third of the MCA territory or greater than one-third of the MCA territory. Of the 624 patients randomized in the NINDS Trial, CT scans of 616 (99%) of the patients were available for review. Early infarct changes were associated with a baseline NIHSS ρ=0.23; p<0.001) and time from stroke onset ( ρ =0.11; p=0.007). The correlation in both cases was not strong ( ρ =0.23). Of significant note, the EICs were not correlated with clinical outcomes after adjustment for baseline variables. This included the composite description of the 3-month favorable outcome along with its component measures of mRS Score, NIHSS, Barthel and Glasgow Outcome Score. There is likewise no correlation between the EICs and deterioration at 24 hours, 3-month lesion volume or death within 90 days. Likewise, the presence or absence of the EICs adjusted for baseline NIHSS was not predictive of symptomatic intracranial hemorrhage within 36 hours. These relationships held true whether the EICs composed less than or greater than one third of the MCA territory.

Roberts et al.205 reported on CT findings and implications from the PROACT II Trial.115 PROACT II92 was conducted between 1996 and 1998 and compared treatment of MCA occlusion within 6 hours by IA pro-urokinase coupled with IV heparin and IV heparin alone. The details of this study have been previously described. The current analysis was limited to the 162 patients (108 pro-UK and 54 controlled) who received the treatment. Seventy-five percent of these patients had infarct changes on their baseline CT scan. A neuro-radiologist at the central facility reviewed all CT scans, and baseline CT volume was correlated with clinical variables and outcome.

The baseline CT abnormality volume did not correlate with the baseline NIHSS. There was, however, a modest correlation between baseline CT volume and outcome at 90 days (r=0.17, p=0.05). Twenty-two of 53 (42%) patients with no CT abnormality at baseline reached a mRS less than or equal to 2 at 90 days. This compared to 2 of 8 (25%) of those with baseline CT changes having a volume of greater than 60 mL. Hemorrhagic infarction was present in 42% of the 108 pro-UK group and 29% of the control group at 24 hours. There was a trend toward increased volume of early CT changes and the presence of infarct. The mean volume in pro-urokinase patients with no bleeding was 11.6 ± 2.7 mL. Those with intracranial hemorrhage had a mean volume of 18.8 ± 3.9 mL while those with hemorrhage and clinical deterioration had a mean early infarct volume of 23.3 ± 8.9 mL.

Intervention H: Do Pretreatment MRI Scoring Systems Affect the Safety and Efficacy of Thrombolytic Therapy for Acute Ischemic Stroke?

Six studies were identified that addressed the effectiveness of an MRI scoring system for ischemic stroke. One multiple prospective cohort study101 and one single prospective cohort study100 were included in our review; they were published in 2002 and 2003, respectively (Summary Table 8). Three non comparative case series reports206208 and one case study were excluded for level of evidence.209

Summary Table 8. Intervention H.


Summary Table 8. Intervention H.

Suarez and colleagues101 reported a single-center cohort in which MRI was used to select patients for IA treatment following IV treatment. Patients enrolled had a diagnosis of ischemic stroke whose onset was less than three hours prior to the onset of treatment. The NIH Stroke Scale Score was greater than or equal to 4. An unenhanced CT scan had demonstrated an absence of hemorrhage. The blood pressure was monitored and treated if greater than 180/110. All patients received IV tPA at a dose of 0.6 mg/kg delivered as a 10% bolus over 1 minute with the remainder over 30 minutes. An emergency MRI was conducted subsequently. T1 weighted, T2 weighted, turbo gradient, spin echo, and echo-planar diffusion weighted axial images and axial time-to-peak maps were obtained. They were processed within 6 minutes and interpreted by the neuroradiologist. MRI changes were categorized into four groups: 1) no evidence of infarct 2) evidence of infarct limited to penetrating artery distributions 3) diffusion imaging (DWI) and perfusion-weighted imaging (PWI) mismatches suggesting infarct involving cortical and subcortical areas, or 4) PWI/DWI matched abnormalities suggesting acute infarction involving cortical and subcortical areas. Patients who had no signs of infarct on the MRI or infarcts involving only perforating artery distributions were not treated further. All others had urgent cerebral angiography. If occlusion was demonstrated on the angiography, patients were treated with either urokinase or tPA. This protocol initially started using urokinase, however, switched to tPA after FDA approval of IA tPA for acute ischemic stroke. The urokinase protocol involved an initial dose of 250,000 units repeated up to three times if the vessel did not recanalize. The tPA protocol involved 5 mg repeated until maximum dose of 0.9 mg/kg was achieved over the vessel recanalized.

A total of 2,180 patients were seen at the center during this period of which 554 presented within 3 hours. Forty-five patients met eligibility criteria including consent for this protocol and were considered for angiography. Of these, 21 patients were treated solely with intravenously administered tPA. Seven of these had normal MRI findings while four had evidence of small subcortical defects and two had complete ICA occlusions. One patient exhibited complete improvement prior to angiography, and two patients had normal angiographic results after abnormal MRI results. The mean delay added by MRI imaging was 17 minutes with mean time to complete IA treatment of 282 ± 41 minutes in the urokinase group and 290 ± 38 in the tPA group. For the 24 patients who received IA tPA, the majority recanalized, 18/24. The pre-specified criterion for good clinical outcome was a Barthel Score of greater than or equal to 95 at 3 months. This was achieved by 92% of those in the IA urokinase group (12/13), 64% in the IA tPA group (7/11), and 66% of those in the IV treatment group (14/21).

Hermier et al.100 reported on the use of MRI characteristics employed prospectively to examine the predictive value of early clinical and MR parameters on recanalization in thrombolytic treatment and late infarct volume. Patients were accrued between 2001 and 2002 at a single center in France. Patients with an ischemic stroke within the carotid territory who could receive MRI scanning and IV tPA within 6 hours were included. An NIHSS greater than 4 and an absence of bleeding on unenhanced CT were required. A baseline MRI scan including time-of-flight MR angiograpy, and DWI/PWI, was obtained. Patients with lacunar syndromes determined either clinically or after MRI, were excluded along with hemodynamically relevant stenoses of the extracranial arteries, which might affect time-to-peak analysis. A neuroradiologist who was unaware of the clinical data carried out interpretation of the MRI scan. All patients received IV tPA 0.8 mg/kg. Of 510 patients diagnosed with a stroke, 61 had stroke in the carotid artery and could receive the MRI within 6 hours. Of these patients, 32 met inclusion criteria and received a baseline MRI scan; three of these patients were excluded since their baseline scans were obscured by a motion artifact.

The correlation between the NIH Stroke Scale Score and recanalization at day 1 was demonstrated with a NIH Stroke Scale Score of less than 15 correlating with canalization in the early time frame (p=0.046). The time-to-peak within the DWI lesion on day 0 was likewise correlated with early recanalization. Thirteen of 15 patients (93%) whose baseline time-to-peak was less than or equal to 36.9 milliseconds recanalized within the first day versus 5 of 15 patients (35.7%) whose time-to-peak was greater than 36.9 milliseconds. The NIH Stroke Scale Score and baseline time-to-peak value were the most powerful predictors of recanalization on multi-variate analysis at day 0. Both the day 0 DWI lesion volume and day one recanalization predicted the 60-day infarct volume. Of note, the extent of day 0 DWI/PWI mismatch had non-predictive value for early recanalization. Recanalization was correlated with a better clinical outcome at day 60.

Intervention I: Do CT Perfusion/Angiography Affect the Safety and Efficacy of Thrombolytic Therapy for Acute Ischemic Stroke?

Three studies (four publications) examining CT perfusion/angiography for ischemic stroke were identified.102, 116, 210, 211 One potentially relevant trial210, 211 was published in abstract form and the authors were contacted to determine if subsequent articles were published. These were excluded following full text screening. Study design could not be determined in two publications and were excluded for level of evidence.212, 213 One single retrospective cohort study102 and one case-control study116 were included in our review (Summary Table 9). They were published in 2001 and 2004, respectively.

Summary Table 9. Intervention I.


Summary Table 9. Intervention I.

Agarwal and colleagues reported on the significance of the hyperdense MCA sign in selecting patients for IA versus intravenous thrombolysis in a cohort of patients collected at a single American center between 1996 and 2001.116 Consecutive patients presenting at the center with acute ischemic stroke were considered; those patients who arrived within 3 hours of symptom onset and had no contraindications to tPA, were treated with 0.9 mg/kg following the NINDS protocol. Patients with a contraindication to intravenous tPA or presenting within 3 to 6 hours underwent IA treatment. In these individuals, a #5 French sheath was placed in the right common femoral artery followed by selective catheterization of the occluded cerebral artery. The guidewire was used to provide mechanical disruption followed by the administration of 14 to 20 mg of tPA by an infusion microcatheter. The total dosage was determined by the presence of recanalization or upon reaching the maximum dose of 20 mg. All patients received evaluation for MCA hyperdensity and other changes of early infarction. A hyperdense MCA was defined by: spontaneous visibility of the whole horizontal part of the MCA; density of the MCA higher than that of the surrounding brain; disappearance on bone windows; unilaterality; and, absence of hemorrhage. The M2 dot sign was defined as hyperdensity of an arterial structure seen as a dot in the sylvian fissure. Obscuration of the lentiform nucleus, loss of the insular ribbon, and hemispheric effacement were also assessed. A 24-hour neurologic improvement, defined as a 4-point NIHSS Score improvement from baseline, was evaluated.

During the course of the study, 66 patients were treated with intravenous tPA and 17 by IA tPA. The presence of the hyperdense MCA sign did not predict neurologic recovery in the IA treated patients with three of eight patients who had the sign achieving 24-hour recovery as defined, along with three of nine patients who did not have the sign. The hyperdense MCA sign, however, did predict recovery, with two of 15 patients having the hyperdense MCA sign achieving 24-hour recovery; conversely, 30 of 51 patients lacking the sign achieved 24-hour neurologic recovery (p=0.005). The M2 dot sign, loss of insular ribbon, obscuration of lenticular nuclei, and sulcal effacement, were not predictive of recovery with either intravenous or IA treatment.

The hyperdense MCA sign was associated with a greater probability of recovery with IA than intravenous treatment (37% versus 13%). This observational data116 suggests that this sign may be used as a tool to triage patients between intravenous and IA treatment. There is a probability that proximal large vessel occlusion may be associated with worse outcomes intravenously. This observation will require testing in a prospective study.

Kirpatrick and colleagues102 reported on a retrospective cohort of patients from a single center between 1997 and 2000. These were selected on the basis of an electronic record search seeking all patients within the period of enrollment who would have had a CT scan, CT angiogram and Xenon CT cerebral blood flow within 24 hours of a stroke. The clinical team ordered the studies at the time of the patient's presentation. Primary intent of the study was to see whether abnormalities on these studies or the NIH Stroke Scale were predictive of infarct on the follow-up CT scan. The NIH Stroke Scale was obtained from the record or calculated from the neurologic examination in the record. The CT scan was reviewed by the investigators in a blinded fashion. The CT angiogram was coded based on the report contained within the record. All CT angiograms were performed on a GE Lightspeed scanner with axial helical images obtained from the level of C6 through the circle of Willis with 3 mm collimation. The CTA was defined as patent if there was no report of occluded or stenotic vessels. It was considered occluded if the ICA or MCA on the symptomatic side were reported to be occluded or heavily stenosed. The Xenon CT cerebral flow image was obtained at four levels and mean CBF values were calculated at 20 standardized cortical regions of interest. The scan level containing the lowest average flow in the MCA territory on the symptomatic side was used. This flow was categorized as normal (greater than or equal to 30 mL/100 g/minute), potentially reversible (7 to 29 mL/100 g/minute) or irreversible (less than 7 mL/100 g/minute). The latter two categories were combined due to the small number of patients for statistical analysis. The groups were further subdivided into those examined within six hours of stroke onset and those greater than six hours. The latter group was too small to draw any conclusions. In the group examined prior to six hours (n=31), the NIH Stroke Scale Score was not predictive of infarct on the follow-up CT scan. Normal cerebral blood flow on the Xenon CT resulted in a rate of infarct of 8% (1/13), while abnormal Xenon CT blood flow resulted in a rate of infarct of 55% (6/11). This difference was statistically significant (p=0.023). A CT angiogram showing patent vessels was associated with a rate of infarct of 7% (1/14 patients) while CT angiogram showing occlusion had an infarct rate of 60% (6/10). This too was statistically significant (p=0.008).

Intervention J: Are Community Education Programs Effective in Reducing Stroke-Related Disability and Mortality?

One controlled clinical trial,103 six before-after studies,214219 and one study for which study design could not be determined,220 investigated the use of community education programs for acute stroke. Subsequently, seven studies were excluded for level of evidence.214220 Only one study103 was included for our review (Summary Table 10). This study was a controlled clinical trial and was published in 2003. The before-after studies did not address relevant clinical outcomes and thus we were unable to extract data from these studies. We did however summarize the studies, including population characteristics and limitations in the Appendix. (Appendix E)

Summary Table 10. Intervention J.


Summary Table 10. Intervention J.

Prior to FDA approval of thrombolytic treatment, acute stroke was not perceived as an emergent reason to present to a healthcare facility. Educational programs for the community and healthcare professionals involved in care of patients with acute stroke were deemed essential, as thrombolytic therapy must be urgently administered in order to be beneficial.

The TLL Temple Foundation Stroke Project was established after approval of thrombolytic therapy by the FDA, with the goal to increase utilization of this therapy in a non-urban community in East Texas. Results of the first and second phases of the study have been published221—phase 1 reported on baseline data in the intervention community and the control community and the development of the intervention; phase 2 included an evaluation of the intervention.

Morgenstern et al. reported on the third phase of the TLL Temple Foundation Stroke Project—a quasi-experimental comparison between two communities to determine if the beneficial effect of the Community and Professional Intervention would be sustained after the active intervention had been completed.103 In this report, comparisons are made with the other phases of the project in a before-after design. The financial sponsors of this project mandated that a specific community receive the intervention, and that another, comparable community be chosen by the project investigator. The intervention and its development were published in a companion publication.221 In summary, the intervention was developed based on the process of Intervention Mapping to create a multi-level program that delivered a community communication campaign combined with professional development and organisation change to increase access to stroke therapy in rural east Texas. Target behaviors of lay community (the “at-risk group”), EMS, ED physicians, neurologists, and community primary care providers were identified, and educational and infrastructure changes were given. For example, to target community members, public service announcements were created using local role models, volunteers were trained to take the message to community groups, and educational pamphlets were distributed. Changes made to EMS included assigning a higher priority for the transport of acute stroke patients, development of protocols, and reinforcement and use of mock stroke codes were performed. To target physicians, changes included providing continuing medical education provided, the use of mock stroke codes, and the distribution of newsletters.

Active surveillance by fellowship stroke-trained neurologists, was used to capture all hospitalized stroke cases in the 10 hospitals (five from each community) using a method developed by WHO in the Monica Study.222

In this third phase of the study, 2,184 patients were screened, and 238 validated cases were documented. Baseline demographics of patients from the two communities were different. Patients from the control community had a higher prevalence of co-morbid illnesses compared with the intervention group community: hypertension (87.5% vs. 75.4%); diabetes (48.2% vs. 27.1%); CAD (56.9% vs. 31.5%); AF (68.4% vs. 10.9%); and, previous stroke (65.2% vs. 44.1%). More patients in the control community had a neurological consultation compared with the intervention community (59.1% vs. 45.0%).

The primary outcome was the proportion of patients treated with IV alteplase. This included, nine out of 13 eligible patients (69.2%) in the intervention community, compared with only one of 5 eligible patients (20%) in the control community. The secondary outcome was the proportion of patients who presented to the hospital within 2 hours of symptom onset. There was no significant difference between the two communities—i.e., 28.6% of patients from the intervention community and 22.6% of patients in the control community.

A comparison in both communities over the three phases of this project was reported for internal hospital delay and physician reluctance to utilize tPA in acute stroke. In the intervention community, there was a significant reduction in the proportion of patients who experienced internal hospital delay and a significant reduction in physician reluctance to utilize tPA in acute stroke. In the control community, there was a similar reduction in hospital delay but no change in physician reluctance to utilize tPA over the three phases of this project.

Intervention K: Are Designated Centers Effective in Reducing Stroke-Related Disability and Mortality?

Initially, no studies meeting eligibility criteria for investigating the use of designated centers as defined by the Brain Attack Coalition were identified by our searches. However, two studies104, 105 were included since they provided valuable end points (numbers treated and time to treatment). Both studies were single prospective cohort designs and were published between 2000 and 2003 (Summary Table 11).

Summary Table 11. Intervention K.


Summary Table 11. Intervention K.

It has been hypothesized that to increase utilization of thrombolytics, a dedicated stroke center strategy should be developed.5 The studies we included were felt to most closely resemble the model of a designated stroke center as defined by the Brain Attack Coalition and detailed by Alberts et al. (2000) in their recommendations for the establishment of primary stroke centers.5

Hill et al.104 reported on building a “brain attack” team to administer thrombolytic therapy to patients with acute stroke and on their initial experience with IV-administered thrombolytics. Although thrombolytic therapy was approved for use in Canada in 1999 (the FDA approved its use in the U.S. in 1996), the Calgary Regional Stroke Program received special permission from the local health authority and began a program of open-label thrombolytic therapy for stroke in 1996 following FDA approval for its use in the U.S. The project was initially approved as a pilot study; after 20 patients, an audit was performed to assess safety and complication concerns. The model of care was organized around five essential elements of acute stroke care (all beginning with the letter “R”): Recognition, Reaction, Response, Reveal, and Reperfusion. The major changes that were instituted to achieve success of this program was the funding of a “blocked bed” on the stroke unit i.e., a bed that was always available for a patient to receive thrombolytic therapy enabling rapid removal of patients with acute stroke from the ED. In addition, a campaign was launched to educate the public to increase awareness and “recognition” of stroke symptoms. EMS were asked to change their dispatch protocols to decrease “response” times by elevating acute stroke to a priority one transfer where all patients with stroke-onset symptoms less than 3 hours were preferentially transferred to the stroke center, bypassing other hospitals. The stroke center was contacted by EMS, and the stroke team contacted on patient's arrival to ED thus decreasing the “reveal” time. The ED staff treated stroke as a life-threatening situation, performed a preliminary assessment, and urgently arranged for CT scanning to decrease the time to “reperfusion”.

The initial audit of 20 patients revealed no safety concerns.104 From the inception of the stroke program through to Jan 31, 1999, 69 patients were treated with IV-administered thrombolytics. A 1-year audit demonstrated that 6% of all patients admitted with acute stroke were treated with thrombolytics. Outcome data, reported in a separate publication,223 compared favorably with data reported in published RCTs, with the exception that when patients were treated beyond the 3-hour window, 25% had a symptomatic hemorrhage with 83% of these patients having died by the 90-day follow-up.

Hill et al.104 reported that EMS had treatment times equal to or less than target times recommended by the NINDS stroke study group,224 with a mean time from symptom onset to ED arrival of 55.8 minutes (range 15–125 minutes). Once the patient arrived at the ED, a mean time of 46.1 minutes (range 5–130 minutes) was required to obtain a CT scan and a further mean time of 55.6 minutes (range 20–315 minutes) was required for initiation of thrombolytic treatment. Treatment times improved significantly over the study period—mean treatment time for the first half of the study period was 63.3 minutes compared with 48.6 minutes for the second half of the study. ED to CT and ED to treatment times did not significantly change. Overall symptom onset to treatment time was significantly decreased from a mean of 167.8 minutes to 147.4 minutes.104

Hill et al.104 concluded that if the public can be taught to recognize the symptoms of stroke and react by calling EMS, EMS will promptly get the patient to the hospital. The authors were sobered by the fact that the improvement in time gain overall, was due primarily to getting patients to the hospital faster.

A 4-month pilot study was started at Suburban Hospital in Bethesda MD to establish a stroke center for the region.105 A stroke critical care pathway was developed which incorporated EMS policies, immediate notification of the stroke team, initiation of urgent diagnostic tests, and medical management for thrombolytics. Community education (i.e., lectures emphasizing the symptoms of stroke and need for rapid response by activation of EMS and risk assessment screenings) targeting local community centers, particularly those with senior citizen populations, was provided by the stroke team. Stroke education was provided to ED personnel, hospital personnel, diagnostic imaging, and laboratory staff as well as regional EMS and the local community.

On January 3, 2000, around-the-clock coverage was instituted for acute stroke.105 The stroke-care critical care pathway mandated contacting the stroke team for any patient presenting with a suspected new stroke and persistent deficit <6 hours in duration. Data was collected prospectively until December 31, 2001 for all patients assessed by the stroke team including demographic data, presentation times, treatment times, reasons for non-treatment, radiological times and findings, stroke scale results (NIHSS, mRS), and disposition. Measured time intervals to action (in minutes) were computed as a running 2-month average.105 Results were compared with benchmarks from the Standard Treatment with Alteplase to Reverse Stroke Study (STARS).225

A total of 511 patients were admitted to the hospital with a diagnosis of suspected ischemic stroke; 420 patients had the diagnosis confirmed and 271 arrived within 3 hours of symptom onset.105 Over the 2-year period the following times decreased: time from patient arrival to paging of the stroke team (median of 24 min to 10 min); time from receiving the stroke-team page to arrival of the stroke team (median of 28 min to 6 min); and, time of triage to time of CT (median of 52 min to 42 minutes). The overall median time to treatment from onset was 134 minutes, and the median door-to-needle time was 88 minutes.105

During the 4-month pilot study, four of 117 patients with ischemic stroke were treated with thrombolytics (3.4%).105 Of the 420 patients diagnosed with suspected ischemic stroke over the 2-year study period, there were 44 patients treated with thrombolytics—10.5% of the total and 16.2% of those arriving to hospital within 3 hours of the onset of symptoms.105 Clinical outcomes and time to treatment were similar to that reported by STARS.225

This study demonstrated that a coordinated stroke strategy and development of a stroke center led to a decrease in treatment and investigation times over the investigation phase in addition to increasing the proportion of patients who received thrombolytics.105

Lattimore and colleagues reported the experience of establishing an acute stroke program at a Maryland hospital.105 The community education program associated with this effort consisted of multiple onsite programs with screening efforts, lectures and engagement of local media that were coupled with an intense reorganisation of ED protocols and designation of an acute stroke response time. Delay time prior to the intervention was not reported but an increase in the treated proportion, from 1.5% prior to the intervention to 10.5% during the subsequent 2 years, was noted. No comment was made regarding the relative contributions of the elements of the intervention to the observed outcome. Thus, the relative contribution of the educational effort cannot be separated from the overall effect.

Intervention L: Are ED Protocols for the Management of Acute Stroke Effective in Reducing Disability and Mortality?

After the publication of NINDS,49 proponents have widely advocated for the implementation of “stroke teams” to increase the proportion of patients who are eligible to receive thrombolytic therapy and increase utilization of thrombolytic therapy in this population.49

From our search we identified one case-control study,226 two single prospective cohort studies,118 two single retrospective cohort studies,106, 117 two non-comparative case series studies,227, 228 and two studies whose design could not be determined.229, 230 The case-control and non-comparative studies were excluded for level of evidence. Four studies106, 117, 118 examining the effect of ED protocols for management of acute stroke met our eligibility criteria and were included in our final analyses (Summary Table 12). These studies were published between 1999 and 2003.

Summary Table 12. Intervention L.


Summary Table 12. Intervention L.

Smith et al.106 proposed an ED model for two main reasons. First, EDs were already staffed for rapid assessment, and second, an ED model was already used for administering thrombolytics for acute myocardial infarction in many centers. Smith and colleagues' ED model included guidelines and checklists to evaluate exclusion and inclusion criteria for treatment with thrombolytics, an informed consent package, treatment guidelines including dosing charts, and post-treatment ICU order sets. Prior to implementation, educational efforts using lectures, small group sessions, and written material were directed at the ED physicians, nursing and ancillary staff and other services including Radiology, Neurology, and the ICU. All ED physicians had access to their local neurologist, and at the same time a regional stroke team was also in development.

They performed a retrospective analysis of patients included in their ED model in four teaching hospitals in Michigan. Pharmacy records were used to identify patients treated with thrombolytics and the information was cross-referenced with medical record searches using diagnosis-related group codes used to identify all patients with acute stroke and thrombolytic use. A study physician reviewed each identified record to abstract patient demographics, medical and social history, physical exam findings, complications, length of stay, and stroke severity using the NIHSS. The physician reviewer estimated within 5-point ranges the NIHSS based on documented physical examination in the medical record prior to treatment if the NIHSS was not documented in the medical record. The primary outcome was length of hospital stay, as well as whether the patient was ICH symptomatic or not, pre-hospital time to ED presentation, and ED time data.

Over the study period, 37 patients received thrombolytic therapy. The average time from onset of symptoms to ED arrival was 64 minutes for patients arriving by EMS and 84 minutes for those arriving by car. This data was given for only two of the hospitals that provided care for 24 of the patients. Times for care in the ED were available for 34 of the 37 patients. Thrombolytic therapy began an average of 97 minutes after arrival in the ED and 166 minutes after stroke onset. Neurology consultation was provided for 23 of the 37 patients (nine in person and 14 by telephone). Treatment protocol violations were identified in seven patients, all relating to administration of thrombolytic therapy after the 180-minute time window. Four of the 37 patients developed a symptomatic hemorrhage within 36 hours after treatment but none of these patients were treated outside the 180-minute time window. Neurological outcomes in the 35 survivors at the time of discharge were normal for four patients, improved for 16 patients, unchanged for 10 patients, and worse for five patients. Fifteen patients were discharged home and 15 to a rehabilitation facility.

Time intervals used in the ED model compared favorably to times reported in publications of dedicated stroke teams.231 ED physicians consulted a neurologist in the majority of cases but often only a telephone discussion was required to support facilitating the timeliness of administering treatment.

Akins et al. reported on a study that also evaluated an ED-based protocol for delivering thrombolytic therapy with transfer of patient care to a neurological service when feasible.117 This group established a prospective stroke registry of patients treated with thrombolytics at five community hospitals within the Mercy Healthcare system of Sacramento. Forty-six consecutive patients who received thrombolytic therapy are reported in this study. The local ED physicians developed their own approach to thrombolytic therapy in collaboration with neurologists and radiologists. Clinical information was abstracted from patient records using a standardized form, and clinical outcomes included the mRS score obtained by telephone contact with the patient or family 3-months after treatment. Complete data was obtained on only 43 patients. The dependent variable for analysis was treatment initiation by an ED physician (n=23 patients) or a neurologist (n=20 patients).

There were no differences between the two groups in terms of age, baseline stroke severity (NIHSS), ED presentation to thrombolytic therapy, recovery to normal function, mortality, or symptomatic hemorrhage rates. A trend was detected for a shorter time interval from CT scan to treatment by a neurologist. This was felt to be secondary to the ability of the neurologist to interpret the CT scans.

Initially, during the first year of the protocol, there were more protocol deviations. Thus, a local education program was given to ED staff and local neurologists. As well, nurse stroke-specialist support was implemented in all five hospitals. Seven protocol deviations occurred before the education endeavor compared with only one after the education program was provided. It was not explicitly reported who made the protocol deviations but it is stated in the discussion that it was not surprising that ED physicians would make errors in the dosing of thrombolytics, as it was a markedly different dose than they were conditioned to use for patients with acute myocardial infarction. Details regarding the education program and data collection were not well presented; hence, an independent assessment of the effectiveness of the program was not feasible.

Jahnke et al. reported on their experience with ED stroke teams and acute stroke pathways, over a 2-year period.118 This group recognized a problem in achieving target times for thrombolytic treatment after a retrospective review that determined that the majority of potential candidates for treatment arrived in the ED more than 2 hours from symptom onset. They developed a stroke team and implemented a written stroke pathway. Prior to implementation, the emergency personnel were educated using group sessions and posters displayed in the ED. The pathway included a standardized set of orders and instructions for the management of acute ischemic stroke. Listed on the pathway were universal stroke symptoms to help staff recognize eligible patients. The pathway gave consideration to the time of stroke onset to determine potential eligibility of the patient for thrombolytic therapy, and urgent activation of the stroke team would occur for patients who presented less than 6 hours from onset of symptoms or non-urgent calls for patients with transient ischemic stroke or symptom with onset >6 hours. The ED physician assessed all patients, and a member of the stroke team would assess and calculate the NIHSS. Documentation of times was performed to allow rating of performance of the stroke team.

Implementation of the pathway identified a number of strategic changes that needed to perform to ensure success. This included identification of two levels of stroke team referral, urgent and non-urgent patients. Also, a major delay identified was time to obtain a CT scan; the CT scanner was geographically distant from the ED and this led to placement of a CT scan in the ED. Another delay identified was time to obtain laboratory test results. Again the laboratory was geographically distant and point of service laboratory equipment was obtained for the ED.

Over the 2-year period, 71 patients were eligible for thrombolytic therapy; 65 patients received treatment and six patients declined. Of the patients who received treatment, one patient died, two required nursing home level of care, 25 were transferred to rehabilitation and 37 were discharged home. Comparing the first year to the second year, there were significant decreases in the following time intervals. First, time from presentation in the ED to ED physician assessment decreased from an average of 33 minutes to 7 minutes. Second, time from presentation to CT scan order decreased from an average of 38 minutes to 7 minutes. Lastly, time from presentation in the ED to completion of the CT scan decreased from an average of 88 minutes to 44 minutes. Compliance for initiating the stroke pathway increased from 40% to 97% of eligible patients, and “door-to-needle” time for the administration of thrombolytic therapy decreased from an average of 111 minutes to 77 minutes. This group still hopes to achieve the NINDS 60 minute goal.

Results of Meta-Analyses

Meta-analyses could only be performed on two of the interventions examined (Intervention A and Intervention C).

Intervention A: Does Surgery Impact the Outcome in Patients with Acute Intracerebral Hematoma?

Death and Disability: The meta-analysis for the mRS included four eligible studies (Figure 2).88, 89, 119, 120 Two studies87, 121 were excluded from our disability analysis (mRS score 0 to 1) because data could not be extracted. Juvela et al.121 reported morbidity of survivors at 6 months after hemorrhage for both groups using an independently derived scale which describes patients as either independent (minimally or moderately disabled) or dependant (severely disabled or vegetative). Morgenstern et al.87 reported median values (Barthel <61) at 6 months that identified patients with either poor outcome or death. The four trials had a total of 246 subjects (range, n=20 to n=100). Note that the Batjer study120 made no contribution to the pooled estimate since in both groups all patients experienced death or severe disability.

FIgure 2. Meta-analysis of the impact of surgery on death and disability in patients with acute intracerebral hematoma.


FIgure 2. Meta-analysis of the impact of surgery on death and disability in patients with acute intracerebral hematoma.

Note that of the three studies that contributed to the above meta-analysis, there was substantial heterogeneity, therefore, the random effects pooled estimate (and associated confidence interval) should be interpreted with caution.

Death: The meta-analysis for death outcomes included six eligible studies (Figure 3).8789, 120, 121, 121 Summary Table 1 includes a description of eligible trials. All studies examined the effectiveness of surgical intervention compared with standard or usual care. The six studies had a total of 298 participants (range, n=20 to n=100). Follow-up interval for studies was 90 days for all studies.

Figure 3. Meta-analysis of the impact of surgery on death in patients with acute intracerebral hematoma.


Figure 3. Meta-analysis of the impact of surgery on death in patients with acute intracerebral hematoma.

Two reports examining surgical intervention for surgical intervention for intracerebral hemorrhage (a meta-analysis published in 2002233 and randomized trial published in 2005122 were also identified. The meta-analysis included two studies which were not captured by our searches. One by McKissock et al. was published prior to our search date of 1964. The second, by Chen (1992) was a non-English language citation, and hence, was excluded by our search strategy. The Surgical Trial in Intracerebral Haemorrhage (STICH)122 was published beyond our search date, but was identified by a reviewer. With regards to the Chen 1992 trial, we were able to extract enough mortality data from this meta-analysis233 to use as a sensitivity check in our current meta-analysis, as well as from the STICH trial.122 (Figure 4). We could not extract data from the STICH trial for death and disability outcomes (mRS) because they used a prognosis-based modified Rankin index which did not permit us to recover the modified Rankin index.

Figure 4. Meta-analysis of the impact of surgery on death in patients with acute intracerebral hematoma including data from the Chen RCT (data extracted from a meta-analysis published in 2002) and STICH trial.


Figure 4. Meta-analysis of the impact of surgery on death in patients with acute intracerebral hematoma including data from the Chen RCT (data extracted from a meta-analysis published in 2002) and STICH trial.

Regardless of whether the Chen study is included, the pooled estimate is not statistically significant. There was, however, a moderate degree of heterogeneity in study outcomes. Note that only the study by Auer119 showed a statistically significant benefit from surgery.

Intervention C: Does IA Thrombolysis Reduce Stroke-Related Mortality and Disability in Adults with Acute Stroke?

Death and disability. Two studies examining the use of IA tPA for acute ischemic stroke were included in our meta-analysis (Figure 5).92, 110 The two trials included 220 subjects. Note that in this case, there is no detectable statistical heterogeneity, and the random effects model provides identical results to a fixed effects model.

Figure 5. Meta-analysis of the impact of IA thrombolysis on death and disability in patients with acute ischemic stroke.


Figure 5. Meta-analysis of the impact of IA thrombolysis on death and disability in patients with acute ischemic stroke.

Death: The meta-analysis for death outcomes associated with the use of IA tPA for acute ischemic stroke included two studies (Figure 6).92, 110 The two trials included 220 subjects. Note that in this case, there is no detectable statistical heterogeneity, and the random effects model provides identical results to a fixed effects model.

Figure 6. Meta-analysis of the impact of IA tPA on death in patients with acute ischemic stroke.


Figure 6. Meta-analysis of the impact of IA tPA on death in patients with acute ischemic stroke.

Statistical significance of pooled results. Neither of the pooled estimates was statistically significant. Note, however, that the wide confidence intervals of the pooled estimates do not rule-out the possibility of substantial benefit from IA thrombolytic therapy.


  • PubReader
  • Print View
  • Cite this Page

In this Page

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...