A Regional System of Stroke Care Provides Thrombolytic Outcomes Comparable With the NINDS Stroke Trial

Introduction 

Tissue plasminogen activator (tPA) was approved by the Food and Drug Administration in 1996, after a randomized clinical trial conducted by the National Institute of Neurological Disorders and Stroke (NINDS) found that patients who received tPA were at least 30% more likely to have minimal or no disability at 3 months than were patients given placebo.1 However, tPA therapy remains controversial for a variety of reasons. Primary concerns relate to short-term risk of hemorrhage and the applicability of this therapy in community practice. The use of well-organized hierarchic systems of acute stroke care have been proposed to link community hospitals to comprehensive stroke centers nationally, but data from such systems, including functional outcomes, have been limited. Evaluation of the stroke literature reveals several summaries of data collected at single sites or in regional, statewide, or international registries.2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 To provide context, selected phase IV studies using various methodologies and reporting outcomes for tPA-treated patients are summarized in Table 1.

Table 1. Selected studies of acute stroke treatment outcomes.

	Study	No. of Centers	No. of Patients Treated With Intravenous tPA	Baseline Severity	Symptomatic Intracerebral Hemorrhage, %	3-Month Outcome (% mRS ≤1)
	Oregon: Egan, 19993	6	33	16.6	9.1	36
	Sacramento: Akins, 20002	5	43	14	7	42
	Cleveland: Katzan, 20007	29	70	12	15.7	NR
	STARS: Albers, 20008	57	389	13	3.3	NR (1-mo outcomes, 35)
	Illinois (OSI): Wang, 20009	20	57	14	5	NR
	Houston: Grotta, 200110	4	269	14	5.6	NR
	Germany: Koennecke, 20014	1	75	(Median) 13	2.7	40
	Indianapolis: Lopez-Yunez, 200111	10	50	11	10	NR
	Melbourne: Szoeke, 20035	1	30	14	7	37
	Cleveland: Katzan, 200312	9	47	NR	6.4	NR
	Carolina Medical Center: Asimos, 20046	1	60	13.5	10	60 for BI ≥95
	CASES (Canada): Hill, 200513	60	1,135	14	4.6	37
	Cleveland: Dick, 200514	3	101	11	2	NR
	Michigan: Deng, 200615	16	43	12	4.6	NR

mRS, Modified Rankin scale; NR, not reported; BI, Barthei Index. These studies have significant differences in methodology. They are listed here to provide an overview of reported tPA treatment outcomes.

In this article, we report the 3-month outcomes of patients treated with tPA through our statewide emergency network during a 9-year period and compare them with the results from the NINDS trial.

Materials and Methods 

Our emergency network for acute stroke care integrates stroke specialty consultation with emergency medical care throughout a wide region.16 The Brain Attack Center is located within our urban academic medical center and is staffed by university-based stroke specialists (the Brain Attack Team), who provide consultations with emergency departments (EDs) in Maryland, south central Pennsylvania, West Virginia, Delaware, and the District of Columbia. The network spans a radius of more than 150 miles, encompassing more than 50 hospitals.

The stroke care system can be activated by an emergency physician in any hospital in the region, who determines that a patient is a possible candidate for lytic therapy. According to protocol, the ED staff ascertains the time of symptom onset and identifies examination findings consistent with stroke to request initial diagnostic studies to ready the patient for acute therapy. Computed tomography (CT) scans of the brain are interpreted by radiologists at the referring hospital. When examination and laboratory results and brain CT findings are consistent with the clinical diagnosis of ischemic stroke, the Brain Attack Team attending physician is contacted through our network coordinating center (Maryland ExpressCare). The coordinators establish an immediate teleconsultation link between the referring emergency physician and the on-call stroke specialist. Teleconsultation was provided by 1 of 2 means: telephone or real-time audiovisual connection between our stroke specialist and a care provider in an ED linked to our brain attack center.16

Patients who are potential thrombolytic candidates are transferred to our stroke center for care. Initiation of intravenous tPA may begin before transfer according to the patient presentation, time of symptom onset, and expertise of the referring ED providers. Maryland ExpressCare coordinators work with the communications component of the Maryland Institute for Emergency Medical Services Systems and resources from our university medical center to establish air or ground transportation, ensure that a bed is ready for the patient being transferred, and facilitate the patient registration process. For patients who are eligible for tPA, treatment can begin at the referring hospital to shorten the time to definitive treatment.

In this study, only stroke neurologists, acute stroke nurse practitioners, and stroke neurology fellows (those completing a stroke neurology fellowship during the period of this study) provided Brain Attack Team consultation. The stroke neurologists included 2 groups. Traditional stroke neurologists received their fellowship training before the tPA era. Acute stroke neurologists trained during the tPA era. Stroke neurology fellow and nurse practitioner fellowship education and supervision followed a standardized training protocol using didactic and observational strategies in addition to evaluation by assessment.17 To further standardize our acute stroke practices, established algorithms, protocols, and order sets were followed by each member of our team.

Using the NINDS stroke trial protocol for intravenous tPA administration, Brain Attack Team consultants confirmed the time of symptom onset, medical history, neurologic examination findings, National Institutes of Health Stroke Scale (NIHSS) score, and indications for and contraindications to tPA therapy. When possible, consultants verified the information with a family member or witness of the onset of the event. Although the NIHSS was scored by each referring emergency physician at the bedside, the NIHSS-certified neurology consultant provided step-by-step confirmation and documentation of scoring to ensure interrater reliability used later for statistical analysis. The baseline NIHSS score used for this study was the NIHSS score that was verified by the Brain Attack Team consultant. The consultant reviewed the technical aspects of intravenous tPA administration with the referring physician and nurse, including verification of total tPA dose, bolus and maintenance infusion doses and rates, and blood pressure parameters. All patients were transferred to our stroke center and admitted to our medical or neurocare ICU, where treatment continued according to an acute stroke protocol modeled after that used in the NINDS trial.

All consultations with the Brain Attack Team are recorded in a prospective database, which is maintained for treatment and operations purposes. As a part of our quality improvement process, all patients who receive tPA are evaluated for adverse events and neurologic outcome at 3 months with the modified Rankin scale, one of the measures used in the NINDS stroke trial.1 After receiving institutional review board approval of our study protocol, we interrogated the Brain Attack Team consultation database for all patients who received intravenous thrombolysis between August 1, 1996, and December 31, 2005. We reviewed medical records for details of treatment, including stroke type, time of symptom onset, time of arrival, time to treatment, baseline NIHSS score, 3-month modified Rankin scale score, evidence of symptomatic intracranial hemorrhage, and demographic information. These data points were consistent with those used in the original NINDS trial, allowing comparison of outcomes.

Results 

Consultations for a total of 2,670 patients were conducted between 1996 and 2005. Eight hundred eighty-two consultations were deemed to be nonacute ischemic stroke diagnoses and therefore not amenable to tPA therapy (including subarachnoid hemorrhage, intracerebral hemorrhage, trauma, seizure, nonacute stroke, stroke mimic diagnoses). The other 1,788 patients were determined to be experiencing ischemic stroke by the Brain Attack Team specialist at consultation. Two hundred forty patients (13.4% of those with acute ischemic stroke; 9% of all consultations) met the criteria for administration and received intravenous tPA (Figure 1). The mean age of the patients receiving tPA was 68 years (SD=14), and the mean pretreatment NIHSS score was 12.9 (SD=6.2; median=12.0). Patient characteristics are summarized in Table 2.

View Large Image Download to PowerPoint

Figure. Treatment pathways resulting from 2670 consultations for possible ischemic stroke provided through a statewide stroke care network between 1996 and 2005. Sixty-seven percent (n = 1788) of patients were deemed eligible for tPA.

Table 2. Characteristics of patients who received intravenous tPA between 1996 and 2005.

	Variables	No. (%) With Various Characteristics
		Everyone (N=240)	No. With mRS Score at 3 Months (n=205)	No. Without mRS Score at 3 Months (n=35)
	Year
	1996–1999	39(16)	37(18)	2(6)
	2000–2002	68(28)	58(28)	10(29)
	2003–2005	133(55)	110(54)	23(66)
	Time between onset and treatment (min)
	0–120	47(20)	35(17)	12(34)
	121–225	192(80)	170(83)	22(63)
	Unknown	1(<1)	0	1(3)
	Baseline NIHSS score
	0–5	34(14)	30(15)	4(11)
	6–10	59(25)	46(22)	13(37)
	11–15	59(25)	53(26)	6(17)
	16–20	52(22)	45(22)	7(20)
	≥21	31(13)	29(14)	2(6)
	Unknown	5(2)	2(1)	3(9)
	Age, y
	0–55	48(20)	38(19)	10(29)
	56–65	54(23)	45(22)	9(26)
	66–75	54(23)	49(24)	5(14)
	≥76	82(34)	73(36)	9(26)
	Unknown	2(1)	0	2(6)
	Sex
	Male	146(61)	122(60)	24(69)
	Female	93(38)	83(40)	10(29)
	Not documented	1(<1)	0	1(3)
	Stroke type
	Large artery(ICA, MCA, basilar)	177(74)	153(75)	24(69)
	Other	59(25)	51(25)	8(23)
	Unknown	4(2)	1(<1)	3(9)

Among the 240 treated patients, intravenous tPA was administered in an outlying ED, with teleconsultation support provided by a Brain Attack Team consultant, in 82 cases (75 by telephone consultations and 7 by real-time audio/video telemedicine consult). One hundred fifty-eight patients received tPA in the ED at the medical center where the Brain Attack Team is based after transfer from an outlying hospital, direct arrival by ambulance, or direct ED admission from home (“walk-ins”).

The average time from symptom onset to treatment was 149 minutes (SD=34 minutes); 47 patients (20%) were treated within 2 hours, 175 (73%) were treated 2 to 3 hours after onset, and 17 (7%) were treated 3 to 4 hours after onset. The average door-to-needle time was 62 minutes (SD=34 minutes); 121 (51%) were treated within 1 hour, 103 (43%) were treated 1 to 2 hours after arriving, and 14 (6%) were treated 2 to 3 hours after arrival. In 2 cases, information about the timing of onset or arrival was missing.

Three-Month Patient Functional Outcomes 

Three-month outcome data for 205 of the 240 tPA recipients (85%) were available for analysis. The remaining 15% were lost to follow-up because of changes in telephone contact information and missed follow-up appointment. Excluding those with missing outcome data, 53% (108/205) achieved a modified Rankin scale score of less than or equal to 1 at 3 months after tPA administration (95% CI 46% to 60%). Patients with missing modified Rankin scale scores tended to be treated more recently (in 2000 or later), tended to have shorter onset-to-treatment times, and tended to be younger. To address our rate of 15% lost to follow-up, we performed a sensitivity analysis, assuming an unfavorable neurologic outcome (modified Rankin scale score of 6) in those missing patients. This analysis revealed that 108 of 240 (45%) (95% CI 39% to 51%) achieved good functional outcome, as defined by a modified Rankin scale score of 0 or 1.

We performed a stratified analysis to evaluate clinical outcomes in relation to year of treatment, time to treatment, baseline NIHSS score, age, sex, stroke type, and provider type (Table 3). A strong association was found with regard to baseline NIHSS score and a trend with regard to patient age.

Table 3. Proportion (%) with 3-month modified Rankin scale score ≤1, overall and by group.

	Subgroups	Proportion (%) With mRS Score ≤1.0	95% CI for Percentage With mRS Score ≤1.0	P
	Overall
	Missing data excluded⁎	108/205(53)	46–60
	Missing data assigned mRS=6	108/240(45)	39–51
	Year			.54
	1996–1999	21/37(57)	39–73
	2000–2002	33/58(57)	43–70
	2003–2005	54/110(49)	39–59
	Time between onset and treatment (min)			.19
	0–120	22/35(63)	45–79
	121–225	86/170(51)	43–59
	Baseline NIHSS score			.0009
	0–5	22/30(73)	54–88
	6–10	33/46(72)	56–84
	11–15	22/53(42)	28–56
	16–20	17/45(38)	23–54
	≥21	14/29(48)	29–68
	Age, y
	0–55	25/38(66)	48–81	.032
	56–65	29/45(64)	48–78
	66–75	23/49(47)	32–62
	≥76	31/73(42)	31–55
	Sex			.028†
	Male	72/122(59)	50–68
	Female	36/83(43)	32–55
	Stroke type (one missing)			.47
	Large artery (ICA, MCA, basilar)	78/153(51)	42–59
	Other	29/51(57)	42–71
	Provider type			NS
	Acute stroke neurologists (2)	45/79(57)	45–68
	Other stroke neurologists (4)	18/39(46)	28–61
	Acute stroke nurse practitioners (3)	27/48(56)	41–71
	Acute stroke neurologist fellows (4)	18/39(46)	28–61

⁎ mRS score was missing for 35 of the 240 patients. † Men are a bit younger in this sample (mean age among men is 66 years vs 71 years for women). Adjusted for age, the P value for the association between sex and mRS score of 1 becomes .098.

Safety Outcomes 

Information on symptomatic intracranial hemorrhage was available for all 240 patients. Eight of them (3.3%) had symptomatic intracranial hemorrhage after treatment with intravenous tPA (95% CI 1% to 7%). Of the patients who received tPA at a remote ED by teleconsultation, 5 of 82 (6%; 95% CI 2% to 14%) experienced symptomatic intracranial hemorrhage versus 3 of 158 (1.9%; 95% CI <.4% to 6%) who received tPA after face-to-face consultations (P = .13). No patients (0 of 7 [0%]; 95% CI 0% to 41%) treated after real-time audio and visual teleconsultation experienced symptomatic intracranial hemorrhage. Symptomatic intracranial hemorrhage rates and predictors are listed in Table 4.

Table 4. Proportion (%) with ICH, overall and by group.

	Subgroups	Proportion (%) With ICH	95% CI for Percentage With ICH
	Overall	8/240(3.3)	1–7
	Year
	1996–1999	2/39(5)	<1–18
	2000–2002	1/68(1)	<1–8
	2003–2005	5/133(4)	1–9
	Time between onset and treatment (min), n=239
	0–120	0/47(0)	0–8
	121–225	8/192(4)	2–8
	Baseline NIHSS score (n=235)
	0–5	0/34(0)	0–11
	6–10	1/59(2)	<1– 9
	11–15	2/59(3)	<1– 12
	16–20	1/52(2)	<1–10
	≥21	3/31(10)	2–26
	Age, y (n=238)
	0–55	1/48(2)	<1–11
	56–65	1/54(2)	<1–10
	66–75	2/54(4)	<1–13
	≥76	4/82(5)	1–12
	Sex
	Male	4/146(3)	1–7
	Female	4/93(4)	1–11
	Stroke type (n=236)
	Large artery (ICA, MCA, basilar)	7/177(4)	2–8
	Other	1/59(2)	<1–9
	Provider type
	Acute stroke neurologists (2)	3/91(3)	<1–10
	Other stroke neurologists (4)	5/55(9)	3–23
	Acute stroke nurse practitioners (3)	0/55(0)	0–7
	Acute stroke neurologist fellows (4)	0/39(0)	0–14

Data were compared between acute stroke–trained providers (ie, acute stroke neurologists, acute stroke nurse practitioners, and acute stroke fellows) and traditional stroke neurologists, as defined above. Patients treated by stroke specialists trained in acute lytic procedures (acute stroke neurologists, nurse practitioners, and stroke neurology fellows) had fewer postlytic intracerebral hemorrhages but no significant difference in 3-month outcomes compared with patients treated by neurologists who did not have formalized training in lytic therapy (Table 3, Table 4).

Comparison of Results With NINDS Trial Outcomes 

The average age of the patients in both studies was 68 years. Baseline NIHSS scores for the 2 groups are compared in Table 5.

Table 5. Comparison of current and NINDS patients (treatment group) with respect to NIHSS scores.

	Baseline NIHSS score	Current Study, No. (%) (n=240)⁎	NINDS, No. (%) (n=310)†
	0–5	34(14)	42(14)
	6–10	59(25)	67(22)
	11–15	59(25)	65(21)
	16–20	52(22)	73(24)
	≥21	31(13)	63(20)

⁎ NIHSS scores were available for 235 of the 240 patients in the current study. † Two patients in the original NINDS groups were excluded from analysis because their treatment was started more than the protocol-specified 180 minutes after stroke onset.17

The percentage of our patients who had a 3-month modified Rankin scale score of less than or equal to 1 (108/205 [53%]) compares favorably with the percentage of such patients in the NINDS treatment group (133/312 [43%]) (P=.04). The 95% CI for the difference between these 2 rates is 1% to 19%. When we adjusted our patients' rate of successful outcome at 3 months (modified Rankin scale score ≤1) to the NINDS baseline NIHSS distribution, the rate of success did not change (53%).

With regard to safety, our rate of symptomatic intracranial hemorrhage was 3.3% (8/240; 95% CI 1% to 7%) versus 6.4% (20/312, 95% CI 4% to 9%) of the tPA recipients in the NINDS trial (P=.15). A 95% CI for the difference between these 2 rates is –6.6% to 0.5%, ie, the data are consistent with the possibility that the rate of hemorrhage in the network is the same as that in the NINDS treatment group, but also with the possibility that the rate is 6.6 percentage points lower. Mortality at 3 months among tPA-treated patients was 13% in this network study versus 17% in the NINDS treatment group.

Limitations 

Our data represent a sample of patients for whom stroke specialty consultation was sought voluntarily by emergency medicine providers within our statewide and surrounding areas. Although this is a relatively large experience compared with many other published reports, the number of patients, adverse events, and the chart review design pose limitations. Therefore, application of our data may be limited to a similarly selected population of acute stroke patients who received stroke specialty consultation. A prospective study would be required to verify the findings and expand the generalizability of results from a comparable well-organized regional stroke system and to explore other questions of interest. For instance, questions about possible miscategorization of symptomatic and asymptomatic hemorrhages would best be clarified by prospective study and CT imaging of all acute stroke patients receiving lytic therapy. Our study may have underestimated symptomatic hemorrhages; however, we attempted to minimize this possibility by routine serial NIHSS scoring throughout tPA-treated patients' hospitalization. We believe the effect of this limitation is countered by serial scoring and matching our definition of symptomatic hemorrhage with that in the NINDS study. Additionally, we did not perform anatomic scoring of ICH, but we do not believe this would be a significant limitation to our conclusions, because all symptomatic hemorrhages of any size were captured as a primary safety outcome.

Our attrition rate of 15% is similar to that from other stroke trials19 but represents a potential limitation because losses to follow-up may not be random events. When we analyzed our data so that all 35 patients with missing 3-month outcome data were assigned the most unfavorable modified Rankin scale score (6), 45% of the patients (108/240) achieved a good outcome (modified Rankin scale score ≤1).

Discussion 

Our emergency network approach to acute stroke care is both effective and safe in light of results obtained within the context of the NINDS clinical trial. To date, this is the largest and most complete data set including functional outcomes from a well-organized regional stroke system.

We believe that our efficacy data and low rate of posttreatment symptomatic intracranial hemorrhage (3.3%) are due in part to our continuous efforts to maintain a cohesive and experienced emergency medicine and stroke specialist team at our hub stroke center. This ensures that interventions are rendered in the shortest possible time with the support of an experienced group of stroke specialists.

We believe that the development of our model of regional stroke care can be applied more generally to other states. To facilitate a statewide approach for stroke care, helicopter transfer was necessary for a subpopulation of our patients. When we were developing our statewide stroke network, some people were concerned by the perception that interfacility transfer using air medical services might delay the care of the ischemic stroke patient, resulting in worse outcomes. To investigate this concern, we analyzed time to treatment, tPA administration rates, and 3-month modified Rankin scale scores for patients transferred from another hospital by helicopter air medical services and those arriving directly at our medical center by ambulance or walk-in. The air-transferred patients (ie, those hypothesized to have a delay in time to treatment, difference in tPA administration rates, and worse 3-month modified Rankin scale scores) and those arriving directly were comparable for sex, age, baseline, and stroke severity. There was no difference in tPA administration rates between the 2 groups (25% versus 17%; NS).20 Analysis of the time from symptom onset to treatment revealed that air and ground transports spanned 208±90 and 187±148 minutes, respectively. Additionally, 3-month outcomes of interfacility transfer patients transported by helicopter were statistically better than those for ground ambulance-transported patients: 36 of 48 helicopter-transported patients reached a good outcome versus 21 of 41 ground-ambulance-transported patients (P=.035). Although this significant difference in 3-month outcomes may be related to referral bias, these data provided reassurance that any delay during interfacility transport was not a substantial risk to patient outcome.

During the study period, acute stroke nurse practitioners, stroke neurology fellows, and neurologists provided around-the-clock, on-call, acute stroke consultation coverage. The nurse practitioners and fellows had identical training and supervision by attending physician faculty members. Our evaluation among providers is part of our brain attack center's continuing performance improvement initiative. Our data serve to support the incorporation of well-trained acute stroke nurse practitioners and fellows as primary acute stroke consultants within a regional system. The analysis suggested a difference in symptomatic intracranial hemorrhage rates between acute stroke–trained providers and traditional neurologists. We attributed this to the fact that acute stroke neurologists had greater experience with lytic therapy during their training period, and we addressed these differences through a quality improvement educational process. As a result, in 2006, among all specialists, no patient (0/41 patients) experienced symptomatic intracranial hemorrhage after tPA therapy for acute ischemic stroke. We believe every brain attack center should have an established process for ongoing surveillance of efficacy and safety data.

Our rate of tPA administration has increased steadily since the year of inception of our emergency medicine stroke network while low rates of complication have been maintained. Factors critical to our success include the following: (1) the expertise of our network of emergency medicine providers, (2) the stability of our core of acute stroke specialists, (3) our continuing dialogue about protocol, (4) continuous quality assurance checks by the program director, and (5) our commitment to delivering educational lectures to referring hospital EDs on a rotating schedule throughout the region.

Many factors contribute to our good 3-month outcomes, such as intravenous tPA administration or admission to our regional stroke center for care.21 Although it is difficult to interpret the data specific to the reason for good outcomes, our observations suggest that care by an expert team increases the rate of tPA administration and improves patient outcome.

Conclusions 

Our study provides a unique opportunity to explore the generalizability of results from the original research as applied in a model pragmatic, well-organized, hierarchic, regional stroke system. The network's outcomes for tPA-treated patients compare favorably with those reported from the NINDS stroke trial. According to these results, we support the establishment of similar regional systems for the care of acute stroke patients.

Summary 

Administration of tissue plasminogen activator (tPA) for acute ischemic stroke remains controversial in community practice. Well-organized hierarchic systems of acute stroke care have been proposed to link community hospitals to comprehensive stroke centers. We report safety and functional outcomes in patients treated with tPA in our regional emergency stroke network and compare them with results reported from the trial conducted by the National Institute of Neurological Disorders and Stroke (NINDS).