Intravenous administration of tissue plasminogen activator (alteplase) is the primary treatment of acute ischemic stroke in eligible patients.Reference Adams, del Zoppo and Alberts 1 , Reference Casaubon, Boulanger and Blacquiere 2 Clinical benefits of stroke thrombolysis depend on the timeliness of alteplase bolus delivery, and earlier treatment has been shown to reduce long-term functional disability, in-hospital mortality, and symptomatic intracranial hemorrhage, and to lower the risk of other medical complications.Reference Saver, Fonarow and Smith 3 Thrombolysis is most effective within 90 minutes of onset-to-needle (OTN) time,Reference Lees, Bluhmki and von Kummer 4 and there is no evidence of benefit beyond 4.5 hours after symptom onset.Reference Lansberg, Schrooten, Bluhmki, Thijs and Saver 5 Reducing the door-to-needle (DTN) time is therefore an essential objective in optimizing hyperacute stroke care.
Interventions aimed at minimizing stroke thrombolysis delays have been suggested in numerous publications,Reference Meretoja, Strbian, Mustanoja, Tatlisumak, Lindsberg and Kaste 6 - Reference Fonarow, Zhao and Smith 10 one of which reported that DTN time was nearly halved following protocol changes.Reference Meretoja, Weir and Ugalde 11 Among the most cited strategies are stroke team prenotification by emergency medical services (EMS), rapid head computed tomography (CT) scan acquisition and alteplase delivery on the CT table.Reference Meretoja, Weir and Ugalde 11 In most of these studies, quality improvement modifications were implemented over a relatively long period, and the long-term persistence of these modifications on DTN time was not presented.
To conform to Canadian guidelines recommending a DTN of ≤60 minutes,Reference Lindsay, Bayley and Smith 12 we implemented some of the most cited DTN time reduction strategies in our acute stroke protocol. To rapidly offer the recommended delays to our patients, these modifications were all applied within 1 month. The primary objective of this quality-improvement, single-center study was to assess the short- and long-term effects of these modifications on our median DTN times and the proportion of patients receiving intravenous alteplase within 60 minutes. Our secondary objective was to compare stroke mimic and symptomatic intracerebral hemorrhage (sICH) rates in both study phases.
METHODS
Health Care Setting
Canada covers most medical services including acute stroke care through a publicly funded health insurance program. During the period covered in this study, in the province of Quebec, EMS transported patients to the nearest hospital and, if necessary, subsequently transferred them to a tertiary center for more specialized treatment such as intravenous alteplase. Notre-Dame Hospital (Centre Hospitalier de l’Université de Montréal) is a tertiary academic center offering thrombolysis to patients residing in the greater Montreal area and parts of the north and south shores of the island-city (with a catchment of approximately 1.5 million inhabitants).
Participants
This study used a prospective registry of consecutive thrombolyzed ischemic stroke patients admitted to Notre-Dame Hospital since 2001. From this registry, we included patients admitted directly to our hospital between January 2012 and March 2015, and excluded individuals transferred from our hospital’s inpatient units or from other hospitals. There was no registry for nonthrombolyzed individuals. Included patients were categorized, based on the date of the stroke event, into the premodification phase (January 1, 2012-May 31, 2013) and the postmodification phase (July 1, 2013-March 31, 2015). June 1 to 30, 2013, constituted the implementation and transition phase. To separate the short- from the long-term effects of our interventions and assess the durability of our quality improvement initiative, the postmodification phase was subdivided into the early postmodification phase (July 1, 2013-December 31, 2013) and the late postmodification phase (January 1, 2014-March 31, 2015). We obtained institutional ethics review board approval to maintain the registry and to conduct this study.
Premodification Hyperacute Stroke Care Model
Until May 2013, rapid notification to the stroke team via a single-call activation system (stroke pager) was enabled by the emergency department (ED) physician after initial assessment of the patient presenting with stroke symptoms. A stroke toolkit was available in the ED and included preprinted CT and laboratory order sets. Approach to care included the following: a system of parallel patient history-taking and neurological evaluation by the resident and neurologist during transport to the CT room, patient blood-drawing upon admission, priority access to standard stat laboratory testing for quick results before treatment, and head CT scan evaluation by stroke neurologist without waiting for radiology report. Stroke patients were followed 3 months postdischarge as part of routine care.
Postmodification Hyperacute Stroke Care Model
In 2013, we completed a quality assessment of the previous model and developed a new model aimed at achieving a median DTN time of ≤60 minutes. Literature describing strategies to improve in-hospital treatment delays were reviewed and locally applicable models were retained. A working group of stroke neurologists, neurology residents, stroke nurses, neuroradiologists, and emergency physicians discussed the feasibility of these strategies and agreed on a restructured management algorithm.
The following changes were made over the course of 1 month: EMS prenotified triage, which immediately alerted the stroke team before EMS hospital arrival via a single-call activation system upon prehospital suspicion of stroke; laboratory results other than a capillary blood glucose level were not waited on to make treatment decisions (except in specific clinical scenarios such as known anticoagulation); the patient was directly transferred from EMS to CT table upon arrival instead; a dedicated key granted access to reserved elevators for faster transfer to CT scan; an in-house radiology technician was made available 24/7; alteplase bolus was administered on the CT table after plain CT; and, in addition to regular informal feedback, treatment delays for all patient were posted in the hyperacute stroke care area to allow direct feedback team-member validation. Table 1 compares strategies used in the pre- and postmodification models.
Table 1 Hyperacute stroke care models during the pre- and postmodification phases (changes in bold type)
CT, computed tomography; ED, emergency department; EMS, emergency medical services; INR, international normalized ratio; tPA, tissue plasminogen activator.
Statistical Analysis
To complete missing data from our prospective registry, medical records were hand-searched to extract date of stroke admission, age, gender, stroke etiology (Trial of Org. 10172 in Acute Stroke Treatment classification),Reference Adams, Bendixen and Kappelle 13 National Institutes of Health Stroke Scale (NIHSS, 0-42 score range, lower is better) on stroke admission, use of CT angiography (CTA), and symptomatic intracranial hemorrhage (sICH, third European Cooperative Acute Stroke Study definitionReference Hacke, Kaste, Bluhmki, Brozman and Dávalos 14 ). Delays were calculated by subtraction and refer to the onset-to-door time (delay between stroke onset and hospital arrival), head CT completion delay time (delay between hospital arrival and first head CT), DTN time (delay between hospital arrival and alteplase bolus administration), and OTN time (delay between stroke onset or last known well time and alteplase bolus administration).
Medians (interquartile range [IQR]) are presented for continuous variables with the assumption that their distributions are skewed. Comparisons across continuous variables were carried out using the Wilcoxon rank-sum and Kruskal-Wallis tests. Pearson chi-squared test was generally used to compare proportions. For the comparison of the probability of stroke mimics, Fisher exact test was used because of small sample sizes. Statistical significance was defined as a p value<0.05 (two-tailed). We used multiple linear regression models to assess for covariates of continuous outcomes and Poisson regression with robust variance for categorical outcomes. Independent variables included in each model, chosen based on their perceived clinical and biological relevance, were gender, age, OTN time, admission NIHSS score, and study group (dichotomized into pre- and postmodification). In the face of missing values, we carried out complete case analyses. The distribution of residuals was confirmed for each multiple linear regression, and goodness of fit was confirmed for each Poisson regression. Statistical analyses were carried out using STATA/SE, version 12.0 (StataCorp LP, College Station, TX).
RESULTS
Baseline Clinical Characteristics
The clinical characteristics of the patient population treated between 2012 and 2015 are shown in Table 2. Forty-eight individuals (median age, 75 years; 58.3% female), were thrombolyzed from January 2012 to May 2013 (premodification phase). Fifty-eight patients (median age, 76 years; 41.4% female) were treated between July 2013 and March 2015 (postmodification phase). Twenty of these were treated between July and December 2013 (early postmodification phase), and 38 between January 2014 and March 2015 (late postmodification phase). There was no difference in median admission NIHSS 12 (IQR 9-18) versus 13 (IQR 7-20) (p=0.841).
Table 2 Baseline characteristics of patient population
NIHSS=National Institutes of Health Stroke Scale.
Values expressed in median (interquartile range) or percentage.
* Overall refers to early and late combined.
† The statistical tests compared premodification phase to overall postmodification phase.
In-Hospital Treatment Delays
As shown in Table 3, median delay to head CT decreased from 25 minutes to 19 minutes (p=0.0204). There was a nonsignificant trend toward more frequent use of CTA, from 52.1% to 69.0% (p=0.076).
Table 3 Imaging and treatment times, symptomatic ICH, and stroke mimic rates during the pre- and postmodification phases
CT=computed tomography; ICH, intracerebral hemorrhage.
Values expressed in median (interquartile range) or percentage.
* Overall refers to early and late combined.
† Premodification phase, early postmodification phase, late postmodification phase.
‡ Premodification phase, overall postmodification phase.
§ Early postmodification phase, late postmodification phase.
|| Third European Cooperative Acute Stroke Study definition of symptomatic ICH.
Median DTN time decreased from 75 to 46 minutes between the pre- and postmodification phases (p<0.0001); no change in DTN time was found between the early and late postmodification phases (p=0.4085). The proportion of patients treated within 60 minutes also increased in the postmodification phase, from 29.2% to 75.9% (p<0.001), and this remained unchanged between the early and late postmodification phase (p=0.161) (Figure 1). There was a nonsignificant trend toward decreased OTN times between the pre- and post- phases, at 151 minutes versus 130 minutes (p=0.0895). The proportion of patients treated within 90 minutes of stroke onset improved (2.1% vs 19.0%, p=0.006) but remained quite low in both phases (Figure 2). The reduction in OTN times was sustained during the early and late postmodification phases (p=0.395).
Figure 1 Proportion of patients with door-to-needle time within 60 minutes (bars) and median door-to-needle times (line) during the study period.
Figure 2 Proportion of patients with onset-to-needle time within 90 minutes (bars) and median onset-to-needle times (line) during the study period.
Stroke Mimic and sICH Rates
Symptomatic ICH (4.2 vs 1.7%, p=0.361) and stroke mimic rates (2.1 vs 5.2%, p=0.625) rates were no different between the pre- and postmodification phases.
DISCUSSION
Our aim was to implement DTN time reduction strategies in our center relatively quickly, to determine whether reductions were achieved, and whether these were durable over time. We showed that it was possible to implement multiple simple interventions in a very short timeframe with no additional financial investment. The median DTN time was significantly shortened from 75 minutes to 46 minutes. This improvement occurred quickly, reaching 41 minutes over the first 6 months and remaining stable over the ensuing months at 46 minutes. The proportion of patients treated within the recommended delays of 60 minutesReference Lindsay, Bayley and Smith 12 nearly tripled, from 29.2% to 75.9%, whereas the number of patients benefiting from treatment within 90 minutes of onset increased ninefold, from 2.1% to 19.0%. Both of these effects were durable. This reduction in treatment times has already been shown in larger cohorts to confer improved clinical and functional outcomes for patients.Reference Saver, Fonarow and Smith 3
Several evidence-based interventions were studied in the conception of our new model,Reference Meretoja, Strbian, Mustanoja, Tatlisumak, Lindsberg and Kaste 6 - Reference Fonarow, Zhao and Smith 10 , Reference Price, Clement, Gray, Donaldson and Ford 15 and we opted for simultaneous implementation of the strategies most applicable to our milieu. Stroke pager notification by ED physician ensures that the stroke team arrives quickly when an acute stroke is suspected.Reference Gomez, Malkoff, Sauer, Tulyapronchote, Burch and Banet 16 , Reference Zweifler, Drinkard, Cunningham, Brody and Rothrock 17 However, stroke team prenotification by the triage nurse enables the stroke physician to begin assessing the patient almost immediately upon ED arrival. Although not shown in our results, others have noted that this strategy is a key driver in DTN time reduction.Reference Belvis, Cocho and Marti-Fabregas 18 , Reference Lin, Peterson and Smith 19 Multiple changes were aimed at shortening the delay from arrival to head imaging, such as direct transfer from EMS stretcher to CT and reserved access to elevators. To mitigate prolonged delays during off-hours,Reference Meretoja, Weir and Ugalde 11 we introduced 24/7 CT technician availability. Commensurate with changing practices, the proportion of patients undergoing CTA increased from 52.1% in May 2013 to 69.0% in July 2013 with no impact on DTN time. The largest reduction in our delays occurred between CT and alteplase bolus and was primarily attributable to not waiting for laboratory results and alteplase delivery on the CT table. Although we analyzed the impact of quality improvement efforts individually, global DTN time reduction results from an overall improvement in the system.Reference Fonarow, Smith and Saver 7 , Reference Kohrmann, Schellinger and Breuer 20
In-hospital strokes were excluded from our study as the systems in place to manage these patients differ from those for patients coming directly to the ED. We also excluded patients transferred from other hospitals because, in these cases, the stroke team was prenotified long before patient arrival and brain imaging was often already completed at the referring site, allowing for much faster local DTN times. Patients treated in June 2013 were excluded because this was an implementation period for our new model, and we arbitrarily divided early and late postmodification phases because we presumed that positive effects from initial change-related enthusiasm were likely to be negligible beyond 6 months.
Other centers in different health care settings have also reported their strategies and successes in reducing DTN times. In Helsinki,Reference Meretoja, Strbian, Mustanoja, Tatlisumak, Lindsberg and Kaste 6 12 measures were implemented over 8 years. The Melbourne Royal Hospital implemented six measures from the Helsinki modelReference Meretoja, Weir and Ugalde 11 and found that DTN time reduction was achieved by maximizing preparation before patient arrival, with hospital prenotification, preacquisition of history, and preorder of tests. Most time reduction strategies were articulated in the Target: Stroke Initiative From the American Heart Association/American Stroke Association, which helped participating hospitals reduce their DTN times.Reference Fonarow, Zhao and Smith 10 Ten recommendations were provided to the hospitals,Reference Fonarow, Smith and Saver 7 and the DTN time was reduced from 77 to 67 minutes postintervention, with improved clinical outcomes and no increase in complications.
Similarly, we found no increased risk of sICH and no significantly increased stroke mimic rates despite more rapid treatment, suggesting that sureness need not be sacrificed for swiftness. However, given the small sample size, functional outcome measures were not analyzed.
Our study also has other limitations. It is a single-center retrospective study that cannot eliminate confounding factors and selection and information biases. We were also unable to accurately assess the contribution of individual interventions because they were interrelated and we did not compare treatment delays between in-hours and out-of-hours. In the absence of a control group associated with a before-and-after type of study, secular trends in DTN time improvement likely exist because of increased awareness of stroke among care provider and accumulated experience and confidence in stroke care.Reference Meretoja, Strbian, Mustanoja, Tatlisumak, Lindsberg and Kaste 6 , Reference Koennecke, Nohr, Leistner and Marx 21 Also, baseline radiological data including ASPECTS (Alberta Stroke Program Early CT Score) and clot burden measurements were not available, although these variables correlate with baseline NIHSS. Finally, despite laudable improvements in DTN times, it is clear that our treatment delays require even further optimization to match those achieved elsewhere.
Although we were able to meet DTN targets proposed in Canadian guidelines from 2014,Reference Lindsay, Bayley and Smith 12 the paramount importance and demonstrated feasibility of even faster times have led the Canadian stroke community to recently adopt new targets recommending a median DTN time of 30 minutes or less and a 90th percentile of 60 minutes.Reference Kamal, Benavente and Boyle 22 To achieve this ambitious but attainable target, all stroke systems must do as much as possible before patient arrival, including acquisition of patient history and demographic information for prehospital transmission to the stroke team and intravenous access insertion by EMS. This would allow preorder of blood tests and CT requests and enable CT bed preemptying. The underlying principle informing all these strategies is to permit parallel workflow and limit sequential processes in which bottlenecks are inevitable. The advent of endovascular therapy has only further reinforced the importance of streamlining hyperacute stroke care trajectories so as to safely achieve the fastest possible door-to-groin puncture and door-to-recanalization times. Furthermore, once implemented, all time reduction strategies need to be continuously reevaluated and optimized to ensure not only durability but also continued safety amidst ever-faster treatment times.
Disclosures
MRK has received grants and personal fees from UCB and Sage Therapeutics that are outside the submitted work. YD has received personal fees from BMS-Pfizer, Bayer, and Novartis that are outside the submitted work. AYP has received personal fees from Medtronic, Biogen, Sanofi, and Pfizer-BMS that are outside the submitted work. BYC and AM have nothing to disclose.
Statement of Authorship
BYC compiled and analyzed the data and drafted the manuscript. AM conceived of the study idea, compiled and analyzed the data, and reviewed the manuscript. MRK provided statistical expertise for data analysis and reviewed the manuscript. YD conceived of the study idea and reviewed the manuscript. AYP
conceived of the study idea, edited the manuscript, and completed revisions.
