Article Text
Abstract
Background In acute stroke due to large vessel occlusion, faster reperfusion leads to better outcomes. We analyzed the effect of optimization steps aimed to reduce treatment delays at our center.
Methods Consecutive patients with ischemic stroke treated with endovascular therapy were prospectively analyzed. We divided the patients into pre-optimization (20 April 2012 to 8 October 2013) and post-optimization (9 October 2013 to 29 July 2014) periods. The main interventions included: (1) continuous feedback; (2) standardized immediate emergency department attending to stroke attending communication with interventional team activation for all potential interventions; (3) pre-notification by the emergency medical service; (4) minimizing additional diagnostic testing; (5) direct transport to the CT scanner; (6) transport directly from the CT scanner to the angiography suite. The main metric used to measure improvement was door to groin puncture time (D2P).
Results We included a total of 286 patients (178 pre-optimization, 108 post-optimization). There were no significant differences between major baseline characteristics between the groups with the exception of higher median CT Alberta Stroke Program Early CT Score in the pre-optimization group (p=0.01). Median D2P improved from 105 min pre-optimization to 67 min post-optimization (p=0.0002). Rates of good clinical outcomes (modified Rankin Scale 0–2 at 3 months) were similar in both groups, with a trend toward a better outcome in the post-optimization group in a subgroup analysis of patients with anterior circulation occlusion who received intravenous tissue plasminogen activator.
Conclusions This pilot study demonstrates that D2P times can be significantly reduced with a standardized multidisciplinary approach. There was no significant difference in the rate of 3-month good outcome, which is most likely due to the small sample size and confounding baseline patient characteristics.
- Intervention
- Stroke
- Thrombectomy
Statistics from Altmetric.com
Introduction
Acute large vessel occlusion (LVO) rapidly leads to permanent tissue injury in the absence of recanalization. First-generation randomized controlled trials of endovascular therapy failed to demonstrate improved outcomes with mechanical thrombectomy.1 ,2 A detailed analysis of the Interventional Management of Stroke III trial provided several potential explanations, including slow treatment time.3 Second-generation trials that emphasized rapid patient workflow with aggressive time metrics and continuous feedback established the superiority of endovascular therapy in LVO stroke.4 ,5 The ESCAPE trial had the fastest imaging to reperfusion times and was the first stroke reperfusion trial to show mortality benefit.4 These recent trials have demonstrated significant positive treatment effects with endovascular therapy,4–8 a fact attributed to dramatic workflow improvements along with the use of stent retrievers and better patient selection. Reproducing trial results in real-world practice requires detailed attention to workflow. Understanding and improving those factors that lead to treatment delays is an important step in optimizing outcomes. In this study we report on a prospective analysis performed on workflow at a single tertiary academic center over 2 years encompassing two phases: (1) an observational phase during which time metrics were captured (pre-optimization phase) as reference; and (2) a post-optimization phase used to measure the effect of the implemented measures (post-optimization phase). These two phases were separated by the implementation of targeted measures designed to improve treatment times. The overall goal of the study was twofold: to characterize those factors leading to delays in providing treatment and to determine whether methods aimed at improving workflow translated into shortened treatment times.
Methods
Patient selection
This prospective single-center quality improvement project was approved by our Institutional Review Board. Patients presenting with acute ischemic stroke were initially evaluated with MRI or CT. Patients with anterior circulation LVO were selected for intra-arterial therapy if their baseline CT showed an Alberta Stroke Program Early CT Score (ASPECTS) of ≥6 or based on assessment of mismatch between infarct and threatened but viable brain. Time from last seen normal was not considered a limiting factor for endovascular therapy. All patients qualifying for intravenous recombinant tissue plasminogen activator (rtPA) were administered thrombolysis. In patients who did not improve after intravenous rtPA or did not qualify for intravenous rtPA (ie, did not meet the inclusion criteria or presented outside the time window), endovascular therapy was pursued.
Prospective data collection
Data were prospectively collected on all stroke patients treated with endovascular therapy including patient demographics, radiographic findings, angiographic data, and clinical data (baseline National Institutes of Health Stroke Scale (NIHSS) and 90-day modified Rankin Scale score (mRS)). Additionally, a formal process was in place to document times between each step from hospital arrival to reperfusion including patient arrival at the hospital until the initial non-contrast CT scan (door to CT), non-contrast CT to arrival in the angiography suite (CT to angiography), arrival in the angiography suite to groin puncture, groin puncture to reperfusion (time that Thrombolysis in Cerebral Infarction (TICI) of ≥2b was achieved), and patient arrival at the hospital until access site puncture was achieved (D2P).
Optimization steps
All consecutive patients with ischemic stroke treated with endovascular therapy were prospectively registered from 20 April 2012 to 29 July 2014. In October 2013 we implemented specific changes in our process to improve D2P. We analyzed two treatment epochs: pre-optimization (20 April 2012 to 8 October 2013) and post-optimization (9 October 2013 to 29 July 2014). Through regular multidisciplinary meetings with stroke, emergency medicine, and neurointerventional attending physicians, we identified five main sources of prolonged D2P: (1) serial rather than parallel workflows; (2) inefficient communication and final decision-making process (emergency department (ED) trainee to ED attending, ED attending to stroke trainee, stroke trainee to stroke attending, stroke attending to interventional trainee, interventional trainee to interventional attending); (3) patient transport delays; (4) delays in obtaining initial brain imaging; and (5) lack of consistent feedback. Our main areas of focus included rapid initiation of imaging, efficient communication and decision making, and continuous feedback (figure 1).
Steps to improve door to CT time
Pre-arrival communication: pre-optimization, only the ED charge nurse and patient care technician received emergency medical service (EMS) stroke alert notifications. Post-optimization, additional recipients included the CT technologist (to clear and prepare the scanner), ED registrar (to initiate registration), ED attending physician, and stroke team.
Direct to CT: EMS transported possible stroke patients directly to the CT scanner on arrival. Team members convened in the CT scanner to receive the patient and initiate care.
Limit non-essential interventions: non-emergent testing and procedures were deferred in favor of rapid neuroimaging (ECG, chest x-ray, additional venous access).
Education and feedback: staff were educated regarding the door to CT goal of 10 min. Feedback on door to CT performance was provided regularly to ED physicians, nurses, CT technicians, and stroke team.
Steps to improve CT to angiography suite time
Early communication: ED attending and stroke physician prioritized early direct communication regarding plan of care. A stroke team member waited for the patient to arrive in the ED for transfers and pre-arrival notifications.
Limit non-essential interventions: intubation, arterial line placement and indwelling bladder catheter placement were deferred unless emergently indicated. Point of care laboratory testing was performed for initial assessment of creatinine and coagulation status in the CT scanner or in the angiography suite.
Direct to angiography: in cases of short interfacility transfer times (<60 min), patients accepted from referral hospitals with proven LVOs based on non-invasive angiography or those with a high likelihood of LVO (NIHSS ≥10 with a favorable ASPECT score) were transferred directly to the angiography suite, bypassing the ED, whenever the patient's condition allowed. The accepting stroke neurologist gave advanced notice to the in-house stroke neurologist (fellow or resident), the neurointerventional team for activation, and the ED as appropriate.
‘No turn around’ approach: in cases in which imaging beyond that obtained from the initial hospital was deemed necessary, a decision to proceed to the angiography suite was made in the CT or MRI scanner and patients were subsequently transferred to the angiography suite without returning to an ED treatment room. Intravenous rtPA was started in the CT scanner or in the angiography suite but did not delay groin access.
Early activation of the neurointerventional team: the on-call IR team was activated early in patients with small core and suspected LVO (NIHSS ≥10 and CT ASPECTS ≥6) before any additional imaging that may have been considered necessary was obtained. The use of MRI was limited to select cases. Overall, during the duration of the study the use of additional imaging has been gradually reduced to the minimum, especially in patients treatable within 6 h of symptom onset.
All these measures were performed accepting a certain percentage of instances in which, despite full activation of the interventional team, treatment was not performed as it was not deemed necessary or appropriate (eg, patient's neurological examination dramatically improved in transfer, LVO was not present, or CT perfusion post-processing revealed a large infarct not evident on CT).
Statistical analysis
Baseline demographics were evaluated using the χ2 test for categorical variables and the Student t test and the Wilcoxon rank-sum test for parametric and non-parametric continuous variables, respectively. Pre-optimization and post-optimization time metrics were compared by the Wilcoxon rank-sum test. The rate of favorable outcome, defined as mRS 0–2 at 3 months, was assessed pre- and post-optimization by the χ2 test. Subsequently, we employed multivariable logistic regression to assess functional outcome while controlling for all independent variables that were statistically significant predictors of outcome. Statistical significance for all tests was predefined as α<0.05. We repeated these steps to perform a post hoc analysis of a more homogeneous (ie, comparable) patient subgroup: those with an anterior circulation occlusion who received intravenous tPA prior to endovascular treatment, similar to the patient population enrolled in SWIFT-PRIME. The statistical analysis was conducted with STATA Software V.13.0 (StataCorp LP, College Station, Texas, USA).
Results
Study population
A total of 286 patients were included in our analysis, 178 patients who were treated pre-optimization and 108 patients treated post-optimization. Patient age and stroke severity were similar in the pre-optimization (mean age 67.8±15.4 years, median NIHSS 17 (IQR 13–21)) and post-optimization (mean age 67.2±14.1 years, median NIHSS 17 (IQR 14–21)) groups. Additional patient demographics (table 1) and stroke characteristics (table 2) were also similar with the exception of a lower (ie, worse) pre-treatment ASPECTS score (8.5 vs 9.0, p=0.01) and lower rate of type 2 diabetes mellitus (11.5% vs 23.8%, p=0.012) in the post-optimization group. There were further notable differences which favored the pre-optimization group, including the number of patients receiving intravenous tPA (37.0% vs 28.6%), internal carotid artery terminus occlusions (6.1% vs 12.1%) and posterior circulation occlusions (9.8% vs 16.8%), that did not attain statistical significance.
Time to treatment and outcomes
Time metrics for each stage of triage and treatment are reported in table 3. Statistically significant improvements were observed in the post-optimization cohort in median arrival in the angiography suite to puncture time (16 (IQR 11–21) vs 10 (IQR 6–15) min, p<0.001) as well as in the median puncture to reperfusion time (63 (IQR 41–92) vs 47 (IQR 31–74) min, p=0.023). ED to CT and CT to angiography suite time metrics were also improved, although the differences were not statistically significant (figure 2). This resulted in an overall decrease in the median D2P time from 105 to 67 min (p<0.001) and door to reperfusion time from 165.0 min to 118.5 min (p<0.001). Although this project was aimed at improving treatment times for endovascular patients, we observed an unexpected significant improvement in the median door to intravenous rtPA times from 56 min to 30 min (p=0.002) during this time period. In addition, 2.8% (5/178) of patients were directly transferred from the helipad or ambulance to the angiography suite in the pre-optimization phase compared with 11% (12/108) of patients in the post-optimization phase. In this subgroup, median door to angiography time was 13.2 vs 80.0 min in the pre-optimization phase and 14 vs 65 min in the post-optimization phase (p<0.001).
Despite a significant reduction in treatment times, the rate of favorable outcome pre- and post-optimization was not significantly different (45.0% vs 43.2%, p=0.834). When controlled for all statistically significant predictors of outcome in our patient population, there remained no difference in changes for a good outcome between the pre- and post-optimization groups (OR 0.91, 95% CI 0.44 to 1.91, p=0.818). In a subgroup analysis of patients with anterior circulation occlusion who received intravenous tPA, a patient population proven to benefit from rapid recanalization with endovascular therapy,4 ,5 there was a non-significant trend towards better outcomes in the post-intervention group (51.7% vs 46.7%, OR 1.33, 95% CI 0.43 to 4.16) when controlling for the pre-treatment ASPECT score and NIHSS.
Discussion
The benefit of endovascular therapy over medical management for acute stroke is most apparent in well selected patients with small core infarcts, large clinical deficit, and proven proximal occlusion who undergo high quality reperfusion with a fast onset to treatment time.4–8 Patient selection approaches continue to be a matter of debate as the largest treatment effect has been observed in patients selected based on CT perfusion,6 whereas patients selected based on CT/CT angiography alone also benefit from endovascular therapy.4 TICI 2b/3 reperfusion rates are as high as 80–90% with stent retriever devices, especially when patients are treated early and still have good collaterals.5 Similar to the cardiology experience, the fundamental requirement of fast treatment time requires multidisciplinary coordination across specialties and geographic space.8
Once patients arrive at the destination hospital, resources must be aligned to achieve rapid diagnosis and definitive treatment. The importance of time and good outcomes has long been recognized in percutaneous coronary intervention (PCI). Indeed, faster door to balloon times lead to lower in-hospital and 6-month mortality (adjusted OR for each 10 min decrease 0.94, 95% CI 0.93 to 0.95; p<0.0001).9 Arguably, the brain may be even more sensitive to timely reperfusion than the myocardium. In patients with acute ischemic stroke due to LVO who received intravenous thrombolysis, every 15 min acceleration in onset to treatment time leads to 13 more patients (out of 1000) being discharged to a more independent environment.10 Similarly, in patients undergoing intra-arterial therapy, every 15 min acceleration in onset to treatment time leads to 34 more patients (out of 1000) having improved functional outcome.11 Across treatments, faster times lead to decreased disability and ultimately lower individual and societal costs. In the EXTEND-IA trial, patients treated with stent retriever thrombectomy were able to return home or back to work significantly sooner than patients who received intravenous rtPA alone (15 vs 73 days (p=0.006)).6 In addition, stent retriever thrombectomy has been shown to be cost effective, so measures aimed at reducing endovascular treatment times in stroke, despite their resource-intensive nature, are likely to lead to better outcomes and also to prove cost effective.12
A comparison of the pre- and post-optimization phase time metrics in our study revealed improvements at several steps including improved door to CT, CT to angiography suite, and groin puncture to reperfusion. The success of the quality initiative is probably multifactorial. A key initial step was defining time metrics. In 2011, multisocietal consensus statement guidelines recommended a maximum 2 h goal for door to treatment time.13 Data from recent randomized endovascular trials suggest that a reappraisal of these values—more closely aligned to those established in ST segment elevation myocardial infarction (STEMI) where door to balloon times of 60 min are the norm—is necessary. The measured times in clinical trials and registries, however, revealed that significant numbers of patients were not treated within these recommended target goals.14 ,15 Once a time metric is established, it is equally critical to establish an iterative feedback mechanism to allow routine auditing and continuous process improvement. The fastest treatment times were noted in the ESCAPE4 and SWIFT Prime5 trials. Both trials set aggressive time goals (ESCAPE: picture to puncture of <60 min and picture to reperfusion of <90 min; SWIFT Prime: picture to puncture <70 min). It should be noted that both ESCAPE and SWIFT Prime took place at large volume endovascular stroke centers that were already attuned to workflow issues and thus it is doubtful that the times achieved in these trials apply to the majority of centers in the USA or Canada. Critically important for the success of these trials was a continuous appraisal of workflow at all enrolling sites. This quality control step allowed for early identification of potential workflow hurdles as well as recognition of best practices. Similar quality initiatives with set time metrics and feedback have been successful at a national level for intravenous therapy in reducing the door to needle times from 77 min to 67 min.16
Early engagement of key players facilitated parallel workflow. Pre-hospital notifications are essential in priming the emergency room for patient arrival and equipment preparedness. In particular, it allowed for clearance of the scanner and ensured availability of appropriate staff to receive the patient. Direct activation of the interventional team by the ED attending was encouraged to minimize delays in communication. Adopting early activation of the neurointerventional team at one academic center led to an improvement in D2P from a median of 143 min to a mean of 107 min.17 Marked improvements have similarly been observed in STEMI when ED physicians directly activated the catheterization laboratory with median door to balloon time decreasing from 113.5 to 75.5 min (p<0.0001).18
Once the patient arrived in our ED, emphasis was placed on deferring and eliminating non-essential testing and procedures in favor of immediate neuroimaging. A retrospective review of ED patients receiving intravenous rtPA showed that an ECG performed before the head CT increased door to CT time by 6 min and a chest x-ray obtained before the head CT increased door to CT time by 13 min.19 In most cases, our patients were transported directly by the EMS to the CT scanner on arrival. Accordingly, we observed faster door to intravenous rtPA times in addition to D2P times. We minimized the use of advanced imaging and rarely used MRI, given the longer times associated with this modality at our center and unproven benefit over CT-based treatment selection. A decision to proceed with endovascular therapy was made while the patient was in the CT scanner and followed by transfer directly to the angiography suite when a small core with LVO was confirmed. Avoiding return to an ED treatment area after scan (‘no turn back approach’) has been associated with a higher rate of CT scan acquisition and door to microcatheter placement time of <90 min (57.6% vs 31.6%, p=0.0007).20
In a subset of transferred patients with short interfacility transport time (<60 min) and available imaging, additional testing was deferred and patients were transferred directly from the helipad to the angiography suite. This approach was associated with dramatically shorter D2P times and was encouraged in the post-optimization phase where time from door to angiography was 14 min compared with 65 min in the 11% (12/108) of patients who were taken directly from the helipad to the angiography suite. In a series of patients undergoing PCI, patients directly admitted to the catheterization laboratory had significantly reduced door to balloon time (58 vs 105 min, p<0.001), with the 90 min target for door to balloon time achieved in 94% of patients compared with 29% of patients triaged from the ED.21
Finally, it should be emphasized that multiple strategies were simultaneously adopted in this quality initiative project to fully optimize workflow. An analysis of hospital strategies aimed to improve door to needle times in patients receiving intravenous rtPA revealed that hospitals that used a greater number of strategies had shorter door to needle times, with 1.3 min (adjusted mean difference) saved for each strategy implemented (14 min if all strategies were used).22 Implementation of multiple strategies hinges on a multidisciplinary effort with involvement of all disciplines involved in the care of stroke patients including EMS, registration, ED personnel, radiology and imaging, nursing, anesthesia, stroke, and interventional services.
One of the limitations of the current study is that it was probably underpowered to detect differences in clinical outcomes. Whereas time to treatment is a well-known determinant of outcome, we did not see a difference in good outcome pre-optimization and post-optimization in this group of patients, presumably due to small sample size and underlying differences in baseline characteristics such as lower ASPECTS, longer time from onset to hospital arrival, higher rate of posterior circulation and internal carotid artery terminus occlusions in the post-optimization group. Although not statistically significant, the lower rate of intravenous rtPA exposure in the post-optimization group may have also influenced the rate of good outcome. An additional explanation is that we treated many patients with favorable physiology in the more extended time window (ie, ‘slow progressors’), whose outcomes are less sensitive to time to reperfusion.23 Our practice consists of a large proportion of patients from outside referring hospitals. In fact, about 30% of patients in each group were treated beyond 6 h from time of onset to last seen well. Although this was similar between the groups, onset to hospital arrival time was slightly longer (35 min) for the post-optimization group (table 3).
To facilitate a more meaningful comparison we then analyzed only those patients with anterior circulation occlusions who received intravenous tPA—a patient population proven to benefit from rapid recanalization with endovascular therapy.4 ,5 In this more homogeneous dataset there was a non-significant trend towards better outcomes in the post-intervention group (51.7% vs 46.7%, OR 1.33, 95% CI 0.43 to 4.16) when controlling for pre-treatment ASPECTS and NIHSS. While this finding did not reach statistical significance, we are underpowered for such an analysis which is beyond the scope of this paper. An additional contributing factor may have been that the door to needle time for intravenous rtPA delivery was also significantly improved in the post-optimization group. However, it is a well-established principle of acute stroke that faster endovascular recanalization times lead to better functional outcomes for patients presenting with a LVO.4
Another limitation of this study is that multiple changes were made at the same time and thus the impact of each individual intervention on the overall outcome metric of shortening D2P time is difficult to gauge. In addition, it is possible that improvements in D2P times were not due to the interventions but simply due to the fact that all metrics were measured (ie, Hawthorne effect). This is plausible but unlikely given that no trends in the positive direction were noted in the pre-optimization phase while times were prospectively collected. In our study 41% of the patients had D2P <90 min compared with 70% post-intervention, but only 23% of our pre-intervention patients and 41% of our post-intervention patients achieved a D2P of <60 min. This suggests that there is continued need for optimization. Additionally, while not specifically addressed in this current quality initiative, future studies will need to focus on patient populations with competing care priorities and delayed recognition of stroke, including trauma patients, in-house strokes,24 and off-hour events.25
Our study focused on hospital workflow whereas future aims will need to target reducing the time from first medical contact to device, as recently proposed for PCI.26 Interfacility transfer times continue to be an important cause of delay. Future technological advancements, including mobile CT scanners, improved pre-hospital assessment tools (stroke severity scales and pre-hospital telemedicine platforms), and customized angiography suites for stroke interventions that incorporate imaging to allow a ‘one-stop shop’ approach can overcome some of the inherent limitations imposed by the current infrastructure. Accumulating evidence suggests that future workflow measures aimed to reduce pre-hospital, inter-hospital, and intra-hospital times to the minimum are likely to result in more patients than currently treated with higher rates of improved clinical outcomes.27
Conclusion
This pilot study demonstrates that D2P times can be significantly reduced with a standardized multidisciplinary approach. The changes that were implemented in this study can be applied to other stroke centers.
References
Footnotes
Twitter Follow Francis Guyette at @Guyettef
Contributors Conception and design: AA, TGJ and APJ. Acquisition of data, analysis and interpretation of data and critically revising the article: all authors. Drafting the article: AA and APJ. Reviewed submitted version of manuscript: AA and APJ. Approved the final version of the manuscript on behalf of all authors: APJ. Administrative/technical/material support: all authors. Study supervision: APJ.
Competing interests TGJ has received consulting and speaker fees from Silk Road, Air Liquide.
Ethics approval Ethics approval was obtained from the Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.