Discussion

Our present analysis points to a unique opportunity to improve efficiencies when patients are transferred for endovascular reperfusion therapy. We have shown that as P2P time elapses, fewer patients have favorable imaging characteristics and good clinical outcomes. The importance of understanding time delays and opportunities to improve transfer processes cannot be understated given the current push toward CSC certification. When this issue is addressed, many similarities and lessons should be drawn from the cardiac literature.

Time delays to primary angioplasty (percutaneous transluminal coronary angioplasty) in the treatment of ST-segment elevation myocardial infarctions have been well characterized in association with increased mortality.¹³ This relationship subsequently spearheaded nationwide efforts to reduce delays to percutaneous transluminal coronary angioplasty through adjustments in institutional practice.¹⁴ In 1999, the American Heart Association published its first performance guidelines, advocating “door-to-balloon” times of <90 minutes.¹⁵ Although ensuing studies substantiated the merits of this metric in reducing patient mortality,¹⁶ a collaborative registry showed that this metric was achieved in 52% of nontransferred patients in 2005.¹⁷ The CathPCI group was able to improve door-to-balloon times to <90 minutes in 76% of patients by 2008 through a concerted effort of education and dissemination of best practices.

Transferred patients represented a challenging group for whom time metrics had not been well established. To improve outcome measures among this unique cohort, the American Heart Association designated a new “door-in door-out” metric in 2008 that captured the time interval from outside hospital admission to EMS departure.¹⁸ Studies since have shown that door-in door-out times of <30 minutes are associated with not only faster times to treatment but also reductions in patient mortality. Fewer than 10% of patients achieved this metric in practice, thus highlighting the importance of streamlining interfacility networks to minimize transfer delays and improve patient outcomes.¹⁹ We believe that these concepts also hold true for the treatment of ischemic strokes with IAT.

As more CSCs become certified, the “hub-and-spoke” model of interfacility transfers from primary stroke centers to CSCs will eventually constitute a large proportion of patients treated with IAT. At our institution alone, 68% of the patients treated during a 21-month period arrived from referring primary stroke centers. Substantial delays existed within this transferring process, during which P2P times were prolonged by 2 hours compared with the times of the local cohort. Such delays were associated with a 22% absolute lower probability of a good outcome. The severity of stroke, as defined by baseline NIHSS and pretreatment THRIVE scores, was analogous in both cohorts and thus did not render the poorer outcomes observed among the transferred patients.

Timely delivery of IAT has been shown previously in the Interventional Management of Stroke I/II trials to increase the probability of good outcomes in patients with acute ischemic stroke.⁸ Our multivariable analysis similarly reveals that every 10-minute delay in P2P times correlates to a 6% relative lower probability of achieving a good outcome. This translates into 1 fewer good outcome per every 5 ischemic stroke patients transferred from outside facilities.

Currently, there are limited data in regard to the time metrics surrounding IAT. As recently as 2011, recommendations were made toward targeting an “arrival-to-treatment” time of <2 hours, although there is still debate in regard to whether successful reperfusion or start of procedure time should be defined as the proper end point. A recent study highlighted the possibility of using procedure time as an appropriate metric for predicting patient outcomes, with target goals of <60 minutes.²⁰ Although the time to successful reperfusion may hold more physiological relevance than puncture time in determining treatment efficacy, difficulty remains in quantifying this end point because of the presence of partial recanalization that occurs before complete revascularization.²¹ Despite advances in device technology to enhance reperfusion rates and reduce procedure times, outcomes have not yet improved,^6,22 which may reflect our current systems of care in transferring patients.

Door-in door-out times have been designated previously as the most appropriate metric for evaluating interfacility transfers in cardiology. However, <50% of the continuum of care from patient presentation to reperfusion is accounted for within this metric. In our study, the P2P time heralds a greater opportunity to minimize delays and enhance outcomes, spanning 74% of the entire continuum of care. Therefore, we propose the utilization of P2P time as a metric for determining the efficiency of work-flow processes in interhospital transfers for strokes, with a target time of 90 minutes. At this threshold, patients would have the greatest opportunity for a positive outcome and an estimated 2-fold increment in the number of patients that qualify for IAT. To achieve this goal, we have identified several opportunities for improvement (Table 4).

The time between initial CT at the outside hospital to CSC notification (decision making) is substantial. This is due to multiple factors including delays in neurological assessments, interpretation of imaging, utilization of advanced modality imaging, and determination of tPA effectiveness. In our study, obtainment of advanced imaging contributed to a 57-minute delay in decision making without substantial benefits in patient outcome. To produce more efficient systems of care, processes such as these must be streamlined with a minimalist approach similar to the ECG for myocardial infarction. Noncontrast head CT imaging can be used as a tool to discern whether a hemorrhage is present but can additionally be used to assess for a thrombus with thin-cut reconstructions²³ as well as for core infarct with the use of the ASPECTS method.⁹ We acknowledge that these tools are not as precise as magnetic resonance imaging or CT perfusion, but time delays diminish the likelihood of benefit from IAT, particularly in transferred patients. A previous study has also demonstrated that obtaining multimodal imaging not only contributes to delays in treatment but fails to reduce hemorrhage rates or improve clinical outcomes.²⁴

Delays in decision making are further compounded by inefficiencies in consultation practices. When referring neurologists were consulted to evaluate an acute stroke patient, the time to CSC notification was delayed by 30 minutes. A number of factors contributed to this delay, including discussions with the emergency department physician, time spent reevaluating the patient, delivery of intravenous tPA, and, in many instances, awaiting the arrival of an off-site neurologist for bedside assessment. Although we recognize the importance of specialty consultations in the setting of difficult clinical presentations, we recommend that emergency department physicians directly upload CT images through regional Picture Archiving and Communication Systems that allow for images to be read directly by CSC physicians. This would potentially reduce the need for unnecessary consultations and expedite the CSC acceptance process. Similar strategies have been described in the cardiology literature, in which emergency physician activation of the catheterization laboratory team reduces door-to-balloon times significantly.²⁵ Given that referring facilities have different thresholds for initiating patient transfers, we are currently working to implement a “rapid-transfer protocol” within our own community that standardizes the minimal requirements needed to contact a CSC for interhospital transfer.

Transfer time itself was also identified as an opportunity for improvement. The time from CSC notification to EMS arrival at the referring facility spanned >30 minutes, with an additional hour to CSC imaging. Simultaneous activation of the EMS team at the time of CSC contact would reduce delays to EMS mobilization, including expedited personnel contact and earlier weather reports. Current practices at many outside facilities have inefficiencies between EMS arrival and hospital departure, including the following: deactivation of the helicopter; transportation of the EMS team to the patient’s room; reevaluation of the patient at bedside; time for awaiting copies of hospital records to be printed; transportation of the patient back to the helipad; and finally, reinitiation of the helicopter engine while loading the patient. In an effort to better streamline this process, direct delivery of patients to the helipad without deactivation of the helicopter engine would substantially mitigate these delays. Electronic or facsimile transmission of records directly to the CSC would reduce the time that EMS is on site. Furthermore, our study reveals that patients transferred from outside facilities beyond 60 miles from the CSC experience delays of up to an hour in transfer time, which was associated with poorer patient outcomes. The development of distance thresholds may eventually be warranted to identify patients who will not benefit from endovascular therapy because of delays in transfer.

Finally, CSC processing time (CSC CT to groin puncture) was noted to be less for outside hospital transfers than for local admissions. This is unique to our institution because the angiography suite is located in the neuro–intensive care unit and next to the CT scanner. We note a reduced time for processing of outside hospital transfers because the entire team awaits the patient, including the anesthesiologist. We suspect that there is an additional time delay in this component in many institutions, and this offers an additional opportunity to implement early activation of the interventional team.

We recognize that there may be additional steps to reduce treatment times for acute stroke patients that fall outside the P2P continuum. For instance, we are currently in the process of training EMS to recognize stroke symptoms earlier using the Los Angeles Motor Score,²⁶ as well as having patients transferred directly to our CSC when located equidistant to referring primary stroke centers at the time of onset. These measures would not only reduce the time from LKN to arrival but also eliminate the delays associated with outside hospital transfers.

In a recent retrospective analysis, higher-volume endovascular stroke centers were correlated with faster treatment times, better reperfusion rates, and improved clinical outcomes.²⁷ We found a similar relationship with interfacility transfers in which centers that transferred more patients had faster P2P times. This demonstrates the importance of hospital experience and the need for streamlined protocols when it comes to achieving lower P2P times.

An ancillary result from our study is the interplay between ASPECTS and time, in which delays to treatment correlate with less favorable imaging. Although we recognize that there is variability in patient collateral status, we believe that transfer delays lead to larger preprocedural core infarcts, which translate into lower ASPECTS over time. Future studies will be needed to better understand the rate of ASPECTS decay as a function of time and collateral status and to further discern the relationship between imaging deterioration and patient outcome.

Our protocol was to accept patients with an ASPECTS >7 and NIHSS score ≥10 from referring facilities. On repeat imaging at our CSC, patients with ASPECTS that deteriorated to 5 to 7 were still considered for treatment as long as their CT images revealed a frank hypodensity that was less than one third of the middle cerebral artery territory. Given that only 13% of patients with ASPECTS ≤7 had good outcomes at 90 days, our CSC has since adopted an ASPECTS threshold of >7 for selection of IAT. By implementing the aforementioned “action items,” our goal is to eventually include P2P times as part of the selection parameter for IAT, with a target goal of <90 minutes for all outside hospital transfers.

We acknowledge there are limitations to our analysis given that this is a consecutive case series from a single CSC. Additionally, the angiographic images were not adjudicated by a central core laboratory. Our limited sample size also risks overfitting of the model, although repeat modeling with increased stringency and likelihood ratio testing confirmed the impact of P2P on outcomes. Third, we had to exclude one fifth of the patients because of a large proportion with posterior infarcts or treatment beyond 9 hours. Patients with posterior infarcts were removed to better homogenize the data set and reflect an anterior stroke population with similarities in disease physiology. We believed that patients with LKN to groin puncture times >9 hours would bias the results of the analysis because they represent a unique cohort with favorable imaging characteristics who are not traditionally studied in standard clinical trials. When analyzed separately, these patients had no significant differences in imaging or outcomes compared with patients treated in <9 hours (Table IV in the online-only Data Supplement), reiterating the notion of their superior collateral systems. Indeed, the significance of time may not be as profound among these patients, whose physiology allows them to overcome the perils of long transfer delays. Nevertheless, inclusion of these patients in our model did not eliminate the overall significance of P2P times in terms of patient outcome because the odds of a good outcome declined by 3% for every 10-minute delay in P2P, regardless of LKN (Table V in the online-only Data Supplement). Moreover, among the patients treated beyond 9 hours, those with good outcomes had shorter times from LKN to groin puncture than those with poor outcomes, although this was not statistically significant because of our small sample size.

Finally, we had 31 patients who were transferred but did not receive IAT because of unfavorable imaging characteristics. It is possible that a subset of these patients would have had shorter P2P times yet would have had poor clinical outcomes, thereby creating a bias in the patients selected for analysis. We believe this to be unlikely, however, given that the time from outside hospital CT to CSC CT was significantly prolonged among patients who were not treated with IAT. As such, their P2P times would also have been substantially greater than those of the patients included in our study. Our data demonstrate that every minute of delay in transfer time correlates with a 1% lower probability that a patient qualifies for IAT, which is comparable to the results of previous studies.⁷

In conclusion, our study defines a unique metric (P2P) that captures systems processes associated with transferring patients to a CSC. Delays in P2P result in lower ASPECTS on preprocedural imaging and worse patient outcomes at 90 days. We have subsequently identified opportunities to minimize these delays and to streamline the transfer process between primary stroke centers and CSCs. We believe that such changes will expand the number of candidates for reperfusion, facilitate higher transfer volumes within the hub-and-spoke network, and ultimately improve patient outcomes.