Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Self-driven Prehospital Triage Decisions for Suspected Stroke—Another Step Closer

Self-driven Prehospital Triage Decisions for Suspected Stroke—Another Step Closer Setting the course for prehospital triage of patients with suspected stroke is challenging. Few would disagree, and to be frank, this is rather old news. Yet the critical question remains unsolved: how do we optimally drive prehospital triage decisions for patients with suspected stroke? Central to this question is the challenge of identifying stroke specifically due to large vessel occlusion (LVO) in the prehospital setting. Furthermore, the “best choice” among the various prehospital LVO screening tools available has not been clear. Although several versions have been developed, many had not been validated in the prehospital setting to date, let alone externally validated with head-to-head, prospective comparisons. In this issue of JAMA Neurology, Nguyen and colleagues1 make a substantial contribution to our understanding of these scales and their relative strengths and feasibilities. The investigators included 7 prehospital stroke prediction scales commonly used worldwide: Cincinnati Stroke Triage Assessment Tool (C-STAT, formerly CPSSS) and Los Angeles Motor Scale (LAMS) from the United States; Rapid Arterial Occlusion Evaluation (RACE) from Spain; Prehospital Acute Stroke Severity (PASS) from Denmark; gaze-face-arm-speech-time (G-FAST) from the international Safe Implementation of Thrombolysis in Stroke investigators; and gaze-facial asymmetry-level of consciousness-extinction/inattention (GACE) from the Netherlands. Six of these scales are ordinal, with a cut point for identification of stroke due to LVO; the seventh scale, GACE, uses a decision tree. In this study, during the pickup and transport of a patient with suspected stroke, emergency medical services paramedics completed an application that enabled reconstruction of each prediction scale for patients with suspected stroke based on initial FAST screening. Two critical questions were considered: (1) How well does each scale perform for identification of stroke due to LVO? (2) How feasible is the scale for routine use? In head-to-head comparisons, with respect to accuracy, the LAMS and RACE scales were the highest performers. The authors appropriately recognized that, in addition to accuracy, consideration of feasibility is critical when evaluating scales for the prehospital setting. Implementation should not place an undue burden on paramedics in the uncontrolled prehospital environment and amidst the various demands of the on-scene evaluation and transport of a patient who is in critical condition. The authors considered feasibility of reconstructing each scale fully, as well as the feasibility of reconstructing each score to a critical cut point that would enable LVO identification. In both scenarios, PASS was the most feasible, and the RACE scale proved to be challenging. These results are important, adding to our understanding of the relative strengths of the various scales. The starting place for this study is with a population of patients with suspected stroke based on a positive finding using initial FAST testing, a screen with its own test characteristics to consider. Taking as a given that the FAST test has a low false-negative rate, and considering the low prevalence of LVO (<1.5%) relative to all ambulance calls,2,3 we quickly recognize that even a hypothetical, nearly perfect prehospital screening tool with 95% sensitivity and 99% specificity could hardly achieve a positive predictive value of 59% for stroke due to LVO. Of course, given their universally low sensitivities by design, application of any of these scales is expected to lead to a high rate of false positives. And the consequence for patients with false positives is a high level of overtriage to thrombectomy stroke centers (TSCs) or to comprehensive stroke centers (CSCs). Seemingly minor differences in test characteristics have the potential to make a substantial difference for patients. Consider the C-STAT and the LAMS scales, which are at the 2 extremes of test accuracy in this study (79% and 89%, respectively). In a hypothetical population of 1000 patients with suspected stroke by FAST screen, application of LAMS rather than C-STAT would result in 120 patients—a full 12% fewer patients—“overtriaged” to a TSC or CSC (Table). Depending on the local geography, such overtriage could lead to substantial increases in transport times and delays in alteplase administration for eligible patients, distancing patients from families and support networks, and diverting patients with stroke unnecessarily away from capable PSCs to crowd TSCs or CSCs. Even so, in the long run, some overtriage is likely justified, and resources may need to be allocated to allow for this, given the unprecedented but time-sensitive benefit of endovascular thrombectomy on patient morbidity and mortality (number needed to treat of 2.6 patients).4 Table. Differences in Expected Prehospital Triage Outcomes at the Extremes of Test Accuracy for Detection of Stroke Due to Large Vessel Occlusiona View LargeDownload The complexities in prehospital decisions for patients with suspected stroke are numerous. First, as highlighted here, is the limited ability to accurately determine patient eligibility for alteplase or thrombectomy in the prehospital setting. Traffic patterns may also substantially impact optimal triage decisions.5 Relatedly, hospitals of varying resources may be located in opposite directions so that, for example, the decision to drive 20 minutes to the PSC rather than 40 minutes to the CSC could mean a 60-minute transfer time should the patient ultimately be eligible for thrombectomy. And another question—should hospital quality be considered in prehospital decision-making? If a particular hospital is located only 10 minutes farther than the closest hospital destination, but that additional 10 minutes would mean access to a higher-performing hospital with faster door-to-needle times and higher quality of care metrics, should we account for this in prehospital triage? With no doubt, our ability to identify patients with stroke due to LVO in the prehospital setting is of paramount importance, and Nguyen and colleagues do tremendous work in advancing the state of this science. Yet, the complexities remain, and one cannot help but wonder: is there something better? Should we lean more heavily on technology in the prehospital setting? Perhaps noninvasive sensors will turn out to be more efficient at identification of large strokes than these simple prehospital screening tools.6 Or perhaps we should consider more mathematically complex decision models easily made widely accessible by smartphone applications; such models may deal with uncertainty, consider probabilities, weigh various transport options, and even incorporate live traffic patterns to drive decision-making using artificial intelligence.7 In addition, is there an increased role for repatriation? Is the answer simply to anticipate an increasingly high volume of acute stroke evaluations sent initially to TSCs and CSCs, and then to rapidly repatriate patients for whom intervention is not appropriate, intervention is uncomplicated and successful, or for whom the resources of the TSC or CSC are no longer required? The solutions are undoubtedly local and any attempt at a one-size-fits-all approach is fraught. The first local factor to be considered is pretest probability of LVO among patients with acute ischemic stroke. The incidence of stroke varies substantially by geography,8,9 and it stands to reason that the same would be true for stroke severity. This variability in the probability of LVO will affect each scale’s performance in a given local context. In a setting with higher prevalence, the negative predictive value of the scale will be lower. In the study’s Dutch cohort, LVOs were identified in 8% of emergency medical services paramedic-suspected strokes, and LAMS showed a negative predictive value of 95%. In a region with a prevalence rates of LVO as high as 33%, the negative predictive value would be 75%, meaning that 25% of patients who were positive for LVO would go undetected in the prehospital setting. Well-designed stroke systems of care should ideally include an optimized distribution of hospitals and resources accounting for regional factors, while ensuring equitable access to high-quality care and including processes for continuous system improvement. Relevant policies for prehospital transport destination will necessarily vary among settings. As we are driven to refine and optimize stroke systems of care, and Nguyen et al1 take us one step closer to incorporating LVO prediction methods in an evidence-based manner, we must now consider incorporating global positioning technology, regional patient characteristics, and local hospital resources and quality measures. We must prespecify these guiding principles in nuanced and nimble ways. What are the rules that we are trying to write? Once these are decided, we can then consider how to code them to create self-driven processes. Back to top Article Information Corresponding Author: Kori S. Zachrison, MD, MSc, Department of Emergency Medicine, Massachusetts General Hospital, 55 Fruit St, Boston, MA 02114 (kzachrison@partners.org). Published Online: November 30, 2020. doi:10.1001/jamaneurol.2020.4425 Conflict of Interest Disclosures: Dr Zachrison reported receiving grants to the Department of Emergency Medicine, Massachusetts General Hospital from the Agency for Healthcare Research and Quality, the National Institutes of Health, CRICO, and the American College of Emergency Physicians. Dr Khatri reported receiving grants from Cerenovus to the Department of Neurology and Rehabilitation Medicine, University of Cincinnati for investigator-initiated study effort, grants from Nervive to the Department of Neurology and Rehabilitation Medicine, University of Cincinnati, and a coinvestigator grant from the National Institutes of Health; receiving personal fees from Lumosa Therapeutics to the Department of Neurology and Rehabilitation Medicine, University of Cincinnati for consulting and for being a member of the data safety monitoring board; personal fees from DiaMedica Therapeutics to the Department of Neurology and Rehabilitation Medicine, University of Cincinnati for being a member of the scientific advisory board; and personal fees from Bayer for being a national leader of a trial outside the submitted work. No other disclosures were reported. References 1. Nguyen TTM, van den Wijngaard IR, Bosch J, et al. Comparison of prehospital scales for predicting large anterior vessel occlusion in the ambulance setting.  JAMA Neurol. Published online November 30, 2020. doi:10.1001/jamaneurol.2020.4418Google Scholar 2. Shah MN, Bazarian JJ, Lerner EB, et al. The epidemiology of emergency medical services use by older adults: an analysis of the National Hospital Ambulatory Medical Care Survey.  Acad Emerg Med. 2007;14(5):441-447. doi:10.1197/j.aem.2007.01.019 PubMedGoogle ScholarCrossref 3. Malhotra K, Gornbein J, Saver JL. Ischemic strokes due to large-vessel occlusions contribute disproportionately to stroke-related dependence and death: a review.  Front Neurol. 2017;8:651. doi:10.3389/fneur.2017.00651 PubMedGoogle ScholarCrossref 4. Goyal M, Menon BK, van Zwam WH, et al; HERMES collaborators. Endovascular thrombectomy after large-vessel ischaemic stroke: a meta-analysis of individual patient data from five randomised trials.  Lancet. 2016;387(10029):1723-1731. doi:10.1016/S0140-6736(16)00163-X PubMedGoogle ScholarCrossref 5. Freyssenge J, Renard F, Schott AM, et al. Measurement of the potential geographic accessibility from call to definitive care for patient with acute stroke.  Int J Health Geogr. 2018;17(1):1. doi:10.1186/s12942-018-0121-4 PubMedGoogle ScholarCrossref 6. Walsh KB. Non-invasive sensor technology for prehospital stroke diagnosis: current status and future directions.  Int J Stroke. 2019;14(6):592-602. doi:10.1177/1747493019866621 PubMedGoogle ScholarCrossref 7. Ali A, Zachrison KS, Eschenfeldt PC, Schwamm LH, Hur C. Optimization of prehospital triage of patients with suspected ischemic stroke.  Stroke. 2018;49(10):2532-2535. doi:10.1161/STROKEAHA.118.022041 PubMedGoogle ScholarCrossref 8. Howard G, Howard VJ. Twenty years of progress toward understanding the stroke belt.  Stroke. 2020;51(3):742-750. doi:10.1161/STROKEAHA.119.024155 PubMedGoogle ScholarCrossref 9. Balamurugan A, Delongchamp R, Bates JH, Mehta JL. The neighborhood where you live is a risk factor for stroke.  Circ Cardiovasc Qual Outcomes. 2013;6(6):668-673. doi:10.1161/CIRCOUTCOMES.113.000265 PubMedGoogle ScholarCrossref http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png JAMA Neurology American Medical Association

Self-driven Prehospital Triage Decisions for Suspected Stroke—Another Step Closer

JAMA Neurology , Volume 78 (2) – Mar 2, 2021

Loading next page...
 
/lp/american-medical-association/self-driven-prehospital-triage-decisions-for-suspected-stroke-another-G1NvnMUV8i
Publisher
American Medical Association
Copyright
Copyright 2020 American Medical Association. All Rights Reserved.
ISSN
2168-6149
eISSN
2168-6157
DOI
10.1001/jamaneurol.2020.4425
Publisher site
See Article on Publisher Site

Abstract

Setting the course for prehospital triage of patients with suspected stroke is challenging. Few would disagree, and to be frank, this is rather old news. Yet the critical question remains unsolved: how do we optimally drive prehospital triage decisions for patients with suspected stroke? Central to this question is the challenge of identifying stroke specifically due to large vessel occlusion (LVO) in the prehospital setting. Furthermore, the “best choice” among the various prehospital LVO screening tools available has not been clear. Although several versions have been developed, many had not been validated in the prehospital setting to date, let alone externally validated with head-to-head, prospective comparisons. In this issue of JAMA Neurology, Nguyen and colleagues1 make a substantial contribution to our understanding of these scales and their relative strengths and feasibilities. The investigators included 7 prehospital stroke prediction scales commonly used worldwide: Cincinnati Stroke Triage Assessment Tool (C-STAT, formerly CPSSS) and Los Angeles Motor Scale (LAMS) from the United States; Rapid Arterial Occlusion Evaluation (RACE) from Spain; Prehospital Acute Stroke Severity (PASS) from Denmark; gaze-face-arm-speech-time (G-FAST) from the international Safe Implementation of Thrombolysis in Stroke investigators; and gaze-facial asymmetry-level of consciousness-extinction/inattention (GACE) from the Netherlands. Six of these scales are ordinal, with a cut point for identification of stroke due to LVO; the seventh scale, GACE, uses a decision tree. In this study, during the pickup and transport of a patient with suspected stroke, emergency medical services paramedics completed an application that enabled reconstruction of each prediction scale for patients with suspected stroke based on initial FAST screening. Two critical questions were considered: (1) How well does each scale perform for identification of stroke due to LVO? (2) How feasible is the scale for routine use? In head-to-head comparisons, with respect to accuracy, the LAMS and RACE scales were the highest performers. The authors appropriately recognized that, in addition to accuracy, consideration of feasibility is critical when evaluating scales for the prehospital setting. Implementation should not place an undue burden on paramedics in the uncontrolled prehospital environment and amidst the various demands of the on-scene evaluation and transport of a patient who is in critical condition. The authors considered feasibility of reconstructing each scale fully, as well as the feasibility of reconstructing each score to a critical cut point that would enable LVO identification. In both scenarios, PASS was the most feasible, and the RACE scale proved to be challenging. These results are important, adding to our understanding of the relative strengths of the various scales. The starting place for this study is with a population of patients with suspected stroke based on a positive finding using initial FAST testing, a screen with its own test characteristics to consider. Taking as a given that the FAST test has a low false-negative rate, and considering the low prevalence of LVO (<1.5%) relative to all ambulance calls,2,3 we quickly recognize that even a hypothetical, nearly perfect prehospital screening tool with 95% sensitivity and 99% specificity could hardly achieve a positive predictive value of 59% for stroke due to LVO. Of course, given their universally low sensitivities by design, application of any of these scales is expected to lead to a high rate of false positives. And the consequence for patients with false positives is a high level of overtriage to thrombectomy stroke centers (TSCs) or to comprehensive stroke centers (CSCs). Seemingly minor differences in test characteristics have the potential to make a substantial difference for patients. Consider the C-STAT and the LAMS scales, which are at the 2 extremes of test accuracy in this study (79% and 89%, respectively). In a hypothetical population of 1000 patients with suspected stroke by FAST screen, application of LAMS rather than C-STAT would result in 120 patients—a full 12% fewer patients—“overtriaged” to a TSC or CSC (Table). Depending on the local geography, such overtriage could lead to substantial increases in transport times and delays in alteplase administration for eligible patients, distancing patients from families and support networks, and diverting patients with stroke unnecessarily away from capable PSCs to crowd TSCs or CSCs. Even so, in the long run, some overtriage is likely justified, and resources may need to be allocated to allow for this, given the unprecedented but time-sensitive benefit of endovascular thrombectomy on patient morbidity and mortality (number needed to treat of 2.6 patients).4 Table. Differences in Expected Prehospital Triage Outcomes at the Extremes of Test Accuracy for Detection of Stroke Due to Large Vessel Occlusiona View LargeDownload The complexities in prehospital decisions for patients with suspected stroke are numerous. First, as highlighted here, is the limited ability to accurately determine patient eligibility for alteplase or thrombectomy in the prehospital setting. Traffic patterns may also substantially impact optimal triage decisions.5 Relatedly, hospitals of varying resources may be located in opposite directions so that, for example, the decision to drive 20 minutes to the PSC rather than 40 minutes to the CSC could mean a 60-minute transfer time should the patient ultimately be eligible for thrombectomy. And another question—should hospital quality be considered in prehospital decision-making? If a particular hospital is located only 10 minutes farther than the closest hospital destination, but that additional 10 minutes would mean access to a higher-performing hospital with faster door-to-needle times and higher quality of care metrics, should we account for this in prehospital triage? With no doubt, our ability to identify patients with stroke due to LVO in the prehospital setting is of paramount importance, and Nguyen and colleagues do tremendous work in advancing the state of this science. Yet, the complexities remain, and one cannot help but wonder: is there something better? Should we lean more heavily on technology in the prehospital setting? Perhaps noninvasive sensors will turn out to be more efficient at identification of large strokes than these simple prehospital screening tools.6 Or perhaps we should consider more mathematically complex decision models easily made widely accessible by smartphone applications; such models may deal with uncertainty, consider probabilities, weigh various transport options, and even incorporate live traffic patterns to drive decision-making using artificial intelligence.7 In addition, is there an increased role for repatriation? Is the answer simply to anticipate an increasingly high volume of acute stroke evaluations sent initially to TSCs and CSCs, and then to rapidly repatriate patients for whom intervention is not appropriate, intervention is uncomplicated and successful, or for whom the resources of the TSC or CSC are no longer required? The solutions are undoubtedly local and any attempt at a one-size-fits-all approach is fraught. The first local factor to be considered is pretest probability of LVO among patients with acute ischemic stroke. The incidence of stroke varies substantially by geography,8,9 and it stands to reason that the same would be true for stroke severity. This variability in the probability of LVO will affect each scale’s performance in a given local context. In a setting with higher prevalence, the negative predictive value of the scale will be lower. In the study’s Dutch cohort, LVOs were identified in 8% of emergency medical services paramedic-suspected strokes, and LAMS showed a negative predictive value of 95%. In a region with a prevalence rates of LVO as high as 33%, the negative predictive value would be 75%, meaning that 25% of patients who were positive for LVO would go undetected in the prehospital setting. Well-designed stroke systems of care should ideally include an optimized distribution of hospitals and resources accounting for regional factors, while ensuring equitable access to high-quality care and including processes for continuous system improvement. Relevant policies for prehospital transport destination will necessarily vary among settings. As we are driven to refine and optimize stroke systems of care, and Nguyen et al1 take us one step closer to incorporating LVO prediction methods in an evidence-based manner, we must now consider incorporating global positioning technology, regional patient characteristics, and local hospital resources and quality measures. We must prespecify these guiding principles in nuanced and nimble ways. What are the rules that we are trying to write? Once these are decided, we can then consider how to code them to create self-driven processes. Back to top Article Information Corresponding Author: Kori S. Zachrison, MD, MSc, Department of Emergency Medicine, Massachusetts General Hospital, 55 Fruit St, Boston, MA 02114 (kzachrison@partners.org). Published Online: November 30, 2020. doi:10.1001/jamaneurol.2020.4425 Conflict of Interest Disclosures: Dr Zachrison reported receiving grants to the Department of Emergency Medicine, Massachusetts General Hospital from the Agency for Healthcare Research and Quality, the National Institutes of Health, CRICO, and the American College of Emergency Physicians. Dr Khatri reported receiving grants from Cerenovus to the Department of Neurology and Rehabilitation Medicine, University of Cincinnati for investigator-initiated study effort, grants from Nervive to the Department of Neurology and Rehabilitation Medicine, University of Cincinnati, and a coinvestigator grant from the National Institutes of Health; receiving personal fees from Lumosa Therapeutics to the Department of Neurology and Rehabilitation Medicine, University of Cincinnati for consulting and for being a member of the data safety monitoring board; personal fees from DiaMedica Therapeutics to the Department of Neurology and Rehabilitation Medicine, University of Cincinnati for being a member of the scientific advisory board; and personal fees from Bayer for being a national leader of a trial outside the submitted work. No other disclosures were reported. References 1. Nguyen TTM, van den Wijngaard IR, Bosch J, et al. Comparison of prehospital scales for predicting large anterior vessel occlusion in the ambulance setting.  JAMA Neurol. Published online November 30, 2020. doi:10.1001/jamaneurol.2020.4418Google Scholar 2. Shah MN, Bazarian JJ, Lerner EB, et al. The epidemiology of emergency medical services use by older adults: an analysis of the National Hospital Ambulatory Medical Care Survey.  Acad Emerg Med. 2007;14(5):441-447. doi:10.1197/j.aem.2007.01.019 PubMedGoogle ScholarCrossref 3. Malhotra K, Gornbein J, Saver JL. Ischemic strokes due to large-vessel occlusions contribute disproportionately to stroke-related dependence and death: a review.  Front Neurol. 2017;8:651. doi:10.3389/fneur.2017.00651 PubMedGoogle ScholarCrossref 4. Goyal M, Menon BK, van Zwam WH, et al; HERMES collaborators. Endovascular thrombectomy after large-vessel ischaemic stroke: a meta-analysis of individual patient data from five randomised trials.  Lancet. 2016;387(10029):1723-1731. doi:10.1016/S0140-6736(16)00163-X PubMedGoogle ScholarCrossref 5. Freyssenge J, Renard F, Schott AM, et al. Measurement of the potential geographic accessibility from call to definitive care for patient with acute stroke.  Int J Health Geogr. 2018;17(1):1. doi:10.1186/s12942-018-0121-4 PubMedGoogle ScholarCrossref 6. Walsh KB. Non-invasive sensor technology for prehospital stroke diagnosis: current status and future directions.  Int J Stroke. 2019;14(6):592-602. doi:10.1177/1747493019866621 PubMedGoogle ScholarCrossref 7. Ali A, Zachrison KS, Eschenfeldt PC, Schwamm LH, Hur C. Optimization of prehospital triage of patients with suspected ischemic stroke.  Stroke. 2018;49(10):2532-2535. doi:10.1161/STROKEAHA.118.022041 PubMedGoogle ScholarCrossref 8. Howard G, Howard VJ. Twenty years of progress toward understanding the stroke belt.  Stroke. 2020;51(3):742-750. doi:10.1161/STROKEAHA.119.024155 PubMedGoogle ScholarCrossref 9. Balamurugan A, Delongchamp R, Bates JH, Mehta JL. The neighborhood where you live is a risk factor for stroke.  Circ Cardiovasc Qual Outcomes. 2013;6(6):668-673. doi:10.1161/CIRCOUTCOMES.113.000265 PubMedGoogle ScholarCrossref

Journal

JAMA NeurologyAmerican Medical Association

Published: Mar 2, 2021

Keywords: cerebrovascular accident,ischemic stroke,triage

References