Immediate Vascular Imaging Needed for Efficient Triage of Patients With Acute Ischemic Stroke Initially Admitted to Nonthrombectomy Centers
Background and Purpose—Current guidelines for endovascular thrombectomy (EVT) used to select patients for transfer to thrombectomy-capable stroke centers (TSC) may result in unnecessary transfers. We sought to determine the impact of simulated baseline vascular imaging on reducing unnecessary transfers and clinical-imaging factors associated with receiving EVT after transfer.
Methods—We identified patients with stroke transferred for EVT from 30 referring hospitals between 2010 and 2016 who had a referring hospitals brain computed tomography and repeat imaging on TSC arrival available for review. Initial Alberta Stroke Program Early CT scores and TSC vascular occlusion level were assessed. The main outcome variable was receiving EVT at TSC. Models were simulated to derive optimal triaging parameters for EVT.
Results—A total of 508 patients were included in the analysis (mean age, 69±14 years; 42% women). Application at referring hospitals of current guidelines for EVT yielded sensitivity of 92% (95% confidence interval, 0.84–0.96) and specificity of 53% (95% confidence interval, 0.48–0.57) for receiving EVT at TSC. Repeated simulations identified optimal selection criteria for transfer as National Institute of Health Stroke Scale >8 plus baseline vascular imaging (sensitivity=91%; 95% confidence interval, 0.83–0.95; and specificity=80%; 95% confidence interval, 0.75–0.83).
Conclusions—Our findings provide quantitative estimates of the claim that implementing vascular imaging at the referring hospitals would result in significantly fewer futile transfers for EVT and a data-driven framework to inform transfer policies.
The effectiveness of endovascular thrombectomy (EVT) in selected patients with ischemic stroke (IS) likely to have large vessel occlusion (LVO) introduced new challenges for facilitated interfacility transport.1 In resources-constrained care networks, the decision to transfer patients with IS from an initial referring hospital (RH) to a thrombectomy-capable stroke center (TSC) can be challenging, and overtriage results in futile transfers while stretching the resources at the TSC.2 The focused update of American Heart Association/American Stroke Association guidelines suggests that EVT be offered only to LVO patients with a favorable imaging profile, National Institute of Health Stroke Scale (NIHSS) ≥6, and who are able to be treated within 6 hours of last-known well.3 These criteria are widely used in the decision to transfer patients for EVT, in the absence of vascular imaging, while they were meant for decision making at the TSC and in the presence of an LVO.
We sought to analyze our real-world experience as a TSC to evaluate the baseline clinical-imaging factors associated with greater likelihood of receiving EVT after transfer and to model the consequences of implementing routine baseline vascular imaging as a criterion for transfer.
Patients and Methods
Using our local prospective Get-with-the-Guidelines Stroke database, we identified adult patients with stroke transferred from 30 RH between 2010 and 2016 in our regional stroke network and performed an observational, single-center cohort study. Institutional review board approval was obtained for all aspects of this study.
Patients with digital images available for review of the RH brain computed tomography and repeat imaging obtained emergently at the TSC were included.
Imaging was reviewed by 2 independent raters (GB, AL) and rated for Alberta Stroke Program Early CT Score, presence/localization of a dense vessel sign, and presence/localization of a vessel occlusion on computed tomographic angiography/magnetic resonance angiography. Reasons for not receiving EVT were implemented by medical record inspection.
Definition of Variables
Patients were defined as the base case when they met current American Heart Association/American Stroke Association guidelines for EVT at RH (Alberta Stroke Program Early CT Score ≥6; NIHSS ≥6; reaching TSC within 6 hours).
The presence of an LVO was then extrapolated 2 ways: first, we considered as LVO+ only patients with a dense vessel sign in the internal carotid or M1 segment on non-contrast computed tomography at the RH (model 1). A more inclusive approach (model 2) added to the prior definition all those cases who, on arrival at the TSC, had evidence of LVO.
The main outcome variable was pre-defined as receiving EVT at the TSC.
Performance of each RH triage model criteria was evaluated using sensitivity, specificity, positive and negative predictive values, and likelihood ratio confidence intervals.
Uni- and multivariable analyses were then conducted to compare each model’s performances. See the Appendix in the online-only Data Supplement for detailed statistical methods.
A total of 508 patients was included (flowchart of patients’ selection in Figure I in the online-only Data Supplement). Among them, 278 (54.7%) would have met current guidelines for EVT eligibility based on the data available at RH, and 79 of those (79 of 278, 28.4%; 79 of 508, 15.6%) ended up receiving EVT at the TSC (Figure 1 and Table for detail and baseline characteristics and group comparisons).
The most frequent reason for not receiving EVT at TSC in patient who met current guidelines at the RH was the absence of an LVO (n=82, 41%) on arrival, followed by imaging evolution to an unfavorable pattern (39, 20%; Figure II in the online-only Data Supplement).
Identifying Optimal Triage Criteria at the RH
Applying the base case (without vascular imaging information) as triage criteria yielded a sensitivity of 92% (95% confidence interval [CI], 0.84–0.96), specificity of 53% (95% CI, 0.48–0.57), positive predictive value of 28% (95% CI, 0.23–0.34), and negative predictive value of 97% (95% CI, 0.94–0.99) for receiving EVT at the TSC.
Using data simulations accounting for the presumed vascular occlusion status based on a restrictive (model 1), an inclusive (model 2) strategy, and simulated varying minimum qualifying NIHSS from 5 to 15 (see Figure 2), we identified the best trade-off between sensitivity and specificity in model 2 with a minimum NIHSS of 9 points (sensitivity=91%; 95% CI, 0.83–0.95; and specificity=80%; 95% CI, 0.75–0.83). Model 1 achieved a higher specificity but at the cost of a significantly lower sensitivity.
These models were then used to analyze the factors associated with delivery of EVT at TSC arrival. Complete models and models comparisons are presented in the Table II and Figure III in the online-only Data Supplement. In all models, younger age, higher NIHSS, and higher initial Alberta Stroke Program Early CT Score were independent predictors of receiving EVT. Simulating vascular imaging under model 2 at the RH resulted in a significantly greater area under the curve for receiving EVT (P=0.024 and P<0.0001 when comparing to model 1 and the base case, respectively).
We present a data-driven framework to inform transfer policies for patients with IS first seen at non-TSC. The main result of our analyses is a quantitative estimate of the claim that patients are optimally sorted for transfer for possible EVT if vascular imaging was to be implemented at RH: in our population sample, if vascular imaging information were available at the RH, for every 10 patients transferred in intent for EVT, 5 would receive EVT and 2 futile transfers would be prevented.
The demonstrated efficacy of thrombectomy has sparked intense debate on the need for reorganization of pre-hospital and interfacility detection and transfer of stroke patients within stroke networks.4 Most notably, the low efficiency of screening referred cases that end up undergoing EVT, such as the 1% rate seen in the EXTEND-IA trial2 (Extending the Time for Thrombolysis in Emergency Neurological Deficits–Intra-Arterial trial) and the 16% rate in our sample in a single-center real-world setting (86 of 508), suggests that increased attention to triage strategies for cases with suspected LVO should be a high priority for communities. Because the Joint Commission is currently seeking input from the field to finalize TSC-advanced certification criteria, RH vascular imaging implementation seems to be a sine qua non condition for efficient triage strategies.
Furthermore, we have shown here that among those patients meeting current guidelines for EVT eligibility at the RH, only a small proportion of cases selected without the benefit for vascular imaging (79 of 278, 28%) will receive thrombectomy. This is in line with previous work reporting such low numbers of patients receiving thrombectomy for various reasons after transfer,2,5 most of which are not related to the delay caused by the initial evaluation at an RH. In resource-constrained healthcare environments, the specificity of the triaging criteria is equally important to the sensitivity mandated by the proven efficacy of the treatment and the important public health consequences of not treating adequate patients.6 Here, we have shown that implementing vascular imaging at RH for patients meeting current guidelines for EVT eligibility was associated with a significant increase and near doubling in the specificity of those criteria, without any loss of sensitivity.
Preliminary data showing that implementing early vascular imaging at RHs can be done in an efficient way and may meliorate clinical outcomes further strengthen our conclusions.7 Namely, in the era of thrombectomy, all suspected patients with stroke should receive immediate vascular imaging at RH arrival.
Limitations of our study, impacting generalization, include the derivation of the study population from a single center with a substantial number of exclusions, retrospective nature, and the fact that it was conducted in a telestroke network where clinicians may have had additional information driving transfer decisions. Moreover, we cannot exclude a temporal trend in EVT decisions (hence transfers) because positive large EVT trials were published during the study period.
In conclusion, we provide a quantitative analytic framework aimed at providing insight into transfer policies and algorithms for those seeking to improve the interfacility transport of cases of IS with possible LVO. Our results strongly suggest the need for systematic vascular imaging at RH for patients with IS.
Dr Schwamm reports being the principal investigator of an investigator-initiated study of extended window intravenous thrombolysis funded by the National Institutes of Neurological Disorders and Stroke (clinicaltrials.gov/show/NCT01282242) for which Genentech provides alteplase free of charge to Massachusetts General Hospital, as well as supplemental per-patient payments to participating sites; serving as chair of the American Heart Association/American Stroke Association Get With the Guidelines stroke clinical work group and hospital accreditation Science Committee; serving as a stroke systems consultant to the Massachusetts Department of Public Health; and serving as a scientific consultant to LifeImage regarding user interface design and usability and regarding trial design and conduct to Lundbeck (international steering committee, DIAS3 [Efficacy and Safety Study of Desmoteplase to Treat Acute Ischemic Stroke trial] 4 trial), Penumbra (data and safety monitoring committee, Separator 3D trial), and Medtronic (Victory AF and Stroke AF trials). The other authors report no conflicts.
Guest Editor for this article was Georgios Tsivgoulis, MD.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.117.017607/-/DC1.
- Received February 16, 2017.
- Revision received May 23, 2017.
- Accepted June 5, 2017.
- © 2017 American Heart Association, Inc.
- Goyal M,
- Menon BK,
- van Zwam WH,
- Dippel DW,
- Mitchell PJ,
- Demchuk AM,
- et al
- Fuentes B,
- Alonso de Leciñana M,
- Ximénez-Carrillo A,
- Martínez-Sánchez P,
- Cruz-Culebras A,
- Zapata-Wainberg G,
- et al
- Powers WJ,
- Derdeyn CP,
- Biller J,
- Coffey CS,
- Hoh BL,
- Jauch EC,
- et al
- Smith EE,
- Schwamm LH
- Leslie-Mazwi TM,
- Hirsch JA,
- Falcone GJ,
- Schaefer PW,
- Lev MH,
- Rabinov JD,
- et al
- McTaggart RA,
- Yaghi S,
- Baird G,
- Cutting S,
- Hemendinger M,
- Siket M,
- et al