Skip main navigation
×

Endovascular Thrombectomy for Acute Ischemic Strokes

Current US Access Paradigms and Optimization Methodology
Originally publishedhttps://doi.org/10.1161/STROKEAHA.120.028850Stroke. 2020;51:1207–1217

Abstract

Background and Purpose—

Timely access to endovascular thrombectomy (EVT) centers is vital for best acute ischemic stroke outcomes.

Methods—

US stroke-treating centers were mapped utilizing geo-mapping and stratified into non-EVT or EVT if they reported ≥1 acute ischemic stroke thrombectomy code in 2017 to Center for Medicare and Medicaid Services. Direct EVT-access, defined as the population with the closest facility being an EVT-center, was calculated from validated trauma-models adapted for stroke. Current 15- and 30-minute access were described nationwide and at state-level with emphasis on 4 states (TX, NY, CA, IL). Two optimization models were utilized. Model-A used a greedy algorithm to capture the largest population with direct access when flipping 10% and 20% non-EVT to EVT-centers to maximize access. Model-B used bypassing methodology to directly transport patients to the nearest EVT centers if the drive-time difference from the geo-centroid to hospital was within 15 minutes from the geo-centroid to the closest non-EVT center.

Results—

Of 1941 stroke-centers, 713 (37%) were EVT. Approximately 61 million (19.8%) Americans have direct EVT access within 15 minutes while 95 million (30.9%) within 30 minutes. There were 65 (43%) EVT centers in TX with 22% of the population currently within 15-minute access. Flipping 10% hospitals with top population density improved access to 30.8%, while bypassing resulted in 45.5% having direct access to EVT centers. Similar results were found in NY (current, 20.9%; flipping, 34.7%; bypassing, 50.4%), CA (current, 25.5%; flipping, 37.3%; bypassing, 53.9%), and IL (current, 15.3%; flipping, 21.9%; bypassing, 34.6%). Nationwide, the current direct access within 15 minutes of 19.8% increased by 7.5% by flipping the top 10% non-EVT to EVT-capable in all states. Bypassing non-EVT centers by 15 minutes resulted in a 16.7% gain in coverage.

Conclusions—

EVT-access within 15 minutes is limited to less than one-fifth of the US population. Optimization methodologies that increase EVT centers or bypass non-EVT to the closest EVT center both showed enhanced access. Results varied by states based on the population size and density. However, bypass showed more potential for maximizing direct EVT-access. National and state efforts should focus on identifying gaps and tailoring solutions to improve EVT-access.

Introduction

Endovascular thrombectomy (EVT) improves clinical outcomes, reduces disability, and saves lives for patients with acute ischemic strokes (AISs) due to anterior circulation large vessel occlusion (LVO). Several randomized clinical trials1–5 have proven thrombectomy efficacy and safety up to 6 hours from last known well (LKW) as compared with medical management only. Recently, the DAWN trial (DWI or CTP Assessment With Clinical Mismatch in the Triage of Wake-Up and Late Presenting Strokes Undergoing Neurointervention With Trevo)6 and DEFUSE 3 trial (Endovascular Therapy Following Imaging Evaluation for Ischemic Stroke)7 extended thrombectomy efficacy and safety up to 24 hours from LKW in selected patients.

Even with EVT efficacy up to 24 hours from LKW, time remains an important factor that affects EVT outcome.8 Thus, timely and direct access to EVT capable centers remains vital to improving clinical outcomes of patients with AIS due to LVO. Current stroke care algorithms largely prioritize initial transport of patients with stroke to the closest hospital equipped with the ability to administer IV tPA (intravenous tissue-type plasminogen activator). Therefore, the majority of patients only have access to EVT through inter-hospital transfers (drip and ship model), which are associated with significant treatment delays and worsen outcomes.9 Strategies to improve current direct access are necessary to achieve optimal clinical outcomes in patients with strokes. Furthermore, there are no clear data on the current distribution or density of EVT-capable centers in the United States, their coverage areas, and, subsequently, the gaps in patient access to timely thrombectomy.

We evaluated current EVT-capable center distribution and identified the current US population with direct EVT access within 15 and 30 minutes utilizing geomapping techniques. Moreover, we attempted to optimize current direct EVT access in all states, with a focused assessment of 4 states, by deploying 2 optimization methodologies to maximize the endovascular coverage for the states’ population.

Methods

US stroke treating centers were identified using MedPAR data, with centers reporting International Classification of Diseases-10 CM codes for IV thrombolysis along with a diagnostic code for AIS. Stroke centers were stratified as EVT-capable if they reported at least one thrombectomy procedure code for International Classification of Diseases-10 codes for AIS in 2017, or non-EVT if they did not report any procedure code to Centers for Medicare and Medicaid Service (CMS). The data used in the manuscript (CMS MedPAR national hospital data) are part of a publicly available database from Center for Medicare and Medicaid Services and can be obtained at https://www.cms.gov/Research-Statistics-Data-and-Systems/Files-for-Order/LimitedDataSets/MEDPARLDSHospitalNational. Stroke centers were mapped utilizing geomapping techniques with geographic information system (ArcGIS Pro 2.4.0, Esri).

The population census (US Census Bureau 201010) was used, and each state was divided into census tracts with its associated population, and then the population-weighted center point (centroid) of each tract was identified. The closest stroke center and the closest EVT center were identified based on the shortest distance using the geographic information system. The means of transportation were restricted to ground transportation using emergency vehicles, with the assumption that emergency vehicles would not cross state borders. All U-turns were allowed. Direct EVT access, defined as a population with the closest facility being an EVT-capable center within 15 or 30 minutes, were calculated at the nation level from validated trauma models adapted for stroke.11 All drive times were calculated as time taken by an EMT vehicle to reach from the population geocentroid to the respective hospital. We calculated current access at the level of all states.

Two optimization models were utilized (Figure 1). Model A, Flipping model, utilized a greedy algorithm to capture the largest population with direct access when flipping (converting) up to 10%, a minimum of one hospital, and 20% non-EVT to EVT centers to maximize the access. If the 10% rule could not identify any hospital in a given state, 1 hospital was converted to an EVT-capable hospital and results were reported. In such cases, no optimization beyond 10% was performed.

Figure 1.

Figure 1. Illustrates the concept of direct endovascular thrombectomy (EVT) access, optimization using flip model, and optimization using bypass model. Drive times were calculated as time taken by an EMT vehicle to reach from the population geocentroid to the respective hospital. A, If an EVT hospital is the closest hospital and within 15 min of ground distance, then that population qualifies as having the direct EVT access. B, Flip model: if the closest non-EVT hospital is within 15 min, flipping it to an EVT hospital allows for direct EVT access. C, Bypass model: if the closest non-EVT hospital is within 15 min and the drive time difference between population and closest EVT hospital and population and closest non EVT center is within 15 min, the EMS bypasses the non-EVT center in favor of EVT center.

Model B, Bypassing model, used a bypassing methodology to directly transport patients to EVT centers instead of the closest non-EVT center when bypass time was <15 minutes. The bypass time was calculated as the difference between drive-time from the population geo-centroid to the closest EVT center and drive time from the geo-centroid to the closest non-EVT center. Sensitivity analysis using cutoffs of 20-, 25-, and 30-minute bypass times was also performed to assess the potential additional gain beyond 15-minute bypass time.

The optimization models were deployed in all states and described in detail in 4 example states (Texas—TX, New York—NY, California—CA, and Illinois—IL), since they provide an opportunity to examine different optimization scenarios with large number of EVT and non-EVT hospitals as well as a significant heterogeneity in their distribution and population distribution and density. State-level estimates were aggregated to obtain national estimates.

Institutional Review Board approval and patients’ consenting were not necessary as no patients’ data were utilized and only publicly available data was used for the analysis.

Results

Nationwide EVT and Non-EVT Stroke Centers and Their Direct Access Coverage

A total of 1941 stroke centers were identified across the United States. Of these centers, 713 (37%) reported one or more EVT for AIS and were considered EVT-capable centers for this study. This represents a growth of 24% from MedPAR data from 2015, which identified 577 EVT capable hospitals. These centers serve 309 million of the US population based on 2010 US census. Among this population, 61 million (19.8%) have direct access to EVT within 15 minutes. Approximately 95 million (30.9%) have 30-minutes direct access. Figure 2 illustrates the current direct access to EVT capable centers within 15 and 30 minutes across the United States.

Figure 2.

Figure 2. Illustrates the current direct access to endovascular thrombectomy (EVT) capable centers within 15 min (green) and 30 min (yellow) across the mainland United States. American National Standards Institute/Federal Information Processing Standards codes for uniform identification of geographic entities through all federal government agencies are used to calculate the access to a given area, which may vary significantly in size and population distribution and density.

State Level EVT and Non-EVT Stroke Centers and Their Direct Access Coverage

Table 1 shows the states’ population, density, number of stroke treating hospitals, proportion of EVT capable centers, and the current 15-minute access in each state.

Table 1. Population Characteristics, Density, Number of Stroke-Treating Hospitals, Proportion of EVT Capable Centers, and Current EVT Access Within 15 Minutes Across the United States

StateState PopulationDensity (Person per m2)All Stroke-Treating HospitalsEVTEVT%Population Within 15 MinAccess for 15 Min
Alabama4 779 73694.4321031.3%477 25910.0%
Alaska710 2311.26116.7%97 24513.7%
Arizona6 392 01756.3391435.9%1 676 24926.2%
Arkansas2 915 9185619736.8%281 0609.6%
California37 253 956239.12027436.6%9 496 17925.5%
Colorado5 029 19648.5301240.0%827 79916.5%
Connecticut3 574 097738.122522.7%595 44016.7%
Delaware897 934460.86233.3%323 89536.1%
District of Columbia601 7239856.56350.0%151 69125.2%
Florida18 801 310350.61346246.3%6 471 43034.4%
Georgia9 687 653168.4591322.0%1 136 64411.7%
Hawaii1 360 301211.811327.3%240 70017.7%
Idaho1 567 582198450.0%286 38418.3%
Illinois12 830 632231.1853338.8%1 965 39315.3%
Indiana6 483 802181521426.9%790 47312.2%
Iowa3 046 35554.521314.3%158 7385.2%
Kansas2 853 11834.919736.8%369 11212.9%
Kentucky4 339 367109.928932.1%536 18612.4%
Louisiana4 533 372104.9381231.6%936 42020.7%
Maine1 328 36143.19111.1%31 1102.3%
Maryland5 773 552594.8381436.8%1 111 14919.2%
Massachusetts6 547 629839.4421023.8%901 36813.8%
Michigan9 883 640174.8602643.3%2 032 07120.6%
Minnesota5 303 92566.6241458.3%1 154 30521.8%
Mississippi2 967 29763.217529.4%213 1557.2%
Missouri5 988 92787.1401845.0%1 242 15520.7%
Montana989 4156.89333.3%106 10210.7%
Nebraska1 826 34123.817635.3%445 34224.4%
Nevada2 700 55124.616743.8%1 042 71538.6%
New Hampshire1 316 47014710330.0%82 6946.3%
New Jersey8 791 8941195.5612337.7%2 350 45826.7%
New Mexico2 059 1791713430.8%258 00912.5%
New York19 378 102411.21053432.4%4 042 98720.9%
North Carolina9 535 483196.1541629.6%1 561 88316.4%
North Dakota672 5919.76583.3%177 29726.4%
Ohio11 536 504282.3732838.4%2 035 23217.6%
Oklahoma3 751 35154.7221150.0%636 36717.0%
Oregon3 831 07439.922731.8%555 76714.5%
Pennsylvania12 702 379283.9913134.1%2 427 79419.1%
Rhode Island1 052 5671018.18225.0%191 12018.2%
South Carolina4 625 364153.9331339.4%767 96516.6%
South Dakota814 18010.75120.0%33 3034.1%
Tennessee6 346 105153.9371848.6%1 220 56919.2%
Texas25 145 56196.31526542.8%5 552 52022.1%
Utah2 763 88533.612541.7%316 40311.4%
Vermont625 74167.92150.0%36 6965.9%
Virginia8 001 024202.6502040.0%1 977 07524.7%
Washington6 724 540101.2401435.0%1 013 04015.1%
West Virginia1 852 99477.110550.0%199 13310.7%
Wisconsin5 686 986105421433.3%659 04911.6%
Wyoming563 6265.84125.0%30 6965.4%
United States308 745 53887.4194171336.7%61 223 82619.8%

EVT indicates endovascular thrombectomy.

The proportion of EVT centers of all stroke treating centers varies among states; 7 states have only 10% to 25% EVT centers, 30 states have 25% to 40%, and only 14 states have >40% of all of their stroke-treating hospitals as EVT centers. Similarly, the 15-minute direct access to EVT centers varies between states, ranging from 2.3% to 38.6% of the states’ populations. Nine states have <10% coverage, 34 states have 10% to -25% coverage, and only 8 states have >25% coverage.

Nationwide and State Level Optimization Models

Flipping the most impactful 10% of the non-EVT hospitals to EVT capable centers resulted in an absolute gain in direct access ranging between 2.8% and 28.1% among all states (Table 2). The majority of the states gained 4.8% to 7.6% by flipping 10%. An additional 10% flip (up to 20%) added less overall value with a range of 2.1% and 9.2%, with the majority of the states gaining between 2.5% and 5.2%. The greedy algorithm utilized in the flip model showed the top 10 hospitals identified by the algorithm to optimize the access in the 4 example states (Table I in the Data Supplement).

Table 2. Illustrates the Results of Optimization Using Both Flipping and Bypass Methods Across All US States

StateDirect Access Within 15 Min—N (%)Flipping 10% HospitalsFlipping 20% Hospitals15 Min Bypass
Absolute Gain—N (%)Total Coverage—(%)Absolute Gain—N (%)Total Coverage—N (%)Absolute Gain—N (%)Total Coverage—N (%)
Alabama477 259 (10%)228 614 (4.8%)705 873 (14.8%)181 634 (3.8%)887 507 (18.6%)370 299 (7.7%)847 558 (17.7%)
Alaska*97 245 (13.7%)65 820 (9.3%)163 065 (23%)N/AN/A116 614 (16.4%)213 859 (30.1%)
Arizona1676 249 (26.2%)531 455 (8.3%)2 207 704 (34.5%)277 106 (4.3%)2 484 810 (38.8%)1 568 211 (24.5%)3 244 460 (50.7%)
Arkansas281 060 (9.6%)82 333 (2.8%)363 393 (12.4%)68 764 (2.4%)432 157 (14.8%)210 228 (7.2%)491 288 (16.8%)
California9 496 179 (25.5%)4 379 065 (11.8%)1 387 5244 (37.3%)3 065 878 (8.2%)16 941 122 (45.5%)10 568 703 (28.4%)20 064 882 (53.9%)
Colorado827 799 (16.5%)245 880 (4.9%)1 073 679 (21.4%)203 556 (4%)1 277 235 (25.4%)523 018 (10.4%)1 350 817 (26.9%)
Connecticut595 440 (16.7%)239 050 (6.7%)834 490 (23.4%)94 464 (2.6%)928 954 (26%)462 494 (12.9%)1 057 934 (29.6%)
Delaware*323 895 (36.1%)79 593 (8.9%)403 488 (45%)N/AN/AN/AN/A
District of Columbia*151 691 (25.2%)168 892 (28.1%)320 583 (53.3%)N/AN/A259 091 (43.1%)410 782 (68.3%)
Florida6 471 430 (34.4%)1 193 864 (6.3%)7 665 294 (40.7%)842 054 (4.5%)8 507 348 (45.2%)2 851 487 (15.2%)9 322 917 (49.6%)
Georgia1 136 644 (11.7%)842 585 (8.7%)1 979 229 (20.4%)508 013 (5.2%)2 487 242 (25.6%)1 281 936 (13.2%)2 418 580 (24.9%)
Hawaii240 700 (17.7%)129 558 (9.5%)370 258 (27.2%)87 821 (6.5%)458 079 (33.7%)301 624 (22.2%)542 324 (39.9%)
Idaho*286 384 (18.3%)88 508 (5.6%)374 892 (23.9%)N/AN/A9199 (0.6%)295 583 (18.9%)
Illinois1 965 393 (15.3%)841 220 (6.6%)280 6613 (21.9%)655 946 (5.1%)3 462 559 (27%)2 478 239 (19.3%)4 443 632 (34.6%)
Indiana790 473 (12.2%)383 562 (5.9%)1 174 035 (18.1%)324 530 (5%)1 498 565 (23.1%)751 094 (11.6%)1 541 567 (23.8%)
Iowa158 738 (5.2%)174 444 (5.7%)333 182 (10.9%)108 523 (3.6%)441 705 (14.5%)67 844 (2.2%)226 582 (7.4%)
Kansas369 112 (12.9%)96 693 (3.4%)465 805 (16.3%)85 637 (3%)551 442 (19.3%)282 920 (9.9%)652 032 (22.8%)
Kentucky536 186 (12.4%)208 244 (4.8%)744 430 (17.2%)166 456 (3.8%)910 886 (21%)128 068 (3%)664 254 (15.4%)
Louisiana936 420 (20.7%)384 167 (8.5%)1 320 587 (29.2%)158 982 (3.5%)1 479 569 (32.7%)864 304 (19.1%)1 800 724 (39.8%)
Maine31 110 (2.3%)63 997 (4.8%)95 107 (7.1%)62 231 (4.7%)157 338 (11.8%)64 720 (4.9%)95 830 (7.2%)
Maryland1 111 149 (19.2%)349 417 (6.1%)1 460 566 (25.3%)421 212 (7.3%)1 881 778 (32.6%)837 480 (14.5%)1 948 629 (33.7%)
Massachusetts901 368 (13.8%)462 371 (7.1%)1 363 739 (20.9%)380 060 (5.8%)1 743 799 (26.7%)1 145 724 (17.5%)2 047 092 (31.3%)
Michigan2 032 071 (20.6%)501 891 (5.1%)2 533 962 (25.7%)462 869 (4.7%)2 996 831 (30.4%)1 185 051 (12%)3 217 122 (32.6%)
Minnesota1 154 305 (21.8%)182 427 (3.4%)1 336 732 (25.2%)112 839 (2.1%)1 449 571 (27.3%)451 037 (8.5%)1 605 342 (30.3%)
Mississippi213 155 (7.2%)94 739 (3.2%)307 894 (10.4%)79 340 (2.7%)387 234 (13.1%)184 341 (6.2%)397 496 (13.4%)
Missouri1 242 155 (20.7%)236 558 (3.9%)1 478 713 (24.6%)205 273 (3.4%)1 683 986 (28%)800 017 (13.4%)2 042 172 (34.1%)
Montana106 102 (10.7%)67 628 (6.8%)173 730 (17.5%)38 527 (3.9%)212 257 (21.4%)15 589 (1.6%)121 691 (12.3%)
Nebraska445 342 (24.4%)142 201 (7.8%)587 543 (32.2%)86 611 (4.7%)674 154 (36.9%)217 369 (11.9%)662 711 (36.3%)
Nevada1 042 715 (38.6%)143 751 (5.3%)1 186 466 (43.9%)141 328 (5.2%)1 327 794 (49.1%)692 175 (25.6%)1 734 890 (64.2%)
New Hampshire82 694 (6.3%)70 560 (5.4%)153 254 (11.7%)68 308 (5.2%)221 562 (16.9%)39 615 (3%)122 309 (9.3%)
New Jersey2 350 458 (26.7%)628 312 (7.1%)2 978 770 (33.8%)532 744 (6.1%)3 511 514 (39.9%)2 264 408 (25.8%)4 614 866 (52.5%)
New Mexico258 009 (12.5%)64 094 (3.1%)322 103 (15.6%)52 754 (2.6%)374 857 (18.2%)116 095 (5.6%)374 104 (18.1%)
New York4 042 987 (20.9%)2 682 906 (13.8%)6 725 893 (34.7%)1 551 134 (8%)8 277 027 (42.7%)5 713 730 (29.5%)9 756 717 (50.4%)
North Carolina1 561 883 (16.4%)618 583 (6.5%)2 180 466 (22.9%)407 209 (4.3%)2 587 675 (27.2%)692 099 (7.3%)2 253 982 (23.7%)
North Dakota*177 297 (26.4%)32 718 (4.9%)210 015 (31.3%)N/AN/AN/AN/A
Ohio2 035 232 (17.6%)565 627 (4.9%)2 600 859 (22.5%)397 942 (3.4%)2 998 801 (25.9%)1 564 428 (13.6%)3 599 660 (31.2%)
Oklahoma636 367 (17%)138 067 (3.7%)774 434 (20.7%)86 627 (2.3%)861 061 (23%)464 887 (12.4%)1 101 254 (29.4%)
Oregon555 767 (14.5%)190 709 (5%)746 476 (19.5%)82 157 (2.1%)828 633 (21.6%)399 121 (10.4%)954 888 (24.9%)
Pennsylvania2 427 794 (19.1%)887 192 (7%)3 314 986 (26.1%)546 071 (4.3%)3 861 057 (30.4%)1 751 254 (13.8%)4 179 048 (32.9%)
Rhode Island191 120 (18.2%)124 740 (11.9%)315 860 (30.1%)96 589 (9.2%)412 449 (39.3%)317 344 (30.1%)508 464 (48.3%)
South Carolina767 965 (16.6%)194 006 (4.2%)961 971 (20.8%)148 275 (3.2%)1 110 246 (24%)397 515 (8.6%)1 165 480 (25.2%)
South Dakota*33 303 (4.1%)55 359 (6.8%)88 662 (10.9%)N/AN/A80 489 (9.9%)113 792 (14%)
Tennessee1 220 569 (19.2%)297 472 (4.7%)1 518 041 (23.9%)165 444 (2.6%)1 683 485 (26.5%)551 114 (8.7%)1 771 683 (27.9%)
Texas5 552 520 (22.1%)2 193 672 (8.7%)7 746 192 (30.8%)1 317 236 (5.2%)9 063 428 (36%)5 895 021 (23.4%)11 447 541 (45.5%)
Utah316 403 (11.4%)224 384 (8.1%)540 787 (19.5%)147 231 (5.3%)688 018 (24.8%)422 337 (15.3%)738 740 (26.7%)
Vermont*36 696 (5.9%)22 881 (3.7%)59 577 (9.6%)N/AN/AN/AN/A
Virginia1 977 075 (24.7%)554 354 (6.9%)2 531 429 (31.6%)442 564 (5.5%)2 973 993 (37.1%)995 528 (12.4%)2 972 603 (37.1%)
Washington1 013 040 (15.1%)412 652 (6.1%)1 425 692 (21.2%)222 644 (3.3%)1 648 336 (24.5%)698 297 (10.4%)1 711 337 (25.5%)
West Virginia*199 133 (10.7%)80 611 (4.4%)279 744 (15.1%)N/AN/AN/AN/A
Wisconsin659 049 (11.6%)405 913 (7.1%)1 064 962 (18.7%)203 534 (3.6%)1 268 496 (22.3%)637 456 (11.2%)1 296 505 (22.8%)
Wyoming*30 696 (5.4%)41 614 (7.4%)72 310 (12.8%)N/AN/AN/AN/A
United States61 223 826 (19.8%)23 094 953 (7.5%)84 318 779 (27.3%)1 528 8117 (5%)99 606 896 (32.3%)51 689 614 (16.7%)11 291 3440 (36.5%)

*In case no hospitals were identified based on 10% threshold, at least 1 hospital was flipped and results were reported. In such cases, further optimization using 20% threshold was not attempted.

†In states where bypass model using 15 min threshold did not identify any population center that would benefit from bypassing, no results were reported.

Bypassing non-EVT centers resulted in additional coverage that ranged from 0.6% to 43.1% for all states (Table 2). Most states gained between 6.7% and 15.8% of coverage. The bypassing model was not feasible in 5 states given the low base numbers of stroke centers and their distribution.

Nationwide, the current direct access of 19.8% increased by 7.5%, approximately an additional 23 million people, to a new access of 27.3% by flipping the top 10% non-EVT hospitals to EVT-capable hospitals in all states (Table 3). Bypassing non-EVT centers by 15 minutes to deliver patients to EVT centers resulted in a 16.7% gain in population coverage, around 52 million, for a 36.5% new total coverage. Sensitivity analysis using bypass cutoffs of 20, 25, and 30 minutes resulted in national incremental coverage gain of 3.1%, 2%, and 1.1%, increasing overall direct EVT access to 42.7% at the 30-minute cutoff. Table II in the Data Supplement demonstrates the incremental coverage gain at 20-, 25-, and 30-minute cutoffs in all states.

Table 3. Describes the Population Access to EVT Capable Centers in the United States and Across 4 States, Both Current and After Optimization Using (1) Flip and (2) Bypass Models

United StatesTexasNew YorkCaliforniaIllinois
Population308 745 53825 145 56119 378 10237 253 95612 830 632
Land area (square miles)3 531 905261 23147 126155 77955 518
Population density (per square mile)87.496.3411.2239.1231.1
EVT capable centers71365347433
Non-EVT centers1228877112852
Current status
 Direct access to EVT within 15 min61 223 826 (19.8%)5 552 520 (22.1%)4 042 987 (20.9%)9 496 179 (25.5%)1 965 393 (15.3%)
Optimization with flip model*
 Additional coverage with 10% flipped2 094 953 (7.5%)*2 193 672 (8.7%)2 682 906 (13.8%)4 379 065 (11.8%)841 220 (6.6%)
 Total coverage84 318 779 (27.3%)7 746 192 (30.8%)6 725 893 (34.7%)13 875 244 (37.3%)2 806 613 (21.9%)
 Additional coverage with 20% flipped15 288 117 (5.0%)*1 317 236 (5.2%)1 551 134 (8.0%)3 065 878 (8.2%)655 946 (5.1%)
 Total coverage99 606 896 (32.3%)9 063 428 (36.0%)8 277 027 (42.7%)16 941 122 (45.5%)3 462 559 (27.0%)
Optimization with bypass model
 Additional coverage51 689 614 (16.7%)5 895 021 (23.4%)5 713 730 (29.5%)10 568 703 (28.4%)2 478 239 (19.3%)
 Total coverage112 913 440 (36.5%)11 447 541 (45.5%)9 756 717 (50.4%)20 064 882 (53.9%)4 443 632 (34.6%)

EVT indicates endovascular thrombectomy.

*Top 10% and 20% hospitals were flipped on state-level (each state individually).

†Bypass to the closest EVT center when drivetimes to EVT center does not exceed the drive time to non-EVT center by 15 min.

Detailed Current and Optimized Access Comparisons in Example States

Figure 3A and 3B represent the EVT coverage optimization using both flipping and bypass models in 4 example states. A total of 152 stroke centers in Texas were identified, 65 of which are recognized as EVT capable, and 87 as non-EVT hospitals. The state population was 25 145 561 people based on 2010 US Census. In Texas, 5.5 million (22.1%) have current direct access to EVT capable centers within 15 minutes. Flipping 10% of the non-EVT to EVT centers (9 hospitals) resulted in an additional 8.7% increase in the 15-minute direct access coverage and a 2.1 million population gain upping the Texans covered to 30.8% of the state population (Table 3; Figure 3A-1). A 15-minute bypass beyond the non-EVT to the EVT center added 23.4%, ≈6 million, of Texas population to EVT direct access for a total of 45.5% (Figure 3B-1).

Figure 3.

Figure 3. Illustrates the current direct access within 15 min at each state and the gain from the 2 hypothetical optimization models in 4 states (Texas, New York, California, and Illinois). Model A utilized a greedy algorithm to capture the largest population with direct access when flipping 10% (sky blue) and 20% (dark blue) non-endovascular thrombectomy (EVT) to EVT centers to maximize access. Model B used bypassing methodology to directly transport patients to EVT centers within 15 min from the closest non-EVT center (orange).

In the state of New York, 105 stroke centers provide stroke care to 19 378 102 individuals, 34 of which are designated as EVT-capable centers. Current direct access within 15 minutes is available to 4 million (20.9%), which increased to 6.7 million (34.7%), a gain of 13.8%, when the top 10% of non-EVT centers (7 hospitals) were flipped in the hypothetical scenario (Table 3; Figure 3A-2). Employing a 15-minute bypass strategy provided direct access to 9.8 million, 50.4% of the population of the state of New York, an increase of 29.5% (Table 3; Figure 3B-2).

There are 202 stroke centers in California, of which 74 are EVT capable, catering to a population of 37 253 956. Current direct access to EVT within 15 minutes is available to 9.5 million (25.5%) individuals. Optimization by flipping the top 10% of non-EVT hospitals (13 hospitals) resulted in an 11.8% additional coverage to increase the direct access to 13.9 million (37.3%) people, whereas optimization with 15-minute bypass resulted in a 28.4% gain over the current access and a new direct access to 20 million, 53.9% of the California population (Table 3; Figure 3A-3 and 3B-3).

The state of Illinois and its 12 830 632 population are served by 33 EVT and 52 non-EVT stroke centers. Two million (15.3%) individuals have current direct access to EVT within 15 minutes, which increased to 2.8 million (21.9%) when the top 10% of non-EVT hospitals (5 hospitals) were converted to EVT capable hospitals, while optimization with 15-minute bypass resulted in direct access to 4.4 million, 34.6% of the population (Table 3; Figure 3A-4 and 3B-4).

Discussion

Current direct EVT access in the United States is suboptimal under predominate EMS routing protocols. Our analysis showed that only one-fifth of the US population has direct access to an EVT capable center within 15 minutes of transportation time. These results were consistent across the nation and by state level. Our data suggested that of all stroke centers, nationwide only 37% are capable to perform thrombectomy based on their procedure code reporting. The proportion of EVT-capable centers varied at the state level. This proportion exceeded 25% in around half of the states and topped 40% in about one quarter of the states. The population direct access coverage similarly varied but was still overall suboptimal. Only in 8 states did the coverage exceed 25% of the population while in 42 states it was <25%, with 9 states having coverage for <10% of the population.

These results reflect a limited access to an effective treatment modality that would improve clinical outcomes in patients with large strokes and prevent potential devastating disability. Current stroke care algorithms result in a large proportion of patients being taken to non-EVT centers and subsequently transferred for EVT which results in significant inter-facility transfer delays and worse outcomes. In a pooled meta-analysis of individual patients’ data from 5 randomized clinical trials assessing thrombectomy efficacy and safety, transfer to EVT capable center was associated with treatment-delay of 95 minutes (time from LKW to procedure: transfer: 260 [215–310] minutes versus direct: 165 [25–226] minutes; P<0.001).8 In an another analysis of a prospective registry, Froehler et al9 found that transfer to an EVT-capable hospital resulted in a median delay of 109.5 minutes of time from LKW to procedure and decreased rates of functional independence (direct: 60% versus transfers: 52%, unadjusted OR, 1.38 [95% CI, 1.06–1.79]; P=0.02).

Strategies to improve current EVT accessibility are needed. We used 2 different methodologies to optimize EVT access and maximize the population coverage. Both methodologies were effective. In the first hypothetical model, 10% and 20% of non-EVT stroke treating hospitals were flipped in all states using a greedy algorithm, which identifies centers with the highest population that would have direct access to thrombectomy should the center be flipped to EVT capable. Flipping 10% of the hospitals resulted in about 7% gain nationwide with similar results across the states. Only 4 states gained >10% additional coverage with this model. Further flipping of an additional 10% of the centers was less effective with the majority of the states gaining <5% additional coverage.

The other optimization methodology of bypassing non-EVT hospitals to the closest EVT-capable hospital within 15 minutes resulted in almost doubling the current direct access nationwide. Thirty states gained >10% additional coverage with this methodology with 9 of them gaining >20% in additional population coverage.

Baseline population density, urban versus suburban areas, and current distribution of thrombectomy centers were the major factors in determining the additional value of flipping and bypassing. Focusing on the 4 large example states, flipping resulted in ≈7% to 14% increase in direct access to an EVT-capable center within 15 minutes while bypassing resulted in additional coverage ranging between 19% and 28%. Overall, for the majority of the states, bypassing resulted in better EVT coverage than flipping.

While some states do employ legislatively directed efforts to direct patients with potential LVO to the closest EVT facility, most of the current systems in stroke care are designed to provide IV thrombolysis at the earliest time point and transfer patients to the EVT capable center in the drip and ship model. This system is associated with significant challenges and delays in EVT delivery as the interhospital transfer process takes a significant time, starting with recognition of a patient who has a large vessel occlusion.12 The lack of persistently utilized and efficient protocols to facilitate identification of patients with LVO at non-EVT centers then transferring to EVT-capable centers is a major source for delay in treatment. Additionally, obtaining a ground or air ambulance unit for the secondary transfer, particularly if IV tPA has been given, is a challenge to efficient transfer in resource poor regions, including rural areas.

Different strategies have been proposed to increase the access to thrombectomy. Enabling direct transfer of patients with potential LVO by bypassing non-EVT centers is one of these strategies. Scoring systems for prehospital screening of LVO have shown a good sensitivity and specificity in small scale nonrandomized studies,13,14 and randomized trials are ongoing to confirm their utility in identifying LVOs and improving clinical outcomes.15 Prehospital care via mobile stroke units that drive to patients’ locations within a certain radius to administer IV tPA can also help to identify patients with potential LVO. This results in patients being transferred for longer distances to EVT-capable centers by bypassing non-EVT centers, which may help bridge the gap in access to EVT centers.16–18 Another strategy is to have mobile endovascular stroke teams available to travel to non-EVT centers in the trip and treat model.19 Proposals for increasing the number of EVT capable centers by converting non-EVT centers to EVT-capable centers have also been made. Importantly, education to improve early detection and efficient secondary transfer of patients with LVO is necessary regardless of which combination of strategies are used to enhance direct access to EVT since a large proportion of patients with acute stroke will inevitably arrive via privately owned vehicles.20 Telestroke services can be proposed as a potential solution to expedite early treatment with tPA and transfer to EVT-capable centers.21

The most recent effort to map EVT access in the US was done using data from almost 10 years ago before the successful thrombectomy trials were conducted.22 These thrombectomy trials resulted in a significant change in EVT indications, utilization, and the need for more accessibility. The number of centers performing thrombectomy has also subsequently increased.23 While proposing that 56% have ground coverage, the prior analysis assessed EVT access within 60 minutes, which is considered too long of an elapsed time for transferring patients with potential LVO AIS. The study also proposed that 85% of the population has 60 minutes coverage by air transportation. The latter, however, is not widely utilized in transportation of patients with AIS from the onset location. Prior attempts at mapping access to stroke care were aimed at assessing simulated outcomes.24 Our study, however, focused on identifying and expanding current EVT access. This is the first attempt in the literature to assess bypassing non-EVT centers in favor of EVT-capable centers using predetermined time limits. Prior studies have attempted to assess the flip model in the United States for thrombolysis25 but not for EVT. This study is also the first attempt to evaluate the utility of this approach to enhance EVT access. We have utilized 15 minute bypass time cutoff as it is consistent with previously established legislations as well as the recommendation from the American Heart Association/ASA quality improvement initiative Mission: Lifeline Stroke.26 This also accounts for the decay in likelihood of functional independence with EVT as time progresses. We calculated bypass access at 20-, 25-, and 30-minute threshold as a sensitivity analysis, which demonstrated low yielding incremental gain over the EVT access coverage obtained using the 15-minute threshold, representing distribution of EVT capable centers closer to the densely populated areas. This also demonstrates that longer bypass times may not provide a significant incremental gain in coverage, which may be of consideration while identification of most appropriate optimization method for a given area or population. These characteristics, while observed across the United States, may not hold for other countries with significantly different population distribution and density and where longer transfer times may be warranted for optimization of coverage.

There are several limitations in this study. First, our analysis focused on increasing access to EVT and did not simulate outcomes. The relative effect of flipping versus bypass on patient outcomes needs further study and needs to be factored in to any regional triage strategy. Next, the results presented in this study are hypothetical and do not account for the costs and logistics involved in flipping a non-EVT center to an EVT-capable center, including the availability of neuro-trained interventionists, equipped angiography suites and technical support logistics. We also did not calculate the potential costs associated with training the EMS staff for in-field identification of large vessel occlusion or savings associated with improvement in outcomes and reducing severe disabilities. For both models, we assumed current access in as-is state, without taking into account state and local legislations affecting EMS directed transfer of patients with stroke. We used an International Classification of Diseases-10-CM code reporting of at least one EVT procedure to identify EVT centers. We cannot account for the possibility of errors in data submission, but a superior source for identification of EVT centers does not currently exist. Using MedPAR data from CMS also excluded patients who are not covered by CMS; however, the likelihood of hospitals providing EVT while not including at least 1 patient that is insured by CMS is low. Since the source list for EVT was available only from the year of 2017 for this analysis, hospitals that started providing EVT after the 2017 reporting period for CMS were also not included in the analysis. Considering a center to be EVT-capable if they reported one thrombectomy procedure for stroke is a liberal inclusive approach. This could have also led to an overestimation of the current EVT access since some of these low volume centers may not provide endovascular thrombectomy on a routine basis. We do not report the characteristics of each center in terms of their coverage hours, number of procedures performed in a year, the quality of stroke care, or patient-level outcomes. The population data were obtained from US Census 2010 and may not accurately represent the current population distribution and locations of the population centers.

Also, we did not explore the distribution of population at high risk of stroke and their effects on potential choice for the flip candidates. Calculated targeting of stroke centers in geographically challenged regions could potentially increase the yield in terms of population coverage, and based on heterogeneity between states and regions, the best strategy must be individualized to the regional distribution of at-risk patients with stroke, hospital facilities, medical expertise, EMS systems, and patient preferences. Bypass protocols also require efficient prehospital identification of patients with potential LVO with special training of EMS responders or wider implementation of mobile stroke units. Strategies to bypass the non-EVT centers in favor of EVT-capable centers using various in-field LVO assessment algorithms have shown to have varied effect on patients’ over-triage, as well as on time taken to reach the EVT capable centers.27 This was not the focus of our analysis and may require further exploration.

Furthermore, while the conditions for transport were considered in the model, we assumed that the clinical status of the patient would allow for the further transfer to an EVT capable center. Thus, our bypass models may not be applicable to patients who are critically unstable and would not tolerate longer transfer times. We did not perform modeling for a combined flipping and bypass approach, which may provide additive additional EVT coverage. Travel times were approximated from a single point within the tract. Although census tracts are small in most cases, population is spread throughout the tract. Population-weighted center points provided by the US Census (2010) were used to provide the best approximations. The assumption of not crossing state borders may not always hold true, especially in certain areas of the country and may have led to conservative estimates of current and optimized EVT access.

In summary, our results showed that for most of states, the bypass approach resulted in better direct access to EVT-capable centers. This approach has the added benefit of ease of implementation and requires less time and resources. However, some states indeed demonstrated better access by flipping. This signifies the heterogeneity among different optimization approaches demonstrated based on population density, urban versus suburban areas, and current distribution of EVT-capable centers in different states. These methodologies, when applied locoregionally, can help identify the best approach to maximize EVT access and in conjunction with the availability of resources to improve local health policy development.

Conclusions

Current direct EVT access within 15 minutes is limited to one-fifth of the US population. Optimization methodologies that increase EVT centers or bypass non-EVT centers to the closest EVT center both showed enhanced access. Results varied by states based on the population size and density. However, bypass showed more potential for maximizing direct EVT access. National and state efforts should focus on identifying gaps and tailoring solutions to improve EVT access.

Acknowledgments

The MedPAR data to identify stroke centers and EVT capable centers based on International Classification of Diseases (ICD)-10-CM treatment codes reporting was acquired from Stryker Neurovascular.

Footnotes

Presented in part at the International Stroke Conference, Los Angeles, CA, February 19–21, 2020.

The Data Supplement is available with this article at https://www.ahajournals.org/doi/suppl/10.1161/STROKEAHA.120.028850.

Correspondence to Amrou Sarraj, MD, Department of Neurology, UT McGovern Medical School, 6431 Fannin St, MSB 7.044; Houston, TX 77030. Email

References

  • 1. Berkhemer OA, Fransen PS, Beumer D, van den Berg LA, Lingsma HF, Yoo AJ, et al.; MR CLEAN Investigators. A randomized trial of intraarterial treatment for acute ischemic stroke.N Engl J Med. 2015; 372:11–20. doi: 10.1056/NEJMoa1411587CrossrefMedlineGoogle Scholar
  • 2. Jovin TG, Chamorro A, Cobo E, de Miquel MA, Molina CA, Rovira A, et al.; REVASCAT Trial Investigators. Thrombectomy within 8 hours after symptom onset in ischemic stroke.N Engl J Med. 2015; 372:2296–2306. doi: 10.1056/NEJMoa1503780CrossrefMedlineGoogle Scholar
  • 3. Goyal M, Demchuk AM, Menon BK, Eesa M, Rempel JL, Thornton J, et al.; ESCAPE Trial Investigators. Randomized assessment of rapid endovascular treatment of ischemic stroke.N Engl J Med. 2015; 372:1019–1030. doi: 10.1056/NEJMoa1414905CrossrefMedlineGoogle Scholar
  • 4. Saver JL, Goyal M, Bonafe A, Diener HC, Levy EI, Pereira VM, et al.; SWIFT PRIME Investigators. Stent-retriever thrombectomy after intravenous t-PA vs. t-PA alone in stroke.N Engl J Med. 2015; 372:2285–2295. doi: 10.1056/NEJMoa1415061CrossrefMedlineGoogle Scholar
  • 5. Campbell BC, Mitchell PJ, Kleinig TJ, Dewey HM, Churilov L, Yassi N, et al.; EXTEND-IA Investigators. Endovascular therapy for ischemic stroke with perfusion-imaging selection.N Engl J Med. 2015; 372:1009–1018. doi: 10.1056/NEJMoa1414792CrossrefMedlineGoogle Scholar
  • 6. Nogueira RG, Jadhav AP, Haussen DC, Bonafe A, Budzik RF, Bhuva P, et al.; DAWN Trial Investigators. Thrombectomy 6 to 24 hours after stroke with a mismatch between deficit and infarct.N Engl J Med. 2018; 378:11–21. doi: 10.1056/NEJMoa1706442CrossrefMedlineGoogle Scholar
  • 7. Albers GW, Marks MP, Kemp S, Christensen S, Tsai JP, Ortega-Gutierrez S, et al.; DEFUSE 3 Investigators. Thrombectomy for stroke at 6 to 16 hours with selection by perfusion imaging.N Engl J Med. 2018; 378:708–718. doi: 10.1056/NEJMoa1713973CrossrefMedlineGoogle Scholar
  • 8. Saver JL, Goyal M, van der Lugt A, Menon BK, Majoie CB, Dippel DW, et al.; HERMES Collaborators. Time to treatment with endovascular thrombectomy and outcomes from ischemic stroke: a meta-analysis.JAMA. 2016; 316:1279–1288. doi: 10.1001/jama.2016.13647CrossrefMedlineGoogle Scholar
  • 9. Froehler MT, Saver JL, Zaidat OO, Jahan R, Aziz-Sultan MA, Klucznik RP, et al.; STRATIS Investigators. Interhospital transfer before thrombectomy is associated with delayed treatment and worse outcome in the STRATIS Registry (Systematic Evaluation of Patients Treated With Neurothrombectomy Devices for Acute Ischemic Stroke).Circulation. 2017; 136:2311–2321. doi: 10.1161/CIRCULATIONAHA.117.028920LinkGoogle Scholar
  • 10. United States Census Bureau. Centers of Population for the 2010 Census.Census.gov. Available at: https://www.census.gov/geographies/reference-files/2010/geo/2010-centers-population.html. Published 2019. Accessed August 15, 2019Google Scholar
  • 11. Branas CC, MacKenzie EJ, ReVelle CS. A trauma resource allocation model for ambulances and hospitals.Health Serv Res. 2000; 35:489–507.MedlineGoogle Scholar
  • 12. Shah S, Xian Y, Sheng S, Zachrison KS, Saver JL, Sheth KN, et al.. Use, Temporal trends, and outcomes of endovascular therapy after interhospital transfer in the United States.Circulation. 2019; 139:1568–1577. doi: 10.1161/CIRCULATIONAHA.118.036509LinkGoogle Scholar
  • 13. Pérez de la Ossa N, Carrera D, Gorchs M, Querol M, Millán M, Gomis M, et al.. Design and validation of a prehospital stroke scale to predict large arterial occlusion.Stroke. 2014; 45:87–91.LinkGoogle Scholar
  • 14. Teleb MS, Ver Hage A, Carter J, Jayaraman MV, McTaggart RA. Stroke vision, aphasia, neglect (VAN) assessment-a novel emergent large vessel occlusion screening tool: pilot study and comparison with current clinical severity indices.J Neurointerv Surg. 2017; 9:122–126. doi: 10.1136/neurintsurg-2015-012131CrossrefMedlineGoogle Scholar
  • 15. Abilleira S, Pérez de la Ossa N, Jiménez X, Cardona P, Cocho D, Purroy F, et al.. Transfer to the Local Stroke Center versus Direct Transfer to Endovascular Center of Acute Stroke Patients with Suspected Large Vessel Occlusion in the Catalan Territory (RACECAT): study protocol of a cluster randomized within a cohort trial.Int J Stroke. 2019; 14:734–744. doi: 10.1177/1747493019852176Google Scholar
  • 16. Walter S, Kostopoulos P, Haass A, Keller I, Lesmeister M, Schlechtriemen T, et al.. Diagnosis and treatment of patients with stroke in a mobile stroke unit versus in hospital: a randomised controlled trial.Lancet Neurol. 2012; 11:397–404. doi: 10.1016/S1474-4422(12)70057-1CrossrefMedlineGoogle Scholar
  • 17. Ebinger M, Winter B, Wendt M, Weber JE, Waldschmidt C, Rozanski M, et al.; STEMO Consortium. Effect of the use of ambulance-based thrombolysis on time to thrombolysis in acute ischemic stroke: a randomized clinical trial.JAMA. 2014; 311:1622–1631. doi: 10.1001/jama.2014.2850CrossrefMedlineGoogle Scholar
  • 18. Parker SA, Bowry R, Wu TC, Noser EA, Jackson K, Richardson L, et al.. Establishing the first mobile stroke unit in the United States.Stroke. 2015; 46:1384–1391. doi: 10.1161/STROKEAHA.114.007993LinkGoogle Scholar
  • 19. Wei D, Oxley TJ, Nistal DA, Mascitelli JR, Wilson N, Stein L, et al.. Mobile interventional stroke teams lead to faster treatment times for thrombectomy in large vessel occlusion.Stroke. 2017; 48:3295–3300. doi: 10.1161/STROKEAHA.117.018149LinkGoogle Scholar
  • 20. Adeoye O, Lindsell C, Broderick J, Alwell K, Jauch E, Moomaw CJ, et al.. Emergency medical services use by stroke patients: a population-based study.Am J Emerg Med. 2009; 27:141–145. doi: 10.1016/j.ajem.2008.02.004CrossrefMedlineGoogle Scholar
  • 21. Demaerschalk BM, Berg J, Chong BW, Gross H, Nystrom K, Adeoye O, et al.. American telemedicine association: telestroke guidelines.Telemed J E Health. 2017; 23:376–389. doi: 10.1089/tmj.2017.0006Google Scholar
  • 22. Adeoye O, Albright KC, Carr BG, Wolff C, Mullen MT, Abruzzo T, et al.. Geographic access to acute stroke care in the United States.Stroke. 2014; 45:3019–3024. doi: 10.1161/STROKEAHA.114.006293LinkGoogle Scholar
  • 23. Mack WJ, Mocco J, Hirsch JA, Chen M, Elijovich L, Tarr RW, et al.. Thrombectomy stroke centers: the current threat to regionalizing stroke care.J Neurointerv Surg. 2018; 10:99–101. doi: 10.1136/neurintsurg-2017-013721Google Scholar
  • 24. Milne MS, Holodinsky JK, Hill MD, Nygren A, Qiu C, Goyal M, et al.. Drip ‘n ship versus mothership for endovascular treatment: modeling the best transportation options for optimal outcomes.Stroke. 2017; 48:791–794. doi: 10.1161/STROKEAHA.116.015321LinkGoogle Scholar
  • 25. Mullen MT, Branas CC, Kasner SE, Wolff C, Williams JC, Albright KC, et al.. Optimization modeling to maximize population access to comprehensive stroke centers.Neurology. 2015; 84:1196–1205. doi: 10.1212/WNL.0000000000001390CrossrefMedlineGoogle Scholar
  • 26. Mission: Lifeline Stroke. www.heart.org.2020. Available at: https://www.heart.org/en/professional/quality-improvement/mission-lifeline/mission-lifeline-stroke. Accessed January 16, 2020Google Scholar
  • 27. Bogle BM, Asimos AW, Rosamond WD. Regional evaluation of the severity-based stroke triage algorithm for emergency medical services using discrete event simulation.Stroke. 2017; 48:2827–2835. doi: 10.1161/STROKEAHA.117.017905LinkGoogle Scholar