Influence of Stroke-Onset Location on the Efficacy of Endovascular Thrombectomy: Comparison of Outcomes between In-Hospital versus Community-Onset Strokes

This study compared clinical outcomes in in-hospital strokes versus community-onset strokes in patients treated with endovascular thrombectomy for large-vessel occlusion. In-hospital strokes had overall faster time metrics from stroke onset to reperfusion. However, the rate of successful reperfusion and safety outcomes (in-hospital mortality and parenchymal hematoma) did not differ between the groups. Ninety-day mRS was also similar in both groups.

Abstract

BACKGROUND AND PURPOSE:

Previous studies have suggested that patients experiencing an in-hospital stroke may face delays in treatment and worse outcomes compared with patients with community-onset strokes. However, most studies occurred when IV thrombolysis was the primary treatment. This study aimed to examine the outcomes of patients experiencing an in-hospital stroke in the endovascular thrombectomy era.

MATERIALS AND METHODS:

This was a single-center retrospective cohort study of patients older than 18 years of age with acute ischemic stroke treated with endovascular thrombectomy within 12 hours of stroke onset from January 1, 2015, to April 30, 2021. Patients were classified into 2 groups: in-hospital strokes and community-onset strokes. We compared the time metrics of stroke care delivery, the rate of successful reperfusion, and functional outcome as scored using the mRS score at 90 days (favorable outcome was defined as mRS 0–2). Differences in proportions were assessed using the Fisher exact and χ² tests as appropriate. For continuous variables, differences in medians between groups were evaluated using Mann-Whitney U tests.

RESULTS:

A total of 676 consecutive patients were included, with 69 (10%) comprising the in-hospital stroke group. Patients experiencing in-hospital stroke were more likely to have diabetes (36% versus 18%, P = .02) and less likely to receive thrombolysis (25% versus 68%, P < .001) than those in the community-onset stroke group, but they were otherwise similar. Patients with in-hospital stroke had significantly faster overall time metrics, most notably from stroke recognition to imaging (median, 70 [interquartile range, 38–141] minutes versus 121 [74–228] minutes, P < .001). Successful recanalization was achieved in >75% in both groups (P = .39), with a median NIHSS score at discharge of <4 (P = .18). The 90-day mRS was similar in both groups, with a trend toward higher in-hospital mortality in the in-hospital stroke group (P = .06).

CONCLUSIONS:

Patients with in-hospital stroke had shorter workflow delays to initiation of endovascular thrombectomy compared with their community counterparts but with a similar rate of successful recanalization and clinical outcomes. Most important, 90-day mortality and mRS scores were equivalent between in-hospital stroke and community-onset stroke groups.

SUMMARY

PREVIOUS LITERATURE:

Studies comparing clinical outcomes in in-hospital strokes (HIS) versus community-onset strokes (COS) have been mainly performed before endovascular thrombectomy (EVT) was established as a standard treatment approach to acute ischemic stroke. These have generally shown worse outcomes in IHS, with delayed interventions and more comorbidities.

KEY FINDINGS:

- IHS had overall faster time metrics from stroke onset to groin puncture and reperfusion.

- The rate of successful reperfusion and safety outcomes (in-hospital mortality and parenchymal hematoma) did not differ between groups.

- At 90 days, favorable outcomes (mRS ≤2) (P = .54) and mRS 5–6 (P = .056) were equivalent between IHS and COS.

KNOWLEDGE ADVANCEMENT:

Contrary to previous findings, IHS can have equivalent outcomes compared with COS in the era of EVT. At our center, time metrics were overall faster in IHS, which may have counteracted any comorbidity-related effects.

Patients experiencing an in-hospital stroke (IHS) face different challenges compared with those with community-onset strokes (COSs). IHS is uncommon, with a reported incidence between 6.5% and 15% of all strokes,^1,2 but likely under-reported due to the lack of stroke-symptom recognition.³ For example, symptoms may be obscured by an acute illness or sedation during a procedure.^4,5 Moreover, patients with IHS are vulnerable to worse outcomes compared with those with COS because of an increased prevalence of comorbidities.³ Consequently, patients with IHS may have a delay in stroke care delivery, be limited in their eligibility for thrombolysis, or have a delay in medical thrombolysis administration and endovascular thrombectomy (EVT).³ Thus, the clinical features, treatment workflow, and outcomes of EVT for acute ischemic stroke (AIS) vary between COS and IHS.

Patients with IHS are a heterogeneous group with different intrinsic prognoses, falling into 3 broad categories in our experience: 1) patients admitted for a procedure that puts them at risk of having a stroke with poor outcomes⁶ (eg, cardiac surgery or endarterectomy); 2) patients admitted for an intercurrent illness that leads to interruption of their blood thinners or increases their thrombogenicity leading to a stroke (eg, coronavirus 2019 [COVID-19] infection or discontinuation of anticoagulants after a fall in patients with atrial fibrillation); and 3) coincidental strokes that were destined to occur regardless of the setting but happened during a hospital admission for an unrelated reason, likely having the same prognosis as COS (assuming the same treatment/time to groin puncture).

Studies comparing clinical outcomes in IHS versus COS have been mainly performed before EVT was established as a standard treatment approach to AIS.^7,8 Thus, there is a paucity of data comparing the outcomes of EVT for large-vessel occlusion (LVO) between these 2 groups,^9,10 forming the basis of the present study. We hypothesized that patients with IHS treated with EVT have worse outcomes than those with COS, reflecting the associated comorbidities and challenges of in-hospital care delays.

MATERIALS AND METHODS

Data Source and Population and Hospital Setting

We conducted a single-center retrospective cohort study of patients with AIS treated with EVT from January 1, 2015, to April 30, 2021. The Ottawa Hospital institutional ethics board approved this study, and the requirement for informed consent was waived.

Participants and Study Groups.

We retrospectively identified consecutive adult participants presenting with AIS symptoms who underwent EVT. Patients were classified into 2 groups: IHS and COS. Patients with IHS were defined as those admitted to our comprehensive stroke center (CSC) for other clinical reasons more than 24 hours before their stroke onset and exhibiting new neurologic deficits consistent with AIS. COS was defined as a patient with an AIS presentation outside a health care facility and transferred to our CSC directly or via a primary stroke center. An IHS subgroup (inpatient transfer) was defined as patients developing AIS while admitted to an outside hospital and subsequently transferred to our CSC for EVT treatment. Our CSC is the sole provider of EVT to a geographic area of 1.3 million people.

Inclusion Criteria.

Inclusion criteria were adult patients (older than 18 years of age) with AIS secondary to LVO treated by EVT within 12 hours of symptoms onset. Exclusion criteria were patients with extracranial arterial occlusion or symptomatic stenosis without intracranial occlusion.

Clinical Parameters and Outcomes Variables

Clinical and Imaging Evaluation.

Stroke severity was assessed by the NIHSS by a stroke neurologist. Patient demographics, medical history, therapeutic interventions, reason for admission (IHS group), stroke etiology, and discharge destination were obtained through an electronic chart review from our hospital database. Premorbid functional independence was defined as a mRS score of ≤2. The subtype of stroke was further classified according to the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification.¹¹

All patients underwent CT/CTA imaging before treatment as per institutional protocol. The ASPECTS and occlusion site were obtained from the diagnostic imaging or endovascular therapy reports. LVO was defined as a thrombus located in the intracranial ICA, distal ICA, and proximal M1 and A1 segments (Carotid T occlusion), and M1 and proximal M2 segments. Tandem occlusion was defined as a thrombus in the extracranial ICA and an intracranial (ICA, M1, M2 segment) thrombus. Posterior circulation LVO included the intracranial vertebral arteries and the basilar artery.

Thrombectomy Procedures.

EVT was performed using any approved device, such as a stent retriever, aspiration device, or both. The number of EVT passes was recorded from the procedural report. Successful reperfusion was defined as a modified TICI (mTICI) score ≥ 2b.¹¹

Timing.

We studied the following process measures: stroke onset to the first neuroimaging procedure, imaging to groin puncture, stroke onset to groin puncture, rtPA to groin puncture, and stroke onset to successful reperfusion.

Outcomes.

The primary end point was the functional neurologic disability scored on the mRS 90 days after EVT, with categories 5 (severe disability) and 6 (death) collapsed into a single category. Secondary outcomes included the successful reperfusion rate after EVT and pre-EVT workflow time delays. Safety outcomes included intracranial hemorrhage, defined as a parenchymal hematoma occupying >30% of the infarcted territory (PH2 according to the Heidelberg Classification¹²) within 5 days after EVT and mortality within 90 days.

Statistical Analysis

Descriptive analyses are summarized using mean (SD) for age, median, and interquartile range (IQR) or other continuous variable counts and proportions for categoric variables. The Wilcoxon rank-sum test or Fisher exact test was used to compare IHS and COS in total patients and the comparisons between the groups in the subgroup analysis of IHS. The Kruskal-Wallis nonparametric test or the Fisher exact test was used to compare cardiac surgery, medical service, and surgery only in the admitted service team subgroup analysis. Two-sided P < .05 was considered statistically significant. A favorable clinical outcome was defined as a 90-day mRS score of ≤2. All analyses were conducted using SAS, Version 9.4 (SAS Institute).

RESULTS

A total of 676 patients were included in the final analysis after 88 exclusions (Fig 1). The IHS group comprised 69 (10%) patients; and the COS group, 607 (90%). The IHS group was more likely to have diabetes (36% versus 18%, P = .02) and less likely to receive thrombolysis (25% versus 68%, P < .001) than the COS group (Table 1). Contraindications for IV thrombolysis were predominantly recent surgery and anticoagulation use; the decision to administer IV therapy was independent of EVT eligibility. The 2 groups had similar rates of independent functional status before stroke presentation, and more than one-half of all patients had a moderate-to-severe stroke (NIHSS >15) (P = .73). Cardioembolic strokes accounted for one-half of the strokes in both groups (57% versus 48%, P = .33). Most strokes involved the anterior intracranial circulation in both groups (≥90%), with a trend to carotid T occlusions in IHS versus COS (25% versus 15%, P = .05). No significant group differences were seen in age, sex, ASPECTS, or other risk factors.

FIG 1.

Patient flow chart. * Patients with recanalization between rtPA and initial angiogram at the time of EVT.

Table 1:

Demographics and baseline characteristics of IHS versus community-onset^a

	IHS (n = 69)	COS (n = 607)	P Value
Mean age (SD) (yr)	70 (SD, 13)	70 (SD, 14)	.87
Sex			.90
Male	34 (49)	305 (50)
Female	35 (51)	302 (50)
Risk factors
HTN	43 (62)	344 (58)	.46
DM	25 (36)	110 (18)	.02
Atrial fibrillation	24 (35)	183 (30)	.34
Smoking	28 (41)	173 (29)	.11
Premorbid mRS score ≤2	61 (88)	539 (89)	.95
Presentation NIHSS (median) (IQR)	18 (12–22)	17 (12–21)	.22
Presentation NIHSS >15	41 (62)	331 (57)	.43
ASPECTS (median) (IQR)	8 (8–10)	8 (8–10)	.51
Etiology			.33
Cardioembolic	38 (57)	293 (48)
Large-artery atherosclerosis	13 (19)	117 (19)
Undetermined or other	16 (24)	197 (33)
Site of occlusion			.05
Carotid T	17 (25)	94 (15)
Intracranial ICA	1 (1)	15 (2)
M1	25 (36)	234 (39)
M1/M2	9 (13)	104 (17)
M2	9 (13)	48 (8)
Posterior circulation occlusion	7 (10)	45 (7)
Tandem occlusion	1 (1)	67 (11)
Thrombolysis given	17 (25)	412 (68)	<.001
Stroke onset to initiation of alteplase (median) (IQR) (min)	105 (62–133)	115 (85–157)	.20

Note:—DM indicates diabetes mellitus; HTN, hypertension.

^aUnless otherwise specified, data are number of patients, with percentages in parentheses.

Compared with those with COS, patients with IHS had shorter time delays after stroke presentation, except for imaging to groin puncture. Patients with IHS were imaged approximately an hour faster after stroke onset (median, 70 [IQR, 38–141] minutes versus 121 [74–228] minutes, P < .001) (Table 2) and had significantly shorter times from stroke onset to groin puncture (median, 156 [IQR, 108–216] minutes versus 184 [126–282] minutes, P = .003) and stroke onset to recanalization times (median, 185 [IQR, 142–248] minutes versus 223 [160–323] minutes; P = .01). On the other hand, patients with IHS had a delay of 15 minutes from imaging to the angiographic suite compared with those with COS (median, 65 [IQR, 37–93] minutes versus 50 [31–77] minutes, P = .01).

Table 2:

Time delays and outcome comparison between IHS versus community-onset^a

	IHS (n = 69)	Community-Onset Stroke (n = 607)	P Value
Stroke onset to imaging (median) (IQR) (min)	70 (38–141)	121 (74–228)	<.001
Imaging to groin puncture (median) (IQR) (min)	65 (37–93)	50 (31–77)	.01
Stroke onset to groin puncture (median) (IQR) (min)	156 (108–216)	184 (126–282)	.003
Initiation of alteplase to groin punctureb (median) (IQR) (min)	56 (38–81)	45 (20–77)	.62
Stroke onset to mTICI ≥2b (median) (IQR) (min)	185 (142–248)	223 (160–323)	.01
Groin puncture to reperfusion (median) (IQR) (min)	26 (17–49)	27 (17–42)	.76
Successful recanalization (mTICI ≥2b)	53 (77)	500 (82)	.33
No. of passes (median) (IQR)	1 (1–2)	1 (1–2)	.72
Type of anesthesia
CS	57 (83)	517 (85)	.57
GA	12 (17)	90 (15)
Safety outcomes
In-hospital death	14 (20)	74 (12)	.059
Parenchymal hematoma	7 (10)	65 (11)	.89
Discharge NIHSS (median) (IQR)	2.0 (6–7)	3.0 (0–8)	.18
Discharge destination			.39
Home	21 (39)	243 (46)
Other facility	33 (61)	285 (54)
Length of stay (median) (IQR) (days)	12 (7–21)	6 (3–12)	<.001
mRS at 90 days (median) (IQR)c	2 (0–5)	1 (0–5)	.54
mRS 5-6 at 90 daysc	20 (38)	111 (25)	.056

Note:—CS indicates conscious sedation; GA, general anesthesia.

^aUnless otherwise specified, data are number of patients, with percentages in parentheses.

^bMissing information from 16 patients in the COS group.

^cMissing information from 187 patients: 16 IHSs; 171 COSs.

Considering safety outcomes, the proportion of patients with IHS who died during hospitalization trended higher than in those with COS (20% versus 12%; P = .06) (Table 2). The proportion of parenchymal hematoma did not differ between groups (10% versus 11%, P = .89). Successful recanalization (mTICI ≥2) was achieved in more than three-quarters of patients in both groups (77% versus 82%, P = .33), with a median NIHSS score at discharge of <4 (P = .18). Patients with IHS had a more extended hospital stay than those with COS (12 versus 6 days, P < .001). At 90 days, the median mRS score was equivalent (P = .54) between the IHS (median, 2 [IQR, 0–5]) and COS (median, 1 [IQR. 0–5]) groups (Fig 2).

FIG 2.

Comparison of outcomes between IHS and COS.

The subgroup analysis of patients with IHS demonstrated that 60% (41/69) had AIS while admitted to the CSC (Online Supplemental Data). Patients with IHS admitted to the CSC were more likely to have a premorbid mRS ≤2 than inpatient transfers (93% versus 82%, P = .04) and to be admitted to a cardiac surgery service at the time of stroke (51% versus 7%). In contrast, the inpatient transfer group was more likely admitted to a medical service (29 versus 68%; P = <.001). The median time after hospital admission to stroke onset was 4 days in both groups (P = .82), and the median traveled distance for inpatient transfers was 7 km. Patients with CSC IHS were transferred 26 minutes faster after imaging to the angiography suite (median, 59 [IQR, 37–84] minutes versus 85 [39–112] minutes, P = .04), had a groin puncture 46 minutes faster after stroke onset (median, 132 [IQR, 95–182] minutes versus 178 [151–221] minutes, P = .02), and had a trend toward achieving a faster recanalization after stroke onset (median, 132 [IQR, 95–182] minutes versus 178 [151–221] minutes, P = .051) than inpatient transfer strokes. Most interesting, inpatient transfers had a shorter time from groin puncture to recanalization than those with CSC IHS (median, 18 [IQR, 14–28] minutes versus 34 [22–56] minutes, P = .002). Successful recanalization was achieved in three-quarters of patients in both groups (78% versus 75%, P = .76), and inpatient transfers showed a trend toward having a higher proportion of parenchymal hematomas (18% versus 5%, P = .07). No differences between groups were noted regarding in-hospital mortality and mRS ≤2 at 90 days.

DISCUSSION

The present study challenged the assumption that patients with IHS necessarily have worse outcomes compared with those with community strokes. Patients with IHS had similar 90-day mRS scores compared with those with COS (P = .54) despite patients with IHS being more likely to have diabetes, less likely to receive thrombolysis, and trending toward having a carotid T occlusion. Patients with IHS had significantly shorter pre-EVT time delays (except for imaging to groin puncture) than those with COS. Moreover, those with IHS trended toward higher in-hospital mortality and an mRS of 5-6 at 90 days and had longer hospital stays, which was likely due to greater comorbidities, recent surgery, and IV thrombolysis ineligibility. However, no differences were noted in the overall rate of successful reperfusion and functional outcomes between groups.

One of the most critical factors associated with poor outcomes in patients with AIS is delayed thrombectomy, as demonstrated across multiple studies.^{13 -17} While some studies have shown delays among patients with IHS before EVT initiation,¹⁴ this finding was not replicated at our institution, highlighting the center-specific nature of the problem. In our study, stroke onset to imaging, groin puncture, and recanalization times were less in the IHS group. Many factors may explain these results, including hospital protocols and stroke training/experience among health care providers to promptly recognize stroke symptoms in admitted patients.

Most interesting, those with IHS had a longer delay to the angiographic suite after imaging than those with COS, possibly due to differences in stroke code activation for hospitalized patients. At our institution, COS codes are often activated by paramedics before the patient arrives, allowing time for the on-call neurointerventional team to prepare (or commute to hospital if after-hours). Because inpatients are imaged faster, less time is available for preparation, especially if the on-call team needs to commute after-hours. Moreover, for transferred inpatients, imaging is often performed at the hospital of origin; therefore, the transfer commute itself would inflate the imaging-to-groin puncture time. Other possible factors include angiographic suites already occupied with routine cases (though rarely a limitation given our capacity to run a second suite as a backup), portering limitations, and the delay caused by the decision-making process to perform EVT in comorbid patients. In the subgroup analysis of IHS, the inpatient transfer group had a delay of 46 minutes from stroke onset to the initiation of EVT compared with inpatient strokes occurring at the CSC, highlighting the need for better interhospital coordination in this subgroup of patients. The shorter time from groin puncture to recanalization in the IHS transfer subgroup is presumably related to a longer duration of tPA exposure before the intervention.¹⁸

Our study aligns with previous publications demonstrating no differences in functional outcomes in patients with IHS compared those with COS. Sano et al¹⁹ had a similar conclusion though the overall time metrics were longer in the IHS group, except for groin-puncture-to-recanalization times, contrary to our findings. The overall delay in IHS thrombectomy times in that study was a consequence of using emergency department entry as opposed to stroke-onset time as the primary metric for the COS group. In fact, their study mirrored our findings when using the “last seen well” to reperfusion metric, which was shorter in IHS. Although diabetes appears unrelated or possibly paradoxically beneficial to the ease of recanalization,²⁰ it is still associated with worse functional outcomes.²¹ In our study, patients with IHS showed 90-day mRS scores similar to those in COS, despite less thrombolysis use and a higher prevalence of diabetes, perhaps offset by faster pre-EVT time intervals.

The strengths of our study include adequately powered sample sizes and the relatively balanced cohorts with similar baseline mRS, NIHSS, ASPECTS, stroke etiology, age, sex, and cardiovascular risk factors (aside from diabetes). The fact that reperfusion rates and hemorrhagic transformation were similar in both groups suggests a balance in case complexity.

Our study has some limitations. First, it is a retrospective study subject to inherent disadvantages, including missing 90-day mRS scores in 23% of those with IHS and 28% of patients with COS. Second, this was a single-center study, reflecting the experience at our institution, which performs an average of about 130 thrombectomies per year and has a uniquely broad catchment area that may not generalize well to other comprehensive stroke centers. For example, because of our wide catchment, a higher proportion of patients with COS may travel longer distances to reach our hospital, thereby inflating the average time from stroke onset to imaging. Comparison with other studies may be limited given that our IHS definition included both transfers and CSC inpatients. However, we showed faster IHS time metrics despite the dilutive effect of including transfers. Therefore, while it is difficult to generalize our findings for those in the transfer subgroup, who have variable distances traversed, outcomes of stationary inpatients would be expected to follow the results of our study (assuming similar demographics and hospital efficiency). Third, while the cohorts had generally balanced characteristics, the disproportionate number of individuals with diabetes mellitus and lower thrombolysis rates in the IHS group serve as confounding factors. Lack of thrombolysis eligibility and diabetes are known risk factors for poor outcomes and so may explain why our IHS group had results like those in the COS group despite faster EVT delivery.

CONCLUSIONS

Those with IHS do not have worse outcomes compared with their community counterparts in the modern EVT era, achieving similar 90-day mRS scores and mortality. Time to EVT for AIS was overall faster for patients with IHS. These equivalent outcomes challenge our hypothesis that patients with IHS do worse; perhaps the inpatient-associated comorbidities are counterbalanced by faster EVT delivery, serving as an equalizer for these previously disadvantaged patients. Future prospective multicenter studies are needed to further capture the impact of in-hospital stroke on EVT outcomes.

ABBREVIATIONS:

AIS

acute ischemic stroke

COS

community-onset stroke

CSC

comprehensive stroke center

EVT

endovascular thrombectomy

IHS

in-hospital stroke

IQR

interquartile range

LVO

large-vessel occlusion

mTICI

modified TICI