a b s t r a c t

Introduction: Our objective was to determine whether acute ischemic stroke patients’ language preference is associated with differences in time from symptom discovery to hospital arrival, activation of emergency med- ical services, door-to-imaging time (DTI), and door-to-needle time.

Methods: We identified consecutive AIS patients presenting to a single urban, tertiary, academic center between 01/2003-05/2014 for whom language preference was available. Data were abstracted from the institution’s Re- search Patient Data Registry and Get with the Guidelines-Stroke registry. Bivariate and regression models eval- uated the relationship between language preference and: 1) time from symptom onset to hospital arrival, 2) use of EMS, 3) DTI, and 4) DTN time.

Results: Of 3190 AIS patients, 300 (9.4%) were non-English preferring (NEP). Comparing NEP to English preferring (EP) patients in unadjusted or adjusted analyses, time from symptom discovery to arrival and rate of EMS utili- zation were not significantly different (overall median time 157 min, IQR 55-420; EMS utilization: 65% vs. 61.3% p = 0.21). There was also no significant differences in DTI or in likelihood of guideline-recommended DTI <= 25 min (overall median 59 min, IQR 29-127; DTI <= 25 min 24.3% vs. 21.3% p = 0.29) or DTN time or in likelihood of guideline-recommended DTN <= 60 min (overall median 53 min, IQR 36-73; DTN <= 60 min 62.5% vs. 58.2% p = 0.60). Conclusion: Consistent with prior reports examining disparities in care, a systems-based approach to acute stroke prevents differences in hospital-based metrics. Reassuringly, NEP and EP patients also had similar speed of symptom recognition and EMS utilization.

(C) 2020

1. Introduction

Recognition of acute stroke symptoms by patients and their families is critical to ensure timely hospital arrival and access to disability- reducing interventions. Public education campaigns have worked to improve stroke recognition and encourage early hospital arrival [1-4]. Ethnic minorities have been shown to have poor levels of stroke knowl- edge [5], and public education campaigns may have less penetration in this population [6]. Even after patients arrive in the emergency depart- ment (ED), language barriers may interfere with clinical communica- tion and introduce challenges into safe, effective, and efficient care delivery [7]. The impact of this is not well studied in acute stroke.

Our group previously reported on acute ischemic stroke care across language differences [8,9]. To complement our prior findings,

* Corresponding author at: Department of Emergency Medicine, Massachusetts General Hospital, 55 Fruit Street, Boston, MA 02114, United States of America.

E-mail address: [email protected] (K.S. Zachrison).

we conducted this study to characterize the association between pa- tients’ language preference and time from stroke symptom discovery to hospital arrival, activation of EMS, door-to-imaging time (DTI), and door-to-needle (DTN) time.

1. Methods
   1. Population and setting

We identified consecutive AIS patients presenting directly to a single urban, tertiary, academic comprehensive stroke center (CSC) between 01/01/2003 and 04/30/2014 using the institution’s Get with the Guide- lines (GWTG) - Stroke database for whom language preference was available. Hospital personnel were trained in data abstraction and col- lected data on consecutively admitted stroke patients. Patients are iden- tified based on International Classification of Diseases-Ninth Revision discharge codes [10]. We excluded patients who did not indicate a lan- guage preference (n = 183), patients transferred from another hospital

https://doi.org/10.1016/j.ajem.2020.10.064

0735-6757/(C) 2020

after initial emergency care (n = 1271), patients whose stroke occurred during an Inpatient hospitalization (n = 1600), and patients with pri- mary residence outside the United States (n = 222). The institutional review board approved this study.

• 1. Variables of interest

The GWTG - Stroke registry includes information on patients’ demo- graphics, medical history, and hospital treatment characteristics and outcomes. We used deterministic linkage to incorporate patients’ lan- guage preference by using the Partners Healthcare System Research Pa- tient Data Registry (RPDR). We also used the RPDR to supplement patients with missing race/ethnicity (n = 312) and insurance status (n = 9) in the GWTG - Stroke data.

Demographic variables were: age, gender, race (Asian, American Indian/Alaskan Native, African American/Black, Multiracial, Native Hawaiian/Pacific Islander, White); Hispanic ethnicity, marital status, insurance status (private, Medicare, Medicaid, uninsured/self-pay). Clinical variables included past medical history of conditions associ- ated with stroke (atrial fibrillation, coronary artery disease or prior myocardial infarction, carotid stenosis, diabetes mellitus, dyslipid- emia, heart failure, hypertension, peripheral vascular disease, previ- ous stroke/TIA, prosthetic heart valve, smoking), and Stroke severity at presentation (based on National Institutes of Health Stroke Scale [NIHSS] score).

• 1. Independent variable of interest

Language preference was determined using patients’ self-identification based on response to the standardized question “In what language do you prefer to receive medical information?” at the time of hospital regis- tration or when updating their information. We categorized patients as English-preferring (EP) or non-EP (NEP). For some patients unable to re- spond, language preference information was obtained from patients’ families and friends.

• 1. Dependent variables of interest

Our dependent variables of interest were 1), time from stroke symp- tom discovery to hospital arrival (minutes); 2), mode of arrival to hos- pital (EMS yes/no); 3), DTI (minutes), and receipt of imaging within guideline-recommended 25 min window (yes/no); 4) DTN time (mi- nutes), and receipt of thrombolysis within 60 min window (yes/no).

• 1. Statistical analysis

We used descriptive statistics to characterize the study population and examined the bivariate association between language preference and patient characteristics using t-tests, chi-square tests, and Wilcoxon rank sum as appropriate.

Table 1

Participant Characteristics, Stratified by Language Preference

	All Patients		Non-English Preferring		English-Preferring	p-value
	N = 3190		n = 300		n = 2890
Age, median (IQR)	72 (60-81)
Female, n (%)	1473 (46.2%)		163 (52.6%)		1310 (45.3%)	0.003
Race, n (%)
Black	192 (6.0%)		43 (14.3%)		149 (5.2%)	<0.001
White	2810 (88.1%)		160 (53.3%)		2650 (91.2%)
Other?	109 (3.4%)		58 (22.3%)		51 (1.8%)
Not documented	79 (2.5%)		39 (13%)		40 (1.4%)
Hispanic, n (%)	177 (5.6%)		105 (39.9%)		72 (3.5%)	<0.001
Insurance Status
Medicare	1937 (60.7%)		152 (50.6%)		1785 (61.8%)	<0.001
Medicaid	34 (1.1%)		9 (3%)		25 (0.9%)
Uninsured/ self-pay	67 (2.1%)		17 (5.7%)		50 (1.7%)
Private	1151 (36.1%)		122 (40.7%)		1029 (35.6%)
Marital Status, n (%)
Single	1468 (64.8%)	137 (48.2%)		1331 (48.3%)		0.14
Divorced/separated	284 (73.7%)	31 (10.9%)		253 (9.2%)
Widowed	686 (21.5%)	73 (25.7%)		613 (22.3%)
Married	599 (18.8%)	43 (15.1%)		556 (20.2%)
Atrial fibrillation, n (%)	623 (19.5%)	46 (15.3%)		577 (20.0%)		0.054
Coronary artery disease, n (%)	686 (21.5%)	54 (18.0%)		632 (21.9%)		0.12
Carotid disease, n (%)	169 (5.3%)	7 (2.3%)		162 (5.6%)		0.02
Diabetes mellitus, n (%)	775 (23.3%)	92 (30.7%)		683 (23.6%)		0.007
Dyslipidemia, n (%)	1337 (41.9%)	128 (42.7%)		1209 (41.8%)		0.78
Hypertension, n (%)	2196 (68.8%)	228 (76%)		1968 (68.1%)		0.005
Prior stroke, n (%)	225 (7.1%)	33 (11%)		192 (6.6%)		0.005
Valvular disease, n (%)	27 (0.9%)	4 (1.3%)		28 (0.8%)		0.33
Peripheral vascular disease, n (%)	159 (5.0%)	5 (1.7%)		154 (5.3%)		0.006
Smoker, n (%)	498 (15.6%)	28 (9.3%)		470 (16.3%)		0.002
NIHSS performed, n (%)	3031 (95.8%)	289 (97.3%)		2742 (95.6%)		0.17
NIHSS, median (IQR)	3 (1-9)	4 (2-12)		3 (1-9)		<0.001
Time from symptom discovery to ED arrival in min, median (IQR)	157.5 (55-420)
ED arrival >3 h from last known well, n (%)	2036 (70.5%)	187 (68.5%)		1849 (70.7%)		0.46
Arrival by EMS, n (%)	1966 (61.6%)	195 (65%)		1771 (61.3%)		0.21
Door-to-imaging time within 25 min, n (%)	982 (30.1%)	88 (29.3%)		894 (30.9%)		0.57
Median door-to-imaging time in minutes (IQR)	59 (29-127)	55 (26-115)		60 (29-128)		0.33
Treated with alteplase, n (%)	321 (10.1%)	40 (13.3%)		281 (9.7%)		0.048
Door-to-needle time within 60 min, n (%)	189 (58.7%)	25 (62.5%)		164 (58.2%)		0.60
Median door-to-needle time in minutes (IQR)	53 (36-73)	51 (36.5-68.5)		53 (36-75)		0.69

Legend: IQR: interquartile range.

* Other includes American Indian/Alaska Native, Asian, and Native Hawaiian/Pacific Islander.

EMS use“>We used linear and logistic regression analyses to determine the in- dependent relationship between language preference and our depen- dent variables of interest. Model co-variates were identified a priori based on previous literature and clinical experience.

1. Results

We identified 3190 AIS patients meeting inclusion criteria. Of these patients, 300 (9.4%) were NEP. Among NEP patients, the most com- monly Preferred languages were Spanish (n = 94, 31.3%), Italian (n = 32, 10.7%), Creole (n = 31, 10.3%), Chinese (n = 28, 9.3%), Portu- guese (n = 26, 8.7%), and Cambodian (n = 11, 3.7%); 78 (26%) patients reported an alternative language preference. NEP patients were more often female and Hispanic, less often white, had higher rates of diabetes, hypertension, and prior stroke than EP patients, were less frequently smokers, and less frequently had peripheral vascular disease or carotid stenosis (Table 1). NEP patients also had slightly more severe strokes on arrival to the ED (median NIHSS 4 [IQR 2-12] vs 3 [IQR 1-9], p < 0.001).

• 1. Time from symptom discovery to ED arrival

Time from symptom discovery to ED arrival was available for 1694 patients (53.1%). There was no significant difference in median time from symptom discovery to ED arrival between NEP and EP patients (128 [IQR 50-495] vs 161 min [IQR 55-415]). This finding remained after adjusting for patient demographic and clinical characteristics

(Table 2). In adjusted analysis, divorced status was associated with slower time from symptom discovery to ED arrival (491 min slower rel- ative to married, p = 0.008), while increasing stroke severity was asso- ciated with faster time from symptom discovery to arrival (26.4 min faster per point increase in NIHSS, p < 0.001).

• 1. EMS use

Overall, 61.6% of patients arrived by EMS. There was no significant difference in arrival by EMS between NEP and EP patients in unadjusted or adjusted analyses. In adjusted analysis, history of atrial fibrillation and increasing stroke severity were both associated with EMS use (Table 2).

• 1. DTI time

DTI time was available for 2371 patients (74.3%) with median time of 59 min (IQR 29-127). Median DTI for NEP patients was not signifi- cantly different than for EP patients (55 min [IQR 26-115] vs 60 min [IQR 29-128], p = 0.33). In adjusted analysis there was no difference in DTI between NEP and EP patients. Faster DTI was associated with in- creased stroke severity as well as history atrial fibrillation and valvular disease (Table 2).

DTI was achieved within the guideline-recommended 25-min win- dow in 982 patients (30.1%) and this rate was not significantly different between NEP and EP patients in unadjusted (29% vs 31%, p = 0.57) or adjusted analyses. In adjusted analysis, the only factor associated with

Table 2

Multivariable Models Examining Relationship between Language Preference, Prehospital characteristics, and Emergency Department-Based Care

Time from symptom discovery to ED arrival (minutes)

n = 1217

EMS use

n = 2075

Door-to-imaging time within

25 min

n = 1525

Door-to-imaging time (minutes)

n = 1525

Door-to-needle time within

60 min

n = 228

Door-to-needle time (minutes)

n = 228

	Change in time per unit change in variable	p-value		OR	95% CI		OR	95% CI		Change in time per unit change in variable	p-value		OR	95% CI		Change in time per unit change in variable	p-value
Non-English Preferring	-74.1	0.68		1.09	0.73-1.64		0.98	0.62-1.56		-0.2	0.97		1.00	0.37-2.67		1.7	0.80
Age	6.4	0.15		1.01	1.00-1.02		1.00	0.99-1.01		0.0	0.96		1.02	0.99-1.05		-0.3	0.11
Female	122.0	0.22		0.89	0.72-1.11		1.17	0.89-1.54		5.7	0.12		0.61	0.32-1.17		7.2	0.11
Race (ref: white)
American Indian/				1	(empty)		1	(empty)			(empty)		1	(empty)
Alaska Native
Asian	476.0	0.77		0.67	0.38-1.18		0.67	0.30-1.53		64.2	0.34		-	(dropped)		-
Black	286.9	0.86		0.75	0.50-1.12		1.00	0.59-1.72		72.0	0.28		0.55	0.02-12.88		11.4	0.55
Nat Hawaiian/Pac	-7.3	0.82		0.18	0.02-1.36		0.83	0.07-9.41		22.6	0.76		-	(dropped)		20.0	0.55
Islander
Hispanic	146.3	0.49		0.87	0.54-1.39		1.03	0.58-1.83		-1.8	0.82		0.71	0.16-3.16		9.1	0.38
Insurance (ref: private)
Medicare	77.3	0.51		0.83	0.65-1.07		0.93	0.67-1.29		-2.2	0.62		0.90	0.40-2.05		1.3	0.81
Medicaid	-50.1	0.90		2.16	0.80-5.88		1.15	0.38-3.49		-1.0	0.95		0.08	0.00-1.52		27.2	0.16
Uninsured/ self	199.1	0.56		0.86	0.42-1.73		0.63	0.25-1.60		-1.5	0.89		2.24	0.21-23.68		-8.9	0.53
Marital Status (ref:
married)
Single	-86.9	0.50		0.91	0.70-1.20		1.24	0.88-1.75		-7.7	0.10		1.35	0.62-2.96		-10.9	0.045
Divorced/Separated	490.9	0.008		1.05	0.72-1.55		1.04	0.64-1.68		-7.8	0.23		1.61	0.48-5.34		-10.6	0.19
Widowed	-250.5	0.13		1.12	0.78-1.59		0.66	0.42-1.03		-4.0	0.50		1.21	0.43-3.41		-6.9	0.33
Atrial fibrillation	-153.9	0.21		1.55	1.15-2.08		1.26	0.92-1.73		-10.1	0.02		0.84	0.41-1.70		5.1	0.29
Coronary artery disease	-57.0	0.63		0.89	0.69-1.15		0.97	0.70-1.33		4.5	0.29		1.32	0.58-2.98		4.4	0.40
Carotid disease	134.1	0.57		0.93	0.58-1.50		1.03	0.54-1.99		13.1	0.13		5.44	0.43-68.7		-29.2	0.04
Diabetes mellitus	21.8	0.19		1.07	0.84-1.36		0.78	0.57-1.07		13.5	0.001		0.85	0.39-1.81		-0.5	0.93
Dyslipidemia	-89.9	0.36		1.11	0.90-1.38		1.03	0.79-1.35		-5.4	0.14		0.78	0.41-1.48		4.6	0.29
Hypertension	-72.2	0.51		0.93	0.73-1.17		0.92	0.68-1.24		0.96	0.81		1.28	0.63-2.59		-2.6	0.59
Prior stroke	-105.5	0.49		1.14	0.81-1.62		1.33	0.89-2.01		-3.3	0.55		1.41	0.39-5.11		-7.7	0.35
Valvular disease	-212.7	0.61		1.41	0.50-3.94		2.01	0.64-6.33		-36.0	0.04		1	omitted		-59.8	0.06
Peripheral vascular	-155.0	0.55		1.30	0.78-2.17		0.61	0.29-1.25		-6.0	0.50		0.20	0.03-1.18		21.1	0.07
disease
Smoker	168.0	0.24	1.25		0.93-1.67	0.90		0.61-1.33	1.7		0.74	1.15		0.46-2.91	1.0		0.87
NIHSS (per point)	-26.4	<0.001	1.23		1.20-1.27	1.12		1.10-1.14	-4.14		<0.001	0.99		0.94-1.03	-0.2		0.46

Legend: EMS: Emergency Medical Services; NIHSS: National Institutes of Health Stroke Scale.

likelihood of achieving DTI within 25 min was increasing stroke severity (OR 1.12 per point increase in NIHSS, 95% CI 1.10-1.14).

• 1. DTN time for thrombolysis

Of the 3190 patients in our sample, 321 were treated with intrave- nous alteplase (10%), and 315 had available DTN times. Median DTN time was 53 min (IQR 36-73), and did not vary between NEP and EP pa- tients in unadjusted or in adjusted analyses.

DTN time within the guideline-recommended 60 min window was achieved in 189 patients (58.7%), and this rate was not significantly dif- ferent between NEP and EP patients in unadjusted 62.5% vs 58.2%, p =

0.60) or adjusted analyses (Table 2).

1. Discussion

In this analysis of AIS patients presenting to our CSC, we found no difference in the time to hospital arrival or in the use of EMS by NEP rel- ative to EP patients. This suggests that these patients had similar recog- nition of stroke symptoms and similar knowledge and means to present urgently. Though we are not aware of prior work examining this specific question, these findings were surprising in the context of previous re- ports of low penetration of stroke preparedness education into ethnic minority populations [6]. One possible explanation for our finding may be that the NEP patients in our sample had better recognition and understanding of stroke because they had higher prevalence of prior stroke relative to EP patients. Another potential explanation is the high prevalence of stroke education materials in Massachusetts that are available in multiple languages [11] coupled with the public ed- ucation awareness campaigns sponsored by the CDC-funded Paul Coverdell Acute Stroke Registry grant.

We also found no evidence of the communication barriers that are

well-understood to characterize clinical interactions with NEP patients [7]. There was no difference in timely Emergency stroke care between NEP and EP patients in our sample, reflected by similar door-to- imaging times and door-to-needle times for intravenous alteplase deliv- ery. While we cannot know the underlying reason for this, one possibil- ity is that particular ED processes are more robust to stresses on the system. For example, stroke is a hyper-acute condition that requires rapid recognition and activation of in-hospital processes, and the ED- based processes for stroke symptom recognition may be strong enough to withstand these kinds of communication challenges.

It is reassuring that within our institution’s stroke patient popula- tion, patients’ Prehospital recognition and timeliness of emergency stroke care received were not negatively influenced by patients’ re- ported language preference. These findings should be validated in other settings, particularly in less urban places where NEP may be more rare and in non-CSC hospitals where acute stroke care processes may be more susceptible to system stresses.

Our study has the usual limitations of a single-institution study and findings may not be generalizable to other settings. While we found no difference in prehospital recognition or in emergency stroke care met- rics, we did not examine patient outcomes between groups. Neverthe- less, we feel that it is important to understand Processes of care independently in order to ensure equitable access to high-quality treat- ment. We also did not examine language concordance between patient and provider, as we did not have provider language reliably documented in our data. Finally, our dataset was through April 2014, and therefore we did not examine differences in Mechanical thrombectomy between groups, as the evidence for the benefit had not yet been published.

1. Conclusion

Within our CSC single-institutional experience, we found no differ- ence in prehospital stroke symptom recognition or in use of EMS

transport between NEP and English-preferring patients. NEP patients also had similar times to imaging and to intravenous thrombolysis de- livery in unadjusted and adjusted analyses, suggesting that robust stroke systems are equipped to deal with language barrier challenges.

Funding

This work was supported by the Agency for Healthcare Research and Quality (K08 HS024561, Zachrison).

Presentations

Abstract presented by Dr. Shaw Natsui at the International Stroke Conference 2017, in Houston, Texas.

Declaration of Competing Interest

KSZ, SN, BML, and NIM have no conflicts of interest to disclose. LHS reports the following relationships relevant to Research Grants or com- panies that manufacture products for thrombolysis or thrombectomy even if the interaction involves non-thrombolysis products: scientific consultant regarding trial design and conduct to Genentech (late win- dow thrombolysis) and Member of steering committee (TIMELESS NCT03785678); consultant on user interface design and usability to LifeImage; stroke systems of care consultant to the Massachusetts Dept of Public Health; member of a Data Safety Monitoring Boards (DSMB) for Penumbra (MIND NCT03342664) and for Diffusion Pharma PHAST- TSC NCT03763929); Serving as National PI for Medtronic (Stroke AF NCT02700945); National Co-PI, late window thrombolysis trial, NINDS (P50NS051343, MR WITNESS NCT01282242; and alteplase provided free of charge to Massachusetts General Hospital as well as supplemen- tal per-patient payments to participating sites by Genentech); Site PI, StrokeNet Network NINDS (New England Regional Coordinating Center U24NS107243).

References

1. Metias MM, Eisenberg N, Clemente MD, et al. Public health campaigns and their ef- fect on stroke knowledge in a high-risk urban population: a five-year study. Vascu- lar. 2017;25(5):497-503. https://doi.org/10.1177/1708538117691879.
2. Kleindorfer D, Khoury J, Broderick JP, et al. temporal trends in public awareness of stroke. Stroke. 2009;40(7):2502-6. https://doi.org/10.1161/STROKEAHA.109. 551861.
3. Sharrief AZ, Johnson B, Urrutia VC. Stroke outreach in an Inner city market: a plat- form for identifying African American males for Stroke prevention interventions. Front Neurol. 2015;6:133. https://doi.org/10.3389/fneur.2015.00133.
4. Williams O, Leighton-Herrmann Quinn E, Teresi J, et al. Improving community stroke preparedness in the HHS (hip-hop stroke) randomized clinical trial. Stroke. 2018;49

(4):972-9. https://doi.org/10.1161/STROKEAHA.117.019861.

1. Jones SP, Jenkinson AJ, Leathley MJ, Watkins CL. Stroke knowledge and awareness: an integrative review of the evidence. Age Ageing. 2010;39(1):11-22. https://doi. org/10.1093/ageing/afp196.
2. Robinson TG, Reid A, Haunton VJ, Wilson A, Naylor AR. The face arm speech test: does it encourage rapid recognition of important stroke warning symptoms? Emerg Med J. 2013;30(6):467-71. https://doi.org/10.1136/emermed-2012-201471.
3. Flores G. Language barriers to health Care in the United States. N Engl J Med. 2006; 355(3):229-31. https://doi.org/10.1056/NEJMp058316.
4. Erfe BL, Siddiqui KA, Schwamm LH, Mejia NI. Relationship between language prefer- ence and intravenous thrombolysis among acute ischemic stroke patients. J Am Heart Assoc. 2016;5(12). https://doi.org/10.1161/JAHA.116.003782.
5. Erfe BML, Siddiqui KA, Schwamm LH, Kirwan C, Nunes A, Mejia NI. Professional medical interpreters influence the quality of acute ischemic stroke care for patients who speak languages other than English. J Am Heart Assoc. 2017;6(9). https://doi. org/10.1161/JAHA.117.006175.
6. Fonarow GC, Reeves MJ, Smith EE, et al. Characteristics, performance measures, and in-hospital outcomes of the first one million stroke and transient ischemic attack ad- missions in get with the guidelines-stroke. Circ Cardiovasc Qual Outcomes. 2010;3 (3):291-302. https://doi.org/10.1161/CIRCOUTCOMES.109.921858.
7. Stroke (Stroke Heroes Act FAST). Massachusetts Health Promotion Clearinghouse. Massachusetts Department of Public Health. https://massclearinghouse.ehs.state. ma.us/category/HDSP.html. Accessed January 21, 2019.