Analysis of Stroke Care Among 2019–2020 National Emergency Medical Services Information System Encounters

Abstract

Objective:

Emergency Medicine Service (EMS) providers play a pivotal role in early identification and initiation of treatment for stroke. The objective of this study is to characterize nationwide EMS practices for suspected stroke and assess for gender-based differences in compliance with American Stroke Association (ASA) guidelines.

Methods:

Using the 2019–2020 National Emergency Medical Services Information System (NEMSIS) Datasets, we identified encounters with an EMS designated primary impression of stroke. We characterized patient characteristics and EMS practices and assessed compliance with eight metrics for “guideline-concordant” care. Multivariable logistic regression modeled the association between gender and the primary outcome (guideline-concordant care), adjusted for age, EMS level of service, EMS geographical region, region type (i.e. urban or rural), and year.

Results:

Of 693,177 encounters with a primary impression of stroke, overall compliance with each performance metric ranged from 18% (providing supplemental oxygen when the pulse oximetry is less than 94%) to 76% (less than 90sec from incoming call to EMS dispatch). 2,382 (0.39%) encounters were fully guideline-concordant. Women were significantly less likely than men to receive guideline-concordant care (adjusted OR 0.82, 95% CI 0.75–0.89; 0.36% women, 0.43% men with guideline-concordant care).

Conclusions:

A minority of patients received prehospital stroke care that was documented to be compliant with ASA guidelines. Women were less likely to receive fully guideline-compliant care compared to men, after controlling for confounders, although the difference was small and of uncertain climical importance. Further studies are needed to evaluate the underlying reasons for this disparity, its impact on patient outcomes, and to identify potential targeted interventions to improve prehospital stroke care.

Background:

The Emergency Medicine Services (EMS) provider is a key link in the “chain of recovery” for acute stroke patients. The vast majority of acute stroke patients will present to the healthcare setting via EMS, a frequent event given that someone has a new or recurrent stroke every forty seconds in the United States(1). In stroke care, great strides are being made to extend the window for therapeutic intervention via intravenous thrombolysis and mechanical thrombectomy. Nonetheless rapid treatment of stroke is critical to improve success of these treatments and improve outcomes. EMS transport of acute stroke patients is associtated with decreased time to presentation for care(2). Additionally, early recognition of acute stroke in patients by EMS is associated with improved emergency department (ED) quality of care measures, including decreased door to computed tomography (CT) times and increased rates of alteplase use even after controlling for age, stroke severity, and symptom duration(3). Short EMS response times are associated with increased odds of survival to discharge after a stroke.

In 2019, the American Heart Association/American Stroke Association (AHA/ASA) updated their guidelines and added clarity to prior recommendations for the early management of acute ischemic stroke to include the following recommendations for individual prehospital providers: use a stroke assessment tool to screen for stroke, provide prehospital notification of suspected strokes and transport to the nearest IV alteplase capable hospital(4). Additional guidelines that remained unchanged in 2019 include recommendations to perform a neurological assessment, obtain a blood glucose and treat hypoglycemia, provide supplemental oxygen to patients only to maintain an oxygen saturation greater than 94%, initiate cardiac monitoring with a 12-lead electrocardiogram, document the last known well time, and limit response times to less than 90 seconds from call to dispatch and less than 15 minutes on-scene(5, 6). Additionally, the 2019 ASA Mission: Lifeline Severity-based Stroke Triage Algorithm suggests that EMS providers should screen for stroke severity in patients with a positive stroke screening scale. In cases where the stroke severity screening tool is suggestive of a large vessel occlusion, the algorithm further suggests that providers consider bypassing the closest facility for a thrombectomy-capable facility when it does not add more than 15 minutes to transport times and does not preclude a patient from receiving IV alteplase. Here we define guideline-concordant care as performing all eight performance metrics for which a patient is eligible with exception of compliance with the stroke severity-based triage. Full adoption of this triage algorithm may not be fully implemented in 2019 and 2020 in some areas.

At the national level, EMS activations with a final impression of stroke did not meet guidelines for EMS dispatch times in 22% of activations and on-scene times in 54% of activations in 2013 (7). Female gender, in addition to age 65–74 years versus ages less than 65 or older than 74, white race and location of non-urban geographic area, has been associated with prolonged response times; having a dispatch code of stroke may shorten response times(2, 8). Furthermore, women are at an increased lifetime risk of stroke, suffer higher stroke morbidity and mortality, and are less likely to receive guideline-concordant in-hospital care(9–12) relative to men, making the finding of prolonged response times for them even more important. To date, however, contemporary EMS nationwide concordance with ASA guidelines for prehospital care have not been analyzed, especially in respect to the potential disparities in care between men and women. The objective of this study is to characterize current EMS practices in suspected stroke patients and to assess for gender-based differences in compliance with ASA guidelines for care.

Methods:

Sample population:

This is a retrospective cohort study of the 2019 and 2020 National Emergency Medical Services Information System (NEMSIS) Version 3 Public-Release Databases. NEMSIS is a publicly available nationwide database of EMS providers in the United States who voluntarily contribute standardized, deidentified patient data with the goal of quality improvement. NEMSIS is supported by the National Highway Traffic Safety Administration and the University of Utah. 2019 and 2020 datasets reflect the most recently available data on prehospital care and allow for analysis of compliance with the 2019 American Stroke Association (ASA) guidelines. In 2019 and 2020 NEMSIS data included a total of 82,691,854 EMS activations covering 50 states and territories and 12,319 EMS agencies. Data is provided by EMS agencies in a standard form and subjected to automated cleaning rules to ensure that data entered is logical (i.e. the time at a patient does not occur before the time a EMS agency has reported to be dispatched and the elapsed time for response is not greater than 24 hours). Specific applied rules are available from NEMSIS. Our cohort included EMS activations that listed a primary impression as determined by the EMS team of stroke or cerebrovascular accident (CVA). Those with patient gender not documented were excluded from analysis (less than 1% of the total population). For this study reported gender and not biological sex was captured. Encounters with a Basic Life Support (BLS) provider were excluded from analysis of primary and secondary outcomes due to the variable and limited scope of practice of these providers which could negatively influence the ability to provide guideline-concordant care. The study was approved by the Colorado Multiple Institutional Review Board as “not human subjects” research.

Definitions of Guideline-Concordant Care:

The 2019 ASA guidelines for the early management of acute ischemic stroke were updated to include multiple class 1 recommendations for individual prehospital providers caring for potential stroke patients. For the purpose of this study, we selected eight metrics that are collected in the NEMSIS Database (Box 1): documentation of a stroke scale score, prearrival notification to a receiving facility in cases of suspected strokes (positive stroke scale), determination of a blood glucose level, provision of supplemental oxygen only to patients with a documented pulse oximetry less than 94%, cardiac monitoring with a 12-lead electrocardiogram, documentation of symptom onset, and response times of less than 90 seconds from call to dispatch and less than 15 minutes on-scene. We defined guideline-concordant care as a binary variable reflecting compliance with all of the eight performance metrics. Our secondary outcome, a composite score, was defined as the total number of performance metrics completed for a given activation, with 8 being the best possible score and 0 the worst. For both guideline-concordant care and the composite score, patients with a negative stroke scale were treated as being compliant with prenotification since prenotification is only recommended in patients who screen positive for a stroke.

Box 1: Eight Performance Metrics Defining Guideline-Concordant Care.

Documentation of a stroke scale score

Prearrival notification of receiving facility in cases of suspected stroke (positive stroke scale)

Determine blood glucose level

Provide supplemental oxygen only to patients with a pulse oximetry less than 94%

Cardiac monitoring with 12-lead electrocardiogram

Documentation of symptom onset

Response time less than 90seconds from incoming call to EMS dispatch

Less than 15minutes on-scene time

Statistical Analysis:

We characterized the cohort using descriptive statistics, reporting frequencies and percentages for categorical variables, and means/medians with interquartile range for continuous variables. Reported percentages represent the percentages out of those with reported information. Variables with high percentages of not documented information are noted in respective tables. The proportion of EMS activations compliant with each of the prehospital guidelines was calculated and stratified by gender. We calculated unadjusted odds ratios with 95% confidence intervals for guideline-concordant care among women compared to men. We used multivariable logistic regression modeling to evaluate potential gender-based differences in the odds of receiving guideline-concordant care, adjusting for age, EMS provider level (basic life support, advanced life support or critical care trained), emergency medicine dispatch (EMD) code (stroke versus other EMD code), EMS geographical service region, EMS region type (urban/suburban vs rural), and year. The secondary outcome, composite score, was analyzed using a proportional odds model to compare composite scores in men and women while controlling for age, EMS level (Advanced Life Support vs specialty care provider), EMD, EMS geographical service region, EMS region type, and dataset year. We were unable to adjust for race, documented barriers to care, and concomitant intoxication with drugs or alcohol due to a high proportion of missing data for these variables. We also report the results from a partial proportional odds model analysis(13). Statistical significance was defined at the 0.05 level. All analysis was performed using SAS v9.4 software (SAS Institute, Cary, NC).

Given the high proportion of missing documentation for the oxygenation and pre-arrival notification metrics, we conducted three separate analyses for the primary and secondary outcomes: excluding cases with missing data from the analysis (Supplemental Table 1), assigning cases without documentation of a metric as the best case scenario (i.e., assuming completed appropriately, Supplemental Table 2), and assigning cases without documentation of a metric as the worst case scenario (i.e., assuming not completed, reported in main manuscript). The worst-case scenario approach was selected as our primary analysis because it is the most conservative.

Results:

Demographics

In 2019 and 2020, there were 693,177 EMS activations within the NEMSIS database with a primary impression of stroke determined by EMS and a documented patient gender. Of these, 359,107 (51.8%) were women (Table 1). Women were slightly older (median age 73 years, Interquartile Range (IQR) 61–83) compared to men (median age 69, IQR 59–79). There were minimal differences in the proportion of men and women transported by Basic Life Support (BLS) crews (12.1% men and 12.4% women) compared to either Advanced Life Support (ALS) (80.0% men and 80.7% women) or Specialty Critical Care (7.8% men and 6.8% women) teams.

Table 1:

Patient and EMS System Characteristics

Patient Characteristic		Men (n=334,070)	Women (n=359,107)
Age (Mean, Stdv); (Median, IQR)*		68 (15); 69 (59–79)	71 (16); 73 (61–83)
Race/Ethnicity - (n, %)1	White	112,169 (63.7%)	119,966 (63.7%)
	African American	43,450 (24.7%)	46,637 (24.8%)
	American Indian/Alaska Native	1,671 (1.0%)	1,823 (1.0%)
	Asian	2,498 (1.4%)	2,696 (1.4%)
	Native Hawaiian or Pacific Islander	535 (0.3%)	558 (0.3%)
	Hispanic	15,770 (9.0%)	16,705 (8.9%)
EMS Level of Service - (n, %)*	BLS	40,509 (12.1%)	44,679 (12.4%)
	ALS	267,439 (80.0%)	289,923 (80.7%)
	Specialty Critical Care	26,122 (7.8%)	24,505 (6.8%)
EMS Region - (n, %)*	West	78,725 (25.6%)	81,550 (22.7%)
	Midwest	64,994 (19.5%)	69,303 (19.3%)
	Northeast	36,919 (11.1%)	40,105 (11.2%)
	South	153,234 (45.9%)	167,951 (46.8%)
Region Type- (n, %)*	Urban	261,638 (80.6%)	285,325 (81.7%)
	Suburban	24,332 (7.5%)	25,257 (7.2%)
	Rural	30,313 (9.3%)	30,425 (8.7%)
	Wilderness	8,388 (2.6%)	8,353 (2.4%)
EMD - CVA - (n, %)*		164,739 (49.3%)	178,450 (49.7%)
1+ Barriers to care present - (n, %)*, 2		44,019 (16.6%)	49,095 (17.2%)
Positive for drugs*, 3		11,648 (5.0%)	6,300 (2.5%)

All with less than 1% missing data except as noted below; Reported percentages represent proportion of those with documented information

^*p-values <0.05 for comparison between EMS activations involving men and women by chi-square or Wilcoxon test

¹Race not documented in 48% of charts

²Barriers to care were not documented in 21% of charts

³Positive drugs included the presence of drugs or alcohol on scene, patient admitted to being under the influence or tested positive; not documented in 31% of charts

Abbreviations: CVA –Cerebrovascular Accident, ALS – Advanced Life Support, BLS- Basic Life Support, Cerebrovascular accident, EMD – emergency medical dispatch, IQR – interquartile range, Stdv – standard deviation

EMS Transport

In terms of EMS care provided (Table 2), EMS reported following a stroke protocol in 26.1% and 26.6% of activations involving men and women respectively. There was no statistically significant difference in proportion of EMS activations involving men or women transported emergently (58.0% men and 57.8% women; p-value=0.08). There was a higher proportion of men transported to a stroke center compared to women (13.1% men, 12.9% women; p-value <0.01) with significantly fewer cases among women being chosen because the receiving facility was a regional specialty center (19.7% men and 18.1% women; p-value <0.01). Rather, a greater proportion of destination choices for women were chosen because of proximity (32.6% men and 33.5% women; p-value<0.01). Men and women had clinically similar total call times with a median time of 67 minutes (IQR 50–91) in men and 66 minutes (IQR 49–88) in women (p-value<0.05).

Table 2:

EMS Care Provided:

Characteristic		Men	Women
Stroke Protocol Used – (n, %)*		87,233 (26.1%)	95,277 (26.6%)
Emergent EMS Transport Mode - (n, %)		193,795 (58.0%)	207,487 (57.8%)
Destination Team Pre-Arrival Notification - (n, %)*, 1	Stroke Alert	73,481 (30.8%)	79,719 (31.3%)
	Other prenotification	3,013 (1.3%)	2,959 (1.2%)
Destination Capability - (n, %)*	Stroke Center	43,854 (13.1%)	46,240 (12.9%)
Reason for Destination Choice - (n, %)*, 2	Closest Facility	108,955 (32.6%)	120,224 (33.5%)
	On-line/On-Scene Medical Direction	3,095 (0.9%)	3,125 (0.9%)
	Regional Specialty Center	65,615 (19.7%)	75,001 (18.1%)
	Other	134,714 (40.4%)	147,813 (41.2%)
Response Times (Average, Stdev; median, IQR)	Dispatch Time (seconds)*	128 (324); 55 (1,120)	126 (319); 55 (0,120)
	Chute Time (minutes)*	2 (5); 1 (0,2)	2 (5); 1 (0,2)
	Response Time (minutes)*	14 (33); 8 (5,13)	13 (30); 7 (5,12)
	Scene Time (minutes)*	19 (13); 16 (12,22)	19 (13); 16 (12,22)
	Transport Time (minutes)*	20 (23); 13 (8,23)	19 (21); 13 (8,22)
	Total Call (minutes)*	78 (55); 67 (50,91)	76 (51); 66 (49,88)

All with less than 1% missing data except as noted below; reported percentages represent proportion of those with documented information

^*p-values <0.05 for comparison between EMS activations involving men and women by chi-square or Wilcoxon test

¹Prenotification – not documented in 41% of EMS activations

²Reason for Destination Choice – not documented in 6% of EMS activations

Definitions: Dispatch time (time from call to notification of EMS team); Chute time (time from dispatch to EMS en route); Response time (time to patient); Scene time (time from arrival on-scene to leaving scene); Transport time (time from scene to destination)

Abbreviations: Stdev – standard deviation; IQR – interquartile range

EMS Assessments

In terms of patient assessments and acuity (Table 3), men and women had clinically similar vital signs with respect to heart rate, systolic blood pressure, respiratory rate, and pulse oximetry. However, there was a significantly higher proportion of women receiving supplemental oxygen for any reason (38.0% men and 39.6% women; p-value<0.01) with a higher proportion of this also being indicated per ASA guidelines (lowest pulse oximetry of less than 94%) (23.6% men and 25.7% women; p-value<0.05). Only 41.4% of activations involving men and women had a positive stroke scale. The most used stroke scale was the Cincinnati Prehospital Stroke Scale (61.0% men, 61.3% women). However, 38% of encounters did not report the type of stroke scale used.

Table 3:

EMS Assessments/Interventions:

Characteristic		Men	Women
Vitals (Mean, StDev; Median, IQR)	Heart Rate (bpm)*	84 (18); 82 (71, 94)	85 (18); 83 (73, 95)
	Systolic Blood Pressure (mmHg)*	152 (30); 150 (132,171)	153 (31); 151 (131,173)
	Respiratory Rate (brpm)*	18 (3); 18 (16,19)	18 (3); 18 (16,19)
	Pulse Oximetry*	96 (3); 97 (95,98)	96 (4); 96 (97,98)
	Lowest Pulse Oximetry*	95 (6); 96 (94,98)	95 (6); 94,98)
Supplemental Oxygen Administered - (n, %)*, 1		57,394 (38.0%)	64,461 (39.6%)
Supplemental Oxygenation per Recommendations - (n, %)*,2	Indicated, Received	22,895 (23.6%)	26,756 (25.7%)
	Indicated, not received	9,660 (10.0%)	9,637 (9.3%)
	Not indicated, Received	31,515 (32.5%)	34,618 (33.3%)
	Not indicated, not received	33,044 (34.0%)	32,951 (31.7%)
Blood Glucose Obtained - (n, %)*,3	Completed	232,653 (69.7%)	255,280 (71.1%)
	refused/unable to complete	3,778 (1.1%)	3,664 (1.0%)
Type of ECG obtained - (n, %)*,4	12lead or better	68,590 (20.5%)	69,426 (19.3%)
	AED/3–5 lead	115,627 (34.6%)	126,411 (35.2%)
Stroke Scale Score - (n, %)*,5	Positive	138,319 (41.4%)	148,501 (41.4%)
	Inconclusive	30,936 (9.3%)	34,847 (9.7%)
	Negative	31,739 (9.5%)	34,183 (9.5%)
	Refused/Unable to complete	7,354 (2.2%)	7,794 (2.2%)
Type of Stroke Scale Used - (n, %)*,5	CPSS	126,667 (61.0%)	137,205 (61.3%)
	LAPSS	17,549 (8.5%)	19,163 (8.6%)
	NIHSS	2,753 (1.3%)	2,691 (1.2%)
	FAST	27,131 (13.1%)	29,000 (13.0%)
	Other	32,848 (15.9%)	35,736 (16.0%)
Time of Symptom Onset Documented*		243,458 (72.9%)	262,563 (73.1%)

^*p-values <0.05 for comparison between EMS activations involving men and women by chi-square or Wilcoxon test

¹Supplemental Oxygen Administration – not documented in 55% of EMS activations

²Supplemental Oxygenation indicated according to AHA guidelines for SpO₂<94%; 71% missing data with regards to oxygen administration and/or pulse oximetry documentation

³Blood Glucose – not documented in 28% of EMS Activations

⁴ECG – not documented in 45% of EMS activations

⁵Stroke Scale Completion Score – not documented in 38% of EMS activations

Abbreviations: AED – automated external defibrillator, bpm – beats per minute; brpm – breaths per minute; CPSS - Cincinnati Prehospital Stroke Scale; ECG- Electrocardiogram; FAST - Face, Arm, Speech, Time; IQR – Interquartile Range; LAPS - Los Angeles Prehospital Stroke Scale; NIHSS - National Institutes of Health Stroke Scale; StDev – Standard Deviation

Guideline-Concordant Care

Assuming that any metric not documented was also not performed (“worst-case” scenario for missing data) and excluding encounters with a BLS provider, compliance with the eight measured performance metrics ranged from 18.1% of all activations administering supplemental oxygen according to guidelines to 75.9% of all calls having a dispatch time of less than 90 seconds (Table 4a). When combining compliance with individual metrics into a single measure of overall compliance, only 0.36% of activations involving women and 0.43% of activations involving men were fully guideline-concordant with all eight recommendations (Table 4b). The unadjusted odds of having guideline-concordant care for EMS activations involving women was 0.83 times those involving men (95% CI 0.76 to 0.90). After controlling for factors previously found to be associated with delays in transport time (age, EMS level of service, EMS geographical region, EMS service region type, EMD) and year, the adjusted odds of having guideline-concordant care for EMS activations involving women was essentially unchanged, at 0.82 times that of EMS activations involving men (95% CI 0.75 to 0.89). Alternatively, excluding missing data or assuming that any metric not documented was performed (Supplementary Tables 1, 2), the adjusted odds of having guideline-concordant care for EMS activations involving women was 0.82 (95% CI 0.75–0.89) and 0.86 (95% CI 0.84–0.89), respectively.

Table 4a:

Proportion of Encounters Completing Individual Performance Metric

Metric	Overall	Men	Women
Documentation of Stroke Scale*	379,569 (62.4%)	182,263 (62.1%)	197,306 (62.8%)
Prenotification*	146,628 (24.1%)	70,3442 (24.0%)	76,286 (24.3%)
Blood Glucose obtained*	450,565 (74.1%)	214,832 (73.2%)	235,733 (75.0%)
Oxygen only for SpO2 <94%	109,755 (18.1%)	53,148 (18.1%)	56,607 (18.0%)
12-lead ECG*	135,328 (22.3%)	67,244 (22.9%)	68,084 (21.7%)
LKW/Symptom onset documented*	452,984 (74.5%)	217,831 (74.2%)	235,153 (74.8%)
<90sec from call to dispatch	461,409 (75.9%)	222,802 (75.9%)	238,607 (75.9%)
<15min on-scene time*	279,897 (46.0%)	137,119 (46.7%)	142,778 (45.4%)

^*p-value<0.05 05 for comparison between EMS activations involving men and women by chi-square

Abbreviations: ECG – Electrocardiogram; LKW – last known well; SpO₂ – pulse oximetry

Table 4b:

Proportion of Encounters with Fully Guideline-Concordant Care or Given Composite Score

Outcome		Overall	Men	Women
Guideline-Concordant Care*		2,382 (0.39%)	1,263 (0.43%)	1,119 (0.36%)
Composite Score*	8	2,382 (0.39%)	1263 (0.43%)	1,129 (0.36%)
	7	22,243 (3.7%)	11,090 (3.8%)	11,153 (3.6%)
	6	84,458 (13.9%)	41,287 (14.1%)	43,171 (13.7%)
	5	160,080 (26.3%)	76,568 (26.1%)	83,512 (26.6%)
	4	170,331 (28.0%)	81,398 (27.7%)	83,512 (26.6%)
	3	110,142 (18.1%)	42,928 (18.0%)	57,214 (18.2%)
	2	42,294 (7.6%)	22,911 (7.8%)	23,383 (7.4%)
	1	11,099 (1.8%)	5,646 (1.9%)	5,453 (1.7%)
	0	960 (0.16%)	470 (0.16%)	490 (0.16%)

^*p-value<0.05 05 for comparison between EMS activations involving men and women by chi-square

Abbreviations: ECG – Electrocardiogram; LKW – last known well; SpO₂ – pulse oximetry

Composite Score for Metrics Completed

To assess guideline compliance in terms of the number of metrics performed during an encounter, we calculated a composite score (Table 4b). This secondary outcome also significantly differed among EMS activations involving men and women over a spread of possible scores from 0 to 8 (again, reporting “worst-case” scenario that missing data was considered not performed). Using a proportional odds model, women had a similar odds of a given composite score compared to men across composite scores (adjusted OR 0.99, 95% CI 0.98–1.0). However, the proportional odds assumption was found to be violated (p-value for score test for the proportional odds assumption <0.001). This suggests that differences between men and women across composites are not constant. Using a partial proportional odds model to evaluate differences between composite scores, the odds of a composite of 8 compared to less than 8 were significantly decreased in EMS activations involving women compared to men (adjusted OR 0.82, 95% CI 0.76–0.89) (Table 5). At higher composite scores, women had a lower odds of compliance compared to men (i.e. composite score 7 or less compared 8 - adjusted OR 0.92, 95% CI 0.90 – 0.94). However, at composite scores of 4 or less compared score of 5 to 8, the difference was either non-significant or it appeared that women had a greater odds of a given composite score compared to men (i.e composite score 2 or less compared to 3 to 8, adjusted OR 1.06, 95% CI 1.02 – 1.10). Similar patterns were seen when missing data was excluded or considered “best case” scenario as performed (Supplementary Tables 1, 2)

Table 5:

Partial Proportional Odds Modeling for Composite Score in Women and Men

Composite Score	Unadjusted OR (95% CI)	Adjusted OR (95% CI)
8	0.83 (0.76 – 0.90)	0.82 (0.76 – 0.89)
7	0.93 (0.90 – 0.95)	0.92 (0.90 – 0.94)
6	0.96 (0.95 – 0.97)	0.95 (0.94 – 0.96)
5	0.99 (0.98 – 1.00)	0.98 (0.97 – 0.99)
4	1.02 (1.01 – 1.03)	1.00 (0.99 – 1.01)
3	1.07 (1.05 – 1.09)	1.04 (1.02 – 1.06)
2	1.10 (1.07 – 1.15)	1.06 (1.02 – 1.10)
1	1.03 (0.91 – 1.17)	0.99 (0.87 – 1.12)

^*Multivariable models adjusted for: age, EMS level of service, EMD, EMS Region, Urban/City, dataset year comparing women to men across all lower levels of composite score.

Discussion:

This study takes a contemporary look at EMS practices with regards to prehospital care of patients with a suspected stroke and identifies low adherence to current evidence-based guidelines for prehospital care and a small but statistically significant discrepancy in care between men and women. In addition to the overall low compliance with all eight ASA guidelines for care of suspected stroke patients, there was also great variability in compliance between individual guidelines. Over 70% of EMS activations involving suspected strokes were compliant with recommendations for documentation of symptom onset, dispatch within 90 seconds from incoming call to EMS, and obtaining a blood glucose measurement. However, EMS activations involving women were less likely to be fully compliant with ASA guidelines. The clinical significance of this requires clarification in future studies.

While this study specifically examines potential gender-based differences in prehospital stroke care, our findings are consistent with prior studies demonstrating that women may experience delays in presentation, be less likely to receive guideline-concordant in-hospital care, and may respond to stroke treatments differently(2, 12, 14, 15). The results of studies assessing gender-based disparities in stroke care are inconsistent. Using 2004 data, a large retrospective study found no difference in-hospital neuroimaging, electrocardiography, or carotid artery ultrasound use between men and women with stroke(16). However, using data from the 2003 to 2008 Get with the Guidelines- Stroke program, women were found to be less likely to receive guideline-concordant in-hospital stroke care compared to men with regard to the ASA ‘Get with the Guidelines-Stroke’ targets for alteplase administration in eligible patients; DVT prophylaxis within 48hours of admission, antithrombotic agents within 48 hours of admission and upon discharge, anticoagulation for atrial fibrillation, treatment of hyperlipidemia, and counseling for smoking cessation(12). Women were also less likely to have advanced imaging and to receive diagnostic and therapeutic interventions(14, 15). These disparities in care may contribute to differences in stroke outcomes between men and women. Several studies have shown that women have worse functional outcomes and increased mortality after adjusting for baseline differences in age and comorbidities(12, 14, 17–19).

Despite the data suggesting gender-based discrepancies in in-hospital stroke care and outcomes, little emphasis in this area has been placed on prehospital care. Our study demonstrates that these disparities begin in the prehospital period. The reason for these disparities are unknown. One possibility is that the small differences in guideline-concordant care (0.36% women and 0.43% in men) may be explained in part by differences in presentations. Women present with atypical symptoms (pain, change in level of consciousness, and unclasiffiable neurologic or nonspecific symptoms) 62% more frequently than men(20). This may prompt EMS providers to focus on other potential interventions. However, in this study, our population was limited to encounters where EMS providers reported that the most likely diagnosis was a suspected stroke. So, while differences in presentations may influence care, in patients with suspected stroke, it is important to emphasize that compliance with established guidelines may improve patient outcomes. Improved understanding of the underlying reasons for differences in care is needed to design further targeted interventions to improve guideline compliance in this population.

Additionally, while the individual differences in compliance with a given metric appear small and may be explained by differences in presenting symptoms, these disparities may translate to even greater differences in patient outcomes when compounded with in-hospital disparities in care and gender-based differences in response to therapies. At the most basic level, EMS on-scene times, there are gender-based discrepancies in care. While the overall median total call time for women patients was less than for men, there were still fewer calls involving women compared to men with an on-scene time less than 15 minutes, the standard per ASA guidelines. The reasons behind this difference in on-scene time are unknown and may reflect underlying differences in patient presentations. These nationwide practice patterns were also reflected in prior studies in Michigan, where less than 50% of all ischemic strokes presenting via EMS had on-scene times of less than 15 minutes on-scene(21). In other areas of stroke care, the findings in Michigan reflected higher compliance with ASA guidelines compared to the nationwide practices reported here. In that study, 78.5% of strokes had a documented Cincinnati Prehospital stroke scale (compared to 61.0% in our sample) and 86% had a documented blood glucose (compared to 70.2% in our sample).

Assessing compliance with ASA guidelines should be evaluated with caution as EMS has been shown to be slow to implement new recommendations in multiple settings. After state-mandated changes in EMS trauma triage and destination plans, EMS providers in North Carolina did not significantly alter their transport patterns in a 6 month post-implementation period that followed a 3 month implementation(22). This is consistent with the average 416 days that it took EMS agencies involved in the Resuscitation Outcomes Consortium to implement the 2005 American Heart Association guidelines to improve out-of-hospital cardiac arrest survival(23). Our study analyzed compliance with the recent 2019 ASA guidelines for prehospital stroke care using 2019 and 2020 NEMSIS data. However, the fundamental principles of the measured metrics are well established dating back to the 2007 and 2013 guidelines for acute stroke management(6, 24). In the 2018 and 2019 ASA guidelines, the recommendations were unchanged; rather, the exact wording was updated to provide further clarity(4, 25). The 2019 prehospital acute stroke care recommendations that required significant practice changes reflect advances in stroke triage and screening for large vessel occlusion. However, due to the nature of the NEMSIS database and previous data suggesting the EMS providers are unaware of these changes, we chose not to assess for gender-based differences in compliance with these recommendations(26). Thus, guideline compliance will continue to increase over time, but delayed implementation of guidelines is unlikely to explain the low proportion of guideline-concordant EMS encounters in men and women we found here.

Interestingly, despite the emphasis on multiple aspects of prehospital care, only prenotification of suspected stroke and emergent transportation with reduced call times have been shown to improve in-hospital care of stroke patients(27, 28). Unfortunately, these are two areas with low compliance in the nationwide sample – EMS notified receiving hospitals of an incoming patient with a positive prehospital stroke scale in only 24.1% of cases, only 46.0% of EMS activations had an on-scene time <15 minutes, and 58% of activations were transported emergently. In our study, even after controlling for factors that previously have been shown to improve response time and likelihood of EMS calls meeting ASA standards (age, geographical region, EMS level of service, and non-urban service region type), the discrepancy between men and women in guideline-concordant care persisted(7).

Limitations:

While our results suggest an important area for improvement in stroke care, they should be interpreted in light of the limitations of this study design. The NEMSIS database is a convenience sample with only a portion of all EMS agencies in the United States voluntarily reporting data. Participation in the NEMSIS database continues to grow but is limited by the voluntary nature of the program and the willingness of EMS agencies to collect and share data, although the information can improve patient care. Given the nature of reporting, this study found some variables with high levels of missing data (race, payer status, potential barriers to care, supplemental oxygen administration, and stroke scale type used). We therefore modified our analysis to provide transparency with regards to missing data in guideline-concordant care and excluded some potential confounders (race, barriers to care, payer status, intoxication) in the multivariable logistic regression model. It is likely that the cause of missing data in transport times compared to electrocardiograph monitoring is not the same. Transport times are automatically documented whereas items like pulse oximetry, blood glucose, and ECG may not be automatically captured. While factors influencing missing data may vary with individual metric, it is unlikely that they significantly vary by sex. In multivariable modeling, EMS provider level (ALS or specialty critical care) were significantly associated with guideline-concordant care. While BLS providers may be more restricted in scope of practice, more advanced providers should be capable of performing the identified performance metrics and are often the primary responders in cases of suspected stroke. Thus, we chose to limit our analysis to ALS or higher levels of EMS providers to minimize this confounder. Additionally, we chose to use three separate analyses – excluding missing data, counting missing data as not performed, and counting missing data as performed. The results of these separate analyses were similar, suggesting that missing data does not fully explain the gender-disparities in care. Finally, without patient outcome data, it is not possible to link the findings directly to changes in patient outcomes. Despite these limitations, the strength of our study lies in its large sample size of EMS activations from a variety of clinical settings that more accurately reflects nationwide practices compared to more detailed analysis of individual practice settings.

Conclusions:

We found that in the prehospital setting, women are less likely to receive guideline-concordant care for suspected stroke or CVA. Combined with known gender-based differences in in-hospital care and response to potential therapies, this has the potential to significantly compound both the morbidity and mortality associated with stroke in women compared to men. Future studies are needed to evaluate the potential causes for this discrepancy in prehospital care and to design targeted interventions to improve care for both men and women suffering from an acute ischemic stroke.

Data Availability:

The data that support the findings of this study are available from the NEMSIS Database at https://nemsis.org/using-ems-data/request-research-data/