- Email: [email protected]
Presenting Symptoms and Onset-to-Arrival Time in Patients With Acute Stroke and Transient Ischemic Attack
Presenting Symptoms and Onset-to-Arrival Time in Patients With Acute Stroke and Transient Ischemic Attack Julia Warner Gargano, PhD,* Susan Wehner, MSN,† and Mathew J. Reeves, PhD*
Delayed arrival to the emergency department (ED) precludes most stroke patients from receiving thrombolytic treatment. Our objective in this study was to examine the association between presenting symptoms and onset-to-arrival time (ie, time between onset of symptoms to arrival at the ED) in a statewide stroke registry. Demographics, clinical data, and presenting symptoms were collected for patients with acute stroke or symptomatic transient ischemic attack (TIA) admitted to 15 Michigan hospitals (n 5 1922). Polytomous logistic regression models were developed to test the association between presenting symptoms and onset-to-arrival time (classified as ,2 hours, 2-6 hours, or .6 hours/unknown). Onset-to-arrival time was ,2 hours in 19% of the patients, 2-6 hours in 22%, and .6 hours/unknown in 59%. Unilateral symptoms (reported by 40%) and speech difficulties (reported by 22%) were associated with increased likelihood of arriving within 2 hours (unilateral: adjusted odds ratio [aOR], 1.5; 95% confidence interval [CI], 1.1-1.9; speech: aOR, 1.6; 95% CI, 1.2-2.2). Difficulty with walking, balance, or dizziness (12%), confusion (9%), loss of consciousness (6.7%) and falls (3.4%) were associated with lower likelihood of arriving within 2 hours (walking: aOR, 0.7; 95% CI, 0.4-1.0; confusion: aOR, 0.5; 95% CI, 0.3-0.8; consciousness: aOR, 0.5; 95% CI, 0.1-0.9; falls: aOR, 0.4; 95% CI, 0.3-0.9). Presenting symptoms were strongly associated with time of arrival; patients with unilateral symptoms and speech difficulties were more likely to seek care early. Future studies should consider including more specific patient-level data to identify psychosocial and behavioral aspects of recognition and action to stroke symptoms. Key Words: Signs and symptoms—emergency treatment—prehospital delay. Ó 2011 by National Stroke Association
Although intravenous (IV) tissue plasminogen activator (t-PA) is an effective treatment for ischemic stroke (IS), only a small minority of patients (3.0%-8.5%) receive it.1 The most important reason for patients not receiving t-PA treatment is late presentation to the emergency department (ED).2,3 Consequently, reducing prehospital From the *Department of Epidemiology, and †Department of Neurology and Ophthalmology, College of Human Medicine, Michigan State University, East Lansing, Michigan. Received September 3, 2009; accepted February 5, 2010. Supported by Centers for Disease Control and Prevention Cooperative Agreement No. U50/CCU520272-01. The authors report no conflicts of interest. Address correspondence to Mathew J. Reeves, PhD, Department of Epidemiology, Michigan State University, B601 West Fee Hall, East Lansing, MI 48824. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2011 by National Stroke Association doi:10.1016/j.jstrokecerebrovasdis.2010.02.022
494
delays for stroke patients has been an important public health goal.4,5 Arrival to the ED within a narrow time interval requires that individuals react to their stroke symptoms and seek care expeditiously. Some studies suggests that people delay seeking care because they wait for symptoms to resolve or do not recognize the symptoms as being important, particularly if the symptoms do not seem severe.6-9 In contrast, individuals whose first impression is that they are having a stroke seek care more quickly.6-8 Only a few epidemiologic studies have reported on the associations between specific symptoms and prehospital delay10-13; of these, the only study conducted in the United States reported on patients in a single metropolitan area in the early 1990s.12 More recent population-based information on the relationship between stroke symptoms and prehospital delay could inform public health messages aimed at educating individuals with stroke symptoms to seek
Journal of Stroke and Cerebrovascular Diseases, Vol. 20, No. 6 (November-December), 2011: pp 494-502
SYMPTOMS AND ONSET-TO-ARRIVAL TIMES
care early. The objective of this study was to examine the association between presenting symptoms and onset-toarrival time (ie, time between onset of symptoms to arrival at the ED) in patients admitted with acute stroke or symptomatic transient ischemic attack (TIA) in a statewide stroke registry.
Methods Study Design The Paul Coverdell National Acute Stroke Registry (PCNASR) was designed to monitor and improve the quality of acute stroke care in the United States.1 The Michigan Acute Stroke Care Overview and Treatment Surveillance System (MASCOTS) was a statewide, hospital-based, acute stroke registry prototype for the PCNASR. Details of the design of the MASCOTS registry have been published previously.1,14 In brief, a single-stage cluster design that used a modified stratified sampling regimen was implemented to obtain a representative statewide sample of 15 Michigan hospitals. Human subject approval was obtained from each hospital’s institutional review board.
Selection of Participants Acute stroke and TIA admissions in the ED at each participating hospital over a 6-month period in 2002 were identified by trained stroke clinical coordinators. Patients with TIA whose symptoms had resolved by the time of presentation to the ED were not included in the registry, because they would not have been considered for t-PA treatment. Thus, only patients with symptomatic TIA are included in this study. Of the 2566 registry subjects, those who did not initially present to the ED (281 direct admissions and 285 hospital transfers) were excluded, leaving 2000 primary ED admissions. Seventy-eight patients (4%) were excluded due to missing time interval data, leaving 1922 subjects for this analysis.
Data Collection and Processing Information on demographics, prehospital care, prestroke ambulatory status, modified Rankin Scale (mRS) score at discharge, and past medical history was abstracted from patient charts using a standardized data collection instrument. Cases were classified as IS (including IS of uncertain duration), hemorrhagic stroke (HS, including intracerebral and subarachnoid hemorrhage), or TIA. Data on time of symptom onset and ED triage were abstracted from the medical charts. The exact date and time of symptom onset were noted when available; if unknown, symptom onset was estimated within a 6-hour interval when possible, or as the date and time the patient was last seen normal.1 Data on presenting symptoms, as recorded as the chief complaint during the initial ED triage, were collected as
495 15
free text. As described previously, a computer program was used to identify patients considered suspected stroke cases on arrival, and to identify specific individual presenting symptoms. Individual symptoms were subsequently aggregated into 5 stroke warning signs (WSs) consistent with the American Heart Association’s (AHA) public health messages:5 numbness/weakness (WS 1), confusion/speech disturbance (WS 2), visual disturbance (WS 3), trouble walking/dizziness/poor balance (WS 4), and headache (WS 5). Other symptoms not included in these 5 WSs (eg, nonheadache pain, loss of consciousness/syncope, falls) were reported individually.
Outcome Measures For subjects with an exact stroke onset time recorded, the onset-to-arrival time was calculated by subtracting the onset date and time from the ED arrival date and time. For subjects with only an estimated time of symptom onset, the beginning of the estimated 6-hour window or the time last seen normal was used to determine the most conservative estimate of onset-to-arrival time. Time intervals were initially categorized as ,2 hours (n 5 357; 18.6%), 2-3 hours (n 5 128; 6.7%), 3-6 hours (n 5 295; 15.3%), $6 hours (n 5 443; 23.0%), and unknown (n 5 699; 36.4%). These categories were aggregated into 3 onset-to-arrival time groups based on eligibility for tPA-related treatments: (1) optimally eligible for IV t-PA (arrived ,2 hours), (2) potentially eligible for IV or intra-arterial t-PA (arrived 2-6 hours), and (3) ineligible for t-PA (arrived $6 hours or unknown).
Primary Data Analysis The demographic and clinical features, suspect stroke, stroke WSs, and self-reported symptoms among the 3 onset-to-arrival groups were compared using the c2 test. The association between onset-to-arrival group and discharge mRS score (0-6) was assessed using the CochranMantel-Haenszel test, with modified Ridit scores to account for ordered categories.16 Patient characteristics independently associated with the onset-to-arrival time groups were identified using multivariate polytomous logistic regression models, using t-PA–ineligible patients (ie, those with arrival time $6 hours or unknown) as the reference group. Age, sex, race, and use of emergency medical services (EMS) were included as a priori variables of interest. Nursing home residence, place (ie, home, work, other) and timing (ie, weekday, weekend, unknown) of stroke onset, prestroke ambulatory status, stroke subtype, mRS score at discharge, and previous medical history (ie, stroke/TIA, heart disease, congestive heart failure, atrial fibrillation, dyslipidemia, and current smoking) also were considered for inclusion (backward-elimination variable selection, with global P ,.05 to stay). Next, presenting symptoms (ie, unilateral symptoms, speech/language difficulty, confusion,
496
walking/dizziness/imbalance, visual disturbance, headache, nonheadache pain, falling, loss of consciousness, and suspected stroke) were added, and were retained if the global P value was ,.10. An additional variable indicating the absence of all WSs or suspected stroke was tested as well. After the final main-effects model was defined, the interaction terms between EMS arrival and symptoms were tested. Model results are presented as adjusted odds ratio (aOR) and 95% confidence interval (CI) for each time interval compared with the t-PA–ineligible group. To illustrate the effect of the polynomial specification of age (ie, age and age2), ORs for 10-year contrasts with age 65 were estimated from the regression model. Finally, the characteristics and influence of subjects with unknown onset-to-arrival time were explored. First, subjects with unknown onset-to-arrival time were compared with those known to have arrived .6 hours after onset using the c2 test. A sensitivity analysis was then performed by excluding subjects with unknown onset-to-arrival time from the final model. The time of day of stroke onset (ie, morning, afternoon, evening, overnight) was explored in this latter model. A sensitivity analysis also was performed by rerunning the final model after excluding all TIA cases.
Results Demographic and clinical characteristics of the 1922 subjects included in this analysis are presented in Table 1. Overall, onset to arrival time was ,2 hours in 18.6% of the subjects, 2-6 hours in 22%, and $6 hours or unknown in 59.4%. Distributions of time intervals by patient demographic and clinical characteristics are presented in Figure 1. There were no statistically significant (P , .05) associations between onset-to-arrival time and sex, race, age, nursing home residence, or insurance status (Fig 1A). Although the difference was not not statistically significant, the oldest and youngest subjects tended to be less likely than those in the middle of the age distribution to arrive early. Subjects with TIA or HS were more likely to arrive within 2 hours of onset than those with IS (Fig 1B), and subjects arriving by EMS were twice as likely to arrive within 2 hours of onset as those who did not (Fig 1C). Having a stroke at work or a stroke occurring between noon and midnight also was associated with earlier arrival (Fig 1C). Distributions of time intervals by suspect stroke, the 5 stroke WSs, and specific individual symptoms are presented in Figure 2. Of the 5 WSs, 3 were significantly associated with onset-to-arrival time in unadjusted analyses (Fig 2A); numbness/weakness (WS 1) was associated with early arrival, whereas visual disturbance (WS 3) and trouble walking/dizziness/balance (WS 4) were associated with later arrival. Neither suspected stroke nor the absence of any WSs were associated with onset-to-arrival time, however (Fig 2A).
J.W. GARGANO ET AL.
The individual symptoms of weakness and facial droop were both significantly associated with earlier arrival, as was the presence of any unilateral symptoms (Fig 2B). Although confusion/speech (WS 2) was unrelated to onset-to-arrival time (Fig 2A), its two individual components were both significantly associated with onset-to-arrival time, but in opposite directions; subjects reporting speech/language difficulties were more likely to arrive early, whereas those reporting confusion were more likely to arrive later (Fig 2B). Dizziness and nonheadache pain were both significantly associated with delayed arrival (Fig 2C). Trouble walking, balance/coordination problems, and falls (Fig 2C) all showed strong trends toward delayed arrival, but the smaller number of subjects reporting these symptoms limited the statistical power of these comparisons. Interestingly, loss of consciousness/syncope was not statistically significantly associated with time of arrival in the unadjusted analysis. Multivariate model results, presented in Table 2, show that many presenting symptoms were significantly associated with onset-to-arrival time intervals. The presence of unilateral symptoms and speech/language difficulties were associated with increased likelihood (OR, 1.5 and 1.6, respectively) of arriving within 2 hours, whereas confusion, nonheadache pain, loss of consciousness/syncope, and falls were all associated with significantly reduced likelihood of early arrival (ORs, 0.3-0.7). The WSs of visual disturbance (WS 3) and difficulty with walking/balance/dizziness (WS 4) also were associated with delayed arrival. After accounting for these presenting symptoms, significant relationships between patient characteristics and onset-to-arrival time remained. There was a quadratic relation between age and early arrival (ie, age2 was significant), with the youngest and oldest subjects having less likelihood of arriving ,2 hours after symptom onset compared with the subjects aged 65 years. Arrival by EMS, being ambulatory, and experiencing stroke while at work were all associated with earlier arrival. Compared with the subjects with IS, those with TIA or HS were both more likely to arrive earlier (ie, ,2 or 2-6 hours after symptom onset). No significant interactions between arrival by EMS and any other variables were seen. Compared with subjects with an unknown onset-toarrival time, subjects known to have arrived $6 hours after symptom onset were more likely to be black (22% vs 18%), less likely to be age $80 years (18% vs 22%), less likely to report confusion (8% vs 12%), and more likely to report speech difficulty (23% vs 17%). However, there were no significant differences between these 2 groups by sex, EMS arrival, stroke subtype, or any other symptoms or WSs. After excluding the 699 subjects with unknown onset time from the adjusted model, ORs were similar, but there were fewer statistically significant findings attributable to diminished statistical power.
SYMPTOMS AND ONSET-TO-ARRIVAL TIMES
497
Table 1. (Continued )
Table 1. Distribution of demographic and clinical characteristics in the study subjects (n 5 1922)
Total Sex Female Male Race Black White Other/not known Age, years ,50 50-59 60-69 70-79 $80 Nursing home resident Yes No Insurance type Public only Some/all private None Previous stroke Yes No Previous myocardial infarction/coronary heart disease Yes No Atrial fibrillation Yes No Prestroke ambulatory status Not independent Independent Stroke subtype TIA IS HS Onset day* Weekday Weekend Unknown Onset timey Morning Afternoon Evening Overnight Unknown Onset place Home Work Other EMS arrival Yes No
n
%
1922
100.0
1041 881
54.2 45.8
340 1414 168
17.7 73.6 8.7
203 269 328 559 563
10.6 14.0 17.1 29.1 29.3
85 1837
4.4 95.6
380 1462 80
19.8 76.1 4.2
731 1191
38.0 62.0
668 1254
34.8 65.2
290 1632
15.1 84.9
167 1755
8.7 91.3
428 1293 201
22.3 67.3 10.5
1280 506 136
66.6 26.3 7.1
357 128 295 443 699
29.2 10.5 24.1 36.2 36.4
1145 48 729
59.6 2.5 37.9
806 41.9 1116 58.1 (Continued )
Onset-to-arrival time ,2 hours 2-6 hours $6 hours/unknown
n
%
358 423 1142
18.6 22.0 59.4
*Weekday: Monday-Friday; weekend: Saturday-Sunday. yMorning: 6 am-noon; afternoon: noon-5:59 pm; evening: 6:00 pm-midnight; overnight: midnight-5:59 am.
When time of day of symptom onset was added to this restricted model, it was strongly associated with onset-to-arrival time (P , .0001); compared with subjects with stroke occurring in the morning (ie, 6 am to noon), those with stroke occurring in the afternoon or evening had an increased likelihood of arriving ,2 hours after onset (afternoon: OR, 1.6; evening: OR, 1.7), whereas those with stroke occurring overnight had less likelhood of arriving ,2 hours after onset (OR, 0.4). Adding time of day did not affect any associations with symptoms or other variables, except that the effect of onset at work was attenuated and lost statistical significance (P 5 .44). Excluding subjects with TIA from the full model also resulted in loss of statistical significance for some symptoms, but the magnitude of the ORs remained similar.
Discussion In this analysis of patients admitted with acute stroke in a statewide hospital-based registry, presenting symptoms were strongly associated with onset-to-arrival time. Unilateral symptoms and speech difficulties were linked with greater likelihood of early arrival (ie, ,2 hours after onset), whereas difficulty with walking/balance/dizziness, vision problems, and confusion were associated with lower likelihood of early arrival. Three symptoms not incorporated in the AHA’s 5 WSs—nonheadache pain, loss of consciousness/syncope, and falls—also were associated with lower likelihood of arriving within 2 hours of symptom onset. Only a few studies have empirically examined associations between specific symptoms and prehospital delay,10-13 and only one of these studies is from the United States.12 This US report was based on data generated from a population-based surveillance study of acute stroke in the Minneapolis–St. Paul area in 1991-1993, in which detailed symptom data were abstracted from patient charts.12 In contrast to our findings, that study found that the presence of syncope and seizures at stroke onset was associated with earlier arrival. However, the study also found that unsteadiness and impaired vision were associated with greater delay, which is concordant with our results. Rossnagel et al11 prospectively interviewed 558 patients with stroke and TIA in 4 Berlin hospitals about their symptoms and found that unilateral numbness was associated with
498
J.W. GARGANO ET AL.
Figure 1. Proportion of subjects arriving at ,2 hours (black) and 2 hours - ,6 hours (white) after symptom onset, by (A) demographics, (B) previous medical history and current stroke subtype, and (C) stroke onset time and place and EMS use. *P ,.05, c2 test across all categories. †Weekday, Monday-Friday; weekend: Saturday-Sunday. ‡Morning: 6 am-noon; afternoon: noon- 5:59 pm; evening: 6:00 pm -midnight; overnight: midnight-5:59 am. Total subjects 5 1922, except xc2 test restricted to subjects with known onset time.
shorter delays, whereas walking disturbance and faintness/weakness were associated with longer delays. These findings also are consistent with our results. More generally, previous studies on stroke-related symptoms and prehospital delay have found that patients who initially perceived their symptoms as not serious,6,7,9,11,13 or did not ascribe their symptoms to stroke,6,8,13,17 tended to delay seeking care. Unfortunately, we were not able to collect data on our subjects’ perceptions or interpretation of their symptoms in the present study. Not surprisingly, previous studies have found that that earlier arrival is associated with more severe stroke9,10,12,18
and with specific features that suggest severe stroke, such as sudden onset of symptoms,10,19 altered level of consciousness,8,12,13 and severe motor impairment.7,10 Consistent with our findings in the present study, previous studies also have found that patients with HS— which typically presents with more severe, fulminant symptoms—tend to arrive earlier than those with IS.12,13,19 We were surprised to find that the patients with TIA were more likely to arrive within 2 hours compared with those with IS, given previous reports indicating that TIA patients often delay seeking medical care.20 However, at least 2 clinical studies have found that
SYMPTOMS AND ONSET-TO-ARRIVAL TIMES
499
Figure 2. Proportion of subjects arriving at ,2 hours (black) and 2 hours - ,6 hours (white) after symptom onset, by (A) stroke warning signs and suspect stroke, (B) symptom components of WSs 1 and 2, and (C) symptom components of WS 4 and other symptoms. *P , .05, c2 test across all categories. †Number (%) of subjects with each symptom. Total subjects 5 1922.
patients with TIA arrived earlier to the hospital than patients with IS.11,19 We suspect that this finding in our study subjects reflects the inclusion criteria used by the PCNASR, which limited TIA cases to those patients who still exhibited symptoms on presentation to the hospital.1 Thus, TIA patients with shorter duration of symptoms would be less likely to be included in our registry data. The registry’s exclusion of nonsymptomatic TIA patients was based on the fact that these patients would not have been considered for t-PA treatment. Consequently, we emphasize that no specific conclusions
can be drawn from our findings regarding subjects’ responses to symptoms of TIA that resolved before presentation. We note that the final model results were similar after we excluded all patients with TIA, and thus we do not believe that their inclusion is a source of bias. There are several important observations to be made regarding the fact that unilateral symptoms and speech/ language difficulties were the 2 symptoms most strongly associated with early arrival. Unilateral symptoms— including facial droop and leg or arm weakness—and speech difficulties are the most commonly reported
J.W. GARGANO ET AL.
500
Table 2. Adjusted ORs for arriving at ED within therapeutically relevant time intervals (,2 hours or 2 hours - ,6 hours vs $6 hours/unknown) ,2 hours
Individual symptoms* Unilateral symptoms Speech/language Confusion Pain (other than headache) Loss of consciousness/syncope Fall Warning signs* Visual disturbance (WS 3) Trouble walking, balance, or dizziness (WS 4) Demographic and clinical characteristics Sex Female Male Race Black Other/not known White Age, yearsy 45 vs 65 55 vs 65 75 vs 65 85 vs 65 95 vs 65 Mode of arrival EMS (air or ambulance) Other Prestroke ambulatory status Independent Not independent Where stroke occurred At home At work Other/not documented Stroke subtype TIA HS IS
2 hours - ,6
aOR
(95% CI)
aOR
(95% CI)
1.5 1.6 0.5 0.3 0.5 0.4
(1.1-1.9) (1.2-2.2) (0.3-0.8) (0.1-1.0) (0.3-0.9) (0.1-0.9)
1.2 1.3 0.7 1.7 0.9 0.6
(0.9-1.5) (1.0-1.8) (0.4-1.0) (1.0-3.1) (0.5-1.3) (0.3-1.2)
0.4 0.7
(0.2-1.0) (0.4-1.0)
0.6 0.6
(0.3-1.1) (0.4-1.0)
0.9 1.0
(0.7-1.2)
0.8
(0.6-1.0)
0.9 1.3 1.0
(0.6-1.3) (0.8-2.0)
0.9 1.2 1.0
(0.6-1.2) (0.8-1.8)
0.7 0.9 0.9 0.7 0.4
(0.5-1.0) (0.8-1.0) (0.8-1.0) (0.5-0.9) (0.3-0.8)
0.9 1.0 1.0 1.0 1.0
(0.7-1.2) (0.9-1.1) (0.9-1.1) (0.8-1.3) (0.6-1.6)
3.6 1.0
(2.7-4.9)
1.9 1.0
(1.5-2.5)
2.2 1.0
(1.3-3.6)
1.3 1.0
(0.9-2.0)
1.0 2.9 0.9 2.7 1.7 1.0
(1.4-6.3) (0.7-1.2) (2.0-3.6) (1.1-2.6)
1.0 2.5 0.9 2.1 1.6 1.0
(1.2-5.2) (0.7-1.1) (1.6-2.8) (1.1-2.3)
*Reference category is symptom not mentioned. yAge was specified as a quadratic polynomial (ie, b1 3 age 1 b2 3 age2). To illustrate the magnitude of the age effect, contrasts versus age 65 were estimated from the model.
symptoms in surveys of acute stroke presentation.12,19,21-23 Unilateral symptoms and speech difficulties are central components of the National Stroke Association’s FAST public education message on stroke symptoms,23 and these symptoms also are more likely to be recognized as stroke WSs in surveys of the general public.24,25 For example, a survey of Cincinnati-area residents found that unilateral numbness and unilateral weakness were identified as WSs of stroke by 31% and 15% of respondents, respectively.24 The combination of high prevalence and widespread public awareness may help explain why the
presence of unilateral symptoms and speech difficulties were associated with earlier arrival in this study. Finally, it is important to note that these 2 symptoms are known to be some of the most useful history and physical examination information for clinically distinguishing between stroke cases and stroke mimics.26,27 For example, the presence of unilateral symptoms was found to be one of the strongest predictors of having a stroke in a series of 350 suspect stroke presentations.26 Stroke public health campaigns have focused on raising awareness of the cardinal stroke symptoms.4,5 The AHA’s
SYMPTOMS AND ONSET-TO-ARRIVAL TIMES
5 WSs, originally developed through a consensus process of experts, include 5 groupings that incorporate most of the recognized stroke WSs.5 Although we found 3 symptoms not included in these 5 WSs to be associated with earlier arrival (ie, nonheadache pain, loss of consciousness/syncope, and falls), none of these were very common; only 3.2%, 6.7%, and 3.4% of patients presented with these complaints, respectively. Other findings from our study indicate that the AHA’s WS of sudden confusion, trouble speaking, or understanding (ie, WS 2) includes symptoms that do not have the same relationship with time of arrival; confusion was found to be associated with increased delay, whereas speech/language difficulties were associated with earlier arrival. The AHA’s WSs have drawn some criticism, because they may not be easy for the public to recall or interpret,23 suggesting that other, simpler educational messages may be more effective in promoting optimal recognition of stroke symptoms and appropriate actions to take. Most previous studies of factors associated with onsetto-arrival time have excluded patients with uncertain time intervals.11,13,28 We included subjects with unknown time of arrival, because these patients were ineligible for acute interventions. Excluding these patients from the multivariate model affected the study’s statistical power, but did not change the ORs. A recent report from 4 Coverdell Registry sites in 2005-2006 indicates that missing onset-to-arrival time information continues to be a major problem, affecting 44.8% of records overall.28 Recommendations for improving and standardizing time interval documentation have been made previously.29 Potential limitations of this study include the fact that the symptom data were generated from the chief complaints recorded in the EDs by triage nurses, and thus might not represent the exact perceptions or understanding of the patients themselves. Second, although severity of stroke is known to be a major determinant of prehospital delay, this information was lacking in this study. The majority of registry subjects did not have a National Institutes of Health Stroke Scale (NIHSS) score recorded, and information necessary to retrospectively generate a score was not included in the registry data elements. We considered 2 variables as crude proxies for severity: stroke subtype, which was included in the model, and mRS score at discharge, which was not associated with onsetto-arrival time and was therefore not included in the model (data not shown). But because patients’ actions are based on their perception of initial symptoms, rather than on a clinically generated severity scale, we do not believe the lack of NIHSS data introduces any bias into our findings. The other findings from our study are generally consistent with previous reports. For example, earlier arrival was found to be strongly associated with use of EMS, being ambulatory before stroke onset, and having a stroke while at work (or during the day).30 The use of EMS remains
501
perhaps the single most important factor decreasing prehospital delay in patients with acute stroke. We found no association between race or sex and prehospital delay, which is consistent with most, although not, all previous reports.30,31 We also found that patients with a previous history of stroke did not arrive earlier to the hospital, which is consistent with other studies showing that individuals who have previously experienced stroke symptoms are at greater risk of delaying care.11,18 This demonstrates that, as was noted in a recent review of prehospital delay, knowledge of symptoms is not sufficient to motivate patients to seek care quickly.30 Future studies evaluating the importance of specific symptoms on prehospital delay should include more specific patient-level data that can address the social and behavioral aspects of recognition of stroke symptoms and necessary actions to take.6,17,22,32 Future studies also should focus on the importance of bystanders in the decision to seek care. Previous studies have demonstrated that the decision to seek care is often moderated by a coworker, friend, or bystander, and that the vast majority of stroke patients rely on bystanders to act for them.17,22,33 Future public health messages aimed at shortening prehospital delay also need to consider the role of psychosocial factors on care-seeking behavior. Acknowledgment: We thank all of the participating institutions and their staff who provided data for this study: Spectrum Health Systems, Grand Rapids; St. Joseph Mercy Hospital, Ann Arbor; University of Michigan Hospital, Ann Arbor; Borgess Medical Center, Kalamazoo; Sparrow Health Systems, Lansing; Ingham Regional Medical Center, Lansing; Detroit Receiving Hospital; Henry Ford Wyandotte Hospital; St. Joseph Mercy of Macomb; Northern Michigan Regional Health System, Petoskey; St. Mary’s Hospital, Saginaw; Bronson Methodist Hospital, Kalamazoo; Harper University Hospital, Detroit; Alpena General Hospital; and St. Joseph Health Systems, Tawas.
References 1. Reeves MJ, Arora S, Broderick JP, et al. Acute stroke care in the US: Results from 4 pilot prototypes of the Paul Coverdell National Acute Stroke Registry. Stroke 2005; 36:1232-1240. 2. O’Connor RE, McGraw P, Edelsohn L. Thrombolytic therapy for acute ischemic stroke: Why the majority of patients remain ineligible for treatment. Ann Emerg Med 1999;33:9-14. 3. Evenson KR, Rosamond WD, Morris DL. Prehospital and in-hospital delays in acute stroke care. Neuroepidemiology 2001;20:65-76. 4. Marler JR, Winters Jones P, Emr M, eds. Proceedings of a National Symposium on Rapid Identification and Treatment of Acute Stroke. Bethesda, MD: National Institute of Neurological Disorders and Stroke, 1997. 5. American Heart Association. Stroke warning signs and risk factors. Available at: http://www.americanheart. org/. Accessed August 20, 2009.
J.W. GARGANO ET AL.
502 6. Barr J, McKinley S, O’Brien E, et al. Patient recognition of and response to symptoms of TIA or stroke. Neuroepidemiology 2006;26:168-175. 7. Zerwic J, Hwang SY, Tucco L. Interpretation of symptoms and delay in seeking treatment by patients who have had a stroke: Exploratory study. Heart Lung 2007;36:25-34. 8. Iguchi Y, Wada K, Shibazaki K, et al. First impression at stroke onset plays an important role in early hospital arrival. Intern Med 2006;45:447-451. 9. Williams LS, Bruno A, Rouch D, et al. Stroke patients’ knowledge of stroke: Influence on time to presentation. Stroke 1997;28:912-915. 10. Casetta I, Granieri E, Gilli G, et al. Temporal trend and factors associated with delayed hospital admission of stroke patients. Neuroepidemiology 1999;18:255-264. 11. Rossnagel K, Jungehulsing GJ, Nolte CH, et al. Out-ofhospital delays in patients with acute stroke. Ann Emerg Med 2004;44:476-483. 12. Smith MA, Doliszny K, Shahar E, et al. Delayed hospital arrival for acute stroke: The Minnesota Stroke Survey. Ann Intern Med 1998;129:190-196. 13. Yu RF, San Jose MC, Manzanilla BM, et al. Sources and reasons for delays in the care of acute stroke patients. J Neurol Sci 2002;199:49-54. 14. Mullard AJ, Reeves MJ, Jacobs BS, et al. Lipid testing and lipid-lowering therapy in hospitalized ischemic stroke and transient ischemic attack patients: Results from a statewide stroke registry. Stroke 2006;37:44-49. 15. Gargano JW, Wehner S, Reeves MJ. Do presenting symptoms explain sex differences in emergency department delays among patients with acute stroke? Stroke 2009; 40:1114-1120. 16. Stokes ME, Davis CS, Koch GG. Categorical Data Analysis Using The SAS System. 2nd ed. Cary, NC: SAS Institute, 2003. 17. Mandelzweig L, Goldbourt U, Boyko V, et al. Preceptual, social, and behavioral factors associated with delays in seeking medical care in patients with symptoms of acute care. Stroke 2006;37:1248-1253. 18. Keskin O, Kalemoglu M, Ulusoy RE. A clinic investigation into prehospital and emergency department delays in acute stroke care. Med Princ Pract 2005;14:408-412. 19. Wester P, Radberg J, Lundgren B, et al. Factors associated with delayed admission to hospital and in-hospital delays in acute stroke and TIA: A prospective, multicenter
20.
21.
22.
23.
24.
25.
26.
27. 28.
29.
30.
31.
32.
33.
study. Seek-Medical-Attention-in-Time Study Group. Stroke 1999;30:40-48. Johnston SC, Fayad PB, Gorelick PB, et al. Prevalence and knowledge of transient ischemic attack among US adults. Neurology 2003;60:1429-1434. Kothari R, Sauerbeck L, Jauch E, et al. Patients’ awareness of stroke signs, symptoms, and risk factors. Stroke 1997; 28:1871-1875. Mosley I, Nicol M, Donnan G, et al. Stroke symptoms and the decision to call for an ambulance. Stroke 2007; 38:361-366. Kleindorfer DO, Miller R, Moomaw CJ, et al. Designing a message for public education regarding stroke: Does FAST capture enough stroke? Stroke 2007;38:2864-2868. Schneider AT, Pancioli AM, Khoury JC, et al. Trends in community knowledge of the warning signs and risk factors for stroke. JAMA 2003;289:343-346. Reeves MJ, Rafferty AP, Aranha AA, et al. Changes in knowledge of stroke risk factors and warning signs among Michigan adults. Cerebrovasc Dis 2008;25:385-391. Hand PJ, Kwan J, Lindley RI, et al. Distinguishing between stroke and mimic at the bedside: The Brain Attack Study. Stroke 2006;37:769-775. Goldstein LB, Simel DL. Is this patient having a stroke? JAMA 2005;293:2391-2402. Frankel M, Hinchey J, Schwamm L, et al. Prehospital and hospital delays after stroke onset—United States, 2005– 2006. MMWR Morb Mortal Wkly Rep 2007;56:474-478. Rosamond WD, Reeves MJ, Johnson A, et al. Documentation of stroke onset time: Challenges and recommendations. Am J Prev Med 2006;31:S230-S234. Moser DK, Alberts MJ, Kimble LP, et al. Reducing delay in seeking treatment by patients with acute coronary syndrome and stroke: A scientific statement from the American Heart Association Council on Cardiovascular Nursing and Stroke Council. Circulation 2006;114:168-182. Reeves MJ, Bushnell CD, Howard G, et al. Sex differences in stroke: Epidemiology, clinical presentation, medical care, and outcomes. Lancet Neurol 2008;7:915-926. Billings-Gagliardi S, Mazor KM. Development and validation of the stroke action test. Stroke 2005; 36:1035-1039. Carroll C, Hobart J, Fox C, et al. Stroke in Devon: Knowledge was good, but action was poor. J Neurol Neurosurg Psychiatry 2004;75:567-571.
Delayed arrival to the emergency department (ED) precludes most stroke patients from receiving thrombolytic treatment. Our objective in this study was to examine the association between presenting symptoms and onset-to-arrival time (ie, time between onset of symptoms to arrival at the ED) in a statewide stroke registry. Demographics, clinical data, and presenting symptoms were collected for patients with acute stroke or symptomatic transient ischemic attack (TIA) admitted to 15 Michigan hospitals (n 5 1922). Polytomous logistic regression models were developed to test the association between presenting symptoms and onset-to-arrival time (classified as ,2 hours, 2-6 hours, or .6 hours/unknown). Onset-to-arrival time was ,2 hours in 19% of the patients, 2-6 hours in 22%, and .6 hours/unknown in 59%. Unilateral symptoms (reported by 40%) and speech difficulties (reported by 22%) were associated with increased likelihood of arriving within 2 hours (unilateral: adjusted odds ratio [aOR], 1.5; 95% confidence interval [CI], 1.1-1.9; speech: aOR, 1.6; 95% CI, 1.2-2.2). Difficulty with walking, balance, or dizziness (12%), confusion (9%), loss of consciousness (6.7%) and falls (3.4%) were associated with lower likelihood of arriving within 2 hours (walking: aOR, 0.7; 95% CI, 0.4-1.0; confusion: aOR, 0.5; 95% CI, 0.3-0.8; consciousness: aOR, 0.5; 95% CI, 0.1-0.9; falls: aOR, 0.4; 95% CI, 0.3-0.9). Presenting symptoms were strongly associated with time of arrival; patients with unilateral symptoms and speech difficulties were more likely to seek care early. Future studies should consider including more specific patient-level data to identify psychosocial and behavioral aspects of recognition and action to stroke symptoms. Key Words: Signs and symptoms—emergency treatment—prehospital delay. Ó 2011 by National Stroke Association
Although intravenous (IV) tissue plasminogen activator (t-PA) is an effective treatment for ischemic stroke (IS), only a small minority of patients (3.0%-8.5%) receive it.1 The most important reason for patients not receiving t-PA treatment is late presentation to the emergency department (ED).2,3 Consequently, reducing prehospital From the *Department of Epidemiology, and †Department of Neurology and Ophthalmology, College of Human Medicine, Michigan State University, East Lansing, Michigan. Received September 3, 2009; accepted February 5, 2010. Supported by Centers for Disease Control and Prevention Cooperative Agreement No. U50/CCU520272-01. The authors report no conflicts of interest. Address correspondence to Mathew J. Reeves, PhD, Department of Epidemiology, Michigan State University, B601 West Fee Hall, East Lansing, MI 48824. E-mail: [email protected]. 1052-3057/$ - see front matter Ó 2011 by National Stroke Association doi:10.1016/j.jstrokecerebrovasdis.2010.02.022
494
delays for stroke patients has been an important public health goal.4,5 Arrival to the ED within a narrow time interval requires that individuals react to their stroke symptoms and seek care expeditiously. Some studies suggests that people delay seeking care because they wait for symptoms to resolve or do not recognize the symptoms as being important, particularly if the symptoms do not seem severe.6-9 In contrast, individuals whose first impression is that they are having a stroke seek care more quickly.6-8 Only a few epidemiologic studies have reported on the associations between specific symptoms and prehospital delay10-13; of these, the only study conducted in the United States reported on patients in a single metropolitan area in the early 1990s.12 More recent population-based information on the relationship between stroke symptoms and prehospital delay could inform public health messages aimed at educating individuals with stroke symptoms to seek
Journal of Stroke and Cerebrovascular Diseases, Vol. 20, No. 6 (November-December), 2011: pp 494-502
SYMPTOMS AND ONSET-TO-ARRIVAL TIMES
care early. The objective of this study was to examine the association between presenting symptoms and onset-toarrival time (ie, time between onset of symptoms to arrival at the ED) in patients admitted with acute stroke or symptomatic transient ischemic attack (TIA) in a statewide stroke registry.
Methods Study Design The Paul Coverdell National Acute Stroke Registry (PCNASR) was designed to monitor and improve the quality of acute stroke care in the United States.1 The Michigan Acute Stroke Care Overview and Treatment Surveillance System (MASCOTS) was a statewide, hospital-based, acute stroke registry prototype for the PCNASR. Details of the design of the MASCOTS registry have been published previously.1,14 In brief, a single-stage cluster design that used a modified stratified sampling regimen was implemented to obtain a representative statewide sample of 15 Michigan hospitals. Human subject approval was obtained from each hospital’s institutional review board.
Selection of Participants Acute stroke and TIA admissions in the ED at each participating hospital over a 6-month period in 2002 were identified by trained stroke clinical coordinators. Patients with TIA whose symptoms had resolved by the time of presentation to the ED were not included in the registry, because they would not have been considered for t-PA treatment. Thus, only patients with symptomatic TIA are included in this study. Of the 2566 registry subjects, those who did not initially present to the ED (281 direct admissions and 285 hospital transfers) were excluded, leaving 2000 primary ED admissions. Seventy-eight patients (4%) were excluded due to missing time interval data, leaving 1922 subjects for this analysis.
Data Collection and Processing Information on demographics, prehospital care, prestroke ambulatory status, modified Rankin Scale (mRS) score at discharge, and past medical history was abstracted from patient charts using a standardized data collection instrument. Cases were classified as IS (including IS of uncertain duration), hemorrhagic stroke (HS, including intracerebral and subarachnoid hemorrhage), or TIA. Data on time of symptom onset and ED triage were abstracted from the medical charts. The exact date and time of symptom onset were noted when available; if unknown, symptom onset was estimated within a 6-hour interval when possible, or as the date and time the patient was last seen normal.1 Data on presenting symptoms, as recorded as the chief complaint during the initial ED triage, were collected as
495 15
free text. As described previously, a computer program was used to identify patients considered suspected stroke cases on arrival, and to identify specific individual presenting symptoms. Individual symptoms were subsequently aggregated into 5 stroke warning signs (WSs) consistent with the American Heart Association’s (AHA) public health messages:5 numbness/weakness (WS 1), confusion/speech disturbance (WS 2), visual disturbance (WS 3), trouble walking/dizziness/poor balance (WS 4), and headache (WS 5). Other symptoms not included in these 5 WSs (eg, nonheadache pain, loss of consciousness/syncope, falls) were reported individually.
Outcome Measures For subjects with an exact stroke onset time recorded, the onset-to-arrival time was calculated by subtracting the onset date and time from the ED arrival date and time. For subjects with only an estimated time of symptom onset, the beginning of the estimated 6-hour window or the time last seen normal was used to determine the most conservative estimate of onset-to-arrival time. Time intervals were initially categorized as ,2 hours (n 5 357; 18.6%), 2-3 hours (n 5 128; 6.7%), 3-6 hours (n 5 295; 15.3%), $6 hours (n 5 443; 23.0%), and unknown (n 5 699; 36.4%). These categories were aggregated into 3 onset-to-arrival time groups based on eligibility for tPA-related treatments: (1) optimally eligible for IV t-PA (arrived ,2 hours), (2) potentially eligible for IV or intra-arterial t-PA (arrived 2-6 hours), and (3) ineligible for t-PA (arrived $6 hours or unknown).
Primary Data Analysis The demographic and clinical features, suspect stroke, stroke WSs, and self-reported symptoms among the 3 onset-to-arrival groups were compared using the c2 test. The association between onset-to-arrival group and discharge mRS score (0-6) was assessed using the CochranMantel-Haenszel test, with modified Ridit scores to account for ordered categories.16 Patient characteristics independently associated with the onset-to-arrival time groups were identified using multivariate polytomous logistic regression models, using t-PA–ineligible patients (ie, those with arrival time $6 hours or unknown) as the reference group. Age, sex, race, and use of emergency medical services (EMS) were included as a priori variables of interest. Nursing home residence, place (ie, home, work, other) and timing (ie, weekday, weekend, unknown) of stroke onset, prestroke ambulatory status, stroke subtype, mRS score at discharge, and previous medical history (ie, stroke/TIA, heart disease, congestive heart failure, atrial fibrillation, dyslipidemia, and current smoking) also were considered for inclusion (backward-elimination variable selection, with global P ,.05 to stay). Next, presenting symptoms (ie, unilateral symptoms, speech/language difficulty, confusion,
496
walking/dizziness/imbalance, visual disturbance, headache, nonheadache pain, falling, loss of consciousness, and suspected stroke) were added, and were retained if the global P value was ,.10. An additional variable indicating the absence of all WSs or suspected stroke was tested as well. After the final main-effects model was defined, the interaction terms between EMS arrival and symptoms were tested. Model results are presented as adjusted odds ratio (aOR) and 95% confidence interval (CI) for each time interval compared with the t-PA–ineligible group. To illustrate the effect of the polynomial specification of age (ie, age and age2), ORs for 10-year contrasts with age 65 were estimated from the regression model. Finally, the characteristics and influence of subjects with unknown onset-to-arrival time were explored. First, subjects with unknown onset-to-arrival time were compared with those known to have arrived .6 hours after onset using the c2 test. A sensitivity analysis was then performed by excluding subjects with unknown onset-to-arrival time from the final model. The time of day of stroke onset (ie, morning, afternoon, evening, overnight) was explored in this latter model. A sensitivity analysis also was performed by rerunning the final model after excluding all TIA cases.
Results Demographic and clinical characteristics of the 1922 subjects included in this analysis are presented in Table 1. Overall, onset to arrival time was ,2 hours in 18.6% of the subjects, 2-6 hours in 22%, and $6 hours or unknown in 59.4%. Distributions of time intervals by patient demographic and clinical characteristics are presented in Figure 1. There were no statistically significant (P , .05) associations between onset-to-arrival time and sex, race, age, nursing home residence, or insurance status (Fig 1A). Although the difference was not not statistically significant, the oldest and youngest subjects tended to be less likely than those in the middle of the age distribution to arrive early. Subjects with TIA or HS were more likely to arrive within 2 hours of onset than those with IS (Fig 1B), and subjects arriving by EMS were twice as likely to arrive within 2 hours of onset as those who did not (Fig 1C). Having a stroke at work or a stroke occurring between noon and midnight also was associated with earlier arrival (Fig 1C). Distributions of time intervals by suspect stroke, the 5 stroke WSs, and specific individual symptoms are presented in Figure 2. Of the 5 WSs, 3 were significantly associated with onset-to-arrival time in unadjusted analyses (Fig 2A); numbness/weakness (WS 1) was associated with early arrival, whereas visual disturbance (WS 3) and trouble walking/dizziness/balance (WS 4) were associated with later arrival. Neither suspected stroke nor the absence of any WSs were associated with onset-to-arrival time, however (Fig 2A).
J.W. GARGANO ET AL.
The individual symptoms of weakness and facial droop were both significantly associated with earlier arrival, as was the presence of any unilateral symptoms (Fig 2B). Although confusion/speech (WS 2) was unrelated to onset-to-arrival time (Fig 2A), its two individual components were both significantly associated with onset-to-arrival time, but in opposite directions; subjects reporting speech/language difficulties were more likely to arrive early, whereas those reporting confusion were more likely to arrive later (Fig 2B). Dizziness and nonheadache pain were both significantly associated with delayed arrival (Fig 2C). Trouble walking, balance/coordination problems, and falls (Fig 2C) all showed strong trends toward delayed arrival, but the smaller number of subjects reporting these symptoms limited the statistical power of these comparisons. Interestingly, loss of consciousness/syncope was not statistically significantly associated with time of arrival in the unadjusted analysis. Multivariate model results, presented in Table 2, show that many presenting symptoms were significantly associated with onset-to-arrival time intervals. The presence of unilateral symptoms and speech/language difficulties were associated with increased likelihood (OR, 1.5 and 1.6, respectively) of arriving within 2 hours, whereas confusion, nonheadache pain, loss of consciousness/syncope, and falls were all associated with significantly reduced likelihood of early arrival (ORs, 0.3-0.7). The WSs of visual disturbance (WS 3) and difficulty with walking/balance/dizziness (WS 4) also were associated with delayed arrival. After accounting for these presenting symptoms, significant relationships between patient characteristics and onset-to-arrival time remained. There was a quadratic relation between age and early arrival (ie, age2 was significant), with the youngest and oldest subjects having less likelihood of arriving ,2 hours after symptom onset compared with the subjects aged 65 years. Arrival by EMS, being ambulatory, and experiencing stroke while at work were all associated with earlier arrival. Compared with the subjects with IS, those with TIA or HS were both more likely to arrive earlier (ie, ,2 or 2-6 hours after symptom onset). No significant interactions between arrival by EMS and any other variables were seen. Compared with subjects with an unknown onset-toarrival time, subjects known to have arrived $6 hours after symptom onset were more likely to be black (22% vs 18%), less likely to be age $80 years (18% vs 22%), less likely to report confusion (8% vs 12%), and more likely to report speech difficulty (23% vs 17%). However, there were no significant differences between these 2 groups by sex, EMS arrival, stroke subtype, or any other symptoms or WSs. After excluding the 699 subjects with unknown onset time from the adjusted model, ORs were similar, but there were fewer statistically significant findings attributable to diminished statistical power.
SYMPTOMS AND ONSET-TO-ARRIVAL TIMES
497
Table 1. (Continued )
Table 1. Distribution of demographic and clinical characteristics in the study subjects (n 5 1922)
Total Sex Female Male Race Black White Other/not known Age, years ,50 50-59 60-69 70-79 $80 Nursing home resident Yes No Insurance type Public only Some/all private None Previous stroke Yes No Previous myocardial infarction/coronary heart disease Yes No Atrial fibrillation Yes No Prestroke ambulatory status Not independent Independent Stroke subtype TIA IS HS Onset day* Weekday Weekend Unknown Onset timey Morning Afternoon Evening Overnight Unknown Onset place Home Work Other EMS arrival Yes No
n
%
1922
100.0
1041 881
54.2 45.8
340 1414 168
17.7 73.6 8.7
203 269 328 559 563
10.6 14.0 17.1 29.1 29.3
85 1837
4.4 95.6
380 1462 80
19.8 76.1 4.2
731 1191
38.0 62.0
668 1254
34.8 65.2
290 1632
15.1 84.9
167 1755
8.7 91.3
428 1293 201
22.3 67.3 10.5
1280 506 136
66.6 26.3 7.1
357 128 295 443 699
29.2 10.5 24.1 36.2 36.4
1145 48 729
59.6 2.5 37.9
806 41.9 1116 58.1 (Continued )
Onset-to-arrival time ,2 hours 2-6 hours $6 hours/unknown
n
%
358 423 1142
18.6 22.0 59.4
*Weekday: Monday-Friday; weekend: Saturday-Sunday. yMorning: 6 am-noon; afternoon: noon-5:59 pm; evening: 6:00 pm-midnight; overnight: midnight-5:59 am.
When time of day of symptom onset was added to this restricted model, it was strongly associated with onset-to-arrival time (P , .0001); compared with subjects with stroke occurring in the morning (ie, 6 am to noon), those with stroke occurring in the afternoon or evening had an increased likelihood of arriving ,2 hours after onset (afternoon: OR, 1.6; evening: OR, 1.7), whereas those with stroke occurring overnight had less likelhood of arriving ,2 hours after onset (OR, 0.4). Adding time of day did not affect any associations with symptoms or other variables, except that the effect of onset at work was attenuated and lost statistical significance (P 5 .44). Excluding subjects with TIA from the full model also resulted in loss of statistical significance for some symptoms, but the magnitude of the ORs remained similar.
Discussion In this analysis of patients admitted with acute stroke in a statewide hospital-based registry, presenting symptoms were strongly associated with onset-to-arrival time. Unilateral symptoms and speech difficulties were linked with greater likelihood of early arrival (ie, ,2 hours after onset), whereas difficulty with walking/balance/dizziness, vision problems, and confusion were associated with lower likelihood of early arrival. Three symptoms not incorporated in the AHA’s 5 WSs—nonheadache pain, loss of consciousness/syncope, and falls—also were associated with lower likelihood of arriving within 2 hours of symptom onset. Only a few studies have empirically examined associations between specific symptoms and prehospital delay,10-13 and only one of these studies is from the United States.12 This US report was based on data generated from a population-based surveillance study of acute stroke in the Minneapolis–St. Paul area in 1991-1993, in which detailed symptom data were abstracted from patient charts.12 In contrast to our findings, that study found that the presence of syncope and seizures at stroke onset was associated with earlier arrival. However, the study also found that unsteadiness and impaired vision were associated with greater delay, which is concordant with our results. Rossnagel et al11 prospectively interviewed 558 patients with stroke and TIA in 4 Berlin hospitals about their symptoms and found that unilateral numbness was associated with
498
J.W. GARGANO ET AL.
Figure 1. Proportion of subjects arriving at ,2 hours (black) and 2 hours - ,6 hours (white) after symptom onset, by (A) demographics, (B) previous medical history and current stroke subtype, and (C) stroke onset time and place and EMS use. *P ,.05, c2 test across all categories. †Weekday, Monday-Friday; weekend: Saturday-Sunday. ‡Morning: 6 am-noon; afternoon: noon- 5:59 pm; evening: 6:00 pm -midnight; overnight: midnight-5:59 am. Total subjects 5 1922, except xc2 test restricted to subjects with known onset time.
shorter delays, whereas walking disturbance and faintness/weakness were associated with longer delays. These findings also are consistent with our results. More generally, previous studies on stroke-related symptoms and prehospital delay have found that patients who initially perceived their symptoms as not serious,6,7,9,11,13 or did not ascribe their symptoms to stroke,6,8,13,17 tended to delay seeking care. Unfortunately, we were not able to collect data on our subjects’ perceptions or interpretation of their symptoms in the present study. Not surprisingly, previous studies have found that that earlier arrival is associated with more severe stroke9,10,12,18
and with specific features that suggest severe stroke, such as sudden onset of symptoms,10,19 altered level of consciousness,8,12,13 and severe motor impairment.7,10 Consistent with our findings in the present study, previous studies also have found that patients with HS— which typically presents with more severe, fulminant symptoms—tend to arrive earlier than those with IS.12,13,19 We were surprised to find that the patients with TIA were more likely to arrive within 2 hours compared with those with IS, given previous reports indicating that TIA patients often delay seeking medical care.20 However, at least 2 clinical studies have found that
SYMPTOMS AND ONSET-TO-ARRIVAL TIMES
499
Figure 2. Proportion of subjects arriving at ,2 hours (black) and 2 hours - ,6 hours (white) after symptom onset, by (A) stroke warning signs and suspect stroke, (B) symptom components of WSs 1 and 2, and (C) symptom components of WS 4 and other symptoms. *P , .05, c2 test across all categories. †Number (%) of subjects with each symptom. Total subjects 5 1922.
patients with TIA arrived earlier to the hospital than patients with IS.11,19 We suspect that this finding in our study subjects reflects the inclusion criteria used by the PCNASR, which limited TIA cases to those patients who still exhibited symptoms on presentation to the hospital.1 Thus, TIA patients with shorter duration of symptoms would be less likely to be included in our registry data. The registry’s exclusion of nonsymptomatic TIA patients was based on the fact that these patients would not have been considered for t-PA treatment. Consequently, we emphasize that no specific conclusions
can be drawn from our findings regarding subjects’ responses to symptoms of TIA that resolved before presentation. We note that the final model results were similar after we excluded all patients with TIA, and thus we do not believe that their inclusion is a source of bias. There are several important observations to be made regarding the fact that unilateral symptoms and speech/ language difficulties were the 2 symptoms most strongly associated with early arrival. Unilateral symptoms— including facial droop and leg or arm weakness—and speech difficulties are the most commonly reported
J.W. GARGANO ET AL.
500
Table 2. Adjusted ORs for arriving at ED within therapeutically relevant time intervals (,2 hours or 2 hours - ,6 hours vs $6 hours/unknown) ,2 hours
Individual symptoms* Unilateral symptoms Speech/language Confusion Pain (other than headache) Loss of consciousness/syncope Fall Warning signs* Visual disturbance (WS 3) Trouble walking, balance, or dizziness (WS 4) Demographic and clinical characteristics Sex Female Male Race Black Other/not known White Age, yearsy 45 vs 65 55 vs 65 75 vs 65 85 vs 65 95 vs 65 Mode of arrival EMS (air or ambulance) Other Prestroke ambulatory status Independent Not independent Where stroke occurred At home At work Other/not documented Stroke subtype TIA HS IS
2 hours - ,6
aOR
(95% CI)
aOR
(95% CI)
1.5 1.6 0.5 0.3 0.5 0.4
(1.1-1.9) (1.2-2.2) (0.3-0.8) (0.1-1.0) (0.3-0.9) (0.1-0.9)
1.2 1.3 0.7 1.7 0.9 0.6
(0.9-1.5) (1.0-1.8) (0.4-1.0) (1.0-3.1) (0.5-1.3) (0.3-1.2)
0.4 0.7
(0.2-1.0) (0.4-1.0)
0.6 0.6
(0.3-1.1) (0.4-1.0)
0.9 1.0
(0.7-1.2)
0.8
(0.6-1.0)
0.9 1.3 1.0
(0.6-1.3) (0.8-2.0)
0.9 1.2 1.0
(0.6-1.2) (0.8-1.8)
0.7 0.9 0.9 0.7 0.4
(0.5-1.0) (0.8-1.0) (0.8-1.0) (0.5-0.9) (0.3-0.8)
0.9 1.0 1.0 1.0 1.0
(0.7-1.2) (0.9-1.1) (0.9-1.1) (0.8-1.3) (0.6-1.6)
3.6 1.0
(2.7-4.9)
1.9 1.0
(1.5-2.5)
2.2 1.0
(1.3-3.6)
1.3 1.0
(0.9-2.0)
1.0 2.9 0.9 2.7 1.7 1.0
(1.4-6.3) (0.7-1.2) (2.0-3.6) (1.1-2.6)
1.0 2.5 0.9 2.1 1.6 1.0
(1.2-5.2) (0.7-1.1) (1.6-2.8) (1.1-2.3)
*Reference category is symptom not mentioned. yAge was specified as a quadratic polynomial (ie, b1 3 age 1 b2 3 age2). To illustrate the magnitude of the age effect, contrasts versus age 65 were estimated from the model.
symptoms in surveys of acute stroke presentation.12,19,21-23 Unilateral symptoms and speech difficulties are central components of the National Stroke Association’s FAST public education message on stroke symptoms,23 and these symptoms also are more likely to be recognized as stroke WSs in surveys of the general public.24,25 For example, a survey of Cincinnati-area residents found that unilateral numbness and unilateral weakness were identified as WSs of stroke by 31% and 15% of respondents, respectively.24 The combination of high prevalence and widespread public awareness may help explain why the
presence of unilateral symptoms and speech difficulties were associated with earlier arrival in this study. Finally, it is important to note that these 2 symptoms are known to be some of the most useful history and physical examination information for clinically distinguishing between stroke cases and stroke mimics.26,27 For example, the presence of unilateral symptoms was found to be one of the strongest predictors of having a stroke in a series of 350 suspect stroke presentations.26 Stroke public health campaigns have focused on raising awareness of the cardinal stroke symptoms.4,5 The AHA’s
SYMPTOMS AND ONSET-TO-ARRIVAL TIMES
5 WSs, originally developed through a consensus process of experts, include 5 groupings that incorporate most of the recognized stroke WSs.5 Although we found 3 symptoms not included in these 5 WSs to be associated with earlier arrival (ie, nonheadache pain, loss of consciousness/syncope, and falls), none of these were very common; only 3.2%, 6.7%, and 3.4% of patients presented with these complaints, respectively. Other findings from our study indicate that the AHA’s WS of sudden confusion, trouble speaking, or understanding (ie, WS 2) includes symptoms that do not have the same relationship with time of arrival; confusion was found to be associated with increased delay, whereas speech/language difficulties were associated with earlier arrival. The AHA’s WSs have drawn some criticism, because they may not be easy for the public to recall or interpret,23 suggesting that other, simpler educational messages may be more effective in promoting optimal recognition of stroke symptoms and appropriate actions to take. Most previous studies of factors associated with onsetto-arrival time have excluded patients with uncertain time intervals.11,13,28 We included subjects with unknown time of arrival, because these patients were ineligible for acute interventions. Excluding these patients from the multivariate model affected the study’s statistical power, but did not change the ORs. A recent report from 4 Coverdell Registry sites in 2005-2006 indicates that missing onset-to-arrival time information continues to be a major problem, affecting 44.8% of records overall.28 Recommendations for improving and standardizing time interval documentation have been made previously.29 Potential limitations of this study include the fact that the symptom data were generated from the chief complaints recorded in the EDs by triage nurses, and thus might not represent the exact perceptions or understanding of the patients themselves. Second, although severity of stroke is known to be a major determinant of prehospital delay, this information was lacking in this study. The majority of registry subjects did not have a National Institutes of Health Stroke Scale (NIHSS) score recorded, and information necessary to retrospectively generate a score was not included in the registry data elements. We considered 2 variables as crude proxies for severity: stroke subtype, which was included in the model, and mRS score at discharge, which was not associated with onsetto-arrival time and was therefore not included in the model (data not shown). But because patients’ actions are based on their perception of initial symptoms, rather than on a clinically generated severity scale, we do not believe the lack of NIHSS data introduces any bias into our findings. The other findings from our study are generally consistent with previous reports. For example, earlier arrival was found to be strongly associated with use of EMS, being ambulatory before stroke onset, and having a stroke while at work (or during the day).30 The use of EMS remains
501
perhaps the single most important factor decreasing prehospital delay in patients with acute stroke. We found no association between race or sex and prehospital delay, which is consistent with most, although not, all previous reports.30,31 We also found that patients with a previous history of stroke did not arrive earlier to the hospital, which is consistent with other studies showing that individuals who have previously experienced stroke symptoms are at greater risk of delaying care.11,18 This demonstrates that, as was noted in a recent review of prehospital delay, knowledge of symptoms is not sufficient to motivate patients to seek care quickly.30 Future studies evaluating the importance of specific symptoms on prehospital delay should include more specific patient-level data that can address the social and behavioral aspects of recognition of stroke symptoms and necessary actions to take.6,17,22,32 Future studies also should focus on the importance of bystanders in the decision to seek care. Previous studies have demonstrated that the decision to seek care is often moderated by a coworker, friend, or bystander, and that the vast majority of stroke patients rely on bystanders to act for them.17,22,33 Future public health messages aimed at shortening prehospital delay also need to consider the role of psychosocial factors on care-seeking behavior. Acknowledgment: We thank all of the participating institutions and their staff who provided data for this study: Spectrum Health Systems, Grand Rapids; St. Joseph Mercy Hospital, Ann Arbor; University of Michigan Hospital, Ann Arbor; Borgess Medical Center, Kalamazoo; Sparrow Health Systems, Lansing; Ingham Regional Medical Center, Lansing; Detroit Receiving Hospital; Henry Ford Wyandotte Hospital; St. Joseph Mercy of Macomb; Northern Michigan Regional Health System, Petoskey; St. Mary’s Hospital, Saginaw; Bronson Methodist Hospital, Kalamazoo; Harper University Hospital, Detroit; Alpena General Hospital; and St. Joseph Health Systems, Tawas.
References 1. Reeves MJ, Arora S, Broderick JP, et al. Acute stroke care in the US: Results from 4 pilot prototypes of the Paul Coverdell National Acute Stroke Registry. Stroke 2005; 36:1232-1240. 2. O’Connor RE, McGraw P, Edelsohn L. Thrombolytic therapy for acute ischemic stroke: Why the majority of patients remain ineligible for treatment. Ann Emerg Med 1999;33:9-14. 3. Evenson KR, Rosamond WD, Morris DL. Prehospital and in-hospital delays in acute stroke care. Neuroepidemiology 2001;20:65-76. 4. Marler JR, Winters Jones P, Emr M, eds. Proceedings of a National Symposium on Rapid Identification and Treatment of Acute Stroke. Bethesda, MD: National Institute of Neurological Disorders and Stroke, 1997. 5. American Heart Association. Stroke warning signs and risk factors. Available at: http://www.americanheart. org/. Accessed August 20, 2009.
J.W. GARGANO ET AL.
502 6. Barr J, McKinley S, O’Brien E, et al. Patient recognition of and response to symptoms of TIA or stroke. Neuroepidemiology 2006;26:168-175. 7. Zerwic J, Hwang SY, Tucco L. Interpretation of symptoms and delay in seeking treatment by patients who have had a stroke: Exploratory study. Heart Lung 2007;36:25-34. 8. Iguchi Y, Wada K, Shibazaki K, et al. First impression at stroke onset plays an important role in early hospital arrival. Intern Med 2006;45:447-451. 9. Williams LS, Bruno A, Rouch D, et al. Stroke patients’ knowledge of stroke: Influence on time to presentation. Stroke 1997;28:912-915. 10. Casetta I, Granieri E, Gilli G, et al. Temporal trend and factors associated with delayed hospital admission of stroke patients. Neuroepidemiology 1999;18:255-264. 11. Rossnagel K, Jungehulsing GJ, Nolte CH, et al. Out-ofhospital delays in patients with acute stroke. Ann Emerg Med 2004;44:476-483. 12. Smith MA, Doliszny K, Shahar E, et al. Delayed hospital arrival for acute stroke: The Minnesota Stroke Survey. Ann Intern Med 1998;129:190-196. 13. Yu RF, San Jose MC, Manzanilla BM, et al. Sources and reasons for delays in the care of acute stroke patients. J Neurol Sci 2002;199:49-54. 14. Mullard AJ, Reeves MJ, Jacobs BS, et al. Lipid testing and lipid-lowering therapy in hospitalized ischemic stroke and transient ischemic attack patients: Results from a statewide stroke registry. Stroke 2006;37:44-49. 15. Gargano JW, Wehner S, Reeves MJ. Do presenting symptoms explain sex differences in emergency department delays among patients with acute stroke? Stroke 2009; 40:1114-1120. 16. Stokes ME, Davis CS, Koch GG. Categorical Data Analysis Using The SAS System. 2nd ed. Cary, NC: SAS Institute, 2003. 17. Mandelzweig L, Goldbourt U, Boyko V, et al. Preceptual, social, and behavioral factors associated with delays in seeking medical care in patients with symptoms of acute care. Stroke 2006;37:1248-1253. 18. Keskin O, Kalemoglu M, Ulusoy RE. A clinic investigation into prehospital and emergency department delays in acute stroke care. Med Princ Pract 2005;14:408-412. 19. Wester P, Radberg J, Lundgren B, et al. Factors associated with delayed admission to hospital and in-hospital delays in acute stroke and TIA: A prospective, multicenter
20.
21.
22.
23.
24.
25.
26.
27. 28.
29.
30.
31.
32.
33.
study. Seek-Medical-Attention-in-Time Study Group. Stroke 1999;30:40-48. Johnston SC, Fayad PB, Gorelick PB, et al. Prevalence and knowledge of transient ischemic attack among US adults. Neurology 2003;60:1429-1434. Kothari R, Sauerbeck L, Jauch E, et al. Patients’ awareness of stroke signs, symptoms, and risk factors. Stroke 1997; 28:1871-1875. Mosley I, Nicol M, Donnan G, et al. Stroke symptoms and the decision to call for an ambulance. Stroke 2007; 38:361-366. Kleindorfer DO, Miller R, Moomaw CJ, et al. Designing a message for public education regarding stroke: Does FAST capture enough stroke? Stroke 2007;38:2864-2868. Schneider AT, Pancioli AM, Khoury JC, et al. Trends in community knowledge of the warning signs and risk factors for stroke. JAMA 2003;289:343-346. Reeves MJ, Rafferty AP, Aranha AA, et al. Changes in knowledge of stroke risk factors and warning signs among Michigan adults. Cerebrovasc Dis 2008;25:385-391. Hand PJ, Kwan J, Lindley RI, et al. Distinguishing between stroke and mimic at the bedside: The Brain Attack Study. Stroke 2006;37:769-775. Goldstein LB, Simel DL. Is this patient having a stroke? JAMA 2005;293:2391-2402. Frankel M, Hinchey J, Schwamm L, et al. Prehospital and hospital delays after stroke onset—United States, 2005– 2006. MMWR Morb Mortal Wkly Rep 2007;56:474-478. Rosamond WD, Reeves MJ, Johnson A, et al. Documentation of stroke onset time: Challenges and recommendations. Am J Prev Med 2006;31:S230-S234. Moser DK, Alberts MJ, Kimble LP, et al. Reducing delay in seeking treatment by patients with acute coronary syndrome and stroke: A scientific statement from the American Heart Association Council on Cardiovascular Nursing and Stroke Council. Circulation 2006;114:168-182. Reeves MJ, Bushnell CD, Howard G, et al. Sex differences in stroke: Epidemiology, clinical presentation, medical care, and outcomes. Lancet Neurol 2008;7:915-926. Billings-Gagliardi S, Mazor KM. Development and validation of the stroke action test. Stroke 2005; 36:1035-1039. Carroll C, Hobart J, Fox C, et al. Stroke in Devon: Knowledge was good, but action was poor. J Neurol Neurosurg Psychiatry 2004;75:567-571.