|Home | About | Journals | Submit | Contact Us | Français|
Few patients arrive early enough at hospitals to be eligible for emergent stroke treatment. There may be barriers specific to underserved, urban populations that need to be identified before effective educational interventions to reduce delay times can be developed.
A survey of respondents’ likely action in a hypothetical stroke situation was given to 253 community volunteers in the catchment areas of a large urban community hospital. Concurrently, 100 structured interviews were conducted in the same hospital with acute stroke patients or proxy.
In this predominantly urban, black population, if faced with a hypothetical stroke, 89% of community volunteers surveyed said they would call 911 first, and few felt any of the suggested potential barriers applied to them. However, only 12% of stroke patients interviewed actually called 911 first (OR 63.9; 95% CI 29.5 to 138.2). Instead, 75% called a relative/friend. Eighty-nine percent of stroke patients reported significant delay in seeking medical attention, and almost half said the reason for delay was thinking the symptoms were not serious and/or they would self-resolve. For those arriving by ambulance, only 25% did so because they thought it would be faster, while 35% cited having no other transportation options.
In this predominantly black urban population, while 89% of community volunteers report the intent of calling 911 during a stroke only 12% of actual stroke patients did so. Further research is needed to determine and conquer the barriers between behavioral intent and actual behavior to call 911 for witnessed stroke.
Stroke has a disproportionate impact on blacks compared to whites reflected in higher incidence and mortality rates.1-5 Time from stroke onset to hospital presentation is a critical factor in acute stroke care because eligibility for thrombolytic therapy requires early arrival. Some studies have found significant racial differences in this delay time6, 7 which likely contributes to the differences in tPA treatment rates by race that have been demonstrated in some studies.8, 9
Previous work has identified factors that may contribute to delay in seeking stroke treatment and has reinforced the complexity of this decision-making process. These include cognitive factors such as considering the symptoms as not being serious,10, 11 and social factors such as calling a relative or physician instead of calling 911.12 Additional racial and cultural factors including embarrassment and perceived control/self-efficacy have not been fully explored.12
Understanding why people make certain decisions during an acute event is crucial for developing effective education interventions to reduce time to hospital arrival. This study surveyed a primarily medically underserved black urban population about stroke knowledge and beliefs that may be barriers to seeking urgent medical attention. Healthy volunteers and hospitalized acute stroke patients or proxies in the same community were interviewed concurrently to compare behavioral intent versus actual behavior and gain insight into reasons for delay. We hypothesized that there are barriers to seeking acute stroke care that are specific to an urban underserved population. These may include 1) informational barriers (knowledge of treatment and recognition of warning signs), 2) attitudinal barriers (mistrust of healthcare system and providers), and 3) institutional barriers (preparedness for acute treatment in a high risk population).
A convenience sample of 20 community-based sites in the high-stroke-risk catchment area of a large urban community hospital, Washington Hospital Center in Washington, D.C., were identified between 2007-2008 for administration of a stroke knowledge and hypothetical action survey. These sites included churches, senior citizen centers, men’s groups, and community health fairs. This survey was modified from one developed by the TLL Temple Project to assess community knowledge prior to educational interventions in East Texas.13 The community survey was administered by trained members of the research team who performed face-to-face interviews with community volunteers. It consisted of open-ended and multiple option questions regarding stroke symptom identification and 5-point Likert scale ratings for attitudinal and self-efficacy statements (Figure S1, please see http://stroke.ahajournals.org). Respondents were also presented with a hypothetical situation of a relative experiencing a possible stroke and asked how likely they would be to take certain actions. This design has been found reliable in studies of health interventions with low literacy adults.14
Concurrently, 100 structured interviews were conducted by a single interviewer with acute stroke patients in the same hospital (Washington Hospital Center) within 48 hours of admission, and consisted of socioeconomic items, initial impression, first action, transport mode,and open-ended questions regarding reasons for transport mode and delay to hospital arrival (Figure S2, please see http://stroke/ahajournals.org). Reasons for delay were further explored with multiple option questions. To include mild and severe strokes, if the patient could not be interviewed because of aphasia, severe dysarthria, or altered mental status, then a relative or friend (proxy) who was with the patient at the time of presentation was interviewed.
Participant characteristics were summarized as percents. For variables collected in both surveys we compared frequencies using chi-square and Fisher’s exact test. Among community participants, stroke symptom recognition from open and multiple option questions were summarized as percents. Among stroke patients, EMS use was summarized as percents and differences in EMS use by patient characteristics were tested using chi-square tests.
Responses regarding actions in the acute stroke setting were compared between the two groups (i.e. hypothetical actions among healthy volunteers and actual actions among patients/proxies). Comparisons were initially done by calculating odds ratios treating volunteer/patient as the outcome and action as predictor (i.e. a case-control analysis). Since the volunteers and patients differed on education and insurance, we next computed adjusted odds ratios controlling for these factors. Because education and insurance were strongly associated, and education was not associated with being a case-control status after adjusting for insurance, our final model included only insurance as a covariate. Health insurance was dichotomized as “private” = private/HMO/PPO and “non-private” = Medicare/Medicaid/DC Healthcare Alliance/none.
Responses for barriers to calling 911 were summarized as frequencies for community volunteers and stroke patients separately. For stroke patients, a bar chart depicting barrier frequency was constructed (Figure 1). Each bar was dichotomized to represent the frequency of actual EMS use among the patients who said the barrier applied to them. Among those arriving by EMS, reasons for use were summarized as percents.
All analyses were performed in SAS version 9.1.3.
This study was approved by the local IRB. Written informed consent was obtained from the community volunteers and a $5 gift card was provided after the survey. Waiver of written informed consent was obtained for the patient/proxy interview.
A total of 253 community volunteers from Washington, D.C. participated in the survey [Table 1]. Two thirds recalled being exposed to stroke information in the previous year with television news, newspapers/magazines, and doctors cited as common sources. Ten percent of respondents had had a stroke themselves, 49% had a relative with stroke, and only 14% had no experience with stroke.
Despite the exposure to stroke information reported by the community volunteers from both public education and personal experience, on open-ended questioning, only 15-48% of respondents were able to cite the classic stroke symptoms (“sudden trouble speaking” and “numbness or weakness of face” were most frequently cited: 36% and 48% of respondents, respectively). Stroke symptom recognition increased to 62-82% of respondents with multiple option questions. However, 69% of respondents also incorrectly identified chest pain as a stroke symptom based on their answers to the open-ended and multiple option questions (18% cited chest pain as a symptom in the open-ended question, 54% on the multiple option question, and 69% on either or both questions).
There were a total of 100 respondents for the patient/proxy interview [Table 1]: 82 were patients themselves, 13 were a combination of patient and a relative/friend, and 5 were a relative/friend only. Half of the patients failed to utilize EMS and instead reached the hospital by car or public transportation. There was no difference in patients’ use of EMS for those whose interviews were completed by a proxy or a combination of patient and proxy compared to the patients themselves. There was no significant association between level of education and use of EMS. In fact, the trend favored more frequent use of EMS by those with less education (56.9% of those with a high school diploma or less used EMS vs. 38.1% of those with more than a high school diploma, p=0.063), a finding consistent with the responses from our healthy volunteers. There was also no difference in EMS use between those who lived alone and those who did not.
If faced with a hypothetical stroke, of community volunteers surveyed, 89% said they would call 911 first. In contrast, only 12% of stroke patients interviewed in fact called 911 first [Table 2]. This was highly significant even when controlling for differences between the two groups in insurance (OR 63.9; 95% CI 29.5 to 138.2). The first action of stroke patients interviewed was more likely instead to call a relative/friend [Table 2].
Barriers for calling 911
Few community volunteers felt any of the suggested barriers applied to them. Those with lower education were actually more likely to plan to call 911 (93% vs. 83%, p=0.012). Meanwhile, 89% of stroke patients felt there was significant delay in seeking medical attention, and in response to open-ended questions, almost half cited thinking the symptoms were not serious and/or the symptoms would resolve as reasons for delay (data not shown). The multiple option format revealed an even higher frequency of these and other barriers (Figure 1). Of those with actual and/or perceived delay in hospital arrival, all attested that one or more of the listed reasons for delay applied to them, including almost one quarter who thought there was nothing that could be done to help. Fewer than half of patients reporting each of the most common barriers ultimately arrived by EMS except for those who cited no transportation or calling a friend/relative first (Figure 1).
Thirty-two percent of patients/proxies reported initially recognizing that the patient may be having a stroke, with no association with level of education (25.9% of those with lower education recognized the stroke symptoms vs. 40.5% of those with higher education, p=0.122). Still only 56.3% of these patients arrived by ambulance compared to 45.6% of those who did not suspect stroke (p=0.320). Despite suspecting stroke, more than two thirds of these patients cited thinking the symptoms were not serious or that the symptoms would resolve as reasons for delay, a similar frequency compared to those who did not suspect stroke.
Of those who arrived by ambulance, only one quarter did so out of belief that utilizing EMS would be the fastest method of transportation to the hospital, while 35% cited not being able to drive or having no other transportation options.
In a high-risk, predominantly black medically underserved urban population, significant discrepancies exist between how healthy volunteers say they would behave during a hypothetical stroke compared to how hospitalized stroke patients actually behaved at the time of their stroke. Whereas most healthy volunteers say they would call 911 immediately, most stroke patients delay taking this action. This paradoxical failure to act promptly contributes to delayed presentation, reduced opportunity for acute therapy, and greater stroke morbidity.
A potential explanation for the observed paradox or disconnect between behavioral intent vs. actual behavior is that patients or witnesses may not recognize stroke signs and symptoms when they occur and, therefore, not trigger a call for EMS. While the DASH II study did not find an association between stroke knowledge and EMS use (in fact, those with greater knowledge were less likely to use EMS),10 perhaps lay people have difficulty translating this knowledge to accurately identify a stroke as it is happening. But, in our study, even among those who suspected stroke, only about half used EMS. Therefore, being able to recognize stroke is not sufficient. The action a patient takes in the acute stroke setting may depend on multiple factors including depth of stroke knowledge, presence of social, cultural, or behavioral barriers, and ability to translate knowledge into appropriate action.15
One factor frequently cited by patients as a contributor to delay was the belief that no medical intervention could help them. That observation underscores the importance of educating the public on not only the availability of a time-sensitive therapy but also motivating the public to utilize it. While current stroke education campaigns focus on symptom recognition and calling 911, most do not explicitly state that calling 911 can result in treatment to improve outcome. If people are taught that prompt treatment of stroke is not just limited to standard supportive care, that knowledge could provide motivation that may overcome social and attitudinal barriers found here and reported in other studies.15-17
Additional misperceptions appear to contribute to delay. Despite the exposure of the study community to national stroke education campaigns, two-thirds of the patients did not recognize they were having a stroke. In addition, the community often believed chest pain to be a common stroke symptom, which may have contributed to patients’ failure to recognize stroke.
The most frequently cited factors for delay included patients’ initial belief that their symptoms were not serious and/or did not require treatment. Even among the one-third of patients who recognized they were experiencing a stroke, utilization of 911 was lower than might be expected given the responses of healthy volunteers to questions about a hypothetical stroke. Some patients may decide to call 911 only if they feel their symptoms are severe.11, 15 Future studies should explore community members’ thresholds for calling 911 depending on hypothetical symptom severity and should examine the correlation between patients’ tendency to call 911 and their stroke severity.
Nevertheless, such findings taken in aggregate indicate that current education campaigns are not achieving their goal in this target community; similar findings might be expected in other urban, underserved communities. Therefore, new approaches to public education about stroke symptoms may be necessary. These might include 1) culturally appropriate descriptions of stroke signs and symptoms with a distinction from cardiac signs and symptoms, 2) education campaigns targeted at audiences other than the potential stroke patient, e.g. children,18, 19 3) an emphasis on the fact that a treatment exists that can improve outcomes and minimize disability, and 4) an emphasis on the fact that stroke symptoms, even when not outwardly severe or dramatic, are manifestations of a serious disease process requiring immediate medical attention.
The current findings indicate that surveying community volunteers does not provide reliable data on what barriers exist in that community to seeking acute stroke care. It was difficult to establish in a hypothetical scenario addressed to community volunteers why patients might delay calling 911 in the setting of an acute stroke.
This study has several limitations regarding participant selection. First, the respondents for both the community survey and patient/proxy interview were self-selected, which introduces selection bias, and the number of refusals were not collected thus preventing the calculation of response rates. Second, some characteristics such as medical history other than stroke were not collected for the community volunteers limiting our ability to ensure the two groups are comparable, and there were socioeconomic differences between the survey and patient respondents. Still, the community participants were recruited at organizations in the catchment area of the hospital, chosen specifically because they serve disproportionately black, lower income and/or elderly populations, similar to the population of stroke patients at this particular hospital. To attempt to mitigate for some of these differences, we evaluated the effects of education and adjusted for insurance type when conducting comparisons between groups. Third, it is possible that the recruited patients had less severe strokes than the hospital population, since the majority of interviews in this study were completed by the stroke patients themselves instead of by proxy. Milder stroke severity may have contributed to their decision not to use EMS. However, our finding that there was no greater use of EMS for those patients who could not complete the interview themselves suggests that stroke severity does not fully account for whether or not a patient will arrive by ambulance. This is an important topic for future investigation.
This study illustrates that while most in this high risk community can correctly answer a survey question to call 911 immediately in case of “stroke,” patients tend to not take this actionwhen faced with the actual situation. Additional studies are needed to explore reluctance to call 911 so that educational interventions can be developed to address these reasons for delay.
The investigators wish to thank the ASPIRE community survey and patient/proxy interview respondents for their time and insights.
Sources of Funding Supported by the National Institute of Neurological Disorders and Stroke (NINDS) and the National Center on Minority Health and Health Disparities (NCMHD) (U54NS057405).
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.