Customize your JAMA Network experience by selecting one or more topics from the list below.
Saltman AP, Silver FL, Fang J, Stamplecoski M, Kapral MK. Care and Outcomes of Patients With In-Hospital Stroke. JAMA Neurol. 2015;72(7):749–755. doi:10.1001/jamaneurol.2015.0284
Copyright 2015 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
A sizeable minority of strokes occur in hospitalized patients. However, little is known about the presentation, care, and outcomes of stroke in this subgroup of patients.
To examine stroke care delivery and outcomes for patients with in-hospital vs community-onset stroke.
Design, Setting, and Participants
Prospective cohort study of all patients older than 18 years with acute stroke seen in the emergency department or admitted to the hospital at participating centers (all regional stroke centers in Ontario, Canada) between July 1, 2003, and March 31, 2012, including those with stroke onset during hospitalization for another cause.
Main Outcomes and Measures
We compared processes of stroke care delivery, including time to neuroimaging and rates of thrombolysis, as well as outcomes, including death and disability, in those with in-hospital vs community-onset stroke. We used multiple logistic regression models to adjust for age, sex, comorbid conditions, and stroke type and severity.
The study sample included 973 patients with in-hospital stroke and 28 837 with community-onset stroke. Patients with in-hospital stroke compared with those with community-onset stroke had significantly longer waiting times from symptom recognition to neuroimaging (median, 4.5 vs 1.2 hours; P < .001; for <2 hours, 32% vs 63%; adjusted odds ratio [AOR] = 0.21; 95% CI, 0.18-0.24), lower use of thrombolysis (12% vs 19% of those with ischemic stroke; AOR = 0.54; 95% CI, 0.43-0.67; P < .001), and longer time from stroke recognition to administration of thrombolysis (median, 2.0 vs 1.2 hours; P < .001). After adjustment for age, stroke severity, and other factors, mortality rates at 30 days and 1 year after stroke were similar in those with in-hospital stroke and community-onset stroke; however, those with in-hospital stroke had a longer median length of stay following stroke onset (17 vs 8 days; P < .001), were more likely to be dead or disabled at discharge (77% vs 65% with modified Rankin Scale score of 3-6; AOR = 1.64; 95% CI, 1.38-1.96; P < .001), and were less likely to be discharged home from the hospital (35% vs 44%; AOR = 0.76; 95% CI, 0.64-0.90; P < .001).
Conclusions and Relevance
Compared with those with community-onset stroke, patients with in-hospital stroke had delays in investigations and treatment, suggesting a need for a standardized approach to the recognition and management of in-hospital stroke, with the aim of ensuring access to rapid acute stroke care.
Quiz Ref IDIn-hospital stroke refers to stroke occurring in a patient hospitalized for another reason. It is estimated that this population represents between 4% and 17% of acute strokes,1 although these events are likely under-recognized and under-reported.2 Unlike most strokes occurring in the community, there is no delay between stroke onset and hospital arrival for this subgroup, and thus there is the potential for rapid diagnosis and treatment. Quiz Ref IDHowever, factors such as recent surgery and acute medical illness in this population may limit the use of thrombolysis and other acute interventions and may adversely affect outcomes.
Existing research in this area suggests that those with in-hospital stroke may have delays in recognition and assessment of stroke symptoms, lower rates of use of thrombolysis, and a greater risk of death or disability compared with those with community-onset stroke.3-6 Consequently, these initial studies emphasized the need for increased awareness by hospital physicians of risk factors for in-hospital stroke and possible interventions to reduce these factors, including the need for strategies to streamline in-hospital acute stroke care.4,7,8
In this study, we used data from the Ontario Stroke Registry (formerly known as the Registry of the Canadian Stroke Network) to evaluate the care and outcomes of patients with in-hospital and community-onset stroke at regional stroke centers in Ontario, Canada.
The Ontario Stroke Registry collects data on consecutive patients with stroke or transient ischemic attack seen in the emergency department or admitted to the hospital in Ontario, Canada, from July 1, 2003, onward. All patients seen at the 11 regional (comprehensive) stroke centers are included in the registry, and periodic audits are also performed at other acute care facilities across the province. The registry collects data on sociodemographic factors, medical conditions, stroke type and severity, in-hospital investigations, treatments, complications, and medications prescribed at discharge. Stroke severity is determined using the Canadian Neurological Scale, a validated tool that measures level of consciousness, orientation, speech, and motor function.9 Functional status at discharge is measured using the modified Rankin Scale.10 Medical record abstraction is performed by trained neurology research nurses, and validation by duplicate medical record abstraction has shown excellent agreement for key variables including age, sex, stroke type, use of thrombolysis, and stroke unit care.11 The data collection software forces medical record abstraction personnel to perform complete data entry before the case record can be submitted for inclusion in the database, ensuring that there are no missing data. The Ontario Stroke Registry is authorized under provincial privacy legislation to perform medical record abstraction without individual patient consent. The registry is housed at the Institute for Clinical Evaluative Sciences, where it is linked to population-based administrative databases using unique encoded patient identifiers. The Ontario Registered Persons Database provides information on any deaths that occur, regardless of location.
This study was approved by the Research Ethics Board of Sunnybrook Health Sciences Centre and by the Ontario Stroke Registry/Registry of the Canadian Stroke Network Publications Committee.
From the Ontario Stroke Registry database, we included all patients who presented to any of the 11 regional stroke centers with acute stroke or were diagnosed as having an in-hospital stroke between July 1, 2003, and March 31, 2012. We excluded those with transient ischemic attack, those without a valid health card number, and those younger than 18 years. For patients with more than 1 stroke event during the study period, only the first event was included in this analysis.
Our primary outcome was the time from stroke presentation (symptom recognition for those with in-hospital stroke and emergency department arrival for those with community-onset stroke) to the first neuroimaging procedure. Secondary outcomes were time from stroke presentation to thrombolysis, use of thrombolysis, stroke unit care, dysphagia screening, carotid imaging, medications for secondary stroke prevention, length of stay, modified Rankin Scale score at discharge, discharge destination, and mortality at 7 days, 30 days, and 1 year following stroke.
We compared baseline characteristics and processes of stroke care for patients with in-hospital stroke and those with community-onset stroke using χ2 tests for categorical variables and t tests and Kruskal-Wallis tests for the mean and median of continuous variables, respectively. In secondary analyses, we compared processes of care for patients who experienced in-hospital stroke by their location at the time of stroke, categorized as occurring during angiography, admission for cardiac surgery, other surgery, or a medical admission. We evaluated the association between in-hospital stroke and length of stay, modified Rankin Scale score at discharge, and mortality at 7 days, 30 days, and 1 year, using multiple logistic regression to adjust for age, sex, vascular risk factors (hypertension, diabetes mellitus, hyperlipidemia, previous stroke or transient ischemic attack, peripheral vascular disease, atrial fibrillation, and smoking status), other medical comorbidities (dementia, heart failure, cancer, cirrhosis, asthma or chronic obstructive pulmonary disease, renal dialysis, gastrointestinal bleeding, depression, and arthritis), stroke type (ischemic or hemorrhagic), and stroke severity (based on the Canadian Neurological Scale). All analyses were performed using SAS version 9.3 statistical software (SAS Institute, Inc).
Our study sample included 29 810 patients, of whom 973 had in-hospital stroke and 28 837 had community-onset stroke. Quiz Ref IDIn-hospital stroke occurred during angiography (15%) or during an admission for cardiac surgery (25%), for noncardiac surgery (22%), or to a medical service (32%); the patient location at the time of stroke was undetermined in 6% (Table 1). Those with in-hospital stroke compared with those with community-onset stroke were older (72% vs 67% aged >65 years) and more likely to have vascular risk factors and other comorbid conditions (32% vs 22% with diabetes mellitus; 69% vs 61% with hypertension; 46% vs 34% with hyperlipidemia; and 21% vs 16% with atrial fibrillation), to have an ischemic stroke as opposed to hemorrhagic stroke (89% vs 71%), and to have greater stroke severity (23% vs 16% with Canadian Neurological Scale score of 0-4) (P < .001 for all comparisons) (Table 1). The 2 populations had similar rates of independent functional status before admission (Table 1).
Patients with in-hospital stroke had significantly longer times from symptom recognition to neuroimaging than patients with community-onset stroke (median, 4.5 vs 1.2 hours; P < .001), with a significantly lower percentage of patients with in-hospital stroke undergoing brain imaging within 2 hours of symptom recognition (32% vs 63%; adjusted odds ratio [AOR] = 0.21; 95% CI, 0.18-0.24; P < .001). Patients with in-hospital stroke also had lower rates of thrombolysis (12% vs 19% of those with ischemic stroke; AOR = 0.54; 95% CI, 0.43-0.67; P < .001) and longer door-to-needle times for thrombolysis (median, 2.0 vs 1.2 hours; P < .001), and fewer of these patients received thrombolysis within 90 minutes of diagnosis compared with their counterparts who had community-onset stroke (29% vs 72%; P < .001) (Table 2). These differences remained significant following adjustment for age, sex, comorbid conditions, stroke type, and severity (Table 3).
In cases for which thrombolysis was not administered, the stated reason was more often a contraindication to thrombolytic therapy (46% vs 9%) or a delay in decision making (2% vs 1%) and less commonly presentation outside the 4.5-hour treatment window (17% vs 62%) or determination that stroke-related deficits were too mild (10% vs 19%) in the in-hospital stroke population compared with the group with community-onset stroke (P < .001 for all comparisons) (Table 2). Those with in-hospital stroke compared with those with community-onset stroke were less likely to be cared for on a stroke unit (28% vs 61%) and had lower rates of neuroimaging (98% vs 99%), carotid imaging (55% vs 65%), Holter monitoring (18% vs 30%), and swallowing assessment (56% vs 63%) (P < .001 for all comparisons) (Table 2). Patients with in-hospital stroke also had a longer median length of stay than those with community-onset stroke (17 vs 8 days; P < .001) and were more likely to be dead or disabled at discharge, even after adjustment for sex, age, comorbid conditions, and stroke type and severity (modified Rankin Scale score of 3-6 in 77% vs 65%; AOR = 1.64; 95% CI, 1.38-1.96; P < .001) (Table 2 and Table 3).
While unadjusted mortality was similar between the in-hospital and community-onset stroke groups at 7 days following stroke (11% vs 10%; P = .77), mortality was higher in those with in-hospital stroke compared with community-onset stroke at 30 days (22% vs 18%; P < .001) and at 1 year (35% vs 27%; P < .001). However, following adjustment for age, sex, comorbid conditions, stroke type, and stroke severity, in-hospital stroke was associated with lower mortality at 7 days following stroke than community-onset stroke (AOR = 0.67; 95% CI, 0.53-0.85), and there were no differences in mortality rates between the 2 groups at 30 days (AOR = 0.89; 95% CI, 0.74-1.07) or at 1 year (AOR = 0.99; 95% CI, 0.85-1.16) (Table 3).
Among those alive at discharge, patients with in-hospital stroke were more likely to be prescribed medications for secondary stroke prevention than those with community-onset stroke (86% vs 75% for antithrombotic therapy and 81% vs 71% for antihypertensive therapy), were less likely to be discharged home (35% vs 44%; AOR = 0.76; 95% CI, 0.64-0.90), and were more likely to be discharged to rehabilitation facilities (40% vs 32%; AOR = 1.42; 95% CI, 1.22-1.66) (P < .001 for all comparisons) (Table 2).
Among patients with in-hospital stroke, there were differences in stroke care by event location, with those who had strokes during angiography having the shortest median time from stroke onset to neuroimaging (1.9 hours) and the highest use of thrombolysis (28%), those admitted to cardiac surgical services having the longest median time to neuroimaging (7.1 hours) and the lowest use of thrombolysis (3%), and those admitted to medical services having intermediate median time to neuroimaging (3.1 hours) and thrombolysis use (16%). A neurologist was the most responsible physician for 15% in the angiography group, none in the cardiac surgery group, 5% in the noncardiac surgery group, and 8% in the medical group. There were also differences in outcomes, with 60% dead or disabled at discharge in the angiography group compared with 75% in the cardiac surgery group, 84% in the noncardiac surgery group, and 83% in the medical group (Table 4).
Quiz Ref IDWe found that patients with in-hospital stroke had higher rates of comorbid illness, experienced more severe strokes, had a longer time from symptom recognition to initial neuroimaging, were less likely to receive thrombolysis, had greater disability after stroke, and had a longer hospital stay compared with their counterparts who had community-onset stroke. However, after adjustment for age, sex, comorbid conditions, and stroke severity, stroke case fatality among patients with in-hospital stroke was lower at 7 days and not significantly different at 30 days or 1 year after stroke than those with stroke onset in the community.
These findings are consistent with those from previous studies in the United States and elsewhere, which have documented delays in diagnosis and treatment of those with in-hospital stroke compared with community-onset stroke.1-5 Because earlier diagnosis and treatment are associated with better outcomes after ischemic stroke, our findings suggest the need for interventions to improve the care of in-hospital stroke.4,7,8,12-15 A 2003 article by Nolan et al15 proposed a so-called code gray for the management of acute stroke. Akin to a code blue or a code STEMI (segment elevation myocardial infarction), this code gray would mandate the training of hospital staff to recognize admitted patients with acute stroke symptoms and to enact a standardized in-hospital stroke protocol to initiate treatment. In addition, a quality improvement initiative in one US hospital, which used a standardized approach to reduce wait times from stroke symptom recognition to arrival in the computed tomographic scanner, was associated with a 57% decrease in median inpatient alert-to–computed tomography time.16
Importantly, we documented significant differences in the baseline characteristics of those with in-hospital vs community-onset stroke, with a greater prevalence of comorbid illness in the in-hospital stroke cohort, higher unadjusted short- and long-term mortality, and a large proportion of patients who were ineligible for thrombolysis because of medical or surgical contraindications, to a greater extent than that observed in other studies.2,5,8 These findings suggest that those with in-hospital stroke represent a distinct subgroup of patients with stroke who may require specific care pathways and whose outcomes may differ from those with community-onset stroke, regardless of the care received.
Even within the in-hospital cohort, there were important differences among patients depending on their location at the time of stroke onset. Those who sustained strokes during angiography had more rapid brain imaging and were more likely to receive thrombolysis than patients admitted to medical or surgical services. Those admitted to surgical services, most of whom would have sustained strokes perioperatively, had the longest time to brain imaging and the lowest use of thrombolysis, as would be expected given the logistical challenges in obtaining imaging and the ineligibility for thrombolysis conferred by recent surgery.
We also found that patients with in-hospital stroke were less likely than those with community-onset stroke to be cared for on an acute stroke unit, possibly related to a need for specific medical or surgical services available on other wards. Only a minority of patients with in-hospital stroke were cared for by a neurologist, with the remainder receiving care from a cardiologist, general internist, family physician, cardiac surgeon, or other surgical specialist. Nonneurologist physicians may have less familiarity and training in stroke care, may be less apt to recognize stroke symptoms, and may require more time to familiarize themselves with appropriate care pathways should a stroke be identified. Whether the use of mobile stroke teams provides an alternative to address these issues for patients with stroke on nonneurology wards or whether transfer to a stroke unit is preferable requires further study.
Our study was strengthened by the inclusion of a large hospital-based sample of consecutive patients from all regional stroke centers in Ontario, thus eliminating the potential biases associated with the inclusion of only self-selected hospitals or an incomplete patient sample. In addition, in contrast to other studies, we had information on key prognostic variables such as stroke severity as well as information on important outcome measures such as functional status at discharge and long-term survival, with minimal missing data.
However, our study has some limitations that merit comment. Our data sources did not contain information on the reasons for delays in diagnosis or treatment, and thus we were unable to identify specific patient or institutional factors that may have contributed to variations in care in those with in-hospital vs community-onset stroke. Although we had some information on the reason for the initial hospitalization in those with in-hospital stroke, we did not have information on specific medical and surgical diagnoses and cannot comment on whether these underlying conditions were associated with differences in care and outcomes. We also do not have access to information on long-term functional status or quality of life following stroke.
Our study confirms that patients with in-hospital stroke have more severe strokes, are less likely to receive timely investigations and treatment, and have worse outcomes compared with their counterparts with community-onset stroke, but it also documents important differences in baseline factors that may influence care and outcomes. These results add to the growing body of evidence in favor of the development of a standardized approach to the recognition and management of in-hospital stroke. Quiz Ref IDIn particular, we advocate for the development of targeted code stroke protocols for the in-hospital stroke population, similar to those used in the emergency department. By providing frontline health care workers with a step-by-step algorithm for diagnosis and treatment in these patients, such protocols (particularly when paired with an educational intervention for physicians and nurses) might serve to address established inefficiencies along the care pathway including time to neuroimaging and time to thrombolysis, thereby facilitating rapid access to best-practice investigations and timely initiation of therapy, with a view toward improving outcomes in this high-needs group.
Corresponding Author: Moira K. Kapral, MD, MSc, Toronto General Hospital, 14EN 215, 200 Elizabeth St, Toronto, ON M5G 2C4, Canada (firstname.lastname@example.org).
Accepted for Publication: February 18, 2015.
Published Online: May 4, 2015. doi:10.1001/jamaneurol.2015.0284.
Author Contributions: Drs Saltman and Kapral had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Saltman, Silver, Kapral.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Saltman, Silver, Kapral.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Fang.
Obtained funding: Silver, Kapral.
Administrative, technical, or material support: Silver, Stamplecoski.
Study supervision: Kapral.
Conflict of Interest Disclosures: None reported.
Funding/Support: The Ontario Stroke Registry is funded by the Canadian Stroke Network and the Ontario Ministry of Health and Long-Term Care. The Institute for Clinical Evaluative Sciences is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care. This work was supported by an educational grant from the Canadian Stroke Network. Dr Kapral is supported by a career investigator award from the Heart and Stroke Foundation, Ontario Provincial Office.
Role of the Funder/Sponsor: The funding agencies had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The opinions, results, and conclusions are those of the authors and are independent from the funding sources. No endorsement by the Canadian Stroke Network, Institute for Clinical Evaluative Sciences, or Ontario Ministry of Health and Long-Term Care is intended or should be inferred.
Create a personal account or sign in to: