BACKGROUND
Although stroke demographics are changing with the aging of western populations, stroke continues to be a significant socioeconomic burden causing disability, and remains the second leading cause of death worldwide.Reference Niessen, Barendregt, Bonneux and Koudstaal 1 - Reference Lozano, Naghavi and Foreman 3 Factors such as hospital type, patient demographics, and stroke type have been shown to affect in-hospital mortality and discharge destination.Reference Mayo, Hendlisz, Goldberg, Korner-Bitensky, Becker and Coopersmith 4 , Reference Fonarow, Reeves and Smith 5 Whereas patient demographics and stroke type are not modifiable, transfer of patients to differently equipped hospitals might influence the patient’s outcome. Transfers within the hyperacute phase can offer advanced treatment options such as endovascular therapy or neurosurgical intervention or, in the subacute phase, patients can return to local community hospitals from higher levels of care.
We analyzed the characteristics and outcomes, by stroke type, of Canadian patients who were and were not transferred between hospitals as part of their acute stroke care.
METHODS
We performed a population-based retrospective cohort study of all stroke patients admitted to hospital, using data from the Canadian Institutes for Health Information (CIHI), a non-profit organization that collects universally available health care information from 28 pan-Canadian databases. The CIHI’s Discharge Abstract Database (DAD) contains demographic, administrative and clinical data (including deaths and transfers) on all inpatient hospital discharges. Health services are administered by each of Canada’s ten provinces and three territories. All hospitals in those provinces and territories (with the exception of Quebec) are required to report to the DAD; Quebec reports their data through DAD but does not allow it to be shared. 6 Coding of stroke in the DAD is known to have high sensitivity and specificity, and because all acute care hospitals in Canada provide universal access to the Canadian population the data are nationally comprehensive.Reference Kokotailo and Hill 7
We analyzed hospital administrative data for the fiscal years 2011/12 through 2013/14 from the DAD. Patients were identified using International Classification of Diseases 10 (ICD-10) codes for acute stroke, including transient ischemic attack (TIA) (G45.x), ischemic stroke (I63.x, I64.x), aneurysmal subarachnoid hemorrhage (I60.x), and intracerebral hemorrhage (I61.x). Stroke risk factors (hypertension, diabetes mellitus and atrial fibrillation) and comorbid illness were identified by ICD-10 code and by their calculated Charlson-Deyo index (grouped 0-1 vs. 2+).
Based on the Heart and Stroke Foundation (HSF) 2014 Stroke Report and the Canadian Stroke Best Practice Recommendations, all hospitals were categorized into four peer groups based on availability of services and resources for stroke care: Comprehensive Stroke Services, Advanced Stroke Services, Basic Stroke Services and General Healthcare Providers. 8 , Reference Casaubon, Boulanger and Blacquiere 9 Advanced and Comprehensive Stroke Services provide access to intravenous thrombolysis and have stroke teams that can be mobilized to rapidly assess, diagnose and implement management strategies to reduce the risk of stroke recurrence or poor outcomes. Comprehensive Stroke Services have a stroke unit, regional support services and a department of neurosurgery. They may also have endovascular interventionists’ service. Advanced Stroke Services do not have a stroke unit and may or may not have a department of neurosurgery. Basic Stroke Services have neuroimaging capability (i.e., a CT scanner on site), do not offer intravenous thrombolysis or other organized stroke care, but in some cases offer basic prevention services. Organizations that are considered General Health Providers do not have neuroimaging capability. HSF Canada provided a list of assigned peer group for every hospital; linkage to individual patient records was then conducted at CIHI before data were anonymized and provided to HSF Canada for analysis. Comprehensive and Advanced Stroke Services were merged to “stroke center” and Basic Stroke Services and General Health Provider were merged to “community hospital”.
A randomly generated unique identifier was assigned to each case by CIHI to allow for tracking patients across multiple admissions or transfers. We defined an interhospital transfer as follows: any admission within 6-48 hours of the previous discharge where the “institution from” matches the “institution to” unique identifier (meaning one case of care) and any admissions within 12 hours of the previous discharge where either the “institution from” or the “institution to” type equals 1 (meaning acute care). This definition excludes readmission to the same institution within that time frame. Discharge location categories were: home without services; home with homecare services; inpatient rehabilitation; other acute care facility; complex continuing care; long term care; died and other. Discharge locations were used as outcome parameters.
Measures of central tendency and measures of variability were calculated according to standard descriptive statistics. We used multivariable logistic regression models to test whether sex or stroke type were associated with transfer to a comprehensive stroke center. We included the following variables in our model: province, age, stroke type and cardiovascular risk factors (namely atrial fibrillation, coronary artery disease, diabetes mellitus, hypertension, previous stroke/transient ischemic attack) according to past medical history. All analyses were conducted with the use of SAS software (SAS Institute Inc, Cary, NC). All reported p-values are two-sided.
RESULTS
During the fiscal years 2011/12 to 2013/14, 84,397 patients were admitted to hospital for stroke syndromes (ischemic stroke, transient ischemic attack, intracerebral hemorrhage, subarachnoid hemorrhage). Data were incomplete in 2.8% of patients. The flowchart of the study population is shown in Figure 1. Less than 5% (n=4030) of patients were transferred between community hospitals and stroke centers in either direction. The proportion of patients transferred by province/territory was: 11.5 % in the Territories, 9.2 % in Saskatchewan, 7.7 % in Prince Edward Island, 6.2 % in Manitoba, 5.9 % in British Columbia and Nova Scotia, 4.8 % in Alberta, 3.9 % in Ontario, 3.3 % in Newfoundland and 2.5 % in New Brunswick.
Flowchart of study population.
Patients transferred to stroke centres had a significantly lower median age. Baseline characteristics of transferred and non-transferred patients are displayed in Table 1. Increasing age was associated with a reduced probability of transfer: 7.6 % in the <45 year old, 6.0 % in the 45–64 year old, 5.2 % in the 65–74 year old, 4.4 % in the 75–84 year old and 3.1 % in the >85 year old. By stroke type, 12.0% of subarachnoid hemorrhages, 6.3 % of intracerebral hemorrhages, 4.9 % of ischemic strokes and 1.1 % of transient ischemic attacks were transferred. The median total length of stay (including both the transferring and receiving facilities) was 6 (3–14) days for non-transferred patients compared to 18 (9–38) days in patients who did get transferred (P<.0001). The median length of stay for different transfer pathways is shown in Table 2.
Clinical characteristics
P values are Pearson’s chi-square test (2-tailed). IQR indicates interquartile range, TIA transient ischemic attack, ICH intracerebral hemorrhage, SAH subarachnoid hemorrhage.
Median length of stay
IQR indicates interquartile range
Detailed results regarding discharge destinations for transferred and non-transferred patients are displayed in Table 3. The highest probability of discharge to home was associated with sole care at a stroke center (43.8%). Transfer to a stroke center from community hospital was associated with the highest probability for discharge to rehabilitation facility (25%) and the lowest one to either long-term (2.1%) or community care center (2.0%). Transferred patients were less likely to die (13.3 % vs. 14.8 %, risk ratio [RR] 0.905, 95 % confidence interval [CI] 0.835–0.981, p=0.015). Discharge destinations for transferred patients divided by specific transfer pathway are reported in Table 4.
Discharge destinations
P values are Pearson’s chi-square test (2-tailed). RR indicates relative risk, CI confidence interval
Discharge destination by transfer pathways
Multivariable regression analysis showed no difference for transfer to a comprehensive stroke center between males and females (odds ratio [OR] 1.03, CI95 0.94–1.11). Patients with intracerebral hemorrhages (OR 1.37, CI951.21–1.56) and subarachnoid hemorrhages (OR 3.39, CI95 2.98–3.85) were more likely to be transferred to a stroke center compared to patients with ischemic strokes.
DISCUSSION
Our study provides evidence that stroke center care is associated with better discharge destination outcomes, a surrogate for final clinical outcome. Sole stroke center care was associated with a high discharge rate to home; transfer to a stroke center was associated with a high discharge rate to rehabilitation facilities and lower mortality rates. A potential explanation is the younger age and lesser comorbidity of the transferred patients with the subsequent stroke unit care (life-saving, sparing large number of severely disabled life years) resulting in better outcomes.
Transfer to a stroke center was initiated faster than transfer to a community hospital when comparing median length of stay in days at the first facility. A possible explanation is that these patients were transferred for (hyper)acute treatment whereas a transfer to a community hospital is likely to represent repatriation to the home community. Common practice is to transfer patients back to their home community, especially those waiting for long-term care beds, rather than take up beds in comprehensive centers. The median length of total stay was prolonged for all transferred patients, perhaps due to higher symptom severity and thus development of complications.
The overall rate of transfers was low, at less than five percent. This is probably the result of multiple factors. First, acute transfer of ischemic stroke (the most common stroke type) for thrombolytic treatment is influenced by distance to the treatment facility and delays in presentation. Up to 50% of rural stroke patients present late to the treatment facility so that their baseline neuroimaging will already demonstrate established infarction, excluding them from hyperacute treatment.Reference Bodechtel, Barlinn and Siepmann 10 Second, stroke syndromes usually affect older patients who might have personal directives arguing against aggressive interventions or even transient life-supporting treatment. In our study younger patients were transferred more frequently to stroke centers and older patients were more likely treated in community hospitals. Third, transfer is not risk-free and the benefit of a transfer must be weighed against the possible acute risks of medical deterioration during transfer. Fourth, although hemorrhagic stroke types are less common, intracerebral and subarachnoid hemorrhages in our study were more likely to be transferred to a comprehensive stroke center than other stroke subtypes. This is probably due to the potential need for neurosurgical or intensive care intervention. Fifth, throughout Canada established bypass protocols in most regions mean that stroke patients are brought directly to primary and comprehensive stroke centers, bypassing smaller community hospitals. These result in fewer patients eligible for transfer, a trend that has been observed in other reports that compare outcome before and after the implementation of bypass agreements.
Although we cannot comment on the indications for transfer initiation in our study, comprehensive stroke centers can benefit from direct viewing of neuroimaging to determine which patients are less likely to benefit from and require transfer. This could lead to optimization of transportation-related costs and resource use. For these cases, televideoconsultations with hospitalists at community hospitals that include recommendations for medical care such as blood pressure goals and treating patients on an intermediate care unit may be a viable alternative to transfer.Reference Nguyen, Mir and Vahidy 11
Mechanical thrombectomy is now the standard of care for ischemic strokes caused by proximal anterior circulation occlusion. 12 Each of the recently published endovascular trials relied on advanced and pre-established stroke systems of care mandating rapid identification, triaging, imaging and initiation of treatment/transfer.Reference Mocco, Fargen and Goyal 13 The ESCAPE trial, which included 11 major centers in Canada, enrolled the first patient in February 2013, mid-way through our analysis period. The trials have underscored the need for faster and better access to comprehensive stroke centers for patients with the diagnosis of acute stroke. The Madrid Stroke Network found that no safety concerns have been indicated for patients with acute ischemic stroke who are transferred between hospitals to undergo reperfusion therapies. Their study also showed for the first time that the frequency of futile transfers for potential endovascular treatment was ~41 %. The most common reason for determining that a transfer was futile was clinical improvement; the second most common was exclusion from endovascular treatment due to the results of a second neuroimaging study performed.Reference Fuentes, Alonso de Lecinana and Ximenez-Carrillo 14 This again underscores the need for speed, with onset-to-reperfusion time as the new bottom-line process metric to achieve favorable clinical outcomes.Reference Hill, Goyal and Demchuk 15 This new treatment option will likely result in a substantive increase in acute transfers to centers equipped with an angiography suite and staffed with stroke neurologists and interventional neurologists. On the other hand, we know that clinical outcomes are more favorable at high-volume centers and it might therefore be a better strategy to transport the patient, if feasible, directly to a specialized center. The higher cost of transfers compared to direct admission, as well as the delay in final treatment initiation (i.e. mechanical thrombectomy) also support this strategy.Reference Gupta, Horev and Nguyen 16 , Reference McNeill, English, Borg, Matta and Menon 17 Optimal regional algorithms for direct admission vs. transfer in the new era of endovascular therapy still need to be determined. These algorithms will need to incorporate that timely access to IV thrombolysis is still achievable for eligible patients within 4.5 hours from symptom onset with a target door-to-needle-time of less than 60 minutes for 90% of treated patients, as per Canadian best practice recommendations.Reference Casaubon, Boulanger and Blacquiere 9
Strengths of our study are the population-based approach with high external validity and a large sample size of more than 80,000 patients. We cannot comment on the timing of presentation to the hospital, the presenting stroke severity or the modalities of treatment provided at the receiving hospital. Other limitations include the fact that the DAD does not capture patient transfers between an Emergency Department and another facility; it is these types of transfers that would be expected for hyperacute endovascular stroke treatment.
CONCLUSION
Patients who were transferred to stroke centers were less likely to die. Further studies are needed to investigate clinical and system predictors to determine who benefits most from transfer and thus avoid increased healthcare costs of unnecessary or futile transfers. Transport pathways might have to be rearranged to ensure a higher proportion of direct admission to a specialized center. Further exploration of the role and utility of transfer is going to be needed specifically for access to endovascular treatment for acute ischemic stroke.
DISCLOSURES
Charlotte Zerna, Noreen Kamal, M. Patrice Lindsay, Jiming Fang, and Michael Hill do not have anything to disclose.
STATEMENT OF AUTHORSHIP
C. Zerna contributed to the study concept and design, interpretation of the data, as well as drafting and revising the manuscript.
N. Kamal contributed to the study concept and design, and revising the manuscript.
J. Fang contributed to the data analysis and interpretation as well as revising the manuscript.
P. Lindsay contributed to data acquisition and management, the study concept and design, and revising the manuscript.
M. Hill contributed to the study concept and design, interpretation of the data, as well as drafting and revising the manuscript and functioned as the senior author for this work.
