Introduction
Stroke is a common disorder and a leading cause of death and disability worldwide. Reference Latchaw, Alberts and Lev1 Transient ischemic attack (TIA) or minor stroke patients (here onwards just called TIA) are at elevated risk of a subsequent stroke in the days to weeks following their initial event. Reference Giles and Rothwell2,Reference Easton and Johnston3 Up to 10% of patients with TIA will have a subsequent stroke within 90 days without urgent care, but with optimized treatment this risk can be reduced to about 1%. Reference Lavallée, Meseguer and Abboud4,Reference Rothwell, Giles and Chandratheva5 The urgency and level of care provided for TIA patients often depends on the perceived risk of subsequent major stroke. Reference Perry, Losier, Stiell, Sharma and Abdulaziz6,Reference Perry, Losier, Stiell, Sharma and Abdulaziz7 Traditionally, and still in some parts of the world, TIA patients were admitted to hospital for rapid investigation and comprehensive management. Reference Ranta and Barber8 In an effort to optimize health resource utilization with improved patient outcomes, outpatient-based models have been introduced. Reference Ranta and Barber8–Reference Sanders, Cadilhac, Srikanth, Chong and Phan11 A stroke prevention clinic (SPC) is a specialized outpatient clinic that provides rapid access to experts, diagnostic tests, and treatments. Such clinics are meant to provide an integrated, comprehensive, and interdisciplinary approach to stroke prevention in a timely manner.
In Canada, the Canadian Stroke Best Practice Recommendations (CSBPR) provide guidelines for the prevention and management of stroke.Reference Gladstone, Lindsay and Douketis12 However, we suspect that they are not always followed, for varying reasons, leading to variation in how patients with suspected TIA are identified and managed. Reference Johnston and Smith13,Reference Goldstein, Bian, Samsa, Bonito, Lux and Matchar14 For example, although CSBPR suggests managing high-risk patients within 48 hours of symptom onset, we suspect that not all organizations meet this recommendation. From information-gathering interviews with local neurologists (key informants), we found that some SPCs might (a) be open limited hours a week, (b) triage patients without assessing and considering risk levels, (c) have protocols in place to have patients complete some tests prior to their clinic appointment, (d) have long wait times for some tests which varies depending on referring source, and (e) lack urgent communication protocols between SPCs and radiology departments. The extent of variation in practice is unknown. Understanding the extent of variation in practice can help develop or modify strategies and guidelines to standardize practices across SPCs which may ultimately reduce the incidence of subsequent stroke.
To understand the extent of variation we conducted a survey of all Canadian SPCs. Prior to implementing the survey, the COVID-19 pandemic started and so we modified our questionnaire to include additional items to understand the impact of the COVID-19 pandemic on TIA patient management.
The aims of this study were twofold: (1) to describe variation in management practices for TIA patients and (2) to characterize the impact of the COVID-19 pandemic on the management of TIA patients including the use of virtual care by the SPCs.
Methods
Study Design and Participants
We conducted a self-administered electronic survey of SPCs in Canada. To be eligible, the respondents must have been physician leads, managers, or coordinators at any of the SPCs in Canada. A list of SPCs was provided by the Heart and Stroke Foundation of Canada from which we identified 80 unique SPCs as of May 5, 2021. We obtained the contact information of the person running each SPC using web searches and by calling the clinics. The study was conducted from May 2021 through November 2021. We designed the survey following Dillman’s techniques. Reference Dillman15 This study has received Ottawa Health Science Network Research Ethics Board approval.
Questionnaire Development
We developed the questionnaire in three stages. First, we conducted telephone interviews with key informants to gather information to prepare a draft survey questionnaire. Second, we conducted cognitive interviews (assessing participants’ understanding of the questionnaire as they complete each question) to evaluate the clarity, comprehensibility, and face validity of each question in the draft survey. Third, we pilot tested the draft survey questionnaire using a smaller subsample of the SPCs to assess the whole questionnaire and the survey process. Reference Dillman15
The final questionnaire included 36 questions. Questionnaires consisted of an eligibility question, information about clinic hours of operation and scheduling (14 items), medical imaging (13 items), bloodwork and medications (three items), impact of the COVID-19 pandemic (five items), and additional information (one item). To improve response rate, most items (including questions on wait times) were designed as closed-ended questions where respondents’ answers were limited to a fixed set of responses. The questionnaire was developed with attention to clarity regarding question applicability to the pre-pandemic period. The questionnaire, landing page, recruitment, and reminder emails were translated into French by a medical translator.
Survey Administration
The survey invitation was personalized for each SPC so that the lead’s name was inserted on all emails along with a unique link for each clinic. Every email and the landing page of the survey included names of the lead principal investigators, contact information, and affiliations, as an indication of a legitimate source.
We pilot tested the survey on 15 SPCs across the provinces and territories. After revising the survey questionnaire for clarity and verifying the feasibility of the process (i.e., our survey questions were answered as intended), invitations were distributed to the remaining 65 SPCs.
The survey was initiated with a recruitment email containing both official languages (English and French) and links to the survey powered by SurveyMonkey (www.surveymonkey.com). A reminder email along with links to the survey questionnaire was emailed every week to non-responders for up to 4 weeks. A final reminder was initiated 2 weeks after the last email reminder by calling the clinic. Several attempts were made to reach clinics by telephone over a period of 3 weeks.
Data Analysis
We calculated descriptive statistics to characterize the SPC responses. Continuous variables were expressed as mean and standard deviation or median with interquartile ranges (IQR), while categorical data were expressed as frequencies and percentages. We used boxplots to help visualize the distribution of some of the continuous variables. Chi-squared tests were conducted to compare the geographic region (Western Canada, Ontario, and Eastern Canada) of respondents and non-respondents to examine the possibility of non-response bias. We used Fisher’s exact test to study the association between seeing high-risk patients within 48 hours and number of days clinics are open. Two-sided significance tests were set at an alpha level of 0.05.
We grouped data based on referring source and other characteristics to organize the results. To reduce the number of items in the tables, we combined categories with sparse numbers. Data were analyzed using SAS version 9.4 (SAS Institute, Cary, North Carolina, USA) and R version 4.0.5 (R Foundation for Statistical Computing, Vienna, Austria).
Results
Of 80 SPCs invited, four indicated they were ineligible (by answering “no” to the question: “are you the physician lead or coordinator/manager of the SPC or answering on his/her behalf?”). Of the remaining 76 SPCs, 38 (50.0%) completed the electronic survey (including eight from the pilot survey and four from final phone reminders). Five respondents provided partial responses.
Table 1 describes the structure of the SPCs. The clinics were open from half a day per week (2.6%) up to 7 days per week (2.6%) with about a quarter (26.3%) being open 2 days or fewer per week. Two-thirds (63.2%) of the clinics were open 5 days a week. Clinic hours varied from less than 4 hours to more than 12 hours per day. Table 2 shows that there is a significant difference in being able to see most high-risk TIA patients within 48 hours and how many days the clinics are open during the week. A small portion (10.5%) of the clinics do not accommodate urgent patients. Patient referrals to the clinics are mostly from their own emergency departments (EDs) (median percentage of patients referred (IQR): 50 (40–70)) followed by family physicians (median (IQR): 20 (15–30)). The initial diagnosis of TIA was confirmed less often when the referral was from family physicians (median (IQR): 30 (20–50)) compared to the organization’s own ED, other EDs, or other specialists. Except for one clinic (2.6%), all the clinics use an appointment system that considers patient risk level. The clinic that does not consider risk levels normally sees all patients within a week. For clinics that do consider patient risk levels, the wait times for low-risk patients are within 30 days for 64.8% of SPCs. Wait times for high-risk patients are within 24 hours for 8.1% of SPCs and between 1 and 2 days for 29.7% of SPCs. We found that 29.0% of the clinics do not a have a protocol in place to consult neurology while the patient is in the ED while 32.3% consult neurology for high-risk patients only. We found 60.0% of the organizations admit patients for revascularization directly from the ED. Two clinics are not able to bypass normal wait times for vascular imaging for urgent cases.
Table 1: Characteristics of stroke prevention clinics
ED, Emergency Department; IQR, interquartile range; TIA, transient ischemic attack.
Table 2: Association between seeing high-risk patients within 48 hours and number of days clinics are open
Data are reported as number (%) of stroke prevention clinics.
* p value obtained using Fisher’s exact test.
To test for the possibility of non-response bias, we compared the geographic region (Western Canada, Ontario, and Eastern Canada) of respondents and non-respondents. Chi-squared analysis showed no indication of a significant difference (p value: 0.17) in responses when we compared the regions (Table 3). We did not have other characteristics to test for non-response bias.
Table 3: Chi-squared tests of non-response bias
a British Columbia, Alberta, Saskatchewan, Manitoba.
b Quebec, New Brunswick, Nova Scotia, Newfoundland, Prince Edward Island.
Figure 1 provides the proportion of patients with completed tests at the SPC assessment. When referrals were from the organization’s own ED, most tests were already completed. There was more variation for family physician and other specialist referrals. MRI, 24-hour heart rate monitoring, and echocardiogram were the least likely to be completed at the time of SPC assessments.
Figure 1: Percentage of patients having already completed a test by the time they arrive at the stroke prevention clinic, by referring source.
Duration of waiting times for test results is as follows: for CT, 77.8% within 24 hours and 11.1% between 1 and 7 days; for MRI, 43.8% within 24 hours and 31.3% between 1 and 7 days; for Doppler ultrasound 51.6% within 24 hours and 25.8% between 1 and 7 days; for echocardiogram 15.2% within 24 hours while 36.4% between 1 and 7 days; for Holter 30.3% between 1 and 7 days and 33.3% between 8 and 14 days; for bloodwork 45.5% within 24 hours and 45.5% between 1 and 7 days (Figure 2).
Figure 2: Time required for results to become available.
Virtual Care and Impact of COVID-19 Pandemic on TIA Stroke Patient Management
Prior to the COVID-19 pandemic, 13 (39.4%) of the 33 SPCs were providing virtual care on a limited basis. During the pandemic, all but 4 (12.1%) SPCs were providing virtual care, with 60.6% providing virtual care for more than half of their patients; 15.2% of the clinics operated only virtual clinics. Among the 29 clinics that provided virtual care to patients during the pandemic, 20.7% of the clinics provided virtual care to all patients, 41.4% provided virtual care for low to medium-risk patients, and 31.0% provided virtual care for follow-ups and new consultations (Table 4). A large percentage (72.7%) of the 33 clinics plan to provide virtual care post-COVID-19 pandemic (Figure 3).
Figure 3: Percent of clinics using virtual care.
Table 4: Extent of virtual care
Five clinics did not answer questions related to COVID-19 and were excluded from the denominator.
Figure 4 shows the impact of the COVID-19 pandemic on SPCs patient management. Compared to pre-COVID period, the most impacted services by the COVID-19 were the number of referrals from family physicians (declined for 36.4% clinics), wait times once a patient was referred to the clinic (increased for 36.4% clinics), number of patients having already completed bloodwork prior to arriving for their appointment (declined for 27.3% clinics), number of referrals from the clinic’s own ED (increased for 24.2% clinics), and proportion of true TIA patients versus mimics (increased for 24.2% clinics).
Figure 4: Impact of COVID-19 pandemic on clinic services (from pre-COVID to COVID period). a. Hours the clinic is physically open/week. b. Referrals from own emergency department. c. Referrals from other emergency departments. d. Referrals from family physicians. e. Referrals from other specialists. f. Wait times for an appointment once a patient is referred to clinic. g. Proportion of true TIA patients versus mimics. h. Patients having completed imaging prior to appointment. i. Patients having completed bloodwork prior to appointment. j. Patients already taking prevention drugs (e.g. anticoagulants and antiplatelet) prior to appointment. k. If patients need medical imaging done, time it takes to have the results. l. If patients need bloodwork done after arriving, time it takes to have the results. *One clinic did not answer items d, h, i, k, l; two clinics did not answer items c, e, g, j.
Discussion
Summary of Main Findings
We found substantial variation among SPCs in management practices such as hours of operation, duration of wait times to see high-risk patients, capacity for add-on urgent patients, tests completed prior to SPC appointment, time for results, and referrals of patients with incorrect diagnosis of TIA. Although time is essential for TIA patient management, most high-risk patients are not seen by the clinics within 2 days, the highest risk period. To compensate for this, most are not seen routinely in the ED by neurology either. Thus, many of the high-risk patients are not being seen quickly enough within 48 hours to meet existing recommendations. SPCs also reported that on average about half of the patients are incorrectly diagnosed and referred to the clinics; thus, impacting wait times for true TIA patients. Wait times for testing are often short except for echocardiogram and Holter. There is a significant variation in the duration of Holter monitoring.
We found a substantial impact of COVID-19 on clinic routines. There was a substantial decline in referrals to SPCs, especially from family physicians, increase in wait times for appointments and test results, and a decline in TIA mimic referrals. While a small portion of SPCs were making use of virtual care prior to COVID-19, a significant portion of the clinics are planning to use virtual care post-COVID-19.
Interpretations
TIA is a medical emergency that requires urgent management to prevent subsequent stroke. The risk of stroke is greatest in the first few days after a TIA with stroke rates as high as 8% within the first 2 days of TIA. Reference Giles and Rothwell2 Hence, there is need for urgent intervention depending on stroke risk level. Admission of all patients is an inefficient use of health resources. Reference Rizos, Ringleb, Huttner, Köhrmann and Jüttler16 Our data show that a significant percentage of SPCs are open 3 days or fewer per week with a quarter only open 2 days or less per week. We suspect that the clinics opening three or fewer days per week are situated in low-populated areas, such as remote or rural areas. Our results show clinics open 3 or fewer days a week are unable to see most of their high-risk patients within 48 hours. A possible explanation for this is that there might be a high volume of referrals to the clinics but not enough resources to open the clinics more often. About half of the clinics open 4 or more days per week are also unable to see most of their high-risk patients within 48 hours.
Referrals to SPCs are initiated from EDs, family physicians, or specialists. Referrals should be triaged based on patient’s risk of subsequent stroke. Advanced clinical prediction rules, such as the Canadian TIA Score, have been developed and validated for identifying low, medium, and high-risk patients. Reference Perry, Sharma and Sivilotti17,Reference Perry, Sivilotti and Émond18 Other clinical prediction rules have been developed and validated for identifying low-risk and high-risk patients including the ABCD Reference Giles and Rothwell2 and ABCD2i Scores. Reference Johnston, Rothwell and Nguyen-Huynh19–Reference Abdulaziz, Perry and Yadav21 However, there are some limitations with clinical prediction rules such as accuracy and non-availability of all the factors during the referral.
In Canada, national guideline recommendations have been developed by Heart and Stroke Foundation of Canada for the management and treatment of TIA patients. Reference Gladstone, Lindsay and Douketis12 All patients suspected with TIA should complete neurologic and cardiac examination including imaging, bloodwork, electrocardiogram, echocardiogram for select patients and, if no etiology found extended cardiac monitoring.Reference Coutts22–Reference Mair and Wardlaw24 TIA patients require optimization of antithrombotic agents (anticoagulants or antiplatelets) and prompt carotid revascularization for symptomatic carotid artery stenosis if present.Reference Coutts22 The Canadian Stroke Best Practice Recommendations are not always followed, for varying and possibly valid reasons, leading to variations of practice. Not adhering to the guidelines might be out of the control of SPC personnel such as competing priorities.
Having tests completed prior to the SPC appointment could expedite diagnosis and treatment. Our data show investigations were often done prior to SPC assessment for referrals from the organization’s own ED. There was less variation among the organization’s own EDs while there was a wide variation among family physicians and specialists. This shows that there are likely better protocols in place for the organization’s own ED while there is less coordination for inter-organizational referrals. Standardization of community referrals may help decrease variability. However, we acknowledge that it would be challenging to reach and impose standardized referrals in some jurisdictions. In addition, unless a solution is found and implemented to reduce TIA mimics, diagnostic imaging departments could be swamped with test requests if all tests were to be completed for all patients prior to SPC appointments. At the clinics, there is also variation on when tests are ordered based on the structure of the clinic. In some clinics, patients are first seen by a doctor who risk stratifies the patients prior to ordering tests with priority given to high-risk patients while in some clinics tests are ordered first, irrespective of patient risk level, prior to seeing a medical doctor. Hence, the variation in test wait times shown by our results could be due to the structure of the clinics.
Misdiagnosis of TIA is common with an estimated 50% of all TIA patients being stroke mimics. Reference Fitzpatrick, Gocan and Wang25–Reference Prabhakaran, Silver, Warrior, McClenathan and Lee27 With such high misdiagnosis rates, there is potential for delay in care for patients at high risk of stroke. However, TIA is a complex medical emergency with multiple risk factors making it challenging to diagnose, especially in resource- and time-limited settings. In addition, reliability of reporting the events experienced by the patients may vary, leading to subjective diagnosis, even among stroke neurologists. Reference Castle, Mlynash and Lee28–Reference Kraaueveld, Gun, Schouten and Staal30 Furthermore, TIA patients with mimics still require timely diagnosis and management. Derivation and guidelines on use of clinical prediction rules to identify stroke mimics could potentially allow for faster assessments for higher risk patients. Several prediction rules for the diagnosis of TIA/stroke mimics (e.g., DOT score, Dawson score) have been derived but not adequately validated. Reference Stanciu, Banciu and Sadighi31–Reference Dawson, Lamb and Quinn33
A significant percentage of the SPCs turned to virtual care during the COVID-19 pandemic. Although most clinics did not use virtual care prior to the pandemic, most clinics are planning to continue using virtual care post-COVID-19. It is likely that the virtual care has helped reduce a load on the SPCs and opened capacity for high-risk patients.
Comparison with Previous Literature
Given the urgency and risk of TIA patients, SPCs need to be accessible. Similarly, all clinics should have a triaging system. Two studies, by Wasserman et al and Martinez-Martinez et al, have shown stroke risk reduction with risk stratification and referrals to SPCs. Reference Martínez-Martínez, Martínez-Sánchez and Fuentes10,Reference Wasserman, Perry and Dowlatshahi34 The study by Martinez-Martinez and colleagues reported a median wait time of 1.5 days to SPCs which is shorter in duration than shown by our study. Although the study by Wasserman and colleagues reported that most patients were seen within 48 hours in the ED, they did not report the time it takes from ED to SPC appointment. Wasserman and colleagues showed that neurology was consulted in the ED for only 5% of patients compared to more than 32% reported by our study.
Our results show that the COVID-19 pandemic had a significant impact on TIA patient management. During the pandemic, the proportion of referrals from family physicians declined while the referrals increased from EDs. This was likely due to family physician clinics being closed. It could have also been due to patients not seeking medical help unless they deemed it an emergency. Other studies have found similar findings in reduction of patients during the pandemic period with similar conclusions. Reference Mann, Swedien and Hansen35–Reference Rodrigues, Grunho and Rachão41 Also in agreement with our study, a study by Dowlatshahi and colleagues showed a drop in presentation rates to a comprehensive stroke centre in Ottawa, Canada at the beginning of the pandemic. Reference Dowlatshahi, Stotts and Bourgoin42 Similarly, a study by D’Anna showed a decline of referrals to North West London, UK SPC clinics during the COVID-19 pandemic. Reference D’Anna, Sheikh and Bathula43 Our study found the proportion of true TIA patients versus mimics increased during the pandemic. Patients with milder stroke symptoms may have intentionally avoided seeking medical care during the COVID-19 pandemic while the patients with more severe cases, such as symptoms due to large vessel occlusion, sought medical help as these more severe symptoms are less likely to be ignored by patients or family members. Reference Siegler, Heslin, Thau, Smith and Jovin38,Reference D’Anna, Sheikh and Bathula43 Another study by D’Anna also observed this marked decrease in mimic diagnoses during the pandemic. Reference D’Anna, Brown and Oishi40
Study Limitations
Despite our efforts to obtain high-quality data, the quality of responses may have been affected by the COVID-19 pandemic. Given that this was a self-reported questionnaire, respondents may have not answered all the questions accurately. Another limitation is that we had a limited set of characteristics to test for non-response bias.
Clinical Implications
Despite the need for urgent assessment and management, there is delay in seeing some high-risk TIA patients within 48 hours. About a quarter of SPCs are open 2 or fewer days a week, leaving the possibility of some high-risk patients having to wait more than the 48-hour critical time for TIA patients depending on referral volumes. A significant percentage of patients referred to SPCs are stroke mimics taking resources from true TIA patients. Using a clinical diagnostic tool, such as the DOT score, after its external validation, could help minimize misdiagnosis of TIA. Stroke prediction tools, such as the Canadian TIA Score, are also important to prioritize high-risk patients.
Research Implications
Our study has shown several gaps in knowledge of SPCs and how they function and interact with other stakeholders. We have also shown the impact of the COVID-19 pandemic on patient management. We have found that referrals from an SPC’s own ED are better managed than from other EDs, family physicians, or specialists. Understanding the reasons behind these discrepancies could help improve referral systems. We have also shown that several clinics are open 3 or fewer days a week with a quarter being open 2 days or fewer each week and not being able to see most high-risk patients within 48 hours. Half of the clinics that do open 4 or more days are also struggling with seeing most of their high-risk patients. The reasons behind this deficiency and the outcomes of patients seeking medical help at these clinics need to be investigated and compared to other clinics. We suspect there would be more stroke outcomes for high-risk patients at these clinics having to wait longer than 48 hours due to limited clinic days. Clinic structure regarding testing first prior to assessment versus assessment first followed by testing on patient outcomes also needs to be investigated.
Conclusion
TIA is a serious condition that requires urgent care, depending on the risk level. Outpatient SPCs have been setup to provide more efficient and effective care. Although there are guidelines on the management of TIA patients, they are not fully implemented leading to variations of practice. We suggest that SPCs investigate delays and attempt to see high-risk patients within 48 hours. We also suggest that better systems are put in place between SPCs and referring sources, especially family physicians, so that patients are risk classified and some test are completed prior to their appointments, as appropriate without causing further delays.
Acknowledgements
The authors thank Dr Dylan Blacquiere, Dr Grant Stotts, Dr Michel Shamy, and Dr Vivien Parker (Neurologists) at The Ottawa Hospital who allowed us to interview them and for their feedback, Dr Christophe Fehlmann who helped with the initial translations of the letters and survey questionnaire into the French language, Angela Marcantonio for her assistance throughout the project, and those individuals who completed our questionnaire.
Funding
This study was funded by a peer-reviewed CIHR Foundation Grant held by Dr Jeffrey Perry.
Conflict of Interest
Dr Jeffrey Perry holds a peer-reviewed mid-career salary support award from the Heart and Stroke Foundation of Ontario. None of the other authors have any conflict of interest or information to disclose in relation to this study.
Statement of Authorship
KEA, JJP, and MT designed the study. KEA conducted interviews, executed the surveys, conducted analyses, and wrote the draft manuscript. DD, IGS, and GAW provided input into the study design. All authors contributed to the final manuscript review.
Open access
