INTRODUCTION
Transient ischemic attacks (TIAs) warn of increased risk of stroke, other vascular events, and death.Reference Furie, Kasner and Adams 1 The risk of recurrent stroke after TIA is 5%–20% within 90 days, with about half of all recurrent strokes occurring in the first two days following initial symptom onset.Reference Furie, Kasner and Adams 1 - Reference Gladstone, Kapral and Fang 4 Early assessment in a dedicated neurovascular clinic and immediate initiation of medical and surgical therapy have been shown to reduce the risk of major stroke after TIA by up to 80%.Reference Rothwell, Giles and Chandratheva 5
Patients with TIA often present to the emergency department (ED) for initial evaluation. The current Canadian Best Practice Recommendations for stroke (and TIA) stipulate that any patient who presents to the ED with a suspected recent stroke or TIA undergo assessment within 24 hours that includes brain imaging, noninvasive vascular imaging (for suspected carotid territory events), electrocardiogram (ECG), initiation or modification of secondary prevention therapy (antihypertensive, lipid-lowering, glucose-lowering, or anticoagulation therapy), and referral to a designated stroke prevention clinic for risk factor assessment.Reference Coutts, Wein and Lindsay 6 A nation-wide audit of stroke and TIA care in Canada (Quality of Stroke Care 2011) showed that many patients with TIA do not receive timely investigations and treatment. 7 We were interested in obtaining a snapshot of the care processes for patients who received a diagnosis of TIA in the EDs of the Capital District Health Authority in Nova Scotia, and to determine the impact of being seen in a dedicated neurovascular clinic on patient outcomes.
METHODS
Study design and subjects
This study was approved by the health system Research Ethics Board. This was a historical cohort study of all patients discharged from the EDs during the 2-year period of January 1, 2011 through December 31, 2012 with a diagnosis of TIA as identified using our ED information system. This regional health authority system serves a population of approximately 400,000, and its four EDs have a combined annual census of approximately 160,000 patients. The ED information system contains the diagnosis at discharge recorded by the attending ED physician for all ED visits in the system. We followed the health records of all identified patients who met study criteria for a 90-day period following the index TIA event. TIA was defined as a sudden, focal neurological deficit of presumed vascular origin lasting <24 hours. Data were not used if a medical record indicated the patient did not want their information used for research purposes. The Division of Neurology, based in the Queen Elizabeth II Health Sciences Centre (the largest hospital in Nova Scotia) provides a 24/7 emergency neurology consultation service, either in person or by telephone. A daily (Monday through Friday) neurovascular clinic, operated by a stroke neurologist and neurology nurse practitioner, evaluates patients referred by emergency (and other) physicians. The standard of care is for patients with suspected TIA or stroke to be seen in the ED by the Neurology services during the same or next day. Our policy is to initiate investigations and start or modify treatment in the ED, and discharge the patient home with arrangements for fasting blood tests and outpatient follow-up in the neurovascular clinic. Patients may be admitted to the Acute Stroke Inpatient Unit if judged necessary by the on-call neurologist, for example, if there has been more than one TIA over the preceding 24 hours, or if there is comorbidity requiring hospitalization, or residence remote from the hospital, or probable or definite severe symptomatic carotid stenosis.
Data collection
An initial search of the ED information system identified patients diagnosed with TIA or transient ischemic attack within our time frame of interest. One person conducted the chart abstraction without a priori knowledge of the content of the records. Source documents included emergency physician notes, paramedic notes, registered nurse notes, consultation notes, outpatient clinic notes, inpatient discharge notes, and death summaries. The health system databases do not include information related to clinical visits outside of affiliated facilities, such as family doctors’ offices. Abstracted data included patient demographics, medical history, care provided throughout the hospital stay, inpatient and outpatient investigations, medication prescriptions, discharge status, hospital readmission within 90 days of the index TIA, and death within 90 days of the index TIA if recorded in the health system clinical databases. Data for clinical variables (history of smoking, hypertension, etc.) not recorded in any of the clinical information were coded as negative.
Outcomes
Process outcomes were the proportion of patients referred to the neurovascular clinic, the proportion of TIA patients seen in the clinic, the proportion of TIA patients that received guide-line recommended assessment (brain CT or magnetic resonance imaging [MRI], carotid imaging, fasting blood tests, ECG, and international normalized ratio [INR]) for those seen in ED only and those seen in the ED and the neurovascular clinic, and the proportion of patients that were on antiplatelet, anticoagulation, antihypertensive, lipid-lowering, or glucose-lowering therapy after being seen in the ED and after follow-up in the neurovascular clinic. Our main clinical outcome was the comparison of readmission rates within 90 days of index TIA for ischemic stroke, myocardial infarction, or vascular death for patients seen in the neurovascular clinic versus those who were not.
Analysis
Population characteristics were summarized as means with standard deviation for continuous data and frequencies with percentage for categorical data. For process outcomes (investigations, treatments), descriptive analyses were conducted. Kaplan-Meier survival analysis and log-rank testing were used to compare clinical outcomes of patients seen in the neurovascular clinic with those who were not. Adjusted Cox proportional hazard models were used to compare time to readmission or death from ischemic stroke or myocardial infarction in those seen in the neurovascular clinic compared to those not who were not. The proportionality assumption was not violated in any of the models considered. Parallel curves resulted in the analysis of the survival function versus time and the log (-log survival) versus log of the survival time. A propensity analysis was used to account for potential confounding and selection biases. The propensity for patients seen in the neurovascular clinic was developed using multivariable logistic regression. Variables included in this derivation were age, sex, location, risk factors, medication use, and symptoms. A greedy matching procedure matched propensity scores to five decimals initially.Reference Parsons 8 This match continued reducing the number of decimals to one. If no match existed at one decimal place, the patient was not used for analysis. Statistical analysis was performed using SAS version 9.4 (Cary, NC).
RESULTS
An initial search of the ED information system identified 697 patients with a discharge diagnosis of TIA. After review of the records, 11 patients were identified that were mislabeled in the information system and did not in fact have an ED discharge diagnosis of TIA. Of the remaining 686 patients, 33% were seen by Neurology in the ED, 51% were referred to the neurovascular clinic, and 36% were actually seen in the clinic (Figure 1). Table 1 compares the clinical characteristics of patients seen in the ED (and not the neurovascular clinic) with those patients seen in the ED and the neurovascular clinic. After propensity matching, there were 215 patients who were seen in the ED only and 215 patients seen in both the ED and neurovascular clinic with statistically similar baseline characteristics (Table 2). Median time from TIA event to triage was 2 hours (1.1–6.6 hours interquartile range [IQR]). The ED discharge diagnosis of TIA agreed with the final diagnosis after being seen in the neurovascular clinic in 86.4% of cases. Of the patients who received an alternative diagnosis from the neurovascular clinic, migraine (24%), syncope/presyncope (7%), seizure (6%), and brain neoplasm (4%) were the most common alternative diagnoses. Patients who did not receive a final diagnosis of TIA after being seen in the neurovascular clinic were included in all subsequent analyses because, at the time of their presentation, they were being managed as cases of TIA. Median time from TIA to time seen in the neurovacular clinic was 32 days (10–60 days IQR) for all patients and 22 days (9–52 days IQR) for those not seen by Neurology in the ED.
Care flow chart showing consults requested in the emergency department, follow-up referrals, and patients seen in the neurovascular clinic as a percentage of the total study population (N=686). Consults in ED and discharge referrals were not mutually exclusive, meaning patients who were seen by Neurology in the ED may also have received a discharge referral to and been seen in the neurovascular clinic. ED=emergency department.
Baseline characteristics of those not seen in neurovascular clinic (N=440) versus those seen in clinic (N=246) before propensity matching
* Multiple symptoms per patient.
† Dizziness may be the same as vertigo.
ASA=acetylsalicylic acid; TIA=transient ischemic attack.
Comparison of baseline characteristics between those not seen in neurovascular clinic (N=215) and those seen in clinic (N=215) after propensity matching
* Multiple symptoms per patient.
† Dizziness may be the same as vertigo.
ASA=acetylsalicylic acid; ED=emergency department; TIA=transient ischemic attack.
In the propensity matched groups, there was no difference in the rates of ECG or INR obtainment in patients seen in the ED only compared to those seen in the ED and neurovascular clinic within 24 hours, 7 days, 30 days, and 90 days of triage (Table 3). There was increased rates of CT scanning in those seen in the neurovascular clinic within 7 days (95.4%) compared to those not seen in the neurovascular clinic (90.2%), and this trend continued at each time point thereafter (Table 3). There were increased rates of vascular imaging with Doppler ultrasound (US) in those seen in the neurovascular clinic versus those who were not seen at 24 hours, 7 days, 30 days, and 90 days following triage (Table 3). There was no difference in rates of magnetic resonance angiography and computed tomography angiography use in patients seen in the neurovascular clinic versus those who were not seen (Table 3). Fasting blood work was obtained in 60.5% of patients seen in the ED and 77.7% of patients who were seen in both the ED and the neurovascular clinic. Among all patients (N=686), ECG was obtained in 87.5%, INR in 59.5%, CT in 92.9%, MRI in 18.7%, Doppler US in 56.6%, magnetic resonance angiography in 16.8%, computed tomography angiography in 7.6%, and fasting blood work in 61.6% within 90 days of triage in the ED.
Investigations in those not seen in neurovascular clinic (N=215) versus those seen in clinic (N=215) after propensity matching
* Patient may have received brain imaging by multiple modalities.
† Patient may have received carotid imaging by multiple modalities; may include non-carotid territory vascular events.
In the propensity matched groups, 86.3% of patients seen in the ED (and not neurovascular clinic) were on antithrombotic therapy (59% received new or modified antithrombotic medication), compared to 94% of patients seen in both the ED and the neurovascular clinic (76% received new or modified antithrombotic medication; Table 4). Of patients with atrial fibrillation (newly or previously diagnosed), 47.6% were on anticoagulant therapy after being seen in the ED, and this increased to 85.7% in those seen in the neurovascular clinic (Table 4). Of the 19 patients who were candidates for carotid endarterectomy, 57.9% received carotid endarterectomy within 2 weeks after the index TIA.
Treatments in those not seen in neurovascular clinic versus those seen in clinic after propensity matching
Of the 686 patients seen in the ED, 4.2% experienced subsequent ischemic stroke, myocardial ischemia, or vascular death within 90 days. After adjustment for age, sex, presence of vascular disease risk factor, and stroke symptoms, the risk of readmission for stroke, myocardial infarction, or vascular death was lower for those seen in the neurovascular clinic (adjusted hazard ratio [HR] 0.28; 95% confidence interval [CI] 0.08–0.99, p=0.048). The risk of readmission or death from ischemic stroke or myocardial infarction at 90 days in the propensity matched cohort was significantly lower in patients who were seen in the neurovascular clinic compared to patients not seen (Table 5 and Figure 2; HR 0.18, 95% CI 0.04–0.8, p=0.02). In the matched cohort, two patients were denied neurovascular clinic services because they were palliative. Both of these patients remained event-free for the study period.
Fewer patients readmitted with ischemic stroke or myocardial infarction, or died from ischemic stroke or myocardial infarction, within 90 days after index TIA when seen in the neurovascular clinic (N=215) compared to those who were not seen in the neurovascular clinic (N=215) after propensity matching.
Outcomes of patients not seen in neurovascular clinic (N=215) and seen in clinic (N=215) after propensity matching
ED=emergency department.
DISCUSSION
The majority of patients diagnosed with TIA in our cohort were treated with antithrombotic agents in the ED and investigated with brain imaging and ECG within 24 hours of triage, but only about half of the patients were referred to the neurovascular clinic. The number of patients who experienced subsequent ischemic stroke, myocardial infarction, or vascular death within 90 days was low, at 4.2%. Although management and outcomes for all patients seen with TIA in the ED were generally favourable, those seen in the neurovascular clinic had lower rates of readmission for stroke, myocardial infarction, or vascular death (adjusted HR 0.28; 95% CI 0.08–0.99, p=0.048).
There is growing evidence that access to outpatient clinics aimed at risk factor management improves rates of investigations, treatments, and outcomes of patients with TIA.Reference Rothwell, Giles and Chandratheva 5 , Reference Webster, Saposnik and Kapral 9 , Reference Dutta, Bowen and Foy 10 Using data from 12 specialized stroke centers in Ontario between 2003 and 2008, Webster et al. (2011) found that referral of patients with ischemic stroke or TIA to a designated neurovascular clinic was associated with a 29% reduction in 1-year mortality (HR, 0.71; 95% CI, 0.65–0.84). Similar to our study, Webster et al. (2011) found that only 47% of the study population was referred to the neurovascular clinic in follow-up, indicating that outpatient neurovascular clinics may be underutilized.Reference Webster, Saposnik and Kapral 9 Although we did not specifically look at reasons for non-referral, some reasons may include referral to a family doctor or community neurologist, failure to document a referral, and lack of awareness of the neurovascular clinic by the ED physician. Reasons for not being seen after referral may include patient noncompliance and miscommunication. The mechanisms underlying the improved outcomes seen with care provided by the neurovascular clinic are likely multifactorial and may include more timely investigations, initiation of appropriate medical therapy for secondary stroke prevention, and access to other professionals with expertise in stroke care. This is supported by our findings that rates of antithrombotic use, lipid-lowering agents, anticoagulants in those with atrial fibrillation, fasting blood tests, CT scan rates, and Doppler US were increased in patients seen in the neurovascular clinic compared to those not seen.
Interestingly, there was morbidity/mortality benefit in the first 7 days following the index TIA among patients who were seen in the neurovascular clinic compared to those who were not, while the median time to being seen in the neurovascular clinic was 32 days. This suggests that there were systemic differences between the two groups. To limit such differences, we used propensity matching, which generated subpopulations with statistically similar baseline characteristics for those seen in ED-only patients, and those seen in both the ED and the neurovascular clinic. A protective effect of attending at the neurovascular clinic was still apparent after this adjustment. The median time to being seen in the neurovascular clinic for patients not seen by Neurology in the ED was 22 days (9–52 days IQR) compared to 32 days (10–60 days IQR) for those not seen by Neurology in the ED. This reflects an effort by practitioners in the neurovascular clinic to see patients not assessed by Neurology in the ED sooner. It is also standard practice for practitioners in the neurovascular clinic to call in prescriptions for secondary stroke medications and request imaging studies and blood work for these patients before being seen in the clinic. Together, these factors may help explain the early protective effect for patients who would later go on to be seen in the neurovascular clinic for follow-up.
It is well known that timely investigations and use of medications for secondary stroke risk factor management improve stroke outcomes. Our rates of CT (88.3%) and ECG (86.3%) investigations within 24 hours were similar to those reported by Perry et al. (2013) and compare favourably to those reported in the Quality of Stroke Care in Canada 2011 report. 7 , Reference Perry, Kerr and Symington 11 Rates of antithrombotic, antihypertensive, and lipid-lowering agents were similar in this study compared to the Quality of Stroke Care 2011 report and other studies. 7 , Reference Perry, Kerr and Symington 11 Vascular imaging was obtained in approximately 55% of patients. Rates of vascular imaging in other studies are also low, ranging from 15%–42%.Reference Gladstone, Kapral and Fang 4 , Reference Chang, Holroyd and Kochanski 12 , Reference Wu, Manns and Hill 13 Approximately 15% of all TIAs are due to large artery atherosclerosis.Reference Kolominsky-Rabas, Weber and Gefeller 14 , Reference Palm, Urbanek and Wolf 15 Timely vascular imaging is needed to identify patients with 50%–99% carotid stenosis so that they can receive carotid endarterectomy surgery within 2 weeks of symptom onset, the time frame in which the greatest benefit is seen.Reference Blacquiere, Sharma and Jetty 16 , Reference Rothwell, Eliasziw and Gutnikov 17 New to the updated Canadian Best Recommendations (2014) is the recommendation to use CT angiography at the time of brain CT as the first-line vascular imaging method.Reference Coutts, Wein and Lindsay 6 For patients in whom a cardioembolic mechanism is suspected, but the ECG does not show atrial fibrillation, prolonged ECG monitoring is recommended in selected patients (older patients with recent embolic stroke of undetermined source who are potential candidates for anticoagulation therapy) for the detection of paroxysmal atrial fibrillation.Reference Coutts, Wein and Lindsay 6 , Reference Gladstone, Spring and Dorian 18 Our results indicate the need for increased investigation into stroke mechanism, specifically carotid imaging, in patients presenting with TIA.
Distinguishing between TIA and ischemic stroke in the acute setting of the ED can be challenging. The traditional definition of TIA is a sudden, focal neurological deficit of presumed vascular origin lasting <24 hours. Decisions regarding thrombolytic therapy need to be made within minutes of presentation to the ED. A 24-hour definition for TIA encourages delay in administration of acute thrombolytic therapy. With advances in MRI, up to three-quarters of TIAs defined by the traditional time-based definition have been found to have evidence of tissue damage.Reference Nah, Kwon and Kang 19 This has sparked a shift towards use of a tissue-based definition of TIA. However, timely access to MRI and expert interpretation is not always available, and MRI is not cost-effective compared to CT.Reference Wardlaw, Brazzelli and Miranda 20 It also remains to be seen whether use of plasminogen activator in patients with recent prior clinical symptoms lasting <24 hours associated with small MRI lesions is safe and effective. For these reasons, a tissue-based definition of TIA has yet to be widely endorsed. Furthermore, from a stroke prevention perspective, the risk of major stroke following a minor event is the same for TIA and mild stroke.Reference Coull, Lovett and Rothwell 21 We found that 4.2% patients diagnosed as TIA in the ED were found to have ischemic culprit lesions identified on CT or MRI at the time of presentation or thereafter. Including or excluding these patients in analyses did not change our results. The majority of TIAs in our study lasted <1 hour, and this is true in other studies.Reference Perry, Kerr and Symington 11 , Reference Giles and Rothwell 22 A more practical definition of TIA in the ED seems to be a sudden-onset, focal neurological deficit lasting less than 1 hour.
Traditionally, diagnostic accuracy of TIA and minor stroke in the ED has been somewhat low, ranging from approximately 50%–64%.Reference Moeller, Kurniawan and Gubitz 23 - Reference Ghia, Thomas and Cordato 25 In our study, the discharge diagnosis of TIA agreed with the diagnosis after being seen in the neurovascular clinic in 86.4% of patients. Including or excluding these patients did not change our results. The most frequent alternative TIA mimics were migraine, syncope/presyncope, seizure, and brain neoplasm. It is possible that consultation of Neurology in the ED for patients with unclear presentation contributed to high rates of diagnostic accuracy in the current study.
Limitations
The health system databases do not include information pertaining to clinical visits outside of this health system (for example, independent family doctors’ offices). It is therefore possible that our main clinical outcome of readmission for stroke, myocardial infarction, or death from stroke or myocardial infarction may have underestimated the actual recurrence rate of stroke, myocardial infarction, or death from stroke or myocardial infarction. Unfortunately, we are not able to estimate how many patients may have presented to a facility outside of our health system. However, we suspect that it would be low, because it is standard practice for the family doctor to send any patients with symptoms of stroke or myocardial infarction directly to the nearest ED, which would presumably be the same ED to which the patient presented initially.
It is possible that a proportion of patients were not referred to the neurovascular clinic because they were palliative or had severe dementia and instead were referred to their family physician for follow-up care. This may have biased the rates of stroke recurrence, myocardial infarction, and death to be higher in those not seen in the neurovascular clinic. In the matched cohort, we identified only two patients who were denied services in the neurovascular clinic because they were palliative. Both of these patients remained event free. However, palliative status for all patients may not have been documented in the clinical databases. Furthermore, we did not analyse presence of dementia in our patient cohort.
A potential confounding factor in our study was that patients who were referred to the neurovascular clinic could have had an adverse event that prevented them from being seen in follow-up. Arguing against this is accrued data (not shown) that rates of adverse events for those referred (but not seen in the clinic) were not different than rates of adverse events for patients seen in the clinic.
Finally, because this was a retrospective chart review, it is not possible to discern whether all patient risk factors, neurological deficits, medications, or the time of TIA event were fully and accurately recorded. By design, we attempted to minimize abstraction bias, with one person conducting the chart abstraction without a priori knowledge of the content of the records, and outcome data were based on objective documentation.
CONCLUSION
Our results support the use of outpatient clinics aimed at management of vascular disease risk factors as one way to improve outcomes of patients with TIA. Use of CT, ECG, and antiplatelet therapy in the ED was high, but there remains a need for more rapid vascular imaging, as well as more aggressive medical management, including oral anticoagulant therapy for those with evidence of atrial fibrillation in the ED after a TIA, in an effort to prevent disabling stroke or death.
Competing interests: None declared.
