A male client in the emergency department has a suspected neurologic disorder

Neurol Clin Pract. 2020 Apr; 10(2): 106–114.

Abstract

Objective

To assess the risk of subsequent stroke among older patients discharged from an emergency department (ED) without a diagnosis of TIA or stroke.

Methods

Using electronic health record data from a large urban, university hospital and a community-based hospital, we analyzed patients aged 60–89 years discharged to home from the ED without an International Statistical Classification of Diseases and Related Health Problems, 9th or 10th Revision diagnosis of TIA or stroke. Based on the presence/absence of a head CT and the presence/absence of a chief complaint suggestive of TIA or stroke (“symptoms”) during the index ED visit, we created 4 mutually exclusive groups (group 1, reference: head CT no, symptoms no; group 2: head CT no, symptoms yes; group 3: head CT yes, symptoms no; and group 4: head CT yes, symptoms yes). We calculated rates of stroke in the 30, 90, and 365 days after the index visit and used multivariable logistic regression to estimate odds ratios (ORs) for subsequent stroke.

Results

Among 35,622 patients (mean age 70 years, 59% women, and 16% African American), unadjusted rates of stroke in 365 days were as follows: group 4: 2.5%; group 3: 1.1%; group 2: 0.69%; and group 1: 0.54%. The adjusted OR for stroke was 3.30 (95% confidence interval [CI], 1.61–6.76) in group 4, 1.56 (95% CI, 1.16–2.09) in group 3, and 0.61 (95% CI, 0.22–1.67) in group 2.

Conclusions

Among patients discharged from the ED without a diagnosis of TIA or stroke, the occurrence of a head CT and/or specific neurologic symptoms established a clinically meaningful risk gradient for subsequent stroke.

As many as 75,000 patients per year in the Unites States have an emergency department (ED) visit at which a TIA or stroke goes unrecognized.1,–4 Such ED visits are missed opportunities to initiate timely preventive strategies to reduce the risk of subsequent stroke, disability, and death since same-day evaluation, diagnosis, and treatment of TIA can lower the relative risk of stroke by 80% compared with delayed or deferred evaluation.5,–7

Accurate diagnosis of TIA is challenging because there is no diagnostic “test” for TIA. A recent meta-analysis of the accuracy of diagnosis of TIA, ischemic stroke, and subarachnoid hemorrhage among patients in the ED found a false-negative rate of 8.7%, with higher rates among TIA and minor stroke.4 The consequence of a missed TIA diagnosis is clear: patients are exposed to an increased and sustained risk of stroke if preventive treatments are not initiated. Improved risk stratification methods may help identify patients in the ED who may have had an unrecognized TIA or stroke. We hypothesized that symptom characteristics may identify patients with elevated risk of stroke after ED visits for neurologic complaints.

Methods

We performed a retrospective cohort study of electronic health record (EHR) data in an enterprise data warehouse for 2 hospitals and their EDs in the Chicago area: (1) a large urban, university hospital designated as a comprehensive stroke center by the Joint Commission and the American Heart Association/American Stroke Association and (2) a community hospital designated as a primary stroke center. The urban hospital (hospital 1) has approximately 88,000 ED visits and the community-based hospital (hospital 2) has approximately 25,000 ED visits, per year.

Study population

Patients aged 60–89 years who were discharged to home from the ED within 24 hours, without an International Statistical Classification of Diseases and Related Health Problems, 9th or 10th Revision (ICD-9/10) diagnosis of TIA or stroke or other cerebral/precerebral artery occlusion (433.xx, 434.xx, 435.xx, or 436.xx, I63.xx or G45.xx), were included. Index ED visits for the study were drawn from the years January 1, 2012, to December 31, 2015. For each patient, we determined the presence or absence during the ED visit of a head CT and/or any of these strings in the (physician-entered, free-text) ED chief complaint (“stroke-like symptoms”): slur, speech, aphasia, confuse, word, difficult, comprehen, weak, clumsy, clumsiness, droop, paralysis, move, moving, face, or facial (but not “facial injury”). These motor and speech symptoms were selected a priori based on their inclusion in the ABCD2 (age, blood pressure, clinical symptoms, duration, diabetes mellitus) score.8 We assessed for negation in multiple ways (e.g., using “no,” “not,” “no evidence of,” or “rule out”). We also searched for other neurologic symptoms such as those that might indicate head trauma or falls: “hit head,” “fell,” “fall,” “lost consciousness,” “LOC,” “passed out,” “concussion,” “head injury,” and “head trauma.” Occurrence of head CT was selected as a surrogate indicator of an ED physician's suspicion about a central neurologic diagnosis. Negation was found in only 1 case in the entire search of chief complaints. Search of terms related to trauma or falls revealed only 5 cases. In a sample of 20 randomly selected records, the retrieved chief complaint symptoms, demographics, and CT occurrence from the query matched exactly the data found in the patient records.

We thereby created 4 mutually exclusive groups of patients discharged from the ED: group 1 (head CT no, stroke-like symptoms no), group 2 (head CT no, stroke-like symptoms yes), group 3 (head CT yes, stroke-like symptoms no), and group 4 (head CT yes, stroke-like symptoms yes). To assess whether hospital admission mitigated stroke risk, we included an additional comparison group (group 5): patients aged 60–89 years who had both a head CT and symptoms as defined above in the ED and who were then admitted to the hospital (and did not have a discharge ICD-9/10 diagnosis of TIA or stroke for that hospital stay). If a patient had more than 1 qualifying index ED visit during the study years (across any of the 5 groups), we included only the first.

Covariates

We recorded demographics (age, sex, insurance status, race, and ethnicity), comorbidities based on ICD-9/10 codes (stroke, TIA, hypertension, diabetes, coronary artery disease, hyper/dyslipidemia, atrial fibrillation, carotid stenosis, seizures, and migraines), previous health care utilization (number of ED visits and number of hospitalizations in the preceding year), and hospital where the index ED visit took place. In each group, we also recorded the percentage of patients with a brain MRI during the index ED visit. In group 5, we also recorded the hospital service to which the patient was admitted.

Ascertainment of subsequent stroke, migraine, and seizures

Stroke outcome was based on ICD-9/10 diagnoses for ischemic stroke (433.x1 or 434.x1 or 436.xx, or I63.xx [except for I63.6x, central venous thrombosis]), during a subsequent ED, observation, or inpatient hospital visit. Ascertainment of stroke in the follow-up period was limited to the 2 index hospitals. In patients without any follow-up encounter at the 2 hospitals, we assumed no stroke outcome occurred. New diagnoses of migraine and seizure disorder were similarly ascertained at subsequent encounters.

Statistical analysis

Primary analyses

We present mean and SD for continuous variables and proportions for categorical data. In each of the 4 groups discharged to home from the ED, we calculated unadjusted rates of stroke in the ensuing 30-, 90-, and 365-day periods. We used logistic regression to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for subsequent stroke in a 365-day period, for each of the prespecified groups, in an unadjusted model and then a multivariable model to adjust for relevant covariates. The reference group was group 1. Identical steps were performed for nonstroke outcomes: (1) new migraine and (2) new seizure/epilepsy disorder.

Secondary analysis (additional comparison group)

To assess the effect of hospitalization vs discharge home after an ED visit among patients who had a head CT and stroke-like symptoms but no stroke/TIA diagnosis during that visit (groups 4 and 5), we performed multivariable logistic regression to calculate adjusted odds for the outcome, any stroke within 365 days after the index ED visit in group 5 vs group 4.

Nontruncated analyses

In sensitivity analyses, we did not truncate the stroke outcome measurement at 365 days postindex. For example, patients with an index ED visit in the first few months of the study period could have a stroke outcome up to 5 years later. We estimated unadjusted cumulative hazards of stroke for each of the 5 groups described above, limited only by the end date of the retrospective data extract (March 2017) using the Nelson-Aalen cumulative hazard risk estimation method. After confirming that analyses met the proportional hazards assumption, we then used Cox proportional hazards modeling for multivariable analysis to explore the replicability of the findings from the primary and secondary multivariable logistic regression analyses described above.

Standard protocol approvals, registrations, and patient consents

This study was approved by the Institutional Review Board of Northwestern University. Informed consent was waived for this retrospective data analysis.

Data availability

All data not presented in this article will be made available in a trusted data repository or shared at the request of other investigators for purposes of replicating procedures and results.

Results

Description of the cohort

The study included 36,301 patients with ED visits and without diagnosis of TIA or stroke at the index visit. Baseline characteristics by group are shown in table 1. The 4 groups discharged to home from the ED (35,622 patients, combined) had a mean age of 70 (SD 8) years; 59% were women, 16% African American, and 7% Hispanic. CT occurrence was more common in Medicare-insured patients (24.3% vs 17.6%, p < 0.001). The additional comparison group (group 5; N = 679) had a mean age of 75 (SD 9) years; 50% were women, 22% African American, and 7% Hispanic. Eleven percent of the patients in this group were admitted to the neurology service, whereas approximately 70% were admitted to medicine services (hospitalists or internal medicine), 10% to medicine subspecialties, and 7% to neurosurgery. At least 1 subsequent contact with this health care system (outpatient, ED, or inpatient visits for any reason) more than 90 days after the index ED visit occurred in 72% of patients, and 65% had at least 1 visit more than 365 days after the index ED visit. Group 4 had slightly higher contact rates (79% at more than 90 days and 71% at more than 365 days) compared with groups 1–3.

Table 1

Baseline characteristics of the cohort

On bivariate analysis of baseline characteristics (table 1), with group 1 as the reference, group 4 was older, had a greater proportion of women and Medicare insurance, and had higher proportions of previous TIA, previous stroke, hypertension, diabetes, carotid stenosis, seizures, and migraines. Compared with group 4, group 5 (the hospitalized group) had higher proportions of Medicare insurance, coronary artery disease, hyper/dyslipidemia, atrial fibrillation, seizures, and migraines. Group 5 was drawn disproportionately from hospital 1 (84% vs 67% among the other 4 groups, p < 0.001). Among all patients in the study, 1.7% had a brain MRI during the index ED visit (group 1: 0.2%, group 2: 1.5%, group 3: 4.0%, group 4: 12.4%, and group 5: 36.4%).

Unadjusted stroke rates in the follow-up period

Rates of stroke at 30, 90, and 365 days are shown in table 2. The 365-day rate of stroke ranged from 0.5% in group 1 to 2.5% in group 4. Compared with group 1, group 4 had a 12-fold higher rate of stroke at 30 days, a 5-fold higher rate of stroke at 90 days, and a 5-fold higher rate of stroke at 365 days. Figure 1 depicts ORs for stroke at 365 days after the index ED visit, based on unadjusted logistic regression, for the 4 groups discharged to home from the ED. In the additional comparison group (group 5), the unadjusted 365-day rate of stroke was 2.5%. There was no relationship between admitting service (medicine vs surgical vs neuroscience) and stroke risk in group 5 patients (p = 0.248). New atrial fibrillation (defined as new diagnosis on subsequent encounter after index ED encounter) was more common (14.9% vs 4.7%; p < 0.001) in those with stroke than without stroke in follow-up.

Table 2

Unadjusted rates of stroke after the ED visit

Unadjusted odds ratios for stroke

Unadjusted odds ratios (with 95% confidence interval bars) for stroke within 365 days, 6 after the index ED visit.

Multivariable analysis

Primary analyses

Table 3 shows the results of multivariable logistic regression modeling for stroke during the 365 days after the index ED visit among the 35,622 patients discharged to home from the ED. After controlling for covariates, the OR for stroke was 3.30 (95% CI, 1.61–6.76) in group 4, 1.56 (95% CI, 1.16–2.09) in group 3, and 0.61 (95% CI, 0.22–1.67) in group 2. Other factors independently associated with stroke within 365 days included age, sex (males), race (nonwhites), previous histories of hypertension, diabetes, coronary artery disease, atrial fibrillation, and seizures.

Table 3

ORs for stroke in the year after the ED visit among those discharged to home

For migraine outcome, group 4 patients were more likely to have migraine in unadjusted (1.4% vs group 1: 0.3%, group 2: 0.1%, group 3: 0.5%; p < 0.001) and adjusted analyses (OR 21.9; 95% CI, 3.7–128.6). For seizure outcome, group 4 patients were more likely to have migraine in unadjusted (2.0% vs group 1: 0.3%, group 2: 0.8%, group 3: 0.7%; p < 0.001) but not in adjusted analysis (OR 3.6; 95% CI, 0.4–36.3).

Secondary analysis (additional comparison group)

In the 1,033 patients with a head CT and a symptom of interest (groups 4 and 5 pooled), the OR for stroke within 365 days among patients admitted to the hospital at the time of the index ED visit was not different from those discharged home in a multivariable model (data not shown).

Nontruncated analyses

When we did not limit the follow-up to 365 days, a Cox proportional hazards model (table e-1, links.lww.com/CPJ/A104) for the 4 groups discharged to home was generally consistent with the results of the primary logistic regression analysis above, although now the risk gradient across the groups was less steep (figure e-1). In a Cox proportional hazards model for the 1,033 patients with head CT and symptoms, hospitalization at the index visit was not associated with an increased risk of stroke in a multivariable model (table e-2).

Discussion

Using a retrospective analysis of readily available EHR data, we observed that patients with high-probability neurologic symptoms who undergo head CT but do not have a TIA or stroke discharge diagnosis have a 2% higher 1-year absolute risk of stroke, whether they were discharged to home or admitted to the hospital from the ED. The 30- and 90-day results suggest pronounced separation of risks between groups early after the index event, when the risk of stroke after an unrecognized TIA or minor stroke would be highest. By inference, we speculate that some patients in the high-risk groups may have had an unrecognized TIA or stroke at the index ED visit.

For patients with vertigo and dizziness—another presentation in which missed diagnosis of TIA or stroke has been well described—others have noted similar findings. A previous study evaluated rates of stroke in the ensuing year among patients discharged to home from the ED with a diagnosis of peripheral vertigo compared with those with a diagnosis of renal colic; the relative risk of stroke in the peripheral vertigo group was 9.3 at 30 days and 2.5 at 365 days.9 Unlike previous studies that investigated only vertigo,10,–12 our study can be applied to a broader group of ED presentations that might result in missed diagnosis of TIA or stroke. A recent study used claims data from New York State and assessed nonfocal transient neurologic attacks (e.g., altered mental status, generalized weakness, and sensory changes) and found that these patients had an increased risk of stroke after ED discharge compared with renal colic (reference) but lower than TIA.13 Our approach using the EHR rather than claims data extends this work to patients who may have had focal neurologic symptoms.

Although both were additive to stroke risk, the presence of head CT order was a stronger marker of stroke risk than was the presence of symptoms of interest. These results suggest that an ED clinician's initial suspicion, based on history and physical examination, about a central neurologic diagnosis has prognostic implications. In a study of patients with chest pain in the ED, the gestalt judgment of clinicians was equivalent to the history, electrocardiogram, age, risk factors, and troponin score14 in discriminating the risk of cardiac ischemic events.15 In TIA or minor stroke, by analogy, clinician judgment as reflected in the head CT imaging order may better capture a set of symptoms and risk profiles that may be poorly documented in or difficult to ascertain from the chief complaint field in the EHR. Apart from the decision to order a head CT, the decision to hospitalize a patient, as an additional manifestation of clinical judgment, did not increase stroke risk in the adjusted analyses.

Our simple 2-factor, 4-level risk model for those presenting to the ED with neurologic symptoms may be useful in retrospective analyses of unrecognized TIA or stroke. Hospitals or health care systems might monitor subsequent rates of stroke among patients discharged to home (or admitted) from the ED after having had a head CT and stroke-like symptoms but in whom stroke or TIA was not initially diagnosed. However, because even in the highest-risk group, the absolute risk of stroke is low, the predictive ability of the model would need to be improved before it would be cost-effective to apply it in system-level interventions such as follow-up phone calls and appointments with primary care physicians and neurologists after the ED visit.

Several clinical and radiographic risk stratification tools have been developed for TIA. The ABCD2 stratification method incorporates clinical features into a validated score.8 Although it was initially developed as a tool to predict the risk of stroke following TIA, it also has some utility in distinguishing true TIA from mimics; higher scores are associated with true (neurologist adjudicated) TIA.16 Although widely used, the ABCD2 score accurately predicts stroke risk in only 60%–70% of patients.17,18 Nonetheless, we applied our age criteria and symptom criteria based on the ABCD2 score elements, in an effort to enrich the cohort with patients with potential TIA. Although we did include diabetes mellitus, we did not include other elements, such as initial blood pressure and the duration of symptoms, which were not readily available in the EHR.

Because TIA is transient, only approximately 4% of patients with a TIA have an acute infarct on noncontrast head CT.19,20 Unlike CT, diffusion-weighted MRI can identify acute infarcts with high sensitivity and specificity.21 Brain MRI studies have confirmed that acute infarction is noted in up to half of patients with TIA and can improve stroke risk prediction.22,23 Therefore, American Heart Association/American Stroke Association guidelines recommend MRI as the diagnostic test of choice, if available.24 We did not analyze MRI as a predictor, however, because relatively few patients had MRIs in the ED compared with the number of patients with a CT during the study period.

This study has other limitations. First, because data from only a single EHR and enterprise data warehouse were available, ascertainment of stroke in the follow-up period was limited to the 2 index hospitals. We may have missed occurrence of strokes during the follow-up period if the patient did not return to the index hospital system for treatment; therefore, we likely underestimated the actual risk of stroke. We also did not ascertain mortality if this occurred outside our health care system. Nevertheless, a strength of the present study is that it includes both an academic medical center and a busy community hospital. We also had relatively high follow-up within the health care system, with roughly two-thirds of the present cohort returning for any type of visit, a year or more after the index ED visit. Although differential health care contact between groups may have led to ascertainment bias, we adjusted for previous health care utilization in our models. Second, we did not analyze access to neurologic consultation in the ED, as this factor was not readily discernible in the enterprise data warehouse. Previous studies have noted that neurologic consultation improves diagnosis of minor stroke and TIA and reduces stroke risk after TIA25,26; yet, it has been noted that neurologic consultation in the ED is suboptimally used.27 Third, we are unable to reliably review discharge instructions and medication prescription and compliance after hospital discharge. Thus, it is possible that the increased risk of stroke in group 4 and group 5 patients may have been attributable to lack of neurologic consultation in the ED or hospital and poor application and/or adherence to secondary prevention strategies. Natural language processing of textual data in the clinical notes is a promising approach to evaluating these questions. Fourth, because we did not include patients younger than 60 years, we cannot comment on the whether the risk of stroke we observed also exists in younger patients.4 Last, we relied on ICD-9/10 codes, which may be less accurate than adjudicated data28,–30 and have been validated for hospital discharge, not ED discharge, diagnoses.

Diagnostic error in stroke and TIA may be shaped, beyond the items discussed above, by a variety of factors including patient census in the ED, access to neuroimaging, availability of neurologist or other types of consultation, and access to outpatient stroke clinics after discharge from the ED.31 More generally, in its 2015 report titled “Improving Diagnosis in Health Care,” the National Academy of Medicine emphasized the pervasiveness of diagnostic error: most Americans are misdiagnosed at least once in their lifetimes.32 In this context, and because false-negative diagnosis of TIA and minor stroke places patients at an increased and sustained risk of stroke, the simple stratification model presented here may prove useful. Applications or refinements of this model may contribute to comparative effectiveness research that evaluates management and triage strategies for patients presenting with neurologic symptoms that are concerning for stroke or TIA.

Author contributions

M.B. Rosenman: drafting and revising the manuscript, study concept or design, and analysis or interpretation of data. E. Oh: data acquisition. C.T. Richards: drafting and revising the manuscript or interpretation of data. S.J. Mendelson: drafting and revising the manuscript and analysis or interpretation of data. J. Lee: drafting and revising the manuscript, study concept or design, analysis or interpretation of data, statistical analysis, and study supervision. J.L. Holl: drafting and revising the manuscript, study concept or design, and analysis or interpretation of data. A.M. Naidech: study concept or design, study supervision, and obtaining funding. S. Prabhakaran: drafting and revising the manuscript, data acquisition, study concept or design, analysis or interpretation of data, and study supervision.

Study funding

No targeted funding reported.

Disclosure

The authors report no disclosures relevant to the manuscript. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/cp.

TAKE-HOME POINTS

• → Among patients discharged from the ED, presence of specific neurologic symptoms and occurrence of CT head confer increased risk of subsequent stroke.

• → Such patients may have had an unrecognized TIA or stroke at the index ED encounter.

• → Health record surveillance using a search for these factors could facilitate future efforts to reduce misdiagnosis of TIA or stroke in the ED.

References

1. Johnston SC, Gress DR, Browner WS, Sidney S. Short-term prognosis after emergency department diagnosis of TIA. JAMA 2000;284:2901–2906. [PubMed] [Google Scholar]

2. Amarenco P, Lavallée PC, Labreuche J. One-year risk of stroke after transient ischemic attack or minor stroke. N Engl J Med 2016;374:1533–2154. [PubMed] [Google Scholar]

3. Kerber KA, Brown DL, Lisabeth LD, Smith MA, Morgenstern LB. Stroke among patients with dizziness, vertigo, and imbalance in the emergency department: a population-based study. Stroke 2006;37:2484–2487. [PMC free article] [PubMed] [Google Scholar]

4. Tarnutzer AA, Lee SH, Robinson KA, Wang Z, Edlow JA, Newman-Toker DE. ED misdiagnosis of cerebrovascular events in the era of modern imaging. Neurology 2017;88:1468–1477. [PMC free article] [PubMed] [Google Scholar]

5. Lavallée PC, Meseguer E, Abboud H, et al.. A transient ischaemic attack clinic with round-the-clock access (SOS-TIA): feasibility and effects. Lancet Neurol 2007;6:953–960. [PubMed] [Google Scholar]

6. Rothwell PM, Giles MF, Chandratheva A, et al.. Effect of urgent treatment of transient ischaemic attack and minor stroke on early recurrent stroke (EXPRESS study): a prospective population-based sequential comparison. Lancet 2007;370:1432–1442. [PubMed] [Google Scholar]

7. Dutta D, Bowen E, Foy C. Four-year follow-up of transient ischemic attacks, strokes, and mimics: a retrospective transient ischemic attack clinic cohort study. Stroke 2015;46:1227–1232. [PubMed] [Google Scholar]

8. Johnston SC, Rothwell PM, Nguyen-Huynh MN, et al.. Validation and refinement of scores to predict very early stroke risk after transient ischaemic attack. Lancet 2007;369:283–292. [PubMed] [Google Scholar]

9. Sidney CL, Grewal K, Lu H, Kapral MK, Kulkarni G, Austin PC. Outcomes among patients discharged from the emergency department with a diagnosis of peripheral vertigo. Ann Neurol 2016;79:32–41. [PubMed] [Google Scholar]

10. Kerber KA, Zahuranec DB, Brown DL, et al.. Stroke risk after nonstroke emergency department dizziness presentations: a population-based cohort study. Ann Neurol 2014;75:899–907. [PMC free article] [PubMed] [Google Scholar]

11. Lee CC, Ho HC, Su YC, et al.. Increased risk of vascular events in emergency room patients discharged home with diagnosis of dizziness or vertigo: a 3-year follow-up study. PLoS One 2012;7:e35923. [PMC free article] [PubMed] [Google Scholar]

12. Kim AS, Fullerton HJ, Johnston SC. Risk of vascular events in emergency department patients discharged home with diagnosis of dizziness or vertigo. Ann Emerg Med 2011;57:34–41. [PubMed] [Google Scholar]

13. Parikh NS, Merkler AE, Kummer BR, Kamel H. Ischemic stroke after emergency department discharge for symptoms of transient neurological attack. Neurohospitalist 2018;8:135–140. [PMC free article] [PubMed] [Google Scholar]

14. Six AJ, Backus BE, Kelder JC. Chest pain in the emergency room: value of the HEART score. Neth Heart J 2008;16:191–196. [PMC free article] [PubMed] [Google Scholar]

15. Visser A, Wolthuis A, Breedveld R, ter Avest E. HEART score and clinical gestalt have similar diagnostic accuracy for diagnosing ACS in an unselected population of patients with chest pain presenting in the ED. Emerg Med J 2015;32:595–600. [PubMed] [Google Scholar]

16. Josephson SA, Sidney S, Pham TN, Bernstein AL, Johnston SC. Higher ABCD2 score predicts patients most likely to have true transient ischemic attack. Stroke 2008;39:3096–3098. [PubMed] [Google Scholar]

17. Stead LG, Suravaram S, Bellolio MF, et al.. An assessment of the incremental value of the ABCD2 score in the emergency department evaluation of transient ischemic attack. Ann Emerg Med 2011;57:46–51. [PMC free article] [PubMed] [Google Scholar]

18. Sanders LM, Srikanth VK, Psihogios H, Wong KK, Ramsay D, Phan TG. Clinical predictive value of the ABCD2 score for early risk of stroke in patients who have had transient ischaemic attack and who present to an Australian tertiary hospital. Med J Aust 2011;194:135–138. [PubMed] [Google Scholar]

19. Douglas VC, Johnston CM, Elkins J, Sidney S, Gress DR, Johnston SC. Head computed tomography findings predict short-term stroke risk after transient ischemic attack. Stroke 2003;34:2894–2898. [PubMed] [Google Scholar]

20. Förster A, Gass A, Kern R, et al.. Brain imaging in patients with transient ischemic attack: a comparison of computed tomography and magnetic resonance imaging. Eur Neurol 2012;67:136–141. [PMC free article] [PubMed] [Google Scholar]

21. Szabo KO, Laubach HJ, Baird AE, et al.. Clinical experience with diffusion-weighted MR in patients with acute stroke. AJNR Am J Neuroradiol 1998;19:1061–1066. [PMC free article] [PubMed] [Google Scholar]

22. Prabhakaran S, Chong JY, Sacco RL. Impact of abnormal diffusion-weighted imaging results on short-term outcome following transient ischemic attack. Arch Neurol 2007;64:1105–1109. [PubMed] [Google Scholar]

23. Kidwell CS, Alger JR, Di Salle F, et al.. Diffusion MRI in patients with transient ischemic attacks. Stroke 1999;30:1174–1180. [PubMed] [Google Scholar]

24. Saver JD, Saver JL, Albers GW, et al.. Definition and evaluation of transient ischemic attack: a scientific statement for healthcare professionals from the American Heart Association/American Stroke Association Stroke Council; Council on Cardiovascular Surgery and Anesthesia; Council on Cardiovascular Radiology and Intervention; Council on Cardiovascular Nursing; and the Interdisciplinary Council on Peripheral Vascular Disease. Stroke 2009;40:2276–2293. [PubMed] [Google Scholar]

25. Schrock JW, Glasenapp M, Victor A, Losey T, Cydulka RK. Variables associated with discordance between emergency physician and neurologist diagnoses of transient ischemic attacks in the emergency department. Ann Emerg Med 2012;59:19–26. [PubMed] [Google Scholar]

26. Giles MF, Rothwell PM. Risk of stroke early after transient ischaemic attack: a systematic review and meta-analysis. Lancet Neurol 2007;6:1063–1072. [PubMed] [Google Scholar]

27. Kim AS, Sidney S, Bernstein AL, Douglas VC, Johnston SC. Urgent neurology consultation from the ED for transient ischemic attack. Am J Emerg Med 2011;29:601–608. [PMC free article] [PubMed] [Google Scholar]

28. Guimarães PO, Krishnamoorthy A, Kaltenbach LA, et al.. Accuracy of medical claims for identifying cardiovascular and bleeding events after Myocardial infarction: a secondary analysis of the TRANSLATE-ACS study. JAMA Cardiol 2017;2:750–757. [PMC free article] [PubMed] [Google Scholar]

29. Birman-Deych E, Waterman AD, Yan Y, Nilasena DS, Radford MJ, Gage BF. Accuracy of ICD-9-CM codes for identifying cardiovascular and stroke risk factors. Med Care 2005;43:480–485. [PubMed] [Google Scholar]

30. Rosenman M, He J, Martin J, et al.. Database queries for hospitalizations for acute CHF: flexible methods and validation based on set theory. J Am Med Inform Assoc 2014;21:345–352. [PMC free article] [PubMed] [Google Scholar]

31. Wilson A, Coleby D, Regen E, et al.. Service factors causing delay in specialist assessment for TIA and minor stroke: a qualitative study of GP and patient perspectives. BMJ Open 2016;6:e011654. [PMC free article] [PubMed] [Google Scholar]

32. Committee on Diagnostic Error in Health Care; Board on Health Care Services; Institute of Medicine; the National Academies of Sciences, Engineering, and Medicine; Balogh EP, Miller BT, Ball JR, editors. Improving Diagnosis in Health Care. Washington: National Academies Press (US); 2015. [Google Scholar]

Articles from Neurology: Clinical Practice are provided here courtesy of American Academy of Neurology