Accuracy in the Diagnosis of Stroke
Background and Purpose The role of emergency physicians in trials of acute stroke therapy is expanding. We investigated the ability of emergency physicians to accurately identify patients with stroke.
Methods We reviewed all 446 patients who were evaluated in the emergency department and had an inpatient admitting or discharge diagnosis of ischemic or hemorrhagic stroke from May 1, 1992, to June 30, 1993. The final admitting diagnosis made by the emergency physician was compared with the final hospital discharge diagnosis, 95% of which were made by neurologists or neurosurgeons.
Results All 76 patients with a final discharge diagnosis of intracerebral or subarachnoid hemorrhage were correctly diagnosed by the emergency physicians (sensitivity, 100%; positive predictive value, 100%). Of the 351 patients with a final discharge diagnosis of ischemic stroke or transient ischemic attack, 346 were correctly identified by the emergency physicians (sensitivity, 98.6%; positive predictive value, 94.8%). Nineteen patients were diagnosed with stroke or transient ischemic attack by the emergency physician but had a final discharge diagnosis other than stroke. Discharge diagnoses included paresthesia or numbness of unknown causes (3), seizure (2), complicated migraine (2), peripheral neuropathy (2), cranial nerve neuropathy (2), psychogenic paralysis (1), and other (7).
Conclusions Emergency physicians at a large urban teaching hospital with a comprehensive stroke intervention program can accurately identify patients with stroke, particularly hemorrhagic stroke. If similar accuracy can be documented in other types of hospitals, emergency physicians may become key providers of urgent stroke intervention.
Emergency physicians are often the first physicians to evaluate and treat patients with an acute stroke.1 2 Proper treatment depends on differentiating stroke patients from patients who have other acute illnesses, as well as correctly identifying whether patients have a hemorrhagic or an ischemic stroke. If thrombolytic therapy is found to be effective in ischemic strokes, diagnostic accuracy will play an even more crucial role in treatment. We investigated the accuracy of emergency physicians in the diagnosis of stroke and stroke subtypes.
Subjects and Methods
The University of Cincinnati Medical Center is a 707-bed teaching institution with a residency training program in emergency medicine. The emergency department is an urban, level-1 trauma center with 65 000 visits per year. All emergency department patients are evaluated by residents who are supervised by residency-trained and board-certified emergency medicine faculty. The decision to admit and the admitting diagnosis are made by the emergency medicine resident and faculty. In questionable cases, telephone consultation by a board-certified neurologist is available. Emergency department consultation by a neurosurgery resident is obtained after the diagnosis of hemorrhagic stroke is made by the emergency physician.
A computerized retrospective search of all hospital admissions from the emergency department was performed from May 1, 1992, to June 30, 1993. Cases were identified if they had a primary admitting or final inpatient discharge diagnosis of ischemic or hemorrhagic stroke (International Classification of Diseases, 9th Revision codes 430, 431, and 433 to 437) and were admitted through the emergency department. To ensure that all potential patients were identified, we manually reviewed all emergency department visits for the study period for clinical findings suggestive of stroke or transient ischemic attack (TIA) or an admitting or discharge diagnosis of stroke or TIA.3 Patients evaluated in the emergency department but not admitted to the hospital were excluded from the study.
Telephone consultations by an attending neurologist were not consistently documented on emergency department records. To estimate the number of stroke-related neurology telephone consultations to the emergency department, a 1-month log was kept of all stroke-related consultations by the on-call attending neurologist and the Cincinnati/Northern Kentucky Stroke Team neurologists (T.B. and J.P.B.).
For patients admitted to the hospital, the admitting diagnosis made by the emergency physician was compared with the final hospital discharge diagnosis. If a patient had two emergency department admitting diagnoses (eg, rule out seizure versus TIA), concordance was assumed if the final hospital discharge diagnosis corresponded to one of the admitting diagnoses. Analysis was also completed assuming discordance between the emergency diagnosis and the discharge diagnosis on patients with two admitting diagnosis. Ninety-five percent of the final hospital discharge diagnoses of stroke or TIA were made by neurologists or neurosurgeons. In three cases, the discharge diagnosis did not specifically mention stroke or TIA but was suggestive of a stroke or TIA (eg, “unilateral arm weakness”). If the discharge summary stated that the findings were consistent with a cerebrovascular event, the case was reviewed by three study investigators. If unanimous agreement was reached that the case was likely a stroke or TIA, the case was placed into the stroke or TIA category.
All patients admitted as stroke or TIA from the emergency department had a noncontrast CT of the head before admission. These scans were interpreted by emergency physicians or radiologists from 8 am to 5 pm weekdays and by emergency physicians alone at all other times. Those patients correctly diagnosed with stroke or TIA by emergency physicians were compared with those incorrectly diagnosed with stroke in terms of age, sex, race, and history of stroke or TIA. Age was analyzed with Student’s t test. All other factors were analyzed with the χ2 test.
Of 77 024 emergency department visits during the 14-month study period, 463 patients (0.6%) had an emergency department diagnosis of stroke or TIA. Twenty-two patients were discharged from the emergency department, and 441 (4% of all admissions) were hospitalized. Twelve patients were admitted with two admitting diagnoses, one of which included ischemic stroke or TIA (eg, rule out stroke versus seizure). Of these 12 patients, 5 had a final diagnosis of stroke or TIA, and 7 had a nonstroke diagnosis. Of 12 057 patients admitted to the hospital from the emergency department during the study period, 427 had a final inpatient discharge diagnosis of stroke or TIA. Three patients had hospital discharge summaries that stated stroke or TIA but had final coded diagnoses other than stroke or TIA. After review by three study physicians, all three cases were included as strokes or TIAs.
The mean age for the 441 patients with an admitting diagnosis of stroke was 63 years. More than half of these patients were female (57%) and black (62%). Of these patients, 422 (96%) were discharged with a final diagnosis of stroke or TIA (Figure⇓). The other 19 patients (4%) were given a diagnosis of stroke or TIA by the emergency physician but had a final hospital discharge diagnosis other than stroke or TIA (Table 1⇓).
All 76 patients with a final discharge diagnosis of intracerebral or subarachnoid hemorrhage were correctly diagnosed by the emergency physician before neurosurgery consultation (sensitivity, 100%; positive predictive value [PPV], 100%; Table 2⇓). In addition, emergency physicians correctly diagnosed 346 of 351 patients with ischemic stroke or TIA (sensitivity, 98.6%; PPV, 94.8%; Table 2⇓). If discordance was assumed on all 12 patients with two admitting diagnosis (eg, stroke versus seizure), the sensitivity is 97.2% and the PPV is 92.6%.
Of the 365 patients initially diagnosed with ischemic stroke or TIA by the emergency physician, 19 (5%) were discharged with a final diagnosis other than stroke or TIA (Table 3⇓). It is not known how many of the 365 initial diagnoses of ischemic stroke or TIA were made in conjunction with a neurological telephone consultation. During a 1-month surveillance period, there was one neurology telephone consultation and eight stroke team consultations (two of which were in-person consultations) regarding a potential stroke patient. In only three of these cases did the neurologist give assistance in the diagnosis. Two of these three patients were eventually diagnosed with stroke. There were 32 patients admitted with a diagnosis of stroke or TIA during this 1-month surveillance period.
Of the 19 patients discharged with a diagnosis other than stroke or TIA (Table 1⇑), the most common final hospital discharge diagnoses included paresthesia or numbness of unknown causes (3), seizure (2), complicated migraine (2), peripheral neuropathy (2), cranial nerve neuropathy (2), and psychogenic paralysis (1). There was a strong trend for these patients to be younger than those correctly diagnosed with stroke or TIA by emergency physicians (mean age of 55 years versus 65 years; P=.06). There was no difference between these two groups in terms of race (P=.59), sex (P=.48), or history of stroke (P=.14).
Five patients (1%) of the 427 patients with a discharge diagnosis of stroke or TIA received diagnoses other than stroke by emergency physicians (Table 4⇓).
Patients discharged home from the emergency department with a diagnosis of stroke or TIA were excluded from the study and had no further follow-up. On review of the emergency department records by study investigators (C.A.H. and R.U.K.), there were no patients with clinical signs or symptoms suggestive of a stroke or TIA who were discharged home from the emergency department with a diagnosis other than stroke.
Clinicians have begun to view stroke as a medical emergency of a magnitude similar to that of trauma victims and patients with acute myocardial infarctions.4 5 6 7 8 The need for rapid identification and treatment of patients with acute stroke has been emphasized by recent therapeutic trials.9 10 11 12 13 Emergency physicians currently play a crucial link in the early evaluation and management of trauma and heart attack victims. Systems have been established that allow treatment within the “golden hour” for these patients. A similar strategy using emergency physicians may be effective in the management of acute stroke.
A major concern regarding the role of emergency physicians in the evaluation and treatment of stroke patients is diagnostic accuracy. We found that emergency physicians with neurological telephone consultation could identify 98% of ischemic strokes (sensitivity, 98.6%; specificity, 99.8%) and were correct 95 of 100 times when they identified a patient as having had a stroke or TIA (PPV, 94.8%). It may be argued that an error rate of 5% is unacceptable if one is giving potentially life-threatening drugs (eg, thrombolytics). This study did not address the accuracy of emergency physicians in identifying patients for therapeutic intervention but rather evaluated emergency physicians’ accuracy in identifying the most likely diagnoses for these patients. The specificity and PPV may improve at the expense of sensitivity if emergency physicians were asked if they were confident enough in their diagnosis of stroke to treat the patient with a potentially life-threatening agent. In addition, this study evaluated only the emergency physicians’ ability to distinguish between the basic stroke subtypes of hemorrhagic versus ischemic infarction. Concordance rates would be significantly lower had emergency diagnoses and hospital discharge diagnoses been compared for other stroke subtypes (eg, lacunar versus nonlacunar infarction).
A second concern is whether emergency physicians can differentiate hemorrhagic from nonhemorrhagic stroke, particularly if thrombolytic therapy is found effective for ischemic stroke.14 Studies suggest that there may be discordance between emergency physicians’ and radiologists’ interpretations of cranial CT scans (for stroke and nonstroke etiologies), but that few management errors occur as a result of these radiographic misinterpretations.14 15 16 Alfaro et al15 found that emergency physicians missed 69% of all infarcts, 62% of parenchymal hematomas, and 50% of subarachnoid hemorrhages on noncontrast cranial CTs. However, they found that none of these patients were managed inappropriately. At our institution, emergency physicians were able to identify all patients with hemorrhagic stroke based on clinical and radiographic findings (sensitivity, 100%; specificity, 100%). The majority of CT scans for this study were initially interpreted by the emergency physician alone. Hemorrhagic stroke patients may have been misdiagnosed if emergency physicians had read all the scans independently. In addition, the discrepancy between our findings and those of Alfaro et al may be due to the fact that there have been ongoing stroke trials at our institution for the last 10 years. Our emergency physicians have heightened interest and experience in urgent interpretation of CT scans performed for stroke. Further studies are needed to determine whether emergency physicians can reliably differentiate hemorrhagic stroke from nonhemorrhagic stroke based on CT findings alone.
Our study was limited to patients admitted to the hospital from the emergency department. We did not follow patients discharged from the emergency department with a diagnosis of stroke or TIA to determine whether subsequent outpatient workup confirmed a diagnosis of stroke or TIA. Potential stroke patients may have been missed if they were diagnosed with something other than stroke and discharged home from the emergency department. On review of the emergency department records, no such cases were identified. If patients with stroke were missed by this screening process, they would likely have had subtle findings or atypical presentations and therefore would not have been likely candidates for therapeutic trials.
Emergency physicians at a large urban teaching hospital with a comprehensive stroke intervention program can accurately identify patients with stroke, particularly hemorrhagic stroke. If similar accuracy can be documented in other types of hospitals, emergency physicians may become key providers of urgent stroke intervention.
- Received June 5, 1995.
- Revision received August 9, 1995.
- Accepted August 31, 1995.
- Copyright © 1995 by American Heart Association
Bratina P, Greenberg L, Pasteur W, Grotta JC. Current emergency department management of stroke in Houston, Texas. Stroke. 1995;26:409-414.
Leibson CL, Naessens JM, Brown RD, Whisnant JP. Accuracy of hospital discharge abstracts for identifying stroke. Stroke. 1994;25:2348-2355.
Barsan WG, Brott TG, Broderick JP, Haley EC Jr, Levy DE, Marler JR. Urgent therapy for acute stroke: effects of a stroke trial on untreated patients. Stroke. 1994;25:2132-2137.
Fisher M, Bogousslavsky J. A glimmer of hope for acute ischemic stroke. Cerebrovasc Dis. 1994;4:177-178.
Gomez CR. Time is brain! J Stroke Cerebrovasc Dis. 1993;3:1-2. Editorial.
Del Zoppo GJ, Poeck K, Pessin MS, Wolpert SM, Furlan AJ, Ferbert A, Alberts MJ, Zivin JA, Wechsler L, Busse O, Greenlee R, Brass L, Mohr JP, Feldmann E, Hacke W, Kase CS, Biller J, Gress D, Otis SM. Recombinant tissue plasminogen activator in acute thrombotic and embolic stroke. Ann Neurol. 1992;32:78-86.
Overgaard K, Sperling B, Boysen G, Pedersen H, Gam J, Ellemann K, Karle A, Arlien-Søborg P, Olsen TS, Videbæk, Knudsen JB. Thrombolytic therapy in acute ischemic stroke: a Danish pilot study. Stroke. 1993;24:1439-1446.
Ringelstein EB, Biniek R, Weiller C, Ammeling B, Nolte PN, Thron A. Type and extent of hemispheric brain infarctions and clinical outcome in early and delayed middle cerebral artery recanalization. Neurology. 1992;42:289-298.
Grotta J, Clark W, Coull B, Pettigrew LC, Mackay B, Goldstein LB, Meissner I, Murphy D, LaRue L. Safety and tolerability of the glutamate antagonist CGS 19755 (Selfotel) in patients with acute ischemic stroke: results of a phase IIa randomized trial. Stroke. 1995;26:602-605.
Kothari R, Brott T, Broderick JP, Gorgas DL. Emergency department thrombolytic therapy for stroke: assessing patient eligibility. Acad Emerg Med. 1994;1:A47. Abstract.
Cordovano S, Gaeta T, Spevack T, Balentine J. Accuracy of emergency department interpretation of computerized tomography of the cranium. Acad Emerg Med. 1994;1:A64. Abstract.