Accuracy of paramedic diagnosis of stroke

Health Services Research

Accuracy of Paramedic Diagnosis of Stroke Richard M. Zweifler, MD, Danny York, BS, Tha Tha U, MD, Jorge E. Mendizabal, MD, and John F. Rothrock, MD

Background and Purpose: Accurate prehospital diagnosis of acute stroke may lead to fewer delays in hospital presentation. In addition, prehospital personnel soon may be administering therapies to patients with presumed stroke. We sought to determine the sensitivity and positive predictive value (PPV) of paramedic diagnosis of stroke in Mobile, Alabama, and to evaluate the impact of an educational program on paramedic diagnostic capability. Methods: We collected data from all paramedicdiagnosed stroke patients transported to a University of South Alabama hospital by Mobile Fire Medics. Final diagnosis was determined by a neurologist and classified as stroke or nonstroke (i.e., PPV). Paramedic diagnoses for all hospitalized stroke patients transported by Mobile Fire Medics were also reviewed (i.e., sensitivity). Sensitivity and PPV were calculated for the period 6/13/95 to 3/13/97. In addition, both indices were calculated for the period before (6/13/95 to 5/5/96) and after (6 / 25 / 96 to 3 / 13 / 97) an 8-week intensive educational program. Results: Seventyone hospitalized stroke patients were transported by Mobile Fire Medics during the study period. Paramedics correctly identified 67 patients in total (94% sensitivity), 29 during the pre-education period (91% sensitivity), and 29 during the posteducation period (97% sensitivity; P ~- .33). Twenty-five patients were incorrectly diagnosed with stroke (73% PPV), 15 during the pre-education period (66% PPV), and 9 during the posteducation period (76% PPV; P = .30). Conclusion: Although paramedics in Mobile misdiagnose few patients with acute stroke, there is a tendency toward overdiagnosis. An educational intervention resulted in a trend toward improved accuracy of diagnosis, but this did not reach statistical significance. Key Words: Cerebral infarction Stroke, Acute--Diagnosis--EMS--Paramedic.

Accurate prehospital diagnosis of acute stroke may reduce delays in hospital presentation and facilitate acute therapeutic intervention, but few data are available regarding the accuracy of paramedic diagnosis of acute stroke. TM In addition, limited data exist that speak to the effect of education on the accuracy of paramedic diagnosis. 3,4 Therefore, we sought to determine the sensitivity and positive predictive value (PPV) of paramedic diagnosis of

From the University of South Alabama Stroke Center, Mobile, AL. Received March 23, 1998;accepted May 19, 1998. Address reprint requests to Richard M. Zweifler, MD, USA Stroke Center, 10th Floor,Suite I, 2451 Fillingim St, Mobile, AL 36617. Copyright 9 1998by National Stroke Association 1052-3057/ 98/ 0706-001053.00/ 0 446

stroke and to evaluate what influence an educational program might have on the ability of paramedics to diagnose acute stroke accurately. Methods We reviewed the run sheets of all patients identified by Mobile Fire Medics (MFM) as having stroke and subsequently transported to a University of South Alabama (USA) hospital between June 13, 1995 and March 13, 1997. MFM account for the majority of emergency medical services (EMS) transports to USA hospitals. The MFM stroke diagnosis was recorded by checking CVA (cerebrovascular accident) on a list of multiple symptoms and conditions. A final diagnosis for those patients was

Journal of Stroke and Cerebrovascular Diseases, Vol. 7, No. 6 (November-December), 1998: pp 446-448

447

P A R A M E D I C D I A G N O S I S OF S T R O K E

determined in the emergency room by a neurologist, typically with the benefit of computed tomography (CT), and specified as stroke or nonstroke. Patients with transient ischemic attack (TIA) or subarachnoid hemorrhage were considered to have stroke. Positive predictive value (PPV) was calculated as the total number of patients transported by MFM with a diagnosis of stroke who were confirmed to have stroke by a neurologist (i.e., true positives) divided by the number of patients with a MFM diagnosis of stroke (i.e., true positives plus false positives). Paramedic diagnoses for all hospitalized stroke patients transported by MFM during the study period were also reviewed. Sensitivity was calculated as the total number of patients transported by MFM with a diagnosis of stroke who were confirmed to have stroke by a neurologist (i.e., true positives) divided by the number of MFM-transported strokes (i.e., true positives plus false negatives). Stroke patients were identified from the USA Stroke Center prospective stroke code log that includes data regarding all patients presenting to a USA hospital with signs and symptoms suggestive of acute (ie, symptom onset within 24 hours) stroke. Sensitivity and PPV were compared for the period before (6 / 13 / 95 to 5 / 5 / 96) and after (6/25/96 to 3/13/97) an intensive educational program. That program consisted of weekly hour-long seminars conducted by one of the investigators (RMZ) that emphasized field diagnosis of stroke and, specifically, the common signs and symptoms of stroke. Selected items from the National Institutes of Health (NIH) Stroke Scale including, but not limited to, facial pals~fi motor arm and leg, sensory, and language were reviewed. In addition, a Paramedic Quick Screen similar to that described by Lyden et al. 5 was developed. All paramedics participated in at least one seminar. Statistical comparisons between the pre-education and posteducation period were made using the X2 test, with a value of P < .05 considered significant. Results

During the study period there were 224 patients identified with acute stroke, 109 (49%) of whom were transported by EMS. Seventy-one patients were transported by Mobile Fire Medics, 32 during the pre-education period, 9 during the 8-week education period, and 30 during the posteducation period. Sixty patients were diagnosed with ischemic stroke, 10 with intracerebral hemorrhage, and 1 with subarachnoid hemorrhage. MFM correctly identified 67 patients in total (94% sensitivity), 29 during the pre-education period (91% sensitivity) and 29 during the posteducation period (97% sensitivity; P = .33) (Fig 1). Of the 4 stroke patients who were misdiagnosed by MFM (i.e., false negatives), 1 each was diagnosed with seizure, decreased mentation, unconsciousness/unresponsive, or unknown; the patient identified with unconsciousness

O0 80 I-I Pre-education 60

9 Post-education

% 40 20 0 PPV

Sensitivity

Figure 1. Effect of education on the accuracy of paramedic diagnosis. P = not significant for both comparisons.

had an intracerebral hemorrhage, and the others had ischemic strokes. Twenty-five patients were incorrectly diagnosed with stroke (73% PPV), 15 during the preeducation period (66% PPV) and 9 during the posteducation period (76% PPV; P = .30) (Fig 1). The stroke misdiagnoses (i.e., mimickers) are listed in Table 1.

Discussion

At this point, successful implementation of acute stroke treatment requires administration of tissue plasminogen activator (TPA) within 3 hours of stroke onset. Vigilant, knowledgeable, and efficient EMS personnel serve as the link between the general public and the site where TPA therapy for ischemic stroke may be available. An on-site "stroke code" system will facilitate the process that leads to potential treatment with TPA, 6 but even so, the success of that system will be dependent to a large extent on the skills and initiative of the EMS team. If paramedics fail to recognize acute ischemic stroke for what it is, afflicted patients may be less likely to receive specific treatment; if the paramedics confuse another condition for stroke, unnecessary and even unsafe use of medical resources may result. Our data appear to indicate that paramedics in Mobile, Alabama, diagnose stroke with a high degree of sensitivTable 1. Stroke misdiagnoses Diagnosis

Number (%)

Seizure Hypoglycemia Migraine Syncope Mass lesion Infection Arrhythmia

19 (76) 1 (4) 1 (4) 1 (4) 1 (4) 1 (4) 1 (4)

448 ity but that a significant number of patients with other conditions are misdiagnosed as having acute stroke. A high stroke diagnostic sensitivity, at the expense of overdiagnosis, is desirable as many stroke mimickers require acute evaluation and intervention. Although the PPV in our study (73%) is similar to values reported by Kothari et al. 2 (72%) and Corry and Smith l (85%), we recorded a higher sensitivity than that reported by Corry (94% vs. 57%). This disparity in sensitivity highlights the importance of local awareness of EMS diagnostic capabilities as educational efforts can then be tailored to the local needs. As was noted by another acute stroke intervention team that included physicians, 7 seizure was the most common stroke mimicker. Our intensive educational program resulted in a trend toward more accurate paramedic diagnosis of stroke, but this did not reach statistical significance. It should be noted that the MFM educational program was initiated approximately 1 year after implementation of a stroke awareness campaign intended for the Mobile community at large, and it is therefore possible that the study design was "contaminated" because of prior, indirect education of some paramedics during that campaign. Furthermore, the study design is suboptimal in that no true control group was studied, and the low number of patients may have invoked a type II error.

R.M. ZWEIFLER ET AL.

Acknowledgment: The authors thank the Mobile Fire Medics for their enthusiasm and support of our acute stroke treatment efforts. We also thank Captain Sam Williams of the Mobile Fire-Rescue Department for organizing the educational seminars.

References 1. Corry MD, Smith WS. Accuracy of paramedic diagnosis of stroke and TIA. Neurology 1996;46:A429 (suppl 1, abstr). 2. Kothari R, Barsan W, Brott T, eta]. Frequency and accuracy of prehospital diagnosis of acute stroke. Stroke 1995;26:937941. 3. Kidwell CS, Saver JL, Eckstein M, et al. High accuracy of emergency medical technician identification of stroke using the Los Angeles Prehospital Stroke Screen (LAPSS). Stroke 1998;29:313 (abstr). 4. Smith WS, Corry MD, Fazackerley J, et al. Paramedic accuracy in the application of the NIH Stroke Scale to victims of stroke. Acad Emerg Med 1997;4:379-380 (abstr). 5. Lyden PD, Rapp K, Babcock T, et al. Ultra-rapid identification, triage, and enrollment of stroke patients into clinical trials. J Stroke Cerebrovasc Dis 1994;4:106-113. 6. Zweifler RM, Drinkard R, Cunningham S, et al. Implementation of a stroke code system in Mobile, AL: Diagnostic and therapeutic yield. Stroke 1997;28:981-983. 7. Libman RB, Wirkowski E, Alvir J, et al. Conditions that mimic stroke in the emergency department: Implications for acute stroke trials. Arch Neuro11995;52:1119-1122.