Is this a stroke? The profile of patients with suspected acute cerebrovascular accident transferred by ambulance to the Neurology Emergency Department

INTRODUCTION

Stroke is the second most common cause of death worldwide. The Global Burden of Disease (GBD) 2019 estimated the mortality rate at 84.69 per 100,000 for all ages, with 66% of all stroke-related deaths occurring in people over the age of 70 [1, 2]. As of 2022, in the US, approximately 795,000 people suffer a new or recurrent stroke each year [3]. According to the National Health Fund (NFZ), in Poland 73,900 patients (225 per 100,000 population) were hospitalized due to stroke in 2022, with a 7-day mortality rate of 8% [4]. In 2017, in-hospital mortality due to stroke varied significantly depending on the admission ward, 6.8% in the stroke ward and up to 13% in other hospital wards [5]. Following the recommendations of the Polish Neurological Society (PNS), stroke patients should be admitted to dedicated hospital wards – stroke units (SU) – which can provide specialized care by experienced staff with dedicated resources at their disposal. As per PNS guidelines, the tasks of the emergency medical team (EMT) involving a patient with a suspected stroke include, among others, gathering information from the patient and their family (about the onset of symptoms, medications taken, recent surgical procedures, comorbidities, and a contact number for next of kin), assessing basic life functions, notifying the hospital emergency department and ensuring safe transportation of the patient to the emergency department/SU [6].

Stroke mimics (SMs) are conditions that resem- ble acute cerebrovascular accidents (CVA) in clinical presentation [7, 8]. SMs are common, with an overall prevalence of 24.9%, with substantial differences between studies, between 1.5% and 76.3% [8]. Their occurrence can lead to unnecessary hospitalizations in SU, causing an overload, additional costs and delaying the diagnosis and treatment of conditions for which SU is not intended [9-12]. They can be also related to unnecessary thrombolysis and its complications [10, 13, 14]. The most common conditions mimicking stroke include vertigo, toxic and metabolic conditions, seizures, functional disorders, migraine attacks, brain tumors, hypertension, and infections [8, 15-17].

This study had several objectives; firstly, it was to determine how common SMs are in the population of patients transferred to the Neurological Emergency Department (NED) by EMT for suspected CVA. Secondly, it was to assess the frequency of specific SMs, the abnormalities on neurological examination, and history that may be suggestive of SMs. In addition, the patients’ emergency medical cards (EMC) were analyzed to determine whether they were filled out in accordance with the recommendations of the PNS [6] and also to compare the differences in the EMC of patients with CVA and SMs.

METHODS

This research is a retrospective analysis of the patients’ data provided by the ambulance to the NED, due to the suspicion of CVA, between August 2021 and January 2022. In our hospital, NED is part of the general emergency system. It is staffed by physicians from different wards, admitting patients transferred by emergency services or referred from medical practices. It is one of the two hospitals, which treat stroke with tissue plasminogen activator (tPA) and mechanical thrombectomy in a city of near 300,000 inhabitants, located in a highly urbanized region of Poland with a few stroke units in the neighboring cities. CVA was defined as acute ischemic or hemorrhagic stroke, transient ischemic attack (TIA) or subarachnoid hemorrhage (SAH). Transient global amnesia (TGA) was not included as a form of CVA, but as one of SMs, because of its unclear etiology and lower risk of stroke or TIA in TGA patients [18, 19]. The study group consisted of 281 patients. Data collection was conducted in two stages. Firstly, from electronic registers to assess and characterize the general study group, and secondly, from EMCs available to assess how well they were completed. The data collected included date of admission, gender, age, abnormalities presented during the neurological examination on admission, comorbidities, final diagnoses, and information about admission to the hospital ward. In the study, active cancer was defined as cancer diagnosed within the previous 6 months; recurrent, regionally advanced; or metastatic cancer which is cancer for which treatment had been administered within 6 months. Also hematological cancer that is not in complete remission, with the exclusion of squamous skin cancer, basal cell carcinoma, benign meningioma and myelodysplastic syndrome [20]. EMCs were assessed in four categories based on the occurrence of specific information, such as the time the stroke’s symptoms began or when the patient was seen operational for the last time, a list of medications taken by the patient (with particular emphasis on anticoagulants with the time of the last dose, and antiepileptic drugs), glucose measurement and next of kin contact, which have to be obligatorily collected by EMT physicians according to PNS recommendations. Other information that should be mandatory for EMC, but which we did not take into account in our analysis because it would be difficult to verify whether its absence is due to the availability of information or its non-applicability, included general assessment of the internal and neurological condition, and history of previous diseases or injuries [6].

All data underwent statistical analysis performed with STATISTICA 13 PL software (Tibco Software Inc.). The normality of distribution was evaluated using the Shapiro-Wilk test. All quantitative variables had non- normal distribution, so they were presented as median (interquartile range). Qualitative variables were shown as absolute values and percentages. The differences between the groups for the quantitative variables were assessed with the U Mann-Whitney. The χ² test or χ² test with Yates correction was used for qualitative variables. In the next step, statistically significant variables (p < 0.05) were analyzed using a logistic regression model to evaluate the relation between selected variables and CVA occurrence. The logistic regression results were shown as odds ratios (OR) with 95% confidence intervals (CI) and Nagelkerke’s R2.

Due to the retrospective character of this study and data anonymization, ethical approval was not required.

RESULTS

Among 281 patients in the study group, 131 (46.6%) were females and 150 (53.4%) males. The median age was 74 (66-82). CVA was diagnosed in 207 (73.7%) patients including 153 (54,4%) patients with ischemic stroke, 26 (9.3%) cases of TIA, 26 (9.3%) with hemorrhagic stroke, and 2 (0.7%) patients with SAH, while the SMs group included 74 (26.3%) patients. No statistical differences were observed between the groups in terms of age (CVA 75 [67-83] vs. SMs 72 [60-82], p = 0.082) and gender (45.4% of CVA patients were males vs. 50% in SMs group, p = 0.497).

SMs were divided into two groups – neurological and non-neurological conditions. Neurological SMs were diagnosed in 43 (58.1%) SMs patients. The most prominent of these conditions were seizures; in 18 (24.3%) SMs cases. Other diseases included brain tumors (6.8%), head traumas (5.4%), vertigo (5.4%), TGA (4.1%), functional neurological disorders (2.7%), neuroinfections (2.7%), spinal cord lesion (1.4%), headache (1.4%), Bell’s palsy (1.4%) and other unclassified neurological conditions (2.7%). Non-neurological conditions were diagnosed in 31 (41.9%) patients among all SMs. In this group, infections other than neuroinfections were most commonly diagnosed, as they were present in 11 patients (14.9%) followed by other cardiovascular diseases present in 7 cases (9.5%), including aortic aneurysm rupture, aortic or femoral arteries thromboembolisms, pulmonary embolism, and supraventricular tachycardia. Other non-neurological SMs included poisonings (4.1%; ethanol, opioids, and carbon monoxide), syncope (4.1%), very high blood pressure (2.7%), dyselectrolytemia (2.7%), throat tumor (1.4%), mandibular dislocation (1.4%) and other unclassified conditions (1.4%). Detailed description of conditions found in SMs groups are presented in Table 1.

The differences in neurological examination were observed between groups diagnosed with CVA and SMs. The first group more often presented symptoms such as paresis (79.7% vs. 36.5%, p < 0.001), pyramidal signs (26.6% vs. 6.8%, p < 0.001), central facial palsy (58% vs. 25.7%, p < 0.001), speech disorders (44% vs. 13.5%, p = 0.043), sensation abnormalities (12.6% vs. 4.1%, p = 0.039) and head or eyes rotation (12.1% vs. 4.1%, p = 0.048). On the other hand, confusion occurred more commonly in patients with SMs rather than CVA patients (24.3% vs. 11.6%, p = 0.008). As for comorbidities in patients admitted to the NED, in the group with CVA, arterial hypertension (AH) (72.5% vs. 50%, p < 0.001) and atrial fibrillation (AF) (22.2% vs. 12.5%, p = 0.016) occurred more frequently, in comparison with SMs, but active cancer was more common in SMs patients (13.5% vs. 4.8%, p = 0.013). Details of neurological signs and comorbidities in the analyzed groups are presented in Table 2.

EMC’s information was collected from 265 patients. The time of the onset of the symptoms was filled out for 222 patients (83.8%). In 193 (73.9%) cards, it was consistent with hospital medical history though it differed in 29 (10.9%) whereas 43 (16.2%) cards were incomplete. The list of medications was filled out for 173 (65.3%) patients, but no information was recorded in 92 (34.7%) cases. Blood glucose measurement was taken for 256 (96.6%) cases. Contact information was noted in 146 (55.1%) of patients’ EMC and 118 (44.9%) had no contact information available for next of kin. The information collected from EMC was compared between CVA and SMs groups. In the study, 199 (75.1%) EMC were from patients with CVA, while 66 (24.9%) belonged to patients with SMs. The statistical difference between the two groups was found for completed information on medication (68.8% vs. 54.5%, p = 0.034) and for all completed information (35.7% vs. 21.2%, p = 0.029). All results are shown in Table 3.

Among the 43 patients with neurological SMs, the most common symptoms were paresis (44.2%), central facial palsy (30.2%) and cerebellar symptoms (25.6%). The most commonly occurring symptom in the non-neurological SMs group was confusion (32.3%).

Statistically significant differences were found in neurological examination between patients with CVA and infections other than neuroinfections, the most common non-neurological SMs (35.5% of all SMs). Patients with infections were significantly less likely to have paresis (79.9% vs. 45.5%, p < 0.021), central facial palsy (58% vs. 18.2%, p = 0.023) and aphasia (23.2% vs. 0%, p = 0.040) than CVA patients. In addition, patients in the group diagnosed with infection were significantly more likely to report confusion (11.6% vs. 54.5%, p < 0.001). Details are presented in Table 4.

Logistic regression analysis revealed a positive association between CVA and the presence of paresis (OR: 3.27, 95% CI: 1.64-6.51), central facial palsy (OR: 2.22, 95% CI: 1.06-4.66), speech disorders (OR: 2.18, 95% CI: 1.02-4.67), pyramidal signs (OR: 4.07, 95% CI: 1.33-12.5) and AH (OR: 2.75, 95% CI: 1.33-5.41). A negative association was observed between CVA and confusion (OR: 0.40, 95% CI: 0.16-0.98) and active cancer (OR: 0.19, 95% CI: 0.06-0.93). Nagelkerke’s R2 was 0.408. The results of logistic regression are presented in Table 5.

Of the 281 patients admitted to the NED, 263 (93.6%) were admitted to hospital wards. A total of 238 (84.7%) patients were referred to the SU or Neurology Department; 202 (71.9%) with CVA and 36 with (12.8%) SMs. Five CVA patients (1.8%) were not admitted to SU or Neurology Department, including 2 cases of SAH and 1 case of hemorrhagic stroke admitted to Neurosurgery Department, 1 case of TIA who refused hospitalization, and 1 case of stroke admitted to COVID-19 ward. Four (1.4%) patients were referred to Neurosurgery Department due to causes other than CVA. Further admissions were made to the Department of Internal Medicine (10, 3.6%), the Department of Vascular Surgery (2, 0.8%), the Department of Cardiosurgery (1, 0.4%), the Department of Toxicology (1, 0.4%), the Intensive Care Unit (1, 0.4%), the Department of Otorhinolaryngology (1, 0.4%). In addition, 1 (0.4%) person refused hospitalization in the Department of Internal Medicine. Departments of Vascular Surgery, Cardiosurgery, Toxicology, COVID-19 and Otorhinolaryngology were located in other hospitals. Thirty-one patients received tPA, including 30/153 (19.6%) patients with ischemic stroke and 1/74 (1.3%) patient with SMs. Nine (5.9%) patients with ischemic stroke had mechanical thrombectomy.

DISCUSSION

According to a comprehensive review by Pohl et al. [8], the overall prevalence of SMs was 24.9%, with considerable differences between studies, varied between 1.5 and 76.3%. However, this review took into consideration very heterogeneous populations of patients in different setups, including pediatric studies. In another adults-only review, the median prevalence of SMs depended on clinical settings – in thrombolysis studies, it was only 10%, in prehospital studies 27%, and in all other studies also 27% [21]. According to a review by Kandiyali et al. [22], positive predictive value for diagnosis of TIA referred to TIA clinics was 12.9-72.5% and for TIA or minor stroke it was 22.0% to 77.9%. In a Polish study, the ambulance physician’s positive predictive value was 34% for TIA [23]. It shows that the prevalence of SMs highly depends on the population assessed, study design, and the conditions of evaluation. Previous studies performed on the Polish adult population showed that 32.2-36.8% of patients referred to the emergency department with suspected CVA had SMs [15, 17, 23]. In the study by Karliński et al. [23], 33% of patients transferred to the Emergency Department with stroke or TIA suspected by ambulance physicians had SMs. In the two newer studies, assessing the prevalence of SMs according to referring entity, SMs were found in 22%, 28% and 46% of patients in case of ambulance physicians, ambulance paramedics and non- ambulance physicians respectively [15, 17]. The prevalence of SMs referred to the emergency department by ambulances in our study was 26.3%, very similar to the prevalence of SMs in other prehospital studies. In comparison to the previously mentioned Polish studies, it was slightly lower than in the population examined in 2006-2007 and similar to the population assessed in 2014, especially as far as the number of SMs referred by paramedics [15, 17, 23]. This is probably related to the prevalence of paramedics in the ambulances that transfer patients to our hospital. The proportion of SMs referred to hospital wards was 57/74 (77%) patients. It is a much larger number in comparison with previous Polish data, where 66 and 59% of patients transferred to the emergency unit by ambulance staff were admitted to hospital wards [15, 17, 23]. A few possible explanations can be offered for these differences. Firstly, it is possible that the performance of ambulance staff has improved since 2014. Secondly, previous studies were conducted in a hospital, where the emergency department is part of the hospital with only psychiatric, neurological, and neurosurgical wards [15, 17, 23]. Patients who were transferred by ambulance to the general hospital, similarly to our study, might have had more severe conditions. Thirdly, this study was conducted during the COVID-19 pandemic, which meant that ambulances tended to avoid transferring patients with less severe conditions to hospitals.

In the review by Pohl et al. [8], the most prevalent SMs were peripheral vestibular disease, toxic/metabolic problems, seizures, functional disorders and migraine. Another review found that in prehospital settings the most common SMs were seizures (19%), non-defined conditions (9%) and ear conditions (5%). For thrombolysis and other studies, these profiles were different with a higher prevalence of headache and psychiatric disorders [21]. In the previous Polish studies, common SMs were vertigo (14-19%), headache (2-10%), seizures (7-11%) and very high blood pressure (7-8%), brain tumor (5-7%), metabolic/electrolyte disturbances (5-12%), infections (4-5%). However, in patients transported by paramedics and ambulance physicians the profile of SMs was more similar. In those cases, seizures, brain tumors and vertigo were the most common neurological SMs, and metabolic/electrolyte disturbance, infections and, in older group of patients, also cardiovascular conditions for non-neurological SMs [15, 17, 23]. The profile of SMs transported to the NED in the hospital was close to other studies but was more dominated by seizures (24%) and infections (15%) with quite a large prevalence of mimics related to other cardiovascular conditions, which is probably mainly associated with the fact that the referring entities were paramedics and ambulance physicians. It is also worth mentioning that in the present study, some of the patients with infections or seizures had a history of brain damage and/or previous focal signs. Sometimes, it is not easy to distinguish old neurological deficits from the new ones. It is a known problem, which causes frequent challenges in the differential diagnosis of CVA and SMs, which probably had some impact on the number of SMs in this study [24]. Peripheral vestibular disorders, psychogenic symptoms, seizures, migraine and toxic-metabolic causes were most common in one study with a strict definition of SMs as a sudden neurological deficit similar to stroke within 24 hours; lack of possibility to exclude SMs based on an initial clinical impression by an emergency physician, basic laboratory tests, and brain CT; need for a vascular neurologist to exclude stroke and TIA; and lack of stroke in MRI. This suggests that these SMs can be most difficult to differentiate from stroke for non-neurologists [25].

Pohl et al. [8], found that SMs patients had a lower burden of stroke risk factors like smoking, AH, dyslipidemia, ischemic heart disease, AF or peripheral vascular disorders, but higher frequency of migraine and cognitive dysfunction. The prevalence of malignancy, seizures and previous stroke were similar in both groups. Another review by McClelland et al. [21], concluded that seizures and medical interviews burdened with psychiatric conditions and migraine were most strongly associated with SMs. That review also suggested, that cardiovascular risk factors were more common in case of stroke patients but some studies reached contradictory results for particular risk factors. Both reviewed articles suggested that stroke patients were older and mostly male in whom focal neurological deficits were more common [8, 21]. In this study, stroke patients, similarly to previous studies, had more commonly AH and focal signs in neurological examination, including paresis, central facial palsy, speech abnormalities, pathological pyramidal signs and head and/or eyeballs rotation. However, confusion and malignancy in medical interviews were more common in SMs. In the present study, the relation of stroke and focal deficits in neurological examination is in agreement with the previously mentioned reviews, but the profile of comorbidities in SMs and stroke is slightly different. The presence of confusion was not associated with stroke in the review by Pohl et al. [8]. This might be related to the inclusion of studies with very heterogeneous designs and populations. However, confusion was described in some previous studies as a symptom associated with SMs, similarly to our study [26, 27]. Isolated altered mental status is relatively rare in stroke, but it is still a frequently encountered stroke chameleon. CVA of the non-dominant inferior parietal lobe, non-dominant temporal gyrus, occipital lobe and vertebrobasilar with thalamic infarcts can present as sudden confusion [10, 28]. According to a large study by Dekker et al. [29], prehospital assessment by paramedics is strongly correlated with an assessment in NIHSS by neurology specialist at the hospital, especially for patients with hemorrhagic stroke and large vessels occlusion. The strongest correlation was for paresis and the weakest for neglect, speech abnormalities and gaze deviation. In this study, the regression analysis revealed differences in speech abnormalities but no differences in neglect and gaze abnormalities. This suggests that speech assessment might be an important educational target in the training of paramedics.

SMs are prevalent and have significant consequences. Some studies reported that SMs occur in over 30% of patients with suspected stroke and 1% to 16% of patients who received thrombolysis [10, 13]. The number of intracranial hemorrhages in this group of patients is lower than in actual strokes, but according to a meta-analysis, it could still occur in 0.5% of them [14]. On the other hand, it should be noted that overdiagnosis of strokes has usually less serious consequences than overlooking a stroke where it is suspected [30]. Due to this fact alone, the main goal of prehospital stroke assessment is to maximize sensitivity to avoid a situation of stroke misdiagnosis [24]. In one study, the authors reported that up to one in seven strokes was missed in the emergency department [31]. SMs also contribute to additional costs, delayed proper diagnosis, and unnecessary diagnostic tests [10-12]. In the study by Dawson et al. [9], patients with SMs were discharged from the stroke unit faster than those with acute strokes. However, admissions of these individuals accounted for 24.2% of all admissions to the SU and occupied approximately 8.2% to 17.2% of bed occupancy. In that study, 19.6% of patients with SMs had to be transferred to another department for treatment, prolonging the time to receive appropriate treatment [9]. Delayed treatment could be extremely dangerous in some cases of SMs, especially those of vascular origin, such as acute limb ischemia or ruptured aortic aneurysm. These vascular SMs investigated in our study in our study and described in the literature require rapid treatment in specialized surgical departments that are not available in many hospitals [32-35].

Accurate prehospital assessment could potentially minimize the negative effects related to SMs and improve the management of stroke patients. In the course of this study, we have also analyzed the availability of the information required by the PNS guidelines in EMCs filled out by EMTs [6]. Most of the cards did not contain all information as required by the Guidelines. The problem was even more common in SMs patients. The most commonly missing or incomplete information in the cards was a contact phone number for patients’ relatives or caregivers which was in fact missing in almost half of all cases. This might cause difficulty in verifying information hence making proper diagnoses and qualifying a patient for reperfusion therapy. Possibly, some of the information that was recorded by ambulance staff and provided to the neurologist but not noted in the cards. The lower number of cards with information completed as required could also be related to the characteristics of SMs patients. In any case, better collection of critical information by EMT might improve the prehospital selection of patients with stroke and SMs. Changes in EMC may be potentially helpful, with consideration given to the symptoms of stroke, making the application of reperfusion therapy by neurologists easier and faster. The information required by the guidelines of the PNS is helpful in qualifying a patient for reperfusion therapy, but not in the identification of SMs other than hypoglycemia [6]. SMs prediction scales might be useful tools helpful in the differential diagnosis of stroke and SMs in the emergency department and prehospital settings [36]. Yet the scales are imperfect and miss approximately 30% of acute strokes [24]. Another potential solution, which could reduce the number of SMs is a usage of mobile stroke units with portable CT scanners, yet so far their availability is low and costs are high [24].

The present study is not without its limitations. Firstly, it is a single-center, retrospective study, with a relatively limited number of SMs patients. More prospective multi- center studies, involving larger sample sizes, are needed for more detailed and objective results. Secondly, some patients with suspected stroke from our catchment area did not come to our center due to the closer location of other stroke units. This could lead to a selection by medical dispatchers, who might direct some of the patients to other centers with different profile, for example due to the absence of cardiology or cardiosurgery wards in our hospital. Thirdly, there may have been a group of patients originally suspected of CVA who suffered from TIA during contact with EMS and when transported to the NEM, but the symptoms subsequently passed, and they were classified as SMs. Potentially few patients with diagnosed TIA could have been misdiagnosed as SMs, but the proportion of TIA in our study is comparable to previous Polish data [23]. Fourthly, the results of our study are difficult to transfer to the general population, due to the regional specificity of operations and transfer of patients to stroke units, as well as the content and requirements of the Polish stroke guidelines, which may differ in detail from other guidelines internationally. However, this seems to be representative for Polish populations of highly urbanized regions. Data assessed in the present study was collected during COVID-19 pandemic, which could have also affected the results, due to factors such as COVID-19 related health issues, changes in healthcare organization and overburden of the healthcare system.

CONCLUSIONS

In our cohort, with 281 subjects suspected of CVA, the diagnosis was confirmed in 207 (73.7%) patients while 74 (26.3%) were SMs. The most common SMs in our population were seizures, infections, and vascular diseases. The findings from the logistic regression analysis indicate that symptoms such as paresis, central facial palsy, speech abnormalities, pyramidal signs and arterial hypertension in medical interviews may suggest the presence of CVA, while confusion and a history of active cancer may serve as indicators of SMs. Furthermore, it was observed that medical emergency cards were completed with consistent frequency across all patients, even though a more frequent completion of the list of medications was noted in the CVA group. It is impossible to completely avoid SMs misdiagnosis, but proper education of ambulance staff and use of tools facilitating fast identification of stroke, and its mimics could potentially help avoiding some obvious cases like Bell’s palsy or mandibular dislocation and improve diagnosis of these less obvious cases. Factors associated with SMs, and stroke identified in the present study might be potentially help in constructing such tools. Moreover, we have concluded that the mandatory data routinely collected by the EMT could be expanded to include the information regarding changes in the patient’s condition during transportation to the hospital (improvement, stability, slow or rapid deterioration).