Prehospital Clinical Presentations and Sex Differences in Stroke Cases and Mimics: A 1-Year Study in a Stroke Unit
Abstract
Introduction: Stroke is a condition demanding prompt treatment. Differentiating stroke cases from mimics poses a challenge in the prehospital setting. An optimal prehospital scale to identify stroke is still not available. The aims of the study were to (i) explore whether dysphagia, visual impairment, skin sensory loss, or combinations of these symptoms could improve diagnostic stroke accuracy beyond FAST (face, arm, speech, and time) scale and (ii) identify sex differences in stroke diagnostic models.
Materials and Methods: We included 319 patients with stroke or transient ischemic attack (TIA) and 119 stroke mimics in a 1-year period in 2013–2014 and 258 stroke/TIA cases and 90 mimics in a validation cohort in 2023, admitted to the Stroke Unit at the University Hospital of North Norway. Retrospective data on clinical presentations were collected from patient records.
Results: Stroke cases were older than mimics and a larger proportion were men. Age explained 7.5% of the variance in odds ratio (OR) for stroke in women and 1.7% in men, while hypertension or coronary heart disease explained 10.2% in women and 3.7% in men. Adding dysphagia to FAST increased OR for stroke from 3.95 (95% confidence interval (CI) 2.00–7.81) to 4.30 (95% CI 2.14–8.64) and explained variance in OR for stroke by 0.5% in women. Adding visual impairment to FAST increased OR from 5.72 (95% CI 2.74–12.0) to 7.69 (95% CI 3.50–16.9) and explained variance in OR for stroke by 1.9% in men. In the validation cohort, the explained variance in OR for stroke did not increase by adding any more clinical presentations to FAST. Stroke mimics accounted for 27.2% and 25.9% in the two cohorts.
Conclusions: By adding clinical presentations to FAST, no meaningful change in diagnostic performance was gained. An optimal scale for prehospital stroke identification is still needed.
1. Introduction
Stroke is a major cause of disability and death worldwide [1, 2]. As the population ages, the number of future stroke cases is expected to increase, despite reduction in stroke rates and improved stroke treatment [1, 3].
Public awareness and prompt prehospital recognition are crucial to improve clinical outcomes, as stroke outcomes are highly time dependent [3, 4]. Although more than 20 prehospital scales have been developed, no optimal prehospital scale exists [3, 5–8]. The FAST (face, arm, speech, and time) scale is widely used as it is simple and easy to communicate [9]. FAST has almost similar test performances as the more complex National Institutes of Health Stroke Scale (NIHSS) [6, 10], which has a sensitivity of 91% and a specificity of 52% in recognizing stroke [11]. NIHSS is considered the gold standard within the hospital setting, especially for identification of large vessel occlusions [5, 11–14].
Stroke cases presenting with subtle symptoms from posterior circulation, which still could benefit from reperfusion treatment, may be overlooked [8]. To enhance diagnostic accuracy, scales incorporating additional symptoms as visual disturbances and balance impairment have been designed [15]. However, none of these scales have demonstrated superior test performances compared to the FAST scale [6, 8, 11, 15, 16].
Several medical or neurological conditions may mimic stroke symptoms, and as many as 26% of these may receive thrombolysis [17, 18]. Approximately 30% of suspected stroke cases, both prehospitally and within a stroke unit, do not meet stroke criteria but may still require urgent medical attention [17, 19, 20].
The aims of this 1-year clinical study were to (i) explore whether dysphagia, visual impairment, skin sensory loss, or combinations of these symptoms could improve diagnostic stroke accuracy beyond FAST, (ii) identify sex differences in stroke diagnostic models, and (iii) investigate sex differences in clinical presentations of stroke cases and mimics admitted to the Stroke Unit at the University Hospital of North Norway.
2. Materials and Methods
2.1. Study Participants
A total of 451 patients were admitted to the Stroke Unit at the University Hospital of North Norway, Tromsø, during a 1-year study period from 1 October 2013 to 30 September 2014. Three patients with subarachnoid hemorrhage and 10 patients with incomplete data were excluded. Of the remaining 438 patients, 219 had an ischemic stroke, 25 had an intracerebral hemorrhage (defined as stroke), and 56 had a transient ischemic attack (TIA) based on the International Classification of Diseases, Tenth Revision (ICD-10) codes. After thorough revision of the patient records (referral notes, in-hospital records, and imaging reports), 19 new ischemic stroke cases were identified. Thus, 319 patients had a stroke or TIA, while 119 patients had no stroke and were defined as stroke mimics. Written informed consent was obtained from all patients or their next of kin on admission. The study was approved by the Regional Committee of Medical and Health Research Ethics (REK 2012/664).
As the data was 10 years old and there have been numerous campaigns focusing on stroke symptoms, the main statistical analyses were repeated in a validation cohort (2023) to test the representativeness. A total of 363 patients were admitted to the Stroke Unit at the University Hospital of North Norway, Tromsø, during a 1-year period (1 January 2023 to 31 December 2023). Fifteen patients did not consent to participate in the study. Of the remaining 348 patients, 258 patients had stroke/TIA (191 had an ischemic stroke, 24 had an intracerebral hemorrhage, and 43 had a TIA). Ninety patients (25.9%) were stroke mimics.
2.2. Validation of Stroke Diagnoses
All stroke and TIA diagnoses in 2013/2014 were validated by H.M.F. using the American Heart Association/American Stroke Association definitions [21, 22]. A random sample of 68 patients with a stroke or TIA diagnoses was validated by T.E. and D.S.H., and the diagnoses were then compared with those set by H.M.F. We calculated the Cohen’s kappa coefficient for agreement between the authors (H.M.F. vs. T.E. and D.S.H.). The coefficient was 0.93 (very strong) for differentiating stroke cases from mimics and 0.85 (strong) for diagnoses of stroke subtypes, indicating very good agreement between the investigators.
All stroke/TIA cases and stroke mimic diagnoses in 2023 were validated by D.S.H., N.J.G., or I.M. Complex cases were further reviewed to reach a consensus by D.S.H. and I.M., both consultants in internal medicine and geriatrics.
2.3. Variables
The clinical symptoms were recorded by general practitioners, paramedics, and/or physicians at the emergency department. The information was collected retrospectively from referral notes and patient records, including age, sex, and dichotomous variables (yes/no) such as previous or current smoking, hypertension, coronary heart disease (CHD), and cognitive impairment. Modified Rankin scale (mRS) range 0–5 was dichotomized as “yes” (Grades 3–5, indicating moderate to severe disability) or “no” (Grades 0–2) [23].
Clinical presentations focused on these dichotomized variables (yes/no): aphasia, dysarthria, one-sided symptoms, sudden onset of symptoms, paralysis, dysphagia (“yes” for Grades 2–4 and “no” for Grade 1) [24], visual impairment (defined as disturbances in gaze, visual field, or vision), sensory skin loss (defined as skin hypoesthesia or altered skin sensation), syncope, confusion, and headache.
The FAST model included a sudden onset of the symptoms and at least one of the following symptoms: aphasia, dysarthria, one-sided symptoms, or paralysis. We designed four new FAST categories by adding dysphagia (FAST2), visual impairment (FAST3), sensory skin loss (FAST4) to FAST, and FAST5 included FAST2, FAST3, or FAST4.
2.4. Statistical Analyses
The age of patients with stroke and mimics is presented as mean and standard deviation (SD). The categorical variables are presented as number and percentages (%). Similarly, we compared the characteristics in women with and without stroke and men with and without stroke, and we tested for sex differences in stroke cases and mimics. Age-adjusted differences between groups were tested using analysis of variance (ANOVA). Characteristics of the patients with ischemic stroke, hemorrhagic stroke, and TIA were compared with mimics (Table S1).
To examine the association between clinical presentations in stroke cases versus mimics, we estimated odds ratio (OR) with 95% confidence interval (CI) using logistic regression analysis in a stepwise backward and forward approach. The initial logistic regression models included age, sex, and all variables with a p value < 0.15 in the ANOVA comparison of groups. The final logistic regression models included exposure variables that remained significantly associated with stroke versus mimics (p < 0.05). We tested the OR by age (5 years), male sex, hypertension or CHD, FAST, and FAST2–FAST5. The variance in OR for stroke explained by age, sex, hypertension or CHD, and FAST before and after inclusion of additional symptoms (FAST2–FAST5) was calculated and presented as squared R (R2). This was tested in all patients and stratified by sex. We calculated sensitivity, specificity, and positive and negative predictive values. The main statistical analyses were repeated in the validation cohort. We used SAS software package, v9.4 (SAS Institute, Cary, North Carolina, United States), for the data analyses. p values < 0.05 were considered significant.
3. Results
Age-adjusted characteristics for 319 stroke cases and 119 mimics are shown in Table 1. Patients with a stroke had a mean age of 71.5 years and were 7.6 years older than mimics. A larger proportion were men (53.6% vs. 42.0%), and a larger proportion had a history of hypertension or CHD than mimics (all p ≤ 0.05). Aphasia, dysarthria, one-sided symptoms, sudden onset of symptoms, paralysis, dysphagia, and visual impairment were present to a larger extent in stroke cases than mimics. The percentage of patients with clinical presentations of all FAST categories was higher in stroke cases (range 86%–92%) than mimics (range 55%–69%), for all patients (Table 1), for women, and for men (Table 2), p < 0.001, for all comparisons.
| Strokea (n = 319) | No stroke (n = 119) | p valuesb | |
|---|---|---|---|
| Characteristics | |||
| Age (years, means ± standard deviation) | 71.5 ± 15.2 | 63.9 ± 18.2 | < 0.001 |
| Male sex | 171 (53.6) | 50 (42.0) | 0.010 |
| Currently or previously smoking | 82 (26.1) | 27 (23.7) | 0.137 |
| Previous ischemic stroke | 191 (59.9) | 50 (42.4) | 0.089 |
| History of hypertension | 171 (53.6) | 41 (34.5) | 0.011 |
| History of coronary heart disease (CHD) | 81 (25.4) | 13 (10.9) | 0.008 |
| History of hypertension or CHD | 202 (63.3) | 45 (37.8) | < 0.001 |
| History of cognitive impairment | 40 (12.5) | 19 (16.0) | 0.061 |
| Modified Rankin scale grade ≥ 3 | 44 (13.8) | 17 (14.3) | 0.138 |
| Clinical presentations | |||
| Aphasia | 91 (28.5) | 21 (17.7) | 0.045 |
| Dysarthria | 112 (35.1) | 19 (16.0) | < 0.001 |
| One-sided symptoms | 257 (80.6) | 77 (64.7) | < 0.001 |
| Sudden onset of symptoms | 297 (93.4) | 82 (68.9) | < 0.001 |
| Paralysis | 225 (70.5) | 62 (52.1) | < 0.001 |
| Dysphagia | 86 (27.0) | 10 (8.4) | < 0.001 |
| Visual impairmentc | 68 (21.5) | 9 (7.6) | 0.002 |
| Sensory skin loss | 72 (22.6) | 34 (28.6) | 0.770 |
| Syncope | 24 (7.6) | 12 (10.1) | 0.364 |
| Confusion | 52 (16.4) | 17 (14.3) | 0.757 |
| Headache | 59 (18.5) | 34 (28.6) | 0.318 |
| FAST | 273 (85.6) | 65 (54.6) | < 0.001 |
| FAST2 | 279 (87.5) | 68 (57.1) | < 0.001 |
| FAST3 | 283 (88.7) | 70 (58.8) | < 0.001 |
| FAST4 | 280 (87.8) | 75 (63.0) | < 0.001 |
| FAST5 | 293 (91.9) | 82 (68.9) | < 0.001 |
- Note: Values are number (percentage) if no other information is provided. FAST included aphasia or dysarthria or one-sided symptoms or paralysis and sudden onset = 1. FAST2 included FAST or dysphagia = 1. FAST3 included FAST or visual impairment = 1. FAST4 included FAST or sensory loss = 1. FAST5 included FAST2, FAST3, or FAST4 = 1. Entries in bold are values of p < 0.05.
- aStroke includes ischemic stroke, hemorrhagic stroke, and transient ischemic attack.
- bAge-adjusted comparison of those with and without stroke using analysis of variance.
- cVisual impairment was defined as disturbances in gaze, visual field, and/or vision.
| Women | Men | Sex differences | ||||||
|---|---|---|---|---|---|---|---|---|
Strokea (n = 148) |
No stroke (n = 69) |
pb | Stroke (n = 171) |
No stroke (n = 50) |
pb | pc | pd | |
| Characteristics | ||||||||
| Age (years, means ± SD) | 74.1 ± 16.4 | 63.6 ± 17.4 | < 0.001 | 69.2 ± 13.8 | 64.3 ± 19.5 | 0.048 | 0.004 | 0.831 |
| Currently or previously smoking | 32 (22.1) | 14 (21.5) | 0.239 | 50 (29.6) | 13 (26.5) | 0.436 | 0.399 | 0.535 |
| Previous ischemic stroke | 84 (56.8) | 27 (39.1) | 0.338 | 107 (62.6) | 23 (46.9) | 0.202 | 0.030 | 0.459 |
| History of hypertension | 82 (55.4) | 23 (33.3) | 0.065 | 89 (52.1) | 18 (36.0) | 0.106 | 0.972 | 0.807 |
| History of coronary heart disease | 21 (14.2) | 1 (1.5) | 0.020 | 60 (35.1) | 12 (24.0) | 0.291 | < 0.001 | < 0.001 |
| History of hypertension or CHD | 91 (61.5) | 23 (33.3) | 0.010 | 111 (64.9) | 22 (44.0) | 0.028 | 0.147 | 0.240 |
| History of cognitive impairment | 20 (13.5) | 11 (15.9) | 0.134 | 20 (11.7) | 8 (16.0) | 0.243 | 0.945 | 0.961 |
| Modified Rankin scale grade ≥ 3 | 27 (18.2) | 12 (17.4) | 0.249 | 17 (9.9) | 5 (10.0) | 0.555 | 0.190 | 0.220 |
| Clinical presentations | ||||||||
| Aphasia | 46 (31.1) | 14 (20.3) | 0.220 | 45 (26.3) | 7 (14.0) | 0.084 | 0.495 | 0.381 |
| Dysarthria | 47 (31.8) | 13 (18.8) | 0.365 | 65 (38.0) | 6 (12.0) | < 0.001 | 0.134 | 0.308 |
| One-sided symptoms | 120 (81.1) | 44 (63.8) | 0.009 | 137 (80.1) | 33 (66.0) | 0.022 | 0.769 | 0.789 |
| Sudden onset of symptoms | 134 (90.5) | 46 (66.7) | < 0.001 | 163 (95.9) | 36 (72.0) | < 0.001 | 0.055 | 0.539 |
| Paralysis | 104 (70.3) | 35 (50.7) | 0.026 | 121 (70.8) | 27 (54.0) | 0.025 | 0.708 | 0.721 |
| Dysphagia | 40 (27.2) | 5 (7.3) | 0.015 | 46 (26.9) | 5 (10.0) | 0.031 | 0.529 | 0.612 |
| Visual impairmente | 33 (22.3) | 7 (10.1) | 0.048 | 35 (20.8) | 2 (4.0) | 0.009 | 0.879 | 0.216 |
| Sensory skin loss | 38 (25.9) | 18 (26.1) | 0.457 | 34 (19.9) | 16 (32.0) | 0.210 | 0.071 | 0.414 |
| Syncope | 13 (8.8) | 8 (11.6) | 0.586 | 11 (6.4) | 4 (8.0) | 0.636 | 0.522 | 0.536 |
| Confusion | 30 (20.4) | 8 (11.6) | 0.544 | 22 (12.9) | 9 (18.0) | 0.241 | 0.186 | 0.342 |
| Headache | 26 (17.6) | 23 (33.3) | 0.339 | 33 (19.3) | 11 (22.0) | 0.989 | 0.766 | 0.168 |
| FAST | 125 (84.5) | 37 (53.6) | < 0.001 | 148 (86.6) | 28 (56.0) | < 0.001 | 0.464 | 0.786 |
| FAST2 | 127 (85.8) | 38 (55.1) | < 0.001 | 152 (88.9) | 30 (60.0) | < 0.001 | 0.271 | 0.585 |
| FAST3 | 129 (87.2) | 41 (59.4) | < 0.001 | 154 (90.1) | 29 (58.0) | < 0.001 | 0.307 | 0.899 |
| FAST4 | 130 (87.8) | 45 (65.2) | < 0.001 | 150 (87.7) | 30 (60.0) | < 0.001 | 0.894 | 0.583 |
| FAST5 | 136 (91.9) | 49 (71.0) | < 0.001 | 157 (91.8) | 33 (66.0) | < 0.001 | 0.814 | 0.585 |
- Note: Values are number (percent) if no other information is provided. FAST included aphasia or dysarthria or one-sided symptoms or paralysis and sudden onset = 1. FAST2 included FAST or dysphagia = 1. FAST3 included FAST or visual impairment = 1. FAST4 included FAST or sensory loss = 1. FAST5 included FAST2, FAST3, or FAST4 = 1. Entries in bold are values of p < 0.05.
- aStroke includes ischemic stroke, hemorrhagic stroke, and transient ischemic attack.
- bp values for comparison of those with and without stroke within each sex adjusted for age using analysis of variance.
- cp values for comparison of women and men with stroke adjusted for age using analysis of variance.
- dp values for comparison of women and men without stroke adjusted for age using analysis of variance.
- eVisual impairment was defined as disturbances in gaze, visual field, and/or vision.
Sex differences in characteristics, and prehospital clinical presentations, for stroke cases and mimics are shown in Table 2. Among stroke cases, women were older than men and had a lower percentage of previous ischemic stroke (56.8% vs. 62.6%) and CHD (14.2% vs. 35.1%). None of the prehospital clinical presentations nor FAST categories differed between sexes. Among mimics, there was no sex difference, except for a higher percentage of CHD in men compared to women. In sensitivity analyses, these results were similar after exclusion of TIA and hemorrhagic stroke cases.
The OR for stroke and the explained variance (R2) in OR for stroke are presented in Table 3. The OR for stroke by age was 1.19 (1.09–1.30) in women and 1.10 (1.00–1.22) in men. R2 by age was 7.5% in women and 1.7% in men. The OR for stroke by male versus female sex was 1.74 (1.12–2.70). For hypertension or CHD, R2 was 10.2% in women and 3.7% in men.
| Characteristics | All (n = 438) | R2 | Women (n = 217) | R2 | Men (n = 221) | R2 |
|---|---|---|---|---|---|---|
| Age (per 5 years) | 1.15 (1.08–1.23)a | 0.040 | 1.19 (1.09–1.30) | 0.075 | 1.10 (1.00–1.22) | 0.017 |
| Male sex | 1.74 (1.12–2.70)b | 0.011 | ||||
| Hypertension or CHD | 2.21 (1.39–3.52)c | 0.077 | 2.31 (1.21–4.41)b | 0.102 | 2.36 (1.24–4.47) | 0.037 |
| FAST | 4.69 (2.86–7.71)d | 0.145 | 3.95 (2.00–7.81)d | 0.165 | 5.72 (2.74–12.0)b | 0.109 |
| FAST2 | 4.91 (2.94–8.18)d | 0.145 | 4.30 (2.14–8.64)d | 0.170 | 5.81 (2.71–12.4)b | 0.104 |
| FAST3 | 5.34 (3.15–9.04)d | 0.149 | 3.93 (1.92–8.05)d | 0.159 | 7.69 (3.50–16.9)b | 0.128 |
| FAST4 | 4.12 (2.44–6.95)e | 0.135 | 3.37 (1.61–7.05)d | 0.144 | 5.53 (2.60–11.8)b | 0.100 |
| FAST5 | 5.01 (2.79–8.98)e | 0.139 | 4.22 (1.83–9.71)d | 0.150 | 6.58 (2.88–15.0)b | 0.102 |
- Note: FAST included aphasia or dysarthria, one-sided symptoms or paralysis, and sudden onset = 1. FAST2 included FAST or dysphagia = 1. FAST3 included FAST or visual impairment = 1. FAST4 included FAST or sensory loss = 1. FAST5 included FAST2, FAST3, or FAST4 = 1.
- Abbreviation: CHD, coronary heart disease.
- aAdjusted for sex.
- bAdjusted for age.
- cAdjusted for sex and age.
- dAdjusted for age and hypertension or CHD.
- eAdjusted for age, sex, and hypertension or CHD.
In women, OR for stroke increased from 3.95 (95% CI 2.00–7.81) to 4.30 (95% CI 2.14–8.64) and R2 increased 0.5% by adding dysphagia to FAST. In men, the OR increased from 5.72 (95% CI 2.74–12.0) to 7.69 (95% CI 3.50–16.9) and R2 increased by 1.9% by adding visual impairment to FAST. Adding skin sensory loss to FAST did not increase OR in neither sex. By including additional prehospital clinical presentations to FAST, the sensitivity increased by 1.9%–6.2%; however, there was a tradeoff for specificity that decreased by 2.5%–14.4% (Table S2).
Stroke mimics accounted for 27.2% of the patients, and their main diagnoses were observation for systemic nervous system disorder, disorder of vestibular function and vertigo, nerve root and plexus disorders, mono- and polyneuropathies, sequelae after previous stroke, and headache (Table 4).
| Diagnosis | ICD-10 codes | n |
|---|---|---|
| Observation for systemic nervous system disorder | R26, R47, Z03 | 19 |
| Disorder of vestibular function and vertigo | H81, R42 | 16 |
| Nerve root and plexus disorders, mono- and polyneuropathies | G51, G54, G56, G62, G69, G72, S04, S14, T79 | 14 |
| Sequelae after previous stroke | G91, G93, I69 | 10 |
| Headache | G44, R51 | 9 |
| Infection | B34, G04, J12, N10, N30, N39 | 6 |
| Epilepsy | G40 | 5 |
| Hypotension/syncope | I95, R55 | 5 |
| Delirium and/or dementia | F05, G30 | 5 |
| Psychiatric diagnosis | F32, F43, F44, F45, R06 | 5 |
| Anesthesia/paresthesia of the skin | R20 | 5 |
| Vision/eye impairment | G90, H53 | 3 |
| Subdural hematoma | I62, S06 | 3 |
| Carotid stenosis or dissection | I65, I72 | 3 |
| Spinal stenosis | M48 | 2 |
| Drug intoxication or side effects | Y4n | 2 |
| Diabetes mellitus | E11, E16 | 2 |
| Malignancy | C56, C90 | 2 |
| Migraine | G43 | 1 |
| Multiple sclerosis | G35 | 1 |
| Rheumatism | M79 | 1 |
- Note: ICD-10 code = the International Classification of Disease 10th revision codes.
The mean age was 3 years higher in both stroke cases and mimics in the validation cohort than the original cohort (Table 1 and Table S3). The differences in mean age and sex distribution between stroke cases and mimics were similar in both cohorts, although the higher percentage of male sex did no longer reach statistical significance. In the validation cohort, a higher proportion had previous ischemic stroke, cognitive impairment, and headache in mimics than the stroke group. A higher proportion of patients had the typical prehospital stroke symptoms and FAST categories in both stroke cases and mimics in the validation cohort than the original cohort (Table 1 and Table S3). The presence of dysphagia and visual impairment as well as FAST categories was significantly different between stroke cases and mimics in both cohorts. The OR for stroke and R2 by age were similar in both cohorts, while OR for sex, hypertension, or CHD was no longer significant in the validation cohort. However, for both sexes in the validation cohort, there was no increase in R2 by adding any of the clinical presentations to FAST (Table S4).
4. Discussion
In the original study cohort, stroke cases were older, a larger proportion were men, and they had more frequent hypertension or CHD than mimics. In stroke cases, women were older than men, and women had a lower percentage of previous ischemic stroke and CHD. Age as well as hypertension or CHD explained more of the variance in OR for stroke in women than in men. By adding dysphagia to FAST in women and visual impairment to FAST in men, the OR for stroke and R2 increased modestly. By combining these clinical presentations with FAST, a modest increase in sensitivity was achieved, while the specificity was reduced to a larger extent. There seems to be a ceiling effect, which is even clearer in the results from the validation cohort, and thus, no meaningful gain was achieved by additional clinical presentations.
To our knowledge, dysphagia has not been previously included as part of a prehospital stroke scale. Dysphagia rarely manifests as a single stroke presentation but may occur in posterior stroke [25]. While several studies have reported a higher incidence of dysphagia in women with acute stroke [26–28], our findings revealed no sex differences in both cohorts, consistent with a Danish study, which may be more comparable to a Norwegian population [29].
In the present study, the OR for stroke increased by adding visual impairment to FAST only in men, and R2 increased correspondingly by 1.9%. This is in line with results from a large comparative study of prehospital scales, where the inclusion of visual impairment enhanced the diagnostic accuracy of stroke [13]. In another study, no sex difference was reported when presenting visual impairment and field defects [30]. However, in the BEFAST study, inclusion of diplopia did not improve diagnostic accuracy compared to FAST [16]. A possible explanation for our findings could be the use of a broader definition of visual impairment, including gaze deviation, diplopia, or visual field loss [16, 30, 31]. Other explanations for the diversity could be the context for assessment of vision or the qualifications of personnel [31]. If visual disturbances were implemented in a prehospital scale, personnel would need more training on these deficits to improve diagnostic stroke accuracy [31].
In two studies on sex differences in acute stroke, sensory skin loss was more common in men [32] and more often part of a final acute stroke/TIA diagnosis in men than women [30]. Interestingly, skin sensory disturbances were strongly associated with prehospital delay [33]. In the current study, a marginally higher percentage of women than men with stroke had sensory skin loss; however, the difference was not significant and OR for stroke did not increase by adding sensory skin loss to FAST in neither sex.
Women tend to have more nonfocal and atypical stroke symptoms than men [11, 34–36], leading to a higher frequency of being diagnosed with stroke mimics [34, 36]. These patients presented with a variety of diagnoses, including some serious medical conditions necessitating urgent clinical treatment [17, 18, 37]. The absence of CHD or lack of cardiovascular risk factors has been previously reported in female stroke mimics [17, 37, 38]. In the current study, mimics accounted for nearly a third of the patients who were admitted to the stroke unit. Our findings are largely in agreement with previous studies. However, among the mimics, women were not younger than men as previously reported, but they had lower percentage of CHD, and this observation was confirmed in the validation cohort.
In Norway, medical services are organized with a first clinical assessment by a paramedic or a general practitioner and next by an emergency doctor or neurologist upon hospital admission, who selects patients to the stroke unit. No specific training for prehospital stroke triage was given before the original study period, but structured training to paramedics was implemented in our region in 2018, prior to the validation study period. Despite this two-step clinical approach, a significant number of patients still enter the stroke unit without meeting the stroke criteria in both cohorts. So far, clinical presentation alone may not be sufficient to differentiate between acute stroke and mimics.
5. Strengths and Limitations
A strength of the study lies in the thorough validation of the stroke diagnoses as well as calculation of OR and explained variance in OR for stroke. Moreover, the main analyses were repeated in a validation cohort using newer data from the same geographic region. The study is relevant for health professionals involved in clinical stroke assessment because of its symptom-based approach. The NIHSS was not implemented during the original study period but was available for all patients in the validation study period, which probably enhanced data quality. Results from dysphagia screening were mainly obtained from intrahospital records. The stroke mimic diagnoses in the original cohort were based on discharge diagnoses and were not validated, in contrast to data in the validation cohort. The study sample size was moderate but represented a typical patient sample admitted to a stroke unit, allowing for generalizability to similar populations.
6. Conclusions
This study confirms and adds knowledge about prehospital clinical presentations and sex differences in acute stroke syndromes. By adding more clinical presentations to FAST, the sensitivity increased at the cost of decreased specificity, and a clear ceiling effect was seen. However, no meaningful change in diagnostic performance was gained when adding items. Notably, 27.2% and 25.9% of the patients admitted to the stroke unit did not have a stroke in both study cohorts, emphasizing the importance of careful clinical attention to all patients with stroke symptoms. Many of the patients presented with medical emergencies requiring prompt medical treatment. An optimal scale for prehospital stroke identification is still needed.
Conflicts of Interest
The authors declare no conflicts of interest.
Funding
The Northern Norway Regional Health Authority funded the study (SFP 1056-12).
Open Research
Data Availability Statement
Due to protection of privacy under General Data Protection Regulation and Norwegian law, the individual-level data can only be made available after approval by the Regional Committee for Medical and Health Research Ethics.
