Sex Differences in Clinical Presentation of Women and Men Evaluated at a Comprehensive Stroke Center for Suspected Stroke

The initial diagnosis of stroke remains primarily clinical and requires accurate interpretation of symptoms and signs. Misdiagnosis is associated with adverse outcomes due to delayed or missed opportunities for acute treatment and secondary prevention [1]. Women are more likely to be diagnosed with a stroke mimic and more often experience nonfocal symptoms when having a stroke [2‒4]. Therefore, women with stroke face a disproportionate risk of being misdiagnosed, and this may contribute to worse outcomes [5].

Due to its time sensitivity, acute stroke management is particularly susceptible to the impact of sex-related disparities. Recently, authors have reported that women with stroke were less likely to be managed with standard prehospital stroke protocols as compared to men in a large population-based analysis [6]. In addition, studies suggest that women are at higher risk of delayed consultation and a protracted diagnosis [7, 8]. Although differences in clinical presentations may contribute to these sex-related disparities and thus recognizing them could help optimize individual patient management, their impact on acute management and final diagnoses of patients with suspected stroke remains uncertain. We therefore aimed to (1) characterize sex differences in clinical characteristics of patients who were treated at our comprehensive stroke center for suspected stroke as well as to (2) investigate the association of presenting symptoms and final diagnosis.

This was a retrospective analysis of consecutive patients with suspected stroke who were treated at the University Hospital Essen between January 2017 and December 2021. Our center is based at a tertiary level hospital in a large metropolitan area in Germany. Alongside with smaller centers, it covers several cities with a total population of more than one million inhabitants and provides comprehensive stroke care, especially for the most severely affected patients. Generally, patients with suspected stroke reach our clinic via emergency medical services (55%), as secondary referrals from other institutions (30%) or on their own initiative (15%). For this study, we included all adult patients who presented to our center due to clinically suspected stroke, as diagnosed in the field, at a referring stroke or non-stroke center or directly at our emergency department (ED). All patients underwent initial neurological evaluation in the ED upon arrival to our hospital. We excluded patients who were under 18 years of age or in whom information on the initial neurological examination or final diagnosis were missing.

We collected data on demographics, vascular comorbidities, signs and symptoms upon presentation to our neurological ED as recorded on the National Institutes of Health Stroke Scale (NIHSS), diagnostic procedures and treatment from our local stroke registry database, ED records, and final discharge letters. In accordance with previous definitions, nonfocal symptoms included impairments in level of consciousness, orientation, and completion of tasks, while aphasia, dysarthria, unilateral sensory deficits, diplopia, visual field deficiency, gaze deviation, ataxia, limb paresis, facial weakness, and neglect were considered as focal symptoms [2]. Final diagnoses as made by the treating neurologist were recorded and classified as cerebrovascular disease (ischemic stroke, transient ischemic attack, intracerebral hemorrhage) or stroke mimics as described previously [9]. Briefly, diagnoses were determined according to standard definitions [10], based on clinical findings and the additional diagnostic workup (mainly imaging). The initial diagnosis in the ED was made by a junior stroke neurologist and validated by a supervising senior neurologist. In case patients were further treated on our stroke unit, the initial diagnosis was additionally validated (or rejected) by the responsible senior stroke neurologist. In patients who were not treated on our stroke unit (mostly stroke mimics), the initial diagnosis was reviewed independently by two stroke neurologists (J.K.E. and B.F.).

Baseline and procedural data in women and men are presented as counts and percentages or median (interquartile range) as appropriate. We used the Kolmogorov-Smirnov test to ascertain the distribution of data. According to the level of measurement, Fisher’s exact test or nonparametric Mann-Whitney U test were used for further comparisons between groups. We conducted a multivariate binary logistic regression analysis using the forced entry method to assess the association of individual NIHSS items with a diagnosis of cerebrovascular disease in women and men. Adjustments were made for known vascular comorbidities (atrial fibrillation, arterial hypertension, diabetes mellitus, prior stroke) as well as age. All statistical tests were two sided, and p values of <0.05 were considered statistically significant. Given the exploratory nature of our analysis, no adjustments for multiple comparisons were made. Statistical analyses were performed with SPSS, version 29 (IBM Corp., Armonk, NY, USA).

We included 6,069 patients with suspected stroke in our final analysis of whom 3,022 (49.8%) were women (Fig. 1). Cerebrovascular disease was diagnosed in 85.2% of women and 88.0% of men (p = 0.002). Women were older (median age 76.0, IQR 64.0–84.0 vs. 69.0, IQR 59.0–80.0, p < 0.001) and less likely to be functionally independent (premorbid modified Rankin Scale ≤2, 77.3% vs. 81.6%, p < 0.001) when compared to men (Table 1). They had a higher prevalence of atrial fibrillation (29.8 vs. 25.5, p < 0.001), whereas diabetes (25.7% vs. 28.5%, p = 0.015) and a history of prior stroke (24.3% vs. 28.2%, p < 0.001) were less common in women. When arriving at our hospital, median time from last known well was shorter in women (6.3 h, IQR 1.7–20.4 vs. 7.1 h, IQR 1.9–27.3, p = 0.002). No differences in total NIHSS scores upon presentation were observed among women and men with suspected stroke. However, when limiting the analysis to patients who were diagnosed with cerebrovascular disease only, the median NIHSS score was slightly higher in women (5.0, IQR 1.0–12.0 vs. 4.0, IQR 1.0–11.0, p = 0.015) (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000543835).

Although the overall diagnoses of ischemic stroke and transient ischemic attack were balanced between groups, women with ischemic stroke were more likely to have large vessel occlusion (LVO) (26.1% vs. 20.1%, p < 0.001). They more often underwent endovascular treatment (19.2% vs. 14.9%, p < 0.001) and tended to be treated with intravenous thrombolysis more frequently (38.9% vs. 35.8%, p = 0.062). Intracerebral hemorrhage was less common in women as compared to men (10.7% vs. 12.6%, p = 0.020). Overall, women were more likely to be diagnosed with a stroke mimic (14.8% vs. 12.0%, p = 0.002). Stroke mimic etiologies were mostly balanced between groups (online suppl. Fig. 1); however, women were more frequently diagnosed with migraine (5.8% vs. 2.7%). Women were less likely to be admitted to our hospital (91.1% vs. 92.9%, p = 0.009).

The most common symptom in our cohort of suspected stroke patients was limb paresis, which occurred in almost half of individuals, without any difference between groups (Fig. 2). Yet, the prevalence of several other individual NIHSS items varied among women and men. Women were more likely to present with aphasia (31.4% vs. 27.7%, p = 0.002), neglect (15.5% vs. 12.8%, p = 0.003), and gaze deviation (21.0% vs. 18.8%, p = 0.034). They also more frequently had nonfocal symptoms including impairments in completion of tasks (31.2% vs. 26.0%, p < 0.001), level of consciousness (17.0% vs. 14.6%, p = 0.012), as well as orientation (42.5 vs. 36.4%, p < 0.001). Limb ataxia (8.1% vs. 11.2%, p < 0.001) and dysarthria (44.0% vs. 46.8%, p = 0.030) were less common in women. Most of these differences persisted and were more pronounced when analyzing patients with cerebrovascular disease only, even in those who had acute ischemic stroke without LVO (online suppl. Tables 2, 3).

Multivariate binary logistic regression adjusted for age and cardiovascular comorbidities revealed that the majority of NIHSS items were significantly associated with a diagnosis of cerebrovascular disease in women as well as in men (Fig. 3). Neglect (aOR, 95% CI: 3.814, 1.147–12.679; p = 0.029 vs. 1.335, 0.603–2.953; p = 0.476) and gaze deviation (aOR, 95% CI: 2.836, 1.477–5.447; p = 0.002 vs. 1.591, 0.945–2.679; p = 0.081) were independent positive predictors of being diagnosed with cerebrovascular disease in women but not in men. Impairments in orientation (aOR, 95% CI: 0.476, 0.341–0.664; p < 0.001 vs. 0.454, 0.321–0.642; p < 0.001) and consciousness (aOR, 95% CI: 0.513, 0.331–0.796; p = 0.003 vs. 0.630, 0.390–1.018; p = 0.059) were negatively associated with cerebrovascular disease diagnosis in both sexes; however, the latter did not reach statistical significance in men.

We analyzed more than 6,000 consultations for suspected stroke at a comprehensive stroke center, and our study revealed several differences in clinical characteristics between women and men. Although focal symptoms were common and largely overlapped between sexes, women were more likely to have cortical as well as nonfocal symptoms. These findings were mainly driven by patients with cerebrovascular disease, who were the majority in our cohort.

Furthermore, women were older, had worse functional status, more atrial fibrillation, and less prevalence of diabetes and prior stroke when compared to men. These baseline differences have been observed in previous analyses, and authors have highlighted their contribution to sex-related disparities in the clinical presentation and outcome of ischemic stroke patients [11‒14]. Therefore, the differences in presenting symptoms that were observed in our study might as well be driven by differences in baseline risk factors, which result in distinct etiologies and subtypes of stroke. The higher prevalence of cortical symptoms in women could, for example, be deriving from a higher rate of cortical infarction due to cardioembolic stroke. The burden of atrial fibrillation is known to be higher among women and higher rates of cardioembolic stroke as compared to men have been reported before [15‒17]. In our study, the overall prevalence of atrial fibrillation was relatively high (29.8% in women and 25.5% in men), which is consistent with previous analyses from our institution [9, 18]. Comparable rates of atrial fibrillation have also been reported in similar cohorts at other centers [11, 19]. However, despite this high overall prevalence of atrial fibrillation, a significant association of cortical symptoms with a diagnosis of cerebrovascular disease prevailed in women but not in men despite adjustments for age and cardiovascular comorbidities. Furthermore, the greater prevalence of cortical symptoms in women seemed not to be solely driven by higher rates of LVO since differences persisted in ischemic stroke patients without LVO. Thus, there might be additional underlying factors. A recent MRI-based study found that women with left parieto-occipital cortical infarction experienced more severe stroke symptoms as compared to men, despite similar infarct volumes [20]. Since the underlying mechanisms remain uncertain, these findings merit further investigation.

We also observed that nonfocal symptoms occurred more frequently in women and were associated with a stroke mimic diagnosis in both sexes. This highlights the risk of potential misdiagnosis in women with stroke, which has been observed previously [21]. As an important limitation due to its nature, our analysis is constrained by the diagnostic accuracy of standard clinical care; thus, we cannot exclude misdiagnosis in some cases. The susceptibility to diagnostic error may be significantly higher among specific subgroups and patients with nonfocal symptoms in particular. However, the higher rate of nonfocal symptoms in women was mainly driven by patients who were actually diagnosed with cerebrovascular disease and had focal symptoms in addition. In contrast, symptoms in patients who were diagnosed with a stroke mimic were mostly balanced between women and men. These findings should remind clinicians that it remains fundamental to consider the complete clinical picture, including full constellation of symptoms, medical history, and circumstances of an event when facing suspected stroke, rather than focusing excessively on individual symptoms and focality. In case of any uncertainty, further investigation must not be discouraged by the presence of nonfocal symptoms [22]. This is particularly important in the preclinical setting, where the need for competence and clinical judgment have been emphasized in order to ensure timely recognition of a potential stroke, including awareness of sex differences in clinical stroke presentations [23].

Additional limitations must be considered when interpreting our study. The analysis of symptoms included items of the NIHSS only, which provided a standardized approach and allowed us to investigate a large number of consultations. However, the score is too superficial to resemble a holistic clinical evaluation. Thus, deficits may have been reflected inadequately, especially in patients with posterior circulation stroke, stroke mimics, and/or those with nonfocal symptoms. Nonfocal symptoms might additionally be underrepresented in our study due to the higher risk of being overlooked and underreported in the acute setting [24]. No adjustments for multiple testing were made which increases the risk of type I error and could have influenced our findings by generating false-positive results. However, we preferred to minimize the risk of overcorrection and type II error since this was an exploratory analysis; therefore, findings should be interpreted as preliminary. In addition, our data lacked a formal classification of stroke etiology which is why we could not investigate the role of different stroke etiologies with regard to our findings. Furthermore, we had no information on sex-specific risk factors such as age at menarche and menopause, hypertensive disorders of pregnancy, or hormonal replacement therapies [25]. Since these can be linked to certain types and etiologies of stroke such as cardioembolism or cervical artery dissection during pregnancy [26, 27], sex-specific risk factors could have influenced the clinical presentations in our study.

Nonetheless, our study revealed several differences in the clinical presentation of women and men with suspected stroke. Since prior analyses have mostly focused on populations with confirmed stroke, our findings provide a broad and more unselected perspective from a large cohort at a comprehensive stroke center. Awareness of these observations may aid in recognizing and addressing sex-specific nuances in acute stroke care. As a practical implication, these findings could be incorporated into educational programs for acute stroke care providers in order to convey competence in clinical judgment. This seems particularly relevant for prehospital evaluation of stroke, where the widespread focus on standardized scales for the detection of LVO may sometimes oversimplify the clinical scenario and thus come at the expense of overlooking subtle or nonfocal symptoms which could disproportionately affect women [23]. Future studies should also aim to provide a deeper understanding of the biological and social factors that underlie sex differences in stroke as well as their association with outcomes in order to shift from a one-size fits all approach towards more individualized strategies for both sexes.

Despite an overlap of focal symptoms, cortical as well as some nonfocal symptoms were more common among women with suspected stroke. Differences in baseline characteristics, stroke etiology, and subtype seem to be important drivers of these disparities; however, additional mechanisms may warrant further investigation. Awareness of sex differences in the presentation of suspected stroke patients and consideration of full symptom constellations are important to optimize care for women and men with suspected stroke.

This study was approved by the Ethics Committee of the Medical Faculty of the University Duisburg-Essen (Approval No. 18-8408-BO). Requirement for written consent was waived by the committee. Our analyses were conducted in agreement with the standards of the local data protection authority and according to the principles of the Declaration of Helsinki.

The authors have no conflicts of interest to declare.

This study was not supported by any sponsor or funder.

B.F. and J.K.E. conceptualized and designed the study. M.K. and C.Kl. supervised the project. B.Bo., L.S.M., D.P., B.Br., Y.L., C.D., C.Ki., P.D., and M.F. made substantial contributions to data acquisition. J.K.E. analyzed the data. B.F., M.K., and C.Kl. made significant contributions to data interpretation. J.K.E. and B.F. drafted the initial version of the manuscript. M.K., C.Kl., B.Bo., L.S.M., D.P., B.Br., Y.L., C.D., C.Ki., P.D., and M.F. critically reviewed the draft for important intellectual content. All authors have approved the final version of the manuscript.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author (B.F.) upon reasonable request.