Service delivery in acute ischemic stroke patients: Does sex matter?

Introduction

While there is mounting evidence that women are older and have higher preexisting handicap and comorbidities at stroke onset than men ^1–3, this only partially explains the poorer long-term functional outcome in female stroke patients. Possibly, acute stroke care treatment, work-up, and secondary prevention could differ between sexes and contribute to these findings and more data are needed on this latter aspect of stroke management as well as on care delivery in the acute and subacute phases. Considering the expected age-dependent increase in stroke incidence in women in the next decades, this issue becomes even more important. In this quality assurance project, we sought to investigate potential sex differences in care delivery in acute ischemic stroke (AIS) in a large consecutive cohort of patients treated at our institution.

Methods

Ethical considerations

ASTRAL is registered in our institution as a clinical and research databank and follows the institutional regulations on clinical and research registries. Patients received a written document from the hospital that their routinely collected clinical data may be used for quality and scientific purposes, and we did not consider a patient’s decision to opt out from data analysis as this was a quality assurance project of the treatment practice in our institution. This type of analysis that aims to evaluate treatment efficiency and safety for quality purposes falls outside of the Swiss Human Research Act of 2011.

Data availability

The raw, anonymized data that support the findings of this study are available from the corresponding author upon reasonable request and after signing a data transfer and use agreement (DTUA). If such data are then used for a publication, the publication methods should be communicated, and internationally recognized authorship rules should be applied.

Statistical Analyses

Study participants’ data were analyzed according to sex classification, as shown in figure 1. We assessed demographic, clinical, and radiological variables from the acute phase of stroke (<24 hours) recorded in the ASTRAL database. Continuous variables were summarized as median with interquartile range (IQR) and categorical variables as percentages and numbers.

Initially, univariate comparisons were performed according to sex and the unadjusted odds ratios (ORs) were evaluated. Subsequently, multiple regression analyses of the aforementioned eight main outcomes were conducted to identify independent associations with sex, after adjusting for the following 23 covariates: age, ethnicity (Caucasian vs. all others), prehospital referral pattern (patients coming from primary vs. secondary hospital catchment areas), time of symptom onset to hospital admission, history of prior cerebrovascular events, stroke severity, pre-hospital admission mRS, cerebrovascular risk factors (hypertension, diabetes, hyperlipidemia, smoking, atrial fibrillation, alcohol abuse, body-mass index), peripheral artery disease, coronary artery disease, cardiac heart valves, cardiac ejection fraction, migraine, active cancer, depression and psychosis. Additional adjustments were made for specific analyses: OTD time to assess the proportion of acute revascularization treatment (IVT and/or EVT), OTD time and carotid stenosis grade to evaluate carotid revascularization treatments, and OTD time to assess length of hospital stay and a CGC.

For continuous response variables, a Cox proportional hazards model was employed to associate covariates with the response. This approach has the advantage that no distributional assumptions are made for the continuous response. For categorical outcomes, with limited values, a proportional odds model was used to identify important links between response and background information.

Missing data were imputed using multivariate imputation by chained equations (MICE) methodology. Five imputed datasets were generated. Stepwise selection methods were employed to identify significant associations of the selected covariates with the specific outcome on each imputed dataset. Results of the five imputed analyses were combined appropriately to formulate the final analysis results that were given as adjusted odds ratios (aORs) with confidence intervals (CIs) for categorical and ordinal outcomes, and adjusted hazard ratios (aHR) with CIs for continous outcomes. All statistical analyses were conducted using the R statistical software version 4.4.2 (R Core Team 2017).

Results

Retrospective data from 5347 patients who fulfilled the inclusion criteria were analyzed (figure 1). Of these patients, 45% were female, 3.8% were of non-Caucasian ethnicity and the median age was 74.6 years. The proportion of women in the population referred from the secondary catchment area was 30.4%, lower than the proportion of men (34.7%). Detailed sex-specific description of the cohort is described in table 1 and supplemental table 1; it resembles closely our previous sex-specific analysis concentrating on clinical presentations and long-term outcomes ².

Unadjusted comparisons (see table 1 and supplemental table 1) showed a significantly higher OTD time (3.4 vs. 3.0 hours; HR 1.02; CI95% 1.01-1.03), a lower proportion of IVT (22.7% vs. 24.6%; OR 0.9; CI95% 0.79-1.02) and a slightly higher proportion of EVT treatments (15.2% vs. 15.1%; OR 0.98; CI95% 0.84-1.14) for women compared to men. Furthermore, women had fewer diagnostic exams (median number 3 vs. 3, OR 0.9 CI95% 0.87-0.93), fewer subacute carotid revascularization treatments (1.7% vs. 4.4%; OR 0.36 CI95% 0.25 - 0.52), longer lengths of hospital stay (9 vs. 8 days; HR 1.01 CI95% 1.00 - 1.01) and a higher proportion of change in goals of care (12.3% vs. 9.1%; OR 1.40 CI95% 1.18 - 1.67) compared to men.

In the adjusted multivariate analysis (see table 2) female sex remained significantly associated with higher OTD (aHR: 1.09, 95%CI 1.04-1.14; p<0.01) and DTP times (aHR: 1.15, 95%CI 1.05-1.25; p<0.01). While not statistically significant, we observed a lower number of diagnostic exams (aOR: 0.94, 95%CI 0.85-1.04; p=0.21), fewer subacute carotid revascularization treatments (aOR: 0.69 95%CI 0.33-1.18; p=0.147) and longer hospital durations (aHR: 1.03, 95%CI 0.99-1.07; p=0.19) in women. Further, in the multivariate analysis we no longer saw trends in the proportions of IVT and/or EVT or changes in goals of treatments between male and female sex patients.

Discussion

In this retrospective study of 5347 consecutive AIS patients admitted to a comprehensive stroke center, we found that female sex was significantly associated with longer OTD and DTP times. Additionally, we noted numerically fewer diagnostic exams in women during the acute hospital stay, fewer subacute carotid revascularization treatments and longer hospital durations, but these differences did not reach statistical significance. Altogether, these indicators of less effective stroke care delivery may contribute to the poorer long-term functional outcomes in female patients ^{2, 11} and warrants further attention.

Existing literature focusing on sex differences in acute stroke care and service delivery is limited and yields inconsistent results. An older analysis from the GWTG-Stroke program ¹² and a recent large retrospective study in China ¹ did not show significant sex differences in acute management and intrahospital and discharge performance measures. A pooled analysis of five randomized controlled trials found that women were more likely to be admitted to an acute stroke unit ¹⁴, whereas other studies from France and Canada showed the contrary.^{14, 15}

This same pooled analysis showed that women were less likely to benefit from intensive care treatment and acute interventions such as intubation ¹³, nasogastric feeding ¹³, and treatment for fever ¹³. The latter finding was confirmed by a recent nationwide study from Canada demonstrating that women and older patients (≥80 years) had lower odds of receiving life-sustaining care, but disparities in intensive care unit admission narrowed over time.¹⁶

Regarding prehospital timing, several other studies reported longer onset-to-hospital times for women ^17–22. whereas others did not find a sex-related difference. ^{23, 24} One plausible reason for increased pre-hospital times is that women are more often widowed and living alone, ²⁵ leading to a later activation of emergency medical services. ²¹ Another explanation might be more non-focal, atypical, and non-traditional symptoms in acute stroke presentations in women, leading to under-recognition of stroke and non-activation of stroke alert pathways. ^{26, 27}

Regarding intravenous thrombolysis, our results confirmed those of other studies showing comparable treatment rates and time metrics for both sexes. ^{1, 22, 23, 28, 31} However, one study of Danish public registries ²¹ and a recent meta-analysis of 17 studies showed 13% lower unadjusted odds of receiving IV rtPA in women. ³⁰ Reassuringly, such differences seem to decrease towards more sex equity over time: a recent nationwide analysis in the US showed that in the last decade, utilization of rtPA increased at a faster pace in women compared to men. ³¹ This trend was confirmed by epidemiological data from France ¹⁵ and by the aforementioned meta-analysis by Strong and co-workers. ³⁰ Recent data from Catalonia for the period from 2016 to 2019 showed even a higher proportion of acute reperfusion therapies in women.²⁴

Regarding the rate of EVT, we found no differences between sexes after adjusting for several factors, whereas other studies showed higher EVT rates in women.^{14, 30, 32} Given that several of these other studies ^{14, 32} adjusted their comparisons with multiple co-variates, there is no simple explanation for these findings such as higher admission NIHSS scores or rates of atrial fibrillation.²

We found longer door-to-groin times in women treated with endovascular therapies, despite adjusting for prehospital delays. This contrasts with other studies that found similar intrahospital times for EVT. ^{14, 24, 33, 34} Our finding does not have an obvious explanation, but a possible explanation is a cognitive sex bias.

Regarding the in-hospital stroke work-up in the subacute phase, our results showed that we performed numerically fewer diagnostic exams in women, but this did not reach statistical significance. Again, prior literature is divided when reporting on sex differences in diagnostic evaluation of stroke, with evidence for less extracranial magnetic resonance angiography (MRA),²³ echocardiography and carotid evaluation in women ^{35, 36} while, on the contrary, older populations ³⁷ and multicentric studies in the US and Canada did not demonstrate significant sex inequities for these exams. ^{38, 39}

A further main result of our study was that after adjusting for age and stenosis grade, we observed numerically fewer subacute carotid vascularization procedures in women, not reaching statistical significance. This result may be explained by evolving sex-specific treatment recommendations for symptomatic carotid stenosis. ^{40, 41} Some studies show a higher stroke risk in male carotid stenosis patients compared to women, probably due to higher plaque instability ⁴², but endarterectomy risk seems similar ⁴³, and the meta-analysis of randomized trial data does not show a differential effect of carotid revascularization depending on biological sex. ⁴¹

As for the lengths of hospital stay, the longer hospital duration for women was not significant after adjusting for the known higher stroke severity and pre-stroke handicap.² Longer hospitalization in women has been reported by Somerford et al ⁴⁴, possibly related to the more frequent absence of a spouse at home who could participate in the care.

Finally, regarding the complex question of sex disparities in CGC in AIS, we did not observe a significant difference in CGC decisions in our adjusted analysis. This confirms what we found in our recent analysis that identified several powerful predictors of GCS, but not sex.⁴⁵

Limitations of our work are its retrospective, non-randomized design in a tertiary stroke center with approximately one-third of patients being referred from other hospitals or stroke units. As a quality assurance project in a single institution, it may not be applicable to other populations, in particular, non-Caucasian and younger populations. As stated above, we did not measure several variables that may have confounded our main outcomes, such as educational level, extent of social networks, isolation and loneliness, and socioeconomic and sociocultural status.

The strengths of our work are the consecutive nature of the collected data over a long period with pre-specified and standardized data collection, and rigorous adjustment of multiple confounders for the eight main outcome analyses.

In conclusion, this study suggests that female patients with AIS arriving at a comprehensive stroke center have longer OTD and DTP times while other indicators for care delivery are not statistically different. Some of these observations may contribute to the explanation of poorer long-term functional outcomes in female patients and require further attention.

Disclosures

F. Medlin: his institution received research grants from the Swiss Heart Foundation.

D. Strambo: none.

D. Lambrou: none.

V. Caso: received speaker fees from Bayer and Boehringer-Ingelheim within the past 3 years

P. Michel: research grants from the Swiss Heart Foundation, Swiss National Science Foundation and the Faculty of Biology and Medicine of the University of Lausanne.

Author contributions

Friedrich Medlin: Study concept and design, data analysis and interpretation, drafting of the manuscript.

Dimitris Lambrou: Statistical analysis, software support, drafting of the manuscript. Davide Strambo: Data collection, data analysis and interpretation.

Valeria Caso: Study concept and design, critical revision of the manuscript.

Patrik Michel: Study concept and design, data collection, collection and handling of funding, critical revision of the manuscript, supervision of study.

Supplemental material

• Download figure
• Open in new tab

Supplemental figure 1.

Study flow diagram