Abstract
Objective To explore the sex differences in outcomes and management after stroke using a large sample with high-quality international trial data.
Methods Individual participant data were obtained from 5 acute stroke randomized controlled trials. Data were obtained on demographics, medication use, in-hospital treatment, and functional outcome. Study-specific crude and adjusted models were used to estimate sex differences in outcomes and management, and then pooled using random-effects meta-analysis.
Results There were 19,652 participants, of whom 7,721 (40%) were women. After multivariable adjustments, women with ischemic stroke had higher survival at 3–6 months (odds ratio [OR] 0.82, 95% confidence interval [CI] 0.70–0.97), higher likelihood of disability (OR 1.20, 95% CI 1.06–1.36), and worse quality of life (weighted mean difference −0.07, 95% CI −0.09 to 0.04). For management, women were more likely to be admitted to an acute stroke unit (OR 1.17, 95% CI 1.01–1.34), but less likely to be intubated (OR 0.58, 95% CI 0.36–0.93), treated for fever (OR 0.82, 95% CI 0.70–0.95), or admitted to an intensive care unit (OR 0.83, 95% CI 0.74–0.93). For preadmission medications, women had higher odds of being prescribed antihypertensive agents (OR 1.22, 95% CI 1.13–1.31) and lower odds of being prescribed antiplatelets (OR 0.86, 95% CI 0.79–0.93), glucose-lowering agents (OR 0.86, 95% CI 0.78–0.94), or lipid-lowering agents (OR 0.85, 95% CI 0.77–0.94).
Conclusions This analysis suggests that women who had ischemic stroke had better survival but were also more disabled and had poorer quality of life. Variations in hospital and out-of-hospital management may partly explain the disparities.
Glossary
- AF=
- atrial fibrillation;
- ASU=
- acute stroke unit;
- BP=
- blood pressure;
- CI=
- confidence interval;
- CV=
- cardiovascular;
- ENCHANTED=
- Enhanced Control of Hypertension and Thrombolysis Stroke study;
- HeadPoST=
- Head Position in Acute Stroke Trial;
- HRQoL=
- health-related quality of life;
- ICH=
- intracerebral hemorrhage;
- ICU=
- intensive care unit;
- INTERACT=
- Intensive Blood Pressure Reduction in Acute Cerebral Hemorrhage Trial;
- IPD=
- individual participant data;
- mRS=
- modified Rankin Scale;
- NIHSS=
- NIH Stroke Scale;
- OR=
- odds ratio;
- SCAST=
- Scandinavian Candesartan Acute Stroke Trial;
- SSS=
- Scandinavian Stroke Scale
Sex differences in disease risk, management, and outcome are an increasing focus of health services and research. Regarding stroke, women are generally older at the time of onset, and a higher proportion have hypertension and atrial fibrillation (AF) than men, but there are ongoing uncertainties over differing aspects of management and outcomes. Studies of ischemic stroke have shown that women are less likely to receive full etiologic assessment and proven therapies for secondary prevention.1 Despite having a higher burden of AF, women are less likely to receive oral anticoagulation,2 especially in those at high risk.3 In studies that have not adjusted for key confounders, women were observed to have poorer poststroke survival than men, which was reversed after adjustment for age, prestroke function, and stroke severity. However, much data on disability and health-related quality of life (HRQoL), overall and according to stroke subtype, are lacking.4,5 Better understanding of the type and degree of inequities in stroke management and outcomes are important for improving health outcomes, and an initial step towards personalized medicine. We aimed to determine sex differences in the management and outcomes of stroke in a pooling analysis of international randomized clinical trial populations.
Methods
Design
A pooled analysis of 5 large, international, multicenter, randomized controlled trials was conducted that included the following: the Intensive Blood Pressure Reduction in Acute Cerebral Hemorrhage Trials (INTERACT-16 and -27 studies; referred to as INTERACT studies), the alteplase-dose arm of the Enhanced Control of Hypertension and Thrombolysis Stroke study (ENCHANTED),8 the Head Position in Acute Stroke Trial (HeadPoST),9 and the Scandinavian Candesartan Acute Stroke Trial (SCAST).10 These acute stroke studies were suitable for pooling because of their similar pragmatic design features and ready availability of individual participant data (IPD) (table 1).
Design of the included trials
Study factors
Details on data collection and definitions have been discussed in detail elsewhere.6,8,–,11 In general, data were collected from the participant or a proxy at the time of presentation to hospital (baseline), over several days after admission, and for at least several months of follow-up.
The factors analyzed for significant sex differences in stroke5 were considered under 4 separate headings: (1) sociodemographic (age, sex, ethnicity, and region of study conduct); (2) premorbid condition (dependence, estimated according to the modified Rankin Scale [mRS],12 comorbidities of hypertension, diabetes mellitus, AF, hypercholesterolemia, prior stroke, and serious cardiovascular [CV] events) and health behaviors such as current cigarette use; (3) clinical data (admission blood pressure [BP] and heart rate); and (4) type of stroke (ischemic or intracerebral hemorrhage [ICH]) and severity of stroke (according to scores on either the NIH Stroke Scale [NIHSS] or Scandinavian Stroke Scale [SSS]).13
Outcomes
These IPD examined the following in ischemic stroke and ICH: (1) in-hospital management (including the use of intubation, antipyretics, venous thromboembolism prophylaxis, admission to an acute stroke unit [ASU] or intensive care unit [ICU], neurosurgery, inpatient rehabilitation, and early withdrawal of medical care); (2) out of hospital management (including preadmission CV prevention and secondary CV prevention medication recorded on 3 months follow-up); (3) death or disability, defined by scores 3–6 on the mRS, stratified by stroke type; (4) CV endpoints (myocardial infarction and recurrent stroke); and (5) HRQoL assessed directly or by proxy, using the EuroQoL 5-Dimension self-report questionnaire.14
Management details were available for all studies except SCAST. For preadmission CV prevention, 4 of the studies (INTERACT studies, ENCHANTED, and HeadPoST) had data on antihypertensives, anticoagulants, antiplatelets, and glucose-lowering medications available for analysis; for secondary CV prevention at follow-up, data on antihypertensives (INTERACT studies, ENCHANTED, and HeadPoST) and aspirin (ENCHANTED and HeadPoST) were available. In 4 of the studies (INTERACT studies, ENCHANTED, and HeadPoST), death and disability were assessed at 3 months; in SCAST, the mRS was assessed at 6 months postrandomization.
Standard protocol approvals, registrations, and patient consents
Clinical trial registration information: clinicaltrials.gov, unique identifier: INTERACT-1, NCT00226096; INTERACT-2, NCT00716079; ENCHANTED, NCT01422616; HeadPoST, NCT02162017; SCAST, NCT00120003. All participants in these studies provided informed consent and received approval from relevant ethics committees.
Statistical analysis
IPD were analyzed using the 2-stage method for meta-analyses.15 The first stage involved building study-specific crude and adjusted models to estimate the differences in outcomes for women compared to men. Consistent with the predefined analysis plans of these studies, logistic regression was used in 4 studies (INTERACTstudies, ENCHANTED, and SCAST), and a generalized linear mixed model with fixed randomized intervention, fixed period, random cluster, and random cluster-period effects was used to adjusted for the cluster crossover design in HeadPoST. A priori, the following rules were applied to determine the covariates for inclusion in the study-specific multivariable models: (1) the covariate was missing in <20% of cases; (2) the covariate was associated with death or disability in unadjusted analysis (p < 0.1); (3) the covariate was associated with sex (p < 0.1); and (4) the inclusion of the covariate in a model with only sex changed the magnitude of the sex coefficient by ≥10% ( figures e-1–e-10, doi.org/10.5061/dryad.4qt8510).
The second stage of analyses involved combining the crude and adjusted study-specific association estimates in a random-effects, inverse-variance–weighted meta-analysis. Statistical heterogeneity was evaluated using I2 statistics.
The adjusted analyses performed in each study used different sets of covariates according to the rules set in the first stage of the IPD, and their availability in each study. Randomized treatment was forced into the models. Hospital management and medication covariates were included in the models for death, disability, recurrent stroke, myocardial infarction, and HRQoL.
INTERACT studies
INTERACT studies reported age, country of origin, ethnicity, history of hypertension, history of stroke, history of myocardial infarction, glucose-lowering treatment, baseline NIHSS score, baseline systolic BP, randomized treatment, baseline hematoma volume, intubation, and ventilation.
ENCHANTED
ENCHANTED reported age, country of origin, ethnicity, history of hypertension, presence of AF, history of myocardial infarction, current smoking, baseline NIHSS score, baseline systolic BP, baseline glucose, randomized treatment, prestroke function, final diagnosis, nasogastric feeding given, subcutaneous heparin used, ASU admission, and any rehabilitation given.
HeadPoST
HeadPoST reported age, country of origin, history of hypertension, presence of AF, history of stroke, history of diabetes, current smoking, baseline NIHSS score, baseline systolic BP, prestroke function, intubation and ventilation, nasogastric feeding given, subcutaneous heparin used, mannitol given, ASU admission, ICU admission, and any rehabilitation given. However, due to a small sample, and to enable model fit, only age and stroke severity were adjusted for the following in HeadPoST: total stroke in intubation, neurosurgery, ICU admission, and all ischemic stroke and ICH in-hospital management and medication outcomes. Fixed intervention (lying flat vs sitting up) and fixed period, random cluster, and random cluster period effects were incorporated into generalized linear mixed models to take into account the study design.
SCAST
SCAST reported age, history of hypertension, presence of AF, baseline SSS score, baseline systolic BP, prestroke function, and randomized treatment.
SAS version 9.3 (SAS Institute, Cary, NC) was used in all analyses.
Data availability
This pooled analysis involves trial data shared among investigators. It is not a data repository and any requests for the source data should be made to the relevant chief investigator of each study.
Results
There were 19,652 participants included in the 5 studies, and 7,721 (40%) were women. The INTERACT studies enrolled the smallest proportion of women (37%), while SCAST enrolled the most (42%) (table 1). Table 2 shows the baseline characteristics of participants in each study population stratified by sex. Women were significantly older at the time of stroke in all studies, and more so for those recruited from Europe/Australia and South America. Women also had higher systolic BP, heart rate, and neurologic impairment (NIHSS scores), and were more likely to have a history of hypertension and AF on admission. Conversely, men were more like to be Asian and to be smokers, and to have had higher diastolic BP, prior stroke, coronary artery disease, diabetes mellitus, and premorbid independence (mRS scores).
Patient characteristics by sex
Figure 1 shows that women were more likely to be admitted to an ASU while in the hospital (odds ratio [OR] 1.17, 95% confidence interval [CI] 1.01–1.34) but were less likely to be intubated (OR 0.58, 95% CI 0.36–0.93), be given treatment for fever (OR 0.82, 95% CI 0.70–0.95), receive nasogastric feeding (OR 0.91, 95% CI 0.82–1.00), and be admitted to an ICU (OR 0.83, 95% CI 0.74–0.93). These results were attenuated after analyzing by stroke subtype due to reduced sample size.
Total stroke numbers in women and men may not equate to the number of patients with ischemic and intracerebral hemorrhage. This is due to the inclusion of undifferentiated stroke and missing data in the total numbers. CI = confidence interval; OR = odds ratio.
On admission to the hospital, women had higher odds of having received antihypertensive medication (OR 1.22, 95% CI 1.13–1.31) and lower odds of having antiplatelet treatment (OR 0.86, 95% CI 0.79–0.93), glucose-lowering therapy (OR 0.86, 95% CI 0.78–0.94), and lipid-lowering therapy (OR 0.85, 95% CI 0.77–0.94) (figure 2). There was no difference in the prescription of anticoagulants between sexes (OR 0.97, 95% CI 0.84–1.12). The treatment effect was attenuated for ICH (except for antihypertensive agents) but remained similar in ischemic stroke (with the exception of glucose-lowering agents). For secondary prevention, there were no sex differences in the prescription of antihypertensive agents (OR 0.92, 95% CI 0.84–1.01) or antiplatelet agents (OR 1.00, 95% CI 0.90–1.10) at 3-month follow-up (figure 2).
Total stroke numbers in women and men may not equate to the number of patients with ischemic and intracerebral hemorrhage. This is due to the inclusion of undifferentiated stroke and missing data in the total numbers. CI = confidence interval; OR = odds ratio.
After multivariable adjustments, women were significantly less likely to die within 3–6 months after an ischemic stroke (OR 0.82, 95% CI 0.70–0.97), though this lower likelihood was not significant for ICH (OR 0.87, 95% CI 0.66–1.13) (figure 3). With respect to disability, women had greater disability (mRS score 3–6) (OR 1.20, 95% CI 1.06–1.36) after ischemic stroke than men, but there was no difference between the sexes for ICH (OR 0.97, 95% CI 0.81–1.16). There were no clear sex differences in the risk of recurrent vascular events for ischemic stroke or ICH (figure 3). The results were consistent in subgroup analyses on thrombolyzed and nonthrombolyzed patients with ischemic stroke (figures e-1 and e-2, doi.org/10.5061/dryad.4qt8510). Overall, women experienced worse HRQoL (weighted mean difference −0.07, 95% CI −0.09 to 0.04). Across all 5 dimensions of HRQoL, women had higher scores (i.e., more problems) than men after ischemic stroke (figure 4). For ICH, conclusions are similar except there was no evidence of a sex difference in mobility (OR 1.06, 95% CI 0.79–1.44) or ability to perform usual activities (OR 0.95, 95% CI 0.79–1.14).
CI = confidence interval; OR = odds ratio.
CI = confidence interval; OR = odds ratio.
Discussion
In this pooled analysis of 19,652 patients with acute stroke, survival among women with ischemic stroke was significantly higher than in men with ischemic stroke but not in those with ICH. In those who survived, women lived with greater disability than men after ischemic stroke, and experienced poorer HRQoL despite adjustments for age and severity. There were important sex-specific differences in management, which may have influenced the disparities in these outcomes.
Our findings are broadly consistent with other studies overall and across the major stroke types.16,–,22 An analysis on sex differences in mortality rates found that men have considerably higher mortality for various causes (including CV disease) across age groups due to interactions with culture and environment.23 Our findings support this analysis, with women having better survival after stroke. Moreover, in our study, women were more likely to be admitted to an acute stroke unit, which have known benefits in improving outcomes.24 Regarding disability, a potential explanation is sex-specific differences in brain injury and repair after stroke.25 Preclinical studies of ischemic stroke suggest that aging brain in women is more sensitive to ischemia than in men, in relation to declining estrogen, increased systemic inflammation, and an alteration in gene expression.16,26 Other studies report that disparities in functional outcome may be influenced by differences in muscle strength27 or motor cortex affected by stroke.28
Our study found that in overall stroke, women were less likely to receive intubation, treatment for fever, nasogastric feeding, or ICU admission, but were more likely to be admitted to ASUs. While differing access to evidence-based treatment may be a determinant of sex differences in functional outcome after stroke,29 these findings may also imply that the women in the study did not require these interventions. Additional research is needed to examine whether there are sex differences in intervention in women and men who have had stroke and experience similar complications.
Older studies have reported sex differences in diagnostic evaluations, such as lipid testing,30 and therapeutic interventions, such as carotid surgery,31 whereas more recent publications have shown no sex differences in hospital stroke care.32,–,35 Studies have indicated that the use of preventative medication for stroke may differ between men and women. The Practice Innovation and Clinical Excellence (PINNACLE) registry,2 for example, showed that women were less likely to receive oral anticoagulant treatment compared with men. Our other findings of women being less likely to be prescribed an antiplatelet agent, glucose-lowering medications, or lipid-lowering medications are also consistent with previous studies.36,37 In addition, while we did not find any sex differences in the use of antihypertensive and antiplatelet agents at 90 days after stroke, the use of both is contingent on prescribing and adherence. More research is needed to better understand the reasons for such disparities in the recommended use and uptake of medications for stroke prevention, and how this influences outcomes.
Strengths of this study include the large sample of stroke participants from a variety of health care settings in countries across the world, which enhances the generalizability of the findings. Moreover, the consistency of the assessment procedures, high rates of follow-up, and rigorous outcomes measurements support the internal validity of these data. However, we recognize that this work has limitations. First, it was post hoc, observational, and restricted in the range of variables assessed (including dose and amount of treatment and marital status), raising the potential for residual confounding. Second, clinical trial participants are likely to receive higher than average standards of care,38 in line with contemporaneous advances in guideline recommendations for ischemic39,40 and hemorrhagic stroke,41,42 which may limit the generalizability of the findings. Third, the higher survival for women may explain their greater disability as a consequence of selection (survivor) bias. Finally, these data on the women and men who met exclusion criteria for the trials have not been analyzed in this pooled analysis.
Sex differences exist regarding survival, disability, and management after stroke. While in-hospital management does not appear to explain the sex differences in outcome, it is possible that the use of preventative medication may influence sex differences in outcome. Continued reporting of treatment and outcomes by sex, as well as further study of risk factor control and secondary prevention in men and women, are required to better appreciate the reasons and solutions for such disparities.
Study funding
Funding principally received from the National Health and Medical Research Council of Australia for INTERACT-1 and INTERACT-2, ENCHANTED, and HeadPoST.
Disclosure
C. Carcel and X. Wang report no disclosures relevant to the manuscript. E. Charlotte Sandset receives fees for speaking at seminars for Bayer and Novartis. C. Delcourt reports no disclosures relevant to the manuscript. H. Arima receives lecture fees from Bayer, Daiichi-Sankyo, and Takeda. R. Lindley is a speaker on behalf of Covidien and Pfizer. M. Hackett receives an NHMRC Career Development Fellowship (APP1141328), NHMRC grant for HeadPoST, and a research grant from the Ramsay Hospital Research Foundation Ltd. P. Lavados received research grants from The George Institute for Global Health and grants from Clínica Alemana de Santiago during the conduct of the study; nonfinancial support from Boehringer Ingelheim; grants and personal fees from Bayer and AstraZeneca; and grants from CONICYT outside the submitted work. T. Robinson is an NIHR Senior Investigator. P. Muñoz Venturelli reports research grants from The George Institute for Global Health and Clínica Alemana de Santiago during the conduct of INTERACT-2 and HeadPoST trial and a research grant from CONICYT outside the submitted work. V. Olavarría reports research grants from The George Institute for Global Health and Clínica Alemana de Santiago during the conduct of the ENCHANTED and HeadPoST studies and research grants from Clínica Alemana de Santiago, CONICYT, and Boehringer Ingelheim outside the submitted work. A. Brunser reports research grants from Clínica Alemana de Santiago. E. Berge reports no disclosures relevant to the manuscript. J. Chalmers received a grant from the NHMRC and from Servier for the ADVANCE trial and the ADVANCE-ON post-trial follow-up study, and served as a speaker about those trials at scientific meetings, and received a grant from Idorsia for the SPIRIT study of resistant hypertension. M. Woodward consults for Amgen and Kirin. C. Anderson received an honorarium from Takeda China and Boehringer Ingelheim and a research grant from Takeda China paid to The George Institute China. Go to Neurology.org/N for full disclosures.
Acknowledgment
The authors thank the hospital research staff and the study participants, their families, and friends.
Appendix Authors
Footnotes
Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article.
↵* These authors contributed equally to this work as co–first authors.
Editorial, page 1025
- Received January 30, 2019.
- Accepted in final form July 29, 2019.
- © 2019 American Academy of Neurology
References
Disputes & Debates: Rapid online correspondence
- Author response: Sex differences in treatment and outcome after stroke: Pooled analysis including 19,000 participants
- Reader response: Sex differences in treatment and outcome after stroke: Pooled analysis including 19,000 participants
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.