Under-representation of women in stroke randomized controlled trials: inadvertent selection bias leading to suboptimal conclusions

Under-representation of women in stroke clinical trials can lead to inadvertent selection bias, and thus suboptimal conclusions for women in both outcomes and stroke care delivery. Here, we provide a comprehensive overview of the most important issues regarding gender differences in stroke treatment and prevention, by providing examples on four different fields of stroke research.

Example 1: the role of gender in the delivery and efficacy of intravenous thrombolysis

According to a meta-analysis of 18 study databases, women with acute ischemic stroke (AIS) have a cumulative 30% lower probability of receiving treatment with intravenous thrombolysis (IVT) compared with men.¹ A report from the Promoting Acute Thrombolysis for Ischaemic Stroke study also confirms that, indeed, women are less likely to receive IVT with tissue plasminogen activator (tPA) than men, because delayed presentation to the emergency department is more common in women. This finding has been attributed to the facts that, on average, women are older at stroke onset and are more likely to live alone or without adequate care monitoring.² However, a recent study reported that after adjusting for multiple covariates, including age and stroke severity, gender was no longer associated with delayed arrival to the emergency department.³ Indeed, a retrospective analysis of 108 IVT-treated octogenarians reported no gender differences regarding either outcome or mortality after IVT.⁴ Moreover, IVT has also been reported to have a greater and faster effect on neurological status in women than in men.^5,6 According to the findings of a small cohort study, the greater efficacy of IVT treatment seen in female patients has been attributed to an increased likelihood of tPA-induced recanalization in women compared with men, suggesting that several anatomical (smaller clot burden due to smaller size of proximal intracranial arteries), pathophysiological and biochemical factors could potentially account for this observed difference.⁷

In a gender-matched cohort analysis, female sex was reported to be independently associated with higher risks of symptomatic intracerebral hemorrhage (sICH) and fatal sICH.⁸ By contrast, data from the Safe Implementation of Treatments in Stroke-International Stroke Thrombolysis Register (SITS-ISTR) not only suggested a possibly larger benefit from treatment effect in women compared with men, but also reported a lower risk of sICH after IVT for women.⁹ Similarly, results from a prospective national registry highlight a significantly lower mortality rate in IVT-treated women than in men.¹⁰ Even though lower mortality rates in females have also been reported in a very recent European cohort study of 9495 IVT-treated patients, female sex was independently associated with a higher probability of poor functional outcome in this cohort.¹¹ Finally, an analysis of five randomized controlled trials (RCTs) suggested a gender interaction on the efficacy of IVT, with women faring better than men in terms of favorable functional outcome.¹² This finding was not replicated in an individual patient data meta-analysis of nine RCTs.¹³ Interestingly, women were under-represented (<45% of participants) in eight out of nine of these RCTs in AIS, as well as in the individual patient data meta-analysis (45% of the included patients). A gender-balanced randomization of the RCTs for AIS would have led to a fairer comparison between males and females, and thus it could also have permitted more definitive conclusions regarding any potential interaction of sex on the efficacy of IVT in AIS.

Example 2: female under-representation in mechanical thrombectomy trials and risk of erroneous conclusions

In a subgroup analysis of the Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands (MR CLEAN) trial, according to gender, no effect of mechanical thrombectomy (MT) was documented in the female subgroup; moreover, in the intervention group women were found to experience more serious adverse events compared to male patients.¹⁴ However, this interaction was neither verified in meta-analysis of MT RCTs^15,16 nor in the individual patient data meta-analysis performed by the HERMES collaboration,¹⁷ thereby suggesting a similar benefit of MT in both genders. The lack of a significant effect on the female subgroup in the MR CLEAN trial could possibly be attributed to an under-representation of women (41.2% of participants),¹⁸ compared with the other MT trials,¹⁶ together with the higher mean age, more prevalent atrial fibrillation and lower antiplatelet pretreatment rates observed in women.

Example 3: gender disparities in carotid revascularization trials

Most carotid endarterectomy (CEA) studies have reported that male patients with symptomatic or asymptomatic carotid artery stenosis not only have both lower stroke and mortality rates but also benefit more from CEA than females.¹⁹ In a pooled data analysis from the ECST and the NASCET trials, the benefit associated with CEA appeared to be greater in males, while females had a significantly higher risk of perioperative stroke.²⁰ Similarly, a subgroup meta-analysis of the VACSP (Veteran Affairs Co-operative Studies Program), the ACST (Asymptomatic Carotid Surgery Trial) and ACAS (Asymptomatic Carotid Atherosclerosis Study) also reported a more pronounced benefit from CEA in asymptomatic carotid artery stenosis for males than for females.²¹ However, it should be highlighted that the proportion of females undergoing CEA remained constant over the 20-year study period and was substantially lower than that of males (34% in ACST, 34% in ACAS and 0% in VACSP).²² These gender disparities may be attributed to differences in plaque characteristics, anatomical differences and differences in coexisting comorbidities between the two genders.¹⁹

Regarding carotid artery stenting (CAS), no gender disparities were observed for the periprocedural risk of stroke or death.¹⁹ Despite this, CAS has been correlated with an increased risk of periprocedural events only in women randomized in the CREST trial (Carotid Revascularization Endarterectomy versus Stenting Trial). This finding was neither significant in multivariate analysis,^23,24 nor in a meta-analysis that combined data from both the ICSS (International Carotid Stenting Study) and CREST databases.²⁵

Given that women have been under-represented in carotid revascularization trials, gender-specific recommendations for either symptomatic or asymptomatic carotid stenoses cannot be reliably provided. The potential reasons for under-representation of women in carotid revascularization trials could be attributed to the increased technical difficulties (smaller ICA size in females) or higher rate of periprocedural (higher microembolization rates and lower cerebrovascular reserves in women) and postprocedural (carotid artery restenosis and ipsilateral stroke) complications reported in females than in males.¹⁹ Thus, there is a need for a separate women’s carotid trial to provide definite answers on potential gender disparities.²⁶

Example 4: under-representation of women in secondary stroke prevention trials

Despite a higher prevalence of atrial fibrillation in women, leading to more cardioembolic strokes,²⁷ trials on novel oral anticoagulants have included <40% of women in their secondary prevention subgroups.²⁸ In fact, the Japanese trial on rivaroxaban (J ROCKET-AF) reported that females made up only 17.4% of the secondary prevention subgroup.²⁹ Likewise, in the WARCEF trial, women made up about 20% of patients randomized in both subgroups.³⁰

Women have also been vastly under-represented (<30% of participants) in most RCTs to date on antiplatelet treatment for primary or secondary AIS prevention,³¹ despite the NIH Revitalization Act (PL-103-143) urging for a balanced inclusion of both genders in RCTs.³² Even though antiplatelets are considered to be equally effective for both genders in clinical practice, an RCT including only women has reported a benefit from aspirin compared with placebo for primary ischemic stroke prevention,³³ whereas a male-only RCT has reported a higher risk of hemorrhagic stroke in primary prevention for men receiving aspirin compared to those receiving placebo.³⁴ Most RCTs and meta-analyses on antiplatelets in cerebrovascular disease have not performed sub-analyses on gender-related differences owing to under-representation of females, thus making it impossible to investigate for gender-specific differences in both primary and secondary ischemic stroke prevention with antiplatelet agents.³¹ Consequently, gender-balanced RCTs for secondary stroke prevention are necessary to investigate for a possible gender role in secondary prevention strategies.

Suggested strategy to reduce the bias

Given that literature evidence suggests potential gender disparities in both acute stroke treatment and prevention, future observational and randomized study protocols should not only ensure a balanced recruitment and equal representation of both genders, but also provide subgroup analyses according to gender for all primary and secondary outcomes.