A Systematic Review of Female Participation in Randomized Controlled Trials of Post-Stroke Upper Extremity Rehabilitation in Low- to Middle-Income Countries and High-Income Countries and Regions Free

Stroke is one of the leading causes of disability worldwide [1]. Post-stroke rehabilitation is an essential part of stroke care to improve functional outcomes and the quality of life in stroke survivors [2]. There are sex differences in stroke epidemiology and outcomes. International reports showed a higher incidence of stroke in males but a higher prevalence and more severe strokes in females, particularly at an older age [3, 4]. Differences in stroke outcomes also exist between countries as stroke mortality and burden of disease are reported to be higher in low-middle-income countries (LMICs) than high-income countries (HICs) [1, 5, 6]. Previous studies showed that female sex was associated with poorer functional outcomes, adjusting for other factors [7, 8], higher levels of dependency in activities of daily living, and lower quality of life after stroke [9, 10].

These sex differences in post-stroke recovery and outcomes highlight the need for developing rehabilitation interventions with attention to females needs [10]. However, female participants are underrepresented in stroke trials [11‒13], and the gap between the female-specific needs in stroke care and their representation in clinical trials has recently been brought to attention [14]. Meta-analysis of stroke randomized controlled trials (RCTs) over almost 3 decades have shown that females accounted for 37–40% of all participants [11, 12], with a lower participation rate than their prevalence rate in the underlying population [12]. Another study showed that females were underrepresented in acute stroke trials by 5.3% relative to their representation in the underlying populations [13].

Multiple reasons have been suggested for the under enrollment of females in stroke clinical trials. Some reasons are specific to stroke characteristics in females, greater severity than males, older age at onset, and higher rates of premorbid and comorbid conditions that may result in exclusion from clinical trials [15]. Other reasons might be related to social or cultural preferences with limited reports that females were less willing to participate [16] and that males showed more willingness to participate in upcoming trials [17]. However, the rationale for lower female enrollment compared to males in stroke rehabilitation RCTs is not fully understood.

The underrepresentation of females in stroke trials raises the concern about the generalizability of results and transferability to clinical practice in females [13, 14]. In clinical trials, it is crucial to have research participants representing the characteristics of underlying population [13, 15], which may differ in disease epidemiology in the region where the study is conducted. Previous studies showed lower female enrollment in the USA than in international cardiovascular trials [18]. More recent studies showed that North America had the highest percentage of females in stroke RCTs with 42% of all participants, compared to Asia (38%) and Europe (31%) from 2008 to 2018 [11]. It is important to understand the female enrollment specifically in post-stroke motor rehabilitation trials in different regions since they experience more severe stroke and poorer functional outcomes than males. However, there is no comprehensive report on female participation in post-stroke rehabilitation RCTs across different countries and regions of the world, over time. This study aimed to provide an overview of female participation in RCTs of post-stroke upper extremity (UE) motor disorder rehabilitation in LMICs and HICs and different regions within the HICs. The results of this study will raise awareness among clinical trial funders, agencies, and investigators and help researchers understand and where necessary adjust female enrollment for the underlying population.

This systematic review was conducted and written using the PRISMA statement [19]. A protocol was not registered prior to conducting this review. The present study uses the same database as our previous review, providing a foundation for a more detailed investigation of female enrollment based on regions and countries of origin. Therefore, this review shares the same search strategy and inclusion criteria with our previous study [20].

The following databases, CINAHL, Embase, PubMed, Scopus, Web of Science, were used to review articles from 1960 to April 1, 2021, with the following search terms: (Stroke OR Cerebrovascular Accident OR CVA) AND (upper extremity OR upper limb OR arm OR hand OR shoulder) AND (Remediation OR Therapy OR Rehabilitation OR Intervention OR Stimulation OR Exercise OR Pharmacotherapy* OR Medications OR Drug OR Pharmaceutical). The filters for “randomized controlled trial,” “English,” and “human” were used where applicable [18]. There were additional articles found by searching the reference lists of the Evidence-Based Review of Stroke Rehabilitation (EBRSR.com).

Publications for this search were included if they (1) were an RCT or RCT crossover design; (2) included adults ≥18 years old; (3) included at least 50% of participants diagnosed and affected by stroke; (4) used an intervention on hemiparetic UEs as the primary focus of the study; and (5) were written in English. Studies were excluded if they were a secondary analysis of a previous RCT, follow-up studies, interim or incomplete studies, narrative reviews, and dissertations [20].

The initial literature search results were imported into Covidence (Veritas Health Innovation, Melbourne, Australia. Available at www.covidence.org), and the duplicated articles were automatically removed. Two independent reviewers completed the title and abstract screening, full-text review, and data extraction. At each step, any disagreements were discussed between reviewers, and consensus was reached by a third reviewer if the disagreement had not been resolved. Data extracted for the aim of this research included publication year, authors’ names, country, absolute numbers, and the proportion of male and female participants. Any information that was missing for these variables was recorded as “not reported.” Studies were recorded by country according to the corresponding author and/or the location site(s) of the trial.

Sex: For the purpose of this review, the term “sex” is used to distinguish biological sex determined by physiological features. In the publications found during this search which fulfilled inclusion criteria, only 2 sexes (male and female) were reported. Therefore, we will only be discussing male and female biological sexes.

Countries and regions: Countries were primarily divided into LMICs and HICs according to “The World Bank” country classification in 2021 with the threshold of USD 13,205 growth national income per capita [21]. Since the HIC group included a large variety of countries from very broad regions of the world (Table 1), the HICs were further divided into three regions to provide a better overview of differences in female participation within the HICs. The high-income regions included North America, Europe, and Asia and Oceania. Three RCTs in HICs were conducted in Chile and were not categorized as part of any of the regions.

The methodological quality of all RCTs in this review was assessed using the Physiotherapy Evidence Database (PEDro) scale and reported in detail in our previous published study on the same dataset [20].

Extracted data were transferred from Covidence to Microsoft Excel for data management. Study characteristics and extracted variables were summarized using standard descriptive statistics. Data analyses were performed for the dataset of 1,209 RCTs that reported sex of participants, using SPSS and RStudio v.4.3.1 (R Core Team, 2023), and figure generation was performed using the ggplot2 v3.4.4 package in RStudio.

Data were assessed for normality using the Kolmogorov-Smirnov, histograms, Q-Q plots, and residual plots. Accordingly, the Mann-Whitney U test and Kruskal-Wallis test with pairwise comparisons were used to compare the mean percentages of female participants between countries and regional groups. A general linear regression of female participation percentage and years was performed to understand the female participation time trend in each group of RCTs. The multi-regression analysis was performed from 1998 to 2020 due to lack of RCTs in LMICs prior to 1998. Female percentage outliers were identified using RStudio and were removed to maintain the data homoscedasticity for linear regression analysis. Sensitivity analyses were performed when (1) considering data from earlier years, (2) including outliers, and (3) considering median female percentage for those RCTs with no sex report. A p value of <0.05 was considered as significant.

The literature search identified 5,408 studies, and 1,276 RCTs met inclusion criteria and were included in this review (Fig. 1), accounting for 76 unique interventions for the hemiparetic UE rehabilitation (online suppl. Table A; for all online suppl. material, see https://doi.org/10.1159/000538610). The methodological quality of all RCTs is provided in our previous published work [20].

From 1972 to 2021, 298 RCTs were conducted in LMICs and 978 RCTs in HICs (Table 2). In HICs, 290 RCTs were conducted in North America, 357 RCTs in Asia and Oceania, 328 RCTs in Europe, and three RCTs in Chile which did not fit in any of three major high-income regions and accounted for 86 participants, with 40.7% females. Table 2 shows the number of participants by sex in LMICs, HICs, and each high-income region. Sex was not reported more often in LMICs. When considering regions in HICs, sex was reported most in European RCTs and least in North America, with the latter similar to rates of sex reporting in LMICs (Table 2).

The average percentage of female participation was 36.9% in RCTs of LMICs and 39.5% in the HIC group. The Mann-Whitney U test showed that the female participation in RCTs of HICs was significantly higher than those in LMICs (t = 2.9, p = 0.003). Comparing the female percentages in RCTs of LMICs with the three regions of HICs group, the difference was significant between RCTs of LMICs and those in HIC-European countries (t = 3.74, p < 0.001) but not in other high-income regions. The analysis within the high-income regions also showed that despite the slightly higher female participation in Europe, the percentages of female participants were statistically comparable in the three regions of HICs.

Female participation did not significantly change over the years in both HICs (Std. β = −0.13, t = −1.75, p = 0.08) and LMICs (Std. β = −0.34, t = −1.76, p = 0.08) with a small negative time trend in both groups (Fig. 2). There was also no significant difference in female participation trends between HICs or LMICs over time (group*time interaction, p = 0.3). When HIC results were considered by region, again, there was no significant difference in female participation trends over time between high-income regions (Fig. 3). Female participation was relatively stable without any significant change over time in Europe (Std. β = −0.05, t = −0.412, p = 0.68) and North America (Std. β = −0.18, t = −1.29, p = 0.20), and Asia and Oceania (Std. β = −0.21, t = −1.38, p = 0.17). The sensitivity analysis did not present any findings that differed in significance from the initial assessment.

Despite the progression of decades of RCTs in post-stroke UE rehabilitation, female participation remained relatively constant, with ongoing underrepresentation of female participants, regardless of country or region. This is consistent with previous research in acute stroke by Strong et al. [13] that showed female under-enrollment in these three regions, even though they reported higher percentages of female participants in studies of North America and Europe compared to the current study. The difference may be due to the different time frames and criteria as their study only included the trials from 2010 to 2020 in the acute phase, and the search was limited to nine major journals [13].

Our study showed that overall female participation was significantly lower in LMICs compared to HICs. Higher female participation was seen with European HICs RCTs followed by North America and then Asia and Oceania HIC RCTs. Despite the overall lower rate of female participation in RCTs of LMICs compared to HICs, female participations did not change significantly over time and showed a small negative trend in both HICs and LMICs.

Several reasons have been suggested for the overall female underrepresentation in stroke trials regardless of geography or country wealth. Females on average are older than men when they suffer a stroke and experience more severe strokes than males, which may result in exclusion from trials with upper age or severity limits [15]. However, a study on stroke rehabilitation with no upper age limit for participants reported that the female sex was still associated with lower likelihood of recruitment in their trials [22], suggesting other factors are important. Another barrier to participate in stroke trials could be related to the less favorable functional outcomes, and greater overall dependency, in female stroke survivors when compared to males, making it more difficult to engage in outside activities such as a clinical trial [15]. Females are also more likely to live alone or live with in home care supports than males at the time of stroke [15, 23] potentially making any travel to attend a clinical trial assessment more difficult.

There is no evidence for differences in any of the proposed factors across different regions or between HICs and LMICs; however, there are anticipated variabilities in underlying populations, resources, and sociocultural factors across countries and regions. Sociocultural factors may affect female participation in clinical trials which may differ across countries and regions. Previous research showed that females are less certain about participating in clinical trials and are less likely to take risks than their male counterparts [24]. Females prefer to have more resources, information, and time before deciding to participate than males [25], which may result in greater reluctance to participate if researchers do not provide adequate time and resources for potential participants.

Recent reports showed a general higher age-standardized stroke-related mortality rate and age-standardized stroke-related disability adjusted life years in LMICs and with a lower percentage of improvement, than HICs, over time for females who have suffered a stroke when compared to their male counterparts [5, 6]. The high stroke incidence, mortality, and burden in LMICs [5, 6], with no improvement in the percentage of female participation in RCTs of LMICs (as well as HICs) over the last 2 decades, raises concern that the generalizability of the research to females who have suffered a stroke and require rehabilitation may be less than for their male counterparts. It highlights the need for clinical trials with representative samples, including a more representative ratio of female subjects.

The importance of encompassing a sample reflective of the target population is a feature of any study. This ensures that the results may be generalizable. The last updated statistics, by the World Stroke Organization, reported that each year, 53% off all strokes occur in women, 49% of all deaths from stroke are in women, and 56% of people who have experienced a stroke and are currently living are women [6]. The relative rate of global stroke incidence, mortality, and burden in both sexes shows the need to try to enroll equal participants in both sexes to ensure that research into interventions being investigated in stroke trials will be equally applicable to both males and females. Having generalizable evidence consistent with the epidemiological characteristics of the target population is particularly important in those countries with higher stroke incidence, mortality, and disability.

The literature is limited on sex-specific stroke statistics and female enrollment in clinical trials comparing different countries and regions. A study of global and regional lifetime risk of stroke showed that global lifetime risk of stroke in 2016 was comparable between the sexes; however, there were reported variations in regions. The risk was greater for females in most areas, except for the LMICs in East Asia, led by China, where the risk was greater in males [26].

While female participation was lower in LMICs, the underrepresentation of females was also found in HICs without a significant change over decades. This finding implies that while economic and geographical differences may influence female participation in clinical trials, there are other multifaceted factors which are more likely to impact the participation of females, leading to the observed under-enrollment in RCTs overall and in different regions.

Several suggestions are proposed to facilitate female enrollment in clinical trials. Providing flexible follow-up visits by considering virtual methods and appointments available outside of working hours or compensating for transportation, loss of work, and caregiving expenses have been suggested to increase female enrollment in cardiovascular trials [27, 28]. Another study of female under-enrollment in stroke trials suggested avoiding upper age limits in the inclusion criteria, given the higher mean age of female stroke survivors [15]. It has also been suggested that having female researchers and leaders in stroke trials may provide an environment where female participants are more likely to join the studies [15]. A recent study on cardiovascular trials showed that having female first or last authors was significantly associated with higher female participants in clinical trials [29].

However, these suggestions may not be feasible in all settings, where funding supports for rehabilitation trials may be less able to provide transportation or financial compensation for participants. Moreover, even with having no age limitation in trials, female stroke survivors who are older may find it difficult to participate as they are more likely to live alone or in home care setting, particularly where active engagement in multiple sessions or follow-up visits are required, which is often the case in rehabilitation trials. For post-stroke rehabilitation RCTs, targeting the home care setting may help recruit more female participants. We also suggest sub-analyzing rehabilitation outcomes based on sex as an alternative to understand the treatment efficacy for females. This can be encouraged by research funders and publishers with clear requirements for post-stroke rehabilitation trials. Examining the reasons for underrepresentation of females in stroke rehabilitation trials will result in better strategies aimed at increasing female participation. More research is needed to discover the reasons why some female (and for comparison male) participants do not become enrolled in clinical trials.

This review provided an overview on female participation in RCTs of post-stroke UE rehabilitation in different regions of the world. We acknowledge there were limitations in this study. The focus of this review was limited to UE stroke rehabilitation trials published in English. Including just English RCTs means these data may not apply to female participation data in those RCTs published in other languages. It is estimated that around 7.4% of the documents indexed in Scopus and 4.6% in Web of Science are not in English, being mostly in Chinese and Spanish, and 13.5% of medical publications in PubMed are published in languages other than English with the most frequent ones in German, French, Japanese, Russian, and Chinese, which are the official languages in both LMICs and HICs [30, 31]. So, although we acknowledge that reviewing only English articles is a limitation, we do not know whether and how this would change the results of our findings. We anticipate that the language limitation would not notably change the main results, given the large scale of our review consisting of 1,276 RCTs. Lastly, due to the large number of RCTs in this review, the rate of female participation was not adjusted according to the overall incidence of stroke in female populations in different regions and countries and at the time of publication. Future research should investigate the female enrollment in clinical trials relative to the stroke epidemiology characteristics of the underlying populations.

Female-to-male ratios in RCTs have not changed significantly over the last decades, and the studies in LMICs had an overall lower percentage of females than studies in HICs. Female underrepresentation in post-stroke UE rehabilitation RCTs in countries with different incomes and across geographical regions highlights the need for sufficient female representation in clinical trials to allow full generalizability of results to both sexes.

An ethics statement is not applicable because this study is based exclusively on the published literature.

The authors have no conflicts of interest to declare. All the co-authors have seen and agreed with the contents of the manuscript. R.T. recently received a research grant from Allergan (now AbbVie) unrelated to the current topic.

This work was supported by the Heart and Stroke Foundation of Canada and the St. Joseph’s Health Care (London) Foundation. The funders had no role in the design, data collection, data analysis, and reporting of this study.

R.T. designed the main database and formulated the research question and the main objective of the review. S.M., L.C., and A.B. were involved in systematic review of the literature, screening, and data extraction. S.M. coordinated the project and was responsible for data management. S.M. and A.B. contributed to the statistical analysis and synthesis of the data. R.T., J.L.F., J.E., and M.B. critically reviewed and revised the manuscript and contributed to discussing the significance of the study. The authorship order reflects the degree of involvement and leadership of each author in the research process. All the authors read and approved the final manuscript.

The data that support the finding of this study are not publicly available due to the ongoing status of the project and lack of public funding to make the data accessible for the publics. However, data are available from the corresponding author in Excel upon reasonable requests.