Sex-Based Differences in the Outcomes of Intracerebral Haemorrhage: A Systematic Review and Meta-Analysis

Intracerebral haemorrhage (ICH) is associated with a high fatality rate and has an estimated global incidence of around 30 per 100,000 person-years [1]. The risk of ICH increases with age, and a comparatively higher incidence is reported in the Asian population [1]. Epidemiological data suggests that the risk of developing ICH is higher in males than females in both developing and developed countries [2]. Previous studies have shown sex-based differences in the incidence and outcomes of cardiovascular diseases [3] and showed that the age-specific risk of cardiovascular diseases and associated mortality were overall higher in men [3]. Similarly, a recent systematic review documented an increased risk of ischaemic stroke in women (incidence risk ratio of 1.44, i.e., 44% more women with ischaemic stroke than men) among the young adult population (aged ≤35 years) [5]. Studies report a comparatively poorer quality of life post-ischaemic stroke and poorer functional outcomes in women, compared to men [6]. While there are studies suggesting that sex-related factors may influence stroke outcomes [8], the data on sex-related outcomes in ICH subjects is scarce. A review by Poorthuis et al. [9] reported different sex-based factors that are associated with the risk of stroke and suggested that these factors need to be considered to provide a tailored assessment of the ICH risk.

A previous meta-analysis that included 7 studies with 4,724 ICH subjects reported on case fatality rates according to sex but did not provide credible data on the differences in functional outcomes, mainly because the included studies varied in their design and time of assessment [10]. The study found similar overall mortality in males and females [10]. Further, only studies that were conducted before the year 2000 were included. However, there has been considerable advancement in the management of ICH in the last 2 decades. A deeper understanding of sex-related differences in ICH subjects is important for medical professionals, including the nursing team, as a tailored approach, based on the sex of the subject, could enhance post-stroke rehabilitation, ensure self-management capabilities, possibly lower the risk of complications, and improve the overall quality of life. This review and meta-analysis aimed to summarize the most recent evidence on the sex-based differences in the mortality risk and functional outcomes in ICH subjects.

We adhered to the recommended PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [11] during the process. To ensure transparency and protocol adherence, we registered our review protocol in PROSPERO with the registration number CRD42023443281 before commencing the meta-analysis.

We conducted a thorough search of three prominent databases, namely PubMed, EMBASE, and Scopus, to identify relevant observational studies. Online supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000535612) provides an overview of our search strategy. Our inclusion criteria focused on studies published in English up until December 31, 2022.

In order to select appropriate studies for analysis, this study implemented specific inclusion criteria. The first criterion was to focus on subjects diagnosed with ICH. For our study, ICH was defined as a brain injury resulting from the leakage of blood into the surrounding brain tissue due to the rupture of a cerebral blood vessel. For the study to be included, trained neurologists should have diagnosed ICH by assessing the subject’s clinical condition and utilizing diagnostic imaging techniques, including brain computed tomography (CT) scans and/or magnetic resonance imaging (MRI). We included only those studies that had subjects with ICH and not subarachnoid haemorrhage (SAH) and/or primary intraventricular haemorrhage (IVH). The second criterion involved the inclusion of studies that documented and compared the risk of mortality and functional outcomes based on the sex of the subjects. The third criterion required the selected studies to have employed an observational design, which could include cohort studies, case-control studies, or cross-sectional studies. Additionally, studies that conducted a secondary analysis of data, utilizing population-based or hospital-based registries, were also considered eligible. Lastly, the publication timeframe was limited to studies published from the year 2000 onwards. By adhering to these inclusion criteria, the study aimed to acquire pertinent and up-to-date evidence regarding sex-related disparities in mortality and functional outcomes among subjects with ICH.

Once the search strategy was implemented, the study proceeded to identify the total number of studies available in each of the three selected databases. Duplicates were then removed to ensure that each unique study was included only once. The remaining studies underwent a screening process in which the title and abstract of each study were independently assessed by two authors (L.Y. and J.H.) of the study. Based on this initial screening, certain studies were excluded. Subsequently, the full texts of the remaining papers were thoroughly examined by the authors independently, and decisions were made regarding their final inclusion in the meta-analysis. Any discrepancies or disagreements that arose during this process were resolved through discussion among the authors (L.Y., J.H., and C.Q.). This rigorous screening and selection procedure aimed to ensure the inclusion of relevant and appropriate studies for the subsequent analysis.

Data extraction from the final selection of studies involved the utilization of a pre-tested electronic sheet. All analyses were performed using STATA 16 software (TX, USA). The pooled effect sizes were presented as odds ratios (OR) with corresponding 95% confidence intervals (CI). Given the variations observed in the subjects’ characteristics, study settings, and characteristics of ICH such as location, aetiology, and duration of follow-up among the included studies, a random-effects model was chosen for all analyses. Publication bias was evaluated using Egger’s test and examination of the funnel plot [11].

The Newcastle-Ottawa Scale was used to assess the risk of bias [12]. Post-hoc subgroup analysis was conducted according to the sample size (<500 and ≥500), study setting (high-income; upper-middle-income), type of haemorrhage (spontaneous or with underlying cause) and location of the haemorrhage (predominantly deep i.e., affecting basal ganglia and/or thalamus; or both deep and lobar i.e., affecting basal ganglia and/or thalamus along with either of the cerebral lobes).

The systematic search across three databases initially yielded 3,243 studies. After removing duplicates, 2,450 unique studies remained. Subsequent screening of titles and abstracts further reduced the number by 2,308 studies. Following a thorough review of the full texts, 110 additional studies were eliminated. Ultimately, a total of 32 studies (14–45) were included in this meta-analysis. Figure 1 provides a visual representation of the selection process.

As shown in Table 1, most of the studies were retrospective (n = 26). Four studies were prospective in design, and the remaining two studies used data collected as a part of a randomized controlled trial. The majority of the studies were conducted in the USA (n = 9), followed by China (n = 7). Three studies were conducted in Sweden and two in Spain. One study each was conducted in Switzerland, Denmark, Qatar, Germany, Japan, Portugal, Singapore, Brazil, Finland, and Turkey. There were 22 studies from high-income countries and 10 studies from upper-middle-income countries. One of the included studies was multicentric. Most of the included studies reported data on the spontaneous ICH (n = 22). In 12 studies, the predominant location of haematoma was in the basal ganglia and/or thalamus (deep), and in 10 studies, it was a combination of deep and lobar (frontal/parietal/temporal/occipital lobes). Categorization based on the sample size resulted in 20 studies with ≥500 subjects and 12 studies with a sample size under 500. Most of the studies that reported on functional outcomes used the modified Rankin Scale (mRS) (n = 16). However, the cut-off used to define poor functional outcomes varied among these studies. Eleven studies used an mRS score of 3–5 (excluding a score of 6, which represents death), three studies used a score of more than 3 and two studies used a mRS score of more than 2. Two studies used “independent ambulation” and one study used “discharge to a skilled nursing facility or hospice” to define poor functional outcomes.

Baseline characteristics differed significantly in male and female subjects across most of the studies. Consistently, female subjects had a higher mean age compared to males and had a higher incidence of neurological deficits at the time of admission. A higher proportion of male subjects have accompanying cardiovascular risk factors such as smoking, alcohol use, hypertension, diabetes mellitus, dyslipidaemia, and previous myocardial infarction. As shown in Table 1, all included studies were of good quality.

Pooled analysis of mortality outcomes showed comparable results in male and female subjects. The risk of mortality at hospital discharge (OR 0.98, 95% CI: 0.90, 1.06, N = 11, I² = 83.9%), at 1-month (OR 0.98, 95% CI: 0.81, 1.18, N = 8, I² = 79.9%), 3-month (OR 1.13, 95% CI: 0.95, 1.33, N = 12, I² = 63.5%), and 12-month (OR 1.04, 95% CI: 0.90, 1.19, N = 12, I² = 70.3%) follow-up was similar in both sexes (Fig. 2). We did not detect publication bias in the studies, except for the risk of mortality at 1-month follow-up (p = 0.039), as shown by Egger’s test and the funnel plots (online suppl. Fig. 1–4).

The risk of mortality was similar at all time points when the sub-group analysis was conducted based on the location of the haematoma and sample size (Table 2). Similarly, pooled analysis of the studies that only reported data on spontaneous ICH did not show any difference in risk of mortality at 1-year follow-up. We then performed the analysis of the mortality in male subjects at hospital discharge based on the study setting. Our results show a reduced mortality risk in studies conducted in high-income countries (OR 0.91, 95% CI: 0.84, 0.99, N = 7, I² = 40.2%), whereas an increased risk was observed in studies from upper-middle-income countries (OR 1.09, 95% CI: 1.06, 1.13, N = 4, I² = 0.0%) (Table 2).

Compared to females, male subjects with ICH had a lower risk of poor functional outcomes at 3 months (OR 0.83, 95% CI: 0.77, 0.89, N = 13, I² = 0.0%) and 12-month (OR 0.87, 95% CI: 0.77, 0.98, N = 4, I² = 0.0%) follow-up (Fig. 3). Although not statistically significant, males had a comparatively lower risk of poor functional outcome at the time of hospital discharge (OR 0.83, 95% CI: 0.68, 1.02, N = 9, I² = 80.8%), compared to their female counterparts. We did not note the possibility of publication bias at any of the three time points, i.e., at hospital discharge, at 3-month, and at 12-month follow-up (online suppl. Fig. 5–7).

When the analysis was conducted with studies having subjects with spontaneous ICH, the reduced risk of poor functional outcome among males was observed at 3-month (OR 0.84, 95% CI: 0.78, 0.90, N = 10, I² = 0.0%) and 12-month (OR 0.87, 95% CI: 0.77, 0.98, N = 4, I² = 0.0%) follow-up (Table 3). In studies from both high-income and upper-middle-income settings, a reduced risk of poor functional outcome was observed in male subjects at 3- and 12-month follow-up. Similarly, irrespective of the location of the haematoma, better functional outcomes were noted for males at 3- and 12-month follow-up. Studies with large sample sizes (≥500) consistently reported a lower risk of poor functional outcome in males at 3-month (OR 0.83, 95% CI: 0.78, 0.90, N = 10, I² = 0.0%) and 12-month (OR 0.87, 95% CI: 0.77, 0.98, N = 4, I² = 0.0%) follow-up (Table 3).

The present meta-analysis was done with the aim to synthesize findings of the recent studies and to compare the risk of mortality and poor functional outcomes in male and female subjects with ICH. Our study synthesized findings from 32 studies and found that the risk of mortality at hospital discharge and at 1-, 3-, and 12-month follow-up was similar in both sexes. However, males had better functional outcomes compared to females at 3-month and 12-month follow-up.

The findings related to the similar risk of mortality in both sexes have also been documented in the previous review by Asch et al. [10]. Their systematic review and meta-analysis reported noted similar case fatality rates in male (35.4; 95% CI: 33. 6–37.1) and female (35.3%; 95% CI: 33. 2–37. 4) ICH subjects [10]. The review, however, could not provide synthesized findings related to the functional outcomes, mainly because the included studies varied in their designs and timing of assessment. Another review by Gokhale et al. [2] reported similar findings. Additionally, they showed that several factors, such as age, race, ethnicity, underlying risk factors, location of the haematoma, management provided, and response to the treatment, can influence the association of sex with survival and functional outcomes in subjects with ICH [2].

Current studies suggest that the association of sex with the outcomes in ICH subjects should be interpreted with caution as several factors may impact this association [7]. These factors include the higher age of female subjects at the time of the first stroke episode compared to males and the higher incidence of pre-stroke disability in females [45‒47]. Our analysis demonstrated that the baseline characteristics differed significantly in male and female subjects in most of the included studies. Female subjects were older than male subjects and had a greater rate of neurological deficits at the time of admission. Epidemiological studies have shown that, on average, women tend to be older than males by 4–5 years at the time of the first occurrence of stroke [48]. This may imply that women may have more co-morbidities compared to their male counterparts. Therefore, in female subjects with stroke, a comparatively higher rate of disability at the time of hospital admission may be reflected as poorer functional outcomes in the post-stroke period. Post-stroke depression has also been mentioned as an important factor influencing the outcomes in subjects with ICH [50]. Studies have documented that the prevalence of mood disorders and depression after an episode of stroke remains comparatively higher in female subjects (around 21%) than in male ones (around 16%) [7] and may contribute to poorer functional outcomes in female ICH subjects.

The study by Phan et al. [53] investigated sex-based variations in the health-related quality of life in subjects that survived an episode of stroke. The findings were indicative of comparatively poorer quality of life in women, compared to men, at one- and 5-year follow-up. Interestingly, when additional factors such as age, the severity of the stroke, pre-stroke disability, and post-stroke depression were put in the analytic model, the sex-based differences in the quality-of-life outcomes became non-significant [53]. Socio-economic factors also may influence the association between sex and stroke outcomes. The support for this comes from a large meta-analysis that looked at the association between sex and the risk of cardiovascular diseases, including stroke, across different socio-economic factors, i.e., education, occupation, and family income [54]. The review found that lower educational levels and income were linked to an increased risk of stroke both in males and females [54]. Findings from cohort-based data suggest that in older adults, lower educational levels may be related to poor functional outcomes, particularly in men [55]. The findings of our review have pragmatic implications for guiding nursing care in subjects with ICH. Nursing personnel can be trained in assessing the risk factors at the time of hospital admission and support the treating physicians in post-stroke rehabilitation.

The findings of the study have practical implications and highlight the need for gender-specific research, particularly from low- to-middle-income settings. The importance of reporting sex-disaggregated data on ICH from low- or middle-income countries appears to be of utmost significance. These regions often grapple with distinct challenges and resource limitations that set them apart from high-income countries. Understanding how ICH affects different genders in these contexts is crucial for several compelling reasons. Firstly, these countries tend to exhibit pronounced healthcare disparities, and gender-specific data helps identify any inequalities in access to healthcare services, treatment, and outcomes. Secondly, as resources for healthcare are typically limited in these settings, reporting data separately for males and females aids in assessing resource allocation to ensure equitable access to treatment and care. Additionally, ICH risk factors may differ between genders due to sociocultural and biological differences. Disaggregated data allows researchers to pinpoint these gender-specific risk factors, informing tailored prevention strategies. Furthermore, gender disparities in ICH treatment outcomes may exist due to factors like differing responses to therapies, access to rehabilitation, or socio-economic conditions. Sex-disaggregated data may identify these disparities, supporting the development of interventions that work for both sexes. Moreover, such data is essential for crafting effective public health policies. It will ensure that health policies are responsive to the specific needs of each gender. The availability of sex-disaggregated data may also provide impetus for more research on gender-specific health disparities, raising awareness about these issues and potentially leading to increased funding for gender-sensitive ICH research.

This meta-analysis has some limitations. First, it mostly included retrospective studies. Therefore, it is possible that data on some important confounders was missing or not adjusted for in the analysis. This may introduce a certain bias into the observed associations. Second, our results could be significantly influenced by the baseline differences in the characteristics of the male and female subjects. Third, for some of the outcomes, there was high heterogeneity. This may be due to differences in baseline characteristics of the subjects, variations in the definitions of functional outcomes used, different geographies in which the studies were conducted, variations in the sample sizes, and predominant location of haematoma. To account for some of these heterogeneities, we conducted subgroup analysis. However, future studies should be more harmonized concerning the study methodology adopted. Fourth, all the studies were done either in high-income or upper-middle-income countries with no representation from low- and low-middle-income countries. Therefore, our findings have limited external generalizability. Additionally, due to the limited number of available studies, our analysis did not take into account relevant data on the established prognostic factors (such as management provided, extent location, and size of haemorrhage, use of anti-coagulants, etc.) of mortality and functional outcomes in subjects with ICH. Further analyses that would include these factors are needed.

This study pooled the findings from more than 30 studies from high-income and upper-middle-income countries and showed a similar risk of mortality in male and female subjects with ICH. Male sex was associated with better functional outcomes of ICH compared to females. Future studies, focussing on the association of sex with survival and functional outcomes, should also include additional factors such as race, ethnicity, socio-economic background, underlying risk factors and management provided, post-stroke depression, and pre-stroke disability status.

An ethics approval is not applicable as this is a systematic review and meta-analysis of published data.

The authors have no conflicts of interest to declare.

The authors declare no funding was received.

L.Y., J.H., and C.Q.: contributed to structuring the review, discussing the literature and the knowledge gaps. W.S.: wrote sections of the manuscript. All authors contributed to the article and approved the submitted version.

All data generated or analyzed during this study are included in this article and its online supplementary material files. Further inquiries can be directed to the corresponding author.