Introduction

Sex and gender significantly shape health outcomes including through their individual or combined influence on a range of factors, such as rates of exposure to risk factors and environments, individual bodily responses to risk exposure, and patterns of health service use [1]. For example, gender, as a social construction, influences how men and women have different exposure to health-harming products such as tobacco, alcohol, and poor nutrition. Sex plays an important role in health outcomes, including by determining physiological responses to gendered risk exposures (e.g., male and female bodies metabolize alcohol differently), while gender influences healthcare access and the quality of care received from health services [2,3].

Understanding differences in risk exposure, health service use, and health outcomes by sex and gender identity (i.e., whether someone identifies as a woman, man or another gender identity) is vital to the identification of interventions which can effectively reduce health inequities, e.g., ensuring that interventions are tailored to the specific needs of women, men, and gender-diverse individuals. However, sex and gender identity are often confused and conflated in health surveys [1]. As a result, analysis of survey data is challenging—were researchers reporting on sex or gender identity? Does the original language of the survey have different terms to distinguish between sex and gender identity? Additionally, very few surveys outside of a few specific country contexts report on gender identity beyond a simple binary (woman/man) [4]. In this study, we report on data from datasets and surveys as being sex-disaggregated, but interpret any observed differences through the lens of gender analysis—which enables identification of modifiable factors for sex- and gender-responsive interventions.

We analyzed global datasets to construct what we call a health pathway—from risk exposure to mortality (illustrated in Fig 1). The health pathway approach can be applied to improve our understanding of where differences (if any) are seen between males/females, and identify opportunities for tailored interventions to reduce inequities.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Illustration of the health pathway.

https://doi.org/10.1371/journal.pmed.1004592.g001

To illustrate the health pathway concept, we present sex-disaggregated data along each step for three conditions with available global data. Disease modeling groups, such as the Institute for Health Metrics and Evaluation (IHME) and NCD Risk Factor Collaboration (NCD-RisC) [5], provide sex- and age-specific estimates for some steps of the health pathway—particularly, risk factor prevalence, disease prevalence, and mortality. The “care cascade” is a framework commonly used to assess how people with a disease/condition are retained through the sequential stages of care (diagnosis, treatment, disease control), emphasizing its role in tracking patient progress and enhancing access to and quality of care [6–8]. While some regional and national studies offer sex-disaggregated data on the care cascade for specific conditions, globally comparable data remains scarce [9–12]. We conducted a scoping review to identify global sex-disaggregated cascade datasets for eight common conditions, and found data on only three: hypertension, diabetes, and HIV and AIDS. Among the global studies that have analyzed these datasets, none have specifically examined sex differences in the care cascade [6,13–17]. Our study addresses this data gap by constructing a pathway from exposure to outcome, including the care cascade, examining sex differences at each stage of the pathway, for three conditions.

Methods

Results

Fig 2A–2C illustrates the distribution of countries with observed sex differences in various stages of the health pathway for the three conditions. We present each pathway step on the x-axis and age groups on the y-axis. For each pathway step-age combination, the figures show whether more countries report males had significantly higher proportions than females or vice versa. Notice that for hypertension, risk factors, prevalence, cascade, and death rates refer to 204 countries and 5-years age groups, but care cascade data refer to 200 countries and similar age groups. For diabetes and HIV and AIDS, risk factors, prevalence, and death rates refer to 204 countries and 5-years age, whereas care cascade data were available for fewer countries and more aggregated age groups. Differences in the number of countries included in each step are specified in the footnotes, and differences in age groups are reflected by aggregated cells across age groups (Panels B and C). The same figures stratified by income groups and regions are reported in the Figs A–F in S1 Text, and percentage of countries with significant sex differences by income groups and regions are in Fig G–M in S1 Text. Significant sex differences in hypertension, diabetes, and HIV and AIDS health pathways across countries, regions, and income groups are reported in Tables 1–3, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Significant sex difference in hypertension health pathway.

https://doi.org/10.1371/journal.pmed.1004592.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Significant sex difference in diabetes health pathway.

https://doi.org/10.1371/journal.pmed.1004592.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Significant sex difference in HIV and AIDS health pathway.

https://doi.org/10.1371/journal.pmed.1004592.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Sex differences (significant when non-overlapping confidence intervals of estimates between females and males) in global health pathways of hypertension, diabetes, and HIV and AIDS.

https://doi.org/10.1371/journal.pmed.1004592.g002

Discussion

Our analysis of sex-disaggregated data along the health pathway is a unique and ambitious attempt to holistically and systematically document sex differences in risk exposure, morbidity, care access, and mortality outcomes, based on existing and standardized datasets. The health pathway approach moves beyond focusing on a single step—whether it be one risk factor or death rate—to view (ill)health that is determined in continuation and accumulation along all steps. By examining the entire health pathway, we can identify the various points at which inequities arise, generate more nuanced hypotheses of the complex network of influences shaping health and mortality outcomes, and thereby design interventions that are more responsive to the roles that sex and gender are playing in outcome differences and arguably improve the health of all. We recognize that progression along the pathway is not inevitable—not every smoker will die from lung cancer, and not everyone who is obese will develop diabetes, for example. The pathway is also not designed to continuously model risks of progression from one step to the next. While acknowledging these shortcomings, this approach has revealed several insights for further exploration and action.

Findings of the study show significant sex differences in each step along pathways. In many countries, males exhibited higher disease prevalence and death rates than females, while in some countries, they also reported lower rates of healthcare seeking, diagnosis, and treatment adherence. For risk factors, males had significantly higher smoking rates in 86% of countries, whereas females had significantly higher smoking rates only in Bhutan. In contrast, females had significantly higher obesity rates in 64% of countries, while males had higher obesity rates only in Malawi and China, highlighting important differences in risk exposure, highlighting important differences in risk exposure [25]. Regarding disease prevalence, males had significantly higher prevalence rates of conditions in 56% of countries for HIV and AIDS, 30% for diabetes, and 4% for hypertension. Females, on the other hand, had significantly higher prevalence rates of conditions in 14% of countries for HIV and AIDS, 5% for diabetes, and only in India for hypertension. Across all the three conditions, death rates were higher in males compared to females in 131 countries (64%) for HIV and AIDS, 107 for hypertension (53%) and 100 countries for diabetes (49%). In contrast, death rates were higher in females compared to males in 25 countries (12%) for HIV and AIDS, 9 countries for diabetes (4%), and only in the United Arab Emirates for hypertension. In the care cascades, females generally had higher rates of diagnosis, treatment, and control. This pattern was observed in 13% of countries for HIV and AIDS (10 countries), 2% for hypertension (3 countries), and only in Cape Verde for diabetes. Males had higher rates of HIV and AIDS treatment and control only in Lebanon. Such findings echo previous studies showing men’s lower use of preventive, diagnostic, and treatment services compared to women, including for HIV, COVID-19, and other conditions [26–28]. Several factors, including rigid constructions of norms relating to masculinity [29], wider societal practices (such as clinic opening hours restricting men’s access) [30], and structural drivers (including healthcare financing) [31] may be contributing to lower healthcare utilization among males. Improving population health, including through universal access to healthcare services, requires consideration of and action on barriers that restrict healthcare access and utilization for all people of all gender identities [32].

The analysis of the pathway data across the three conditions identified some opportunities for gender-responsive interventions to address some risk exposures. For example, smoking rates were higher in males almost worldwide, while obesity prevalence was higher among females in two-thirds of the countries. Such findings are not totally unexpected, but despite widespread and long-standing evidence of higher tobacco-smoking rates in men (and LGBTQ+ populations) [25], frequently associated with the socio-cultural construction of gender norms, and manipulated and promoted by the tobacco industry, relatively little work has been undertaken by the public health community to design gender-responsive tobacco control programmes. This represents a missed opportunity for improving population health and reducing gender gaps in health.

Previous studies have reported on the association between the cascade of care and factors associated with care progression across countries [6,13,17] but none of the studies discussed sex differences. In our study, while many countries showed significant sex differences in risk factors, disease prevalence, and death, fewer exhibited significant differences between males and females within the care cascade for hypertension, diabetes, and HIV/AIDS. For example, for HIV and AIDS, while almost 60% of the countries show significant sex differences in drug use, unsafe sex, disease prevalence, and death rates, only 13% of the countries report significant sex differences in diagnosis, treatment, and control. There are several potential explanations to these findings. We are only reviewing three conditions, which may happen to have smaller sex differences along the pathway than other leading causes of morbidity and mortality. Other conditions, such as mental health disorders, and infant and neonatal health [9,10,21,22], or cardiovascular diseases [33,34] have exhibited more pronounced sex-differences in care-seeking patterns in some countries. Once in the care system, studies have shown differences in the quality of care received— nationwide cohort study in England and Wales found that women suffering a heart attack were less likely to receive care that followed clinical guidelines [35]. Other studies have found higher mortality rates for female heart attack patients treated by male physicians, suggesting gender bias [36]. These differences illustrate the importance of having sex-disaggregated data along the care cascade for all conditions—findings will vary by condition and country, and the underlying reasons for inequities will vary, meaning that interventions need to be data-driven and context specific.

Another key finding of the study is a notable absence of globally available sex- and age-disaggregated care cascade data. We identified only three conditions, among the eight major conditions we explored, that have sex-disaggregated cascade data for more than five countries. The absence of this data for some major contributors to morbidity and mortality was surprising. For example, we were unable to identify global sex-disaggregated care cascade datasets for tuberculosis. A previous study reported that while males are biologically more vulnerable to tuberculosis than females, traditional gender norms may discourage women from using prevention and treatment services [37]. However, without sex-disaggregated data on the tuberculosis cascade, it is difficult to identify who is left behind and where to intervene most strategically on the pathway. Although global data groups such as NCD RisC include data disaggregated by sex, studies analyzing trends in risk factors, prevalence, and treatment for the three conditions we reviewed often provide limited discussion on sex differences or focus primarily on aggregate populations without explicitly assessing sex-based disparities. While several studies report estimates separately for females and males, they mostly do not include a detailed assessment of sex differences in outcomes (e.g.,[6,38–40]). For example, a recent analysis of global trends in hypertension prevalence and treatment from 1990 to 2019 using data from the NCD Risk Factor Collaboration presented estimates stratified by sex but did not deeply examine sex differences in trends and outcomes [6,38]. Other studies, such as those by Rahim and colleagues [41] and Lee and colleagues [42], focus on total population estimates without sex-specific analyses. Additionally, studies have limitations in geographical coverage [39,42,43]. For example, three recent papers examined diabetes risk and preventive measures across low- and middle-income countries and the diagnostic testing for hypertension, diabetes, and hypercholesterolemia within similar economic groups [17,40,41], while another study explored trends in hypertension awareness, treatment, and control across 12 high-income countries [21]. Moreover, sex-disaggregated analyses, particularly regarding the care cascade for specific diseases, remain limited [11].

Among the identified datasets we used in this study, their coverage and completeness are far from satisfactory. While complete hypertension care cascade data was available for about 200 countries, for diabetes we were only able to construct the cascade for 39 countries in the last decade, and for most countries data were available for only one time period thus excluding the possibility of monitoring trends over time. Similarly, although data on the HIV and AIDS care cascade were available for 76 countries, these were not stratified by age. Other studies have found that age-specific data are essential for designing effective HIV and AIDS programmes that can adapt to the changing healthcare needs of populations as they age [14]. These limitations highlight the need for the creation of comprehensive and harmonized cascade reporting systems on a wider range of conditions, disaggregated by sex and age, with broader geographical and temporal coverage.

Among 204 countries, our analysis found significantly higher mortality rates in males compared to females in more than half of the countries. Our analysis cannot explain these differences (or similarities), but can highlight areas for further exploration with the lenses of both gender and sex. For example, are men subject to the constructions of masculinities that often discourage prevention and care-seeking, presenting later in the disease progression at health facilities compared to women (and hence more at risk of mortality from the diagnosed disease)? Are women’s causes of death being under-reported or mis-attributed? Are there biological reasons why males have a higher mortality rate from diabetes compared to females—and, if so, does this warrant sex-specific clinical guidelines? We cannot address these questions within the pathway, but the pathway can pinpoint questions for deeper analysis and future research.

Despite a strong evidence base confirming the importance of sex and gender as determinants of disease, health policies often do not include sex-specific responsive approaches. Global Health 50/50 found that 91% of COVID-19 national health-related policies (of approximately 450 reviewed) did not consider gender [44]. The absence of gender-responsive policies in many areas of health results in gender-based inequities from risk exposure to care cascades that are not being adequately addressed in many settings. Acknowledging and addressing the unique health needs of women, men, and gender-diverse people through gender-responsive interventions is needed if we are to reduce health inequities across the whole population and better health outcomes for all.

Our study has some limitations including, the lack of available datasets and their own respective limitations. For example, on the diabetes and HIV cascade data, we observed variations in data quality and completeness across countries. Secondly, the data available may be affected by sample bias including the underrepresentation of certain populations, and inconsistencies in data collection methods. Reliance on national surveys may result in the exclusion or under-representation of some populations such as those marginalized by sexual or gender identity, or people living in geographically remote or in challenging conditions such as conflict zones. Self-reported data, inconsistencies in measurement techniques and varying definitions of health conditions further complicate comparability. Thirdly, although we identify gender as determinant of inequities along the pathway, gender is expressed through identity, experienced through socio-cultural norms, and embedded in inequitable systems and structures [32]. Therefore, our ability to reduce health inequities will require acknowledging and addressing the role of gender across all these levels. Fourthly, we cannot empirically link the steps in the health pathway with each other. For example, we know that for any given condition the data on risk factor exposure and prevalence are related, but we cannot causally associate risk factor exposure and prevalence based on the datasets at hand, all of which were collected or modeled separately. This highlights the importance of longitudinal studies for a more nuanced understanding of causal relationships and sequential progress along the health pathway.

Based on our findings, we urge the collection and publication by government and research entities of comparable global data along the health pathway that is disaggregated by sex and age, and that it starts by focusing on the leading causes of death and morbidity. Without such data, we will be navigating the health pathway without clear insights into health inequities, thus limiting our ability to identify which population is suffering inequity (e.g., in rates of exposure to risk factors, low rates of access to health services, quality of care received, etc.) and thus where gender-responsive interventions might reduce inequities.

In order to tackle health inequities, we need more policy directives for data disaggregation across more intersectional characteristics and more health conditions and for the reporting of such data in the public domain. We call on policymakers to attempt to make greater use of such data to identify and respond to those population groups stumbling along or falling off the pathway. By integrating our understanding of the sex and gender differences through analyses such as this, policymakers can be encouraged to develop and implement gender-responsive strategies and interventions. Moreover, by comparing countries along the health pathways, our approach allowed us to identify countries that do not follow the general trends in sex differences, encouraging further investigation into the reasons behind the discrepancies. Disaggregating by other characteristics—urban/rural location, socio-economic position, race/ethnicity, disability, etc. —would add even greater specificity and permit even more effective policy and programme responses.

In addition to more disaggregated data, there is a need for inclusive data collection tools and harmonized reporting guidelines. This would enable more robust secondary analyses, including cross-country comparisons and in-depth exploration of subpopulations. Achieving this requires leadership from large international organizations or consortia, such as WHO or GBD, to ensure global alignment and coordination. To assemble and publish standardized datasets for additional health pathways, several options present themselves. As directing and coordinating authority for international cooperation on health, WHO can standardize data across countries, but it faces resource limitations and many competing priorities. Entities such as Global Health 50/50, Data 2× or Equal Measures (https://equalmeasures2030.org/members/data2x/) can seek private and other funding and operate independently, but they lack the convening power and generally must rely on publicly available data. Condition-specific epistemic communities offer deep expertise in their areas, yet they focus on one condition at a time, making comparisons challenging. Therefore, a collaborative approach might be the most effective solution to more joined up data coalition and analysis along condition-specific pathways. Finally, future iterations of country pathways should incorporate more upstream legal and policy environments, as these significantly influence risk exposures.

Supporting information

Acknowledgments

All authors thank IHME for providing prevalence data and NCD-RisC for providing country-level data, and Sonja Tanaka for her administrative support.