Introduction

Life expectancy has continuously increased since the Industrial Revolution due to improved life conditions, public health interventions, and scientific findings [1,2]. This trend has led to a steadily aging global population. In Mexico, the Law of the Rights of Older Adults (Ley de los Derechos de las Personas Adultas Mayores) defines an older adult as an individual aged 60 years or older [3]. This population grew from 6% of the total population in 1990 to 12% in 2020 [4,5]. The rise in the number of older adults has created an epidemiological scenario characterized by chronic diseases [6] and a reduction of functionality [7]. A research study reported an increase in disability among Mexican older adults, from 13.1% to 15.4% between 2012 and 2015 [8]. In 2020, 21% of older adults in Mexico were living with a disability [9].

The loss of functionality varies based on economic, social, ethnic, or cultural factors, thus leading to inequity [10,11]. It’s well known that lower socioeconomic status [12], having minor levels of education [13], or being a woman [8,13] increases the probability of having a disability in older adults. However, the processes or mechanisms from which these differences exist are not totally clear [14]. Certain limitations of classical epidemiological research overlook the complete understanding of health inequities—unjust disparities in health outcomes rooted in non-biological factors. It often assumes that the characteristics involved in the genesis of inequality are independent. However, these attributes must be fully studied since they are closely interrelated, and this interrelation significantly shapes inequality [15,16]. Analyzing one attribute at a time leads to identifying risk profiles at the individual level, resulting in interventions focused on the person, leaving behind the contextual factors that significantly impact personal health. Such studies also rely on group averages for comparisons, which offer limited predictive power in distinguishing cases and non-cases. These comparisons are typically made between the most and least advantaged individuals or groups, overlooking the differences within the middle groups, often excluding them from the public health agenda [17,18]. Consequently, inequity must be analyzed using more comprehensive frameworks that thoroughly explain how individual, or group attributes define and maintain health inequities.

Intersectionality has been proposed to address these limitations. As a theoretical framework, it posits that individuals belong to multiple social categories based on characteristics, such as sex, religion, race, education, income, etc. Each of these characteristics is linked to power relations, which are expressed through societal advantages or disadvantages (e.g., racism, sexism, xenophobia, homophobia). These attributes are interrelated and shape an individual’s “social location” within the social structure. This location determines differential access to resources and opportunities through complex, interconnected systems of privilege and oppression at the structural level. As an analytical framework, intersectionality argues that a comprehensive understanding of inequity can only be achieved by examining the intersections of these attributes at both individual and group levels [19–23]. Therefore, applying intersectionality in epidemiology allows for a more accurate capture of the structural mechanisms that produce health inequities, and facilitates an understanding of how social systems shape the inequities experienced by each individual [24]. Since its conceptualization in 1989 [25], intersectionality has been predominantly explored through qualitative studies, while quantitative intersectional analyses have only recently emerged and remain in early stages of use. Additionally, key variables often analyzed include sex, ethnicity, and socioeconomic factors, though inequities in disability have been poorly explored [26].

By using the intersectional quantitative approach, it has been possible to gain a deeper understanding of inequity. For example, in the United States, opioid addiction has traditionally been associated with white, middle-class individuals. However, intersectional analysis has revealed that other ethnic groups—such as African American and Latino individuals with higher incomes—also experience significant levels of opioid addiction [27]. These groups have often been overlooked in public policies aimed at addressing this problem. Additionally, intersectionality has provided insights into the “Mexican paradox” which refers to the phenomenon where Latino-Mexican individuals in the U.S. exhibit better health outcomes for certain conditions compared to white Americans. It has been suggested that the type of employment among Latino-Mexican individuals contributes significantly to their lower levels of disability [28].

Consequently, the intersectional framework can be used to gain a comprehensive understanding of the complexity of health inequities [15]. Applying an intersectional analysis is essential when studying functionality among Mexican older adults, as they face significant inequities [29] and a growing prevalence of functionality loss [8,9]. Therefore, this research aims to apply the intersectional quantitative approach to analyze health inequities in functional limitation among older adults.

Materials and methods

Results

Fig 1 shows the sample selection of the 2021 MHAS survey, from the 15,248 observations, 13,982 were considered to define the 192 strata. Table 1 shows the size distribution of these observations in the strata. The strata with 5 or fewer observations were eliminated. Consequently, from the 192 strata, 157 remain and represent 13,903 observations (Fig 1).

Download:

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

Table 1. Size distribution of the 192 constructed strata.

https://doi.org/10.1371/journal.pone.0325211.t001

Download:

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

Fig 1. Sample selection of the 2021 MHAS survey.

BADLs: Basic activities of daily living. IADLs: Instrumental activities of daily living.

https://doi.org/10.1371/journal.pone.0325211.g001

Table 2 reports the descriptive statistics of the final sample. From the 13,903 observations, 3,426 (24.64%) have functional limitations. Younger adults, males, with social engagement, and good socioeconomic status tended to have less functional limitations. Supporting Information 2 presents the strata ranking and its probability of functional limitations (S2 File). The consistent characteristics related to the outcome were age, social engagement, and economic status. The strata ranked 1 and 157 had a likelihood of functional limitation of 8.31% and 77.95%, respectively. The functional limitation probability of the most vulnerable strata was 9 times higher than that of the most protected strata.

Download:

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

Table 2. Descriptive analysis of the final sample.

https://doi.org/10.1371/journal.pone.0325211.t002

Table 3 shows results from Model 1, Model 2A-F, and Model 3 from MAIHDA analysis. The VPC in Model 1 indicated that less than 20% of the total individual differences related to whether someone lost or did not have functional limitations were attributed to the strata to which they belonged. This value indicates a moderate to high contextual influence on the outcome. The PCV value from the variables included in Models 2A-F means that age, social engagement, and economic status were the most important factors in explaining variation in the probability of having functional limitations in older adults.

Download:

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

Table 3. Regression Models 1, 2A-2F, and 3.

https://doi.org/10.1371/journal.pone.0325211.t003

After adjusting for the six variables, the VPC and PCV were 0.68% and 96.7%, respectively. These findings suggest that the differences in functional limitations across strata are largely explained by the additive effects of the variables—that is, each factor independently contributes to the observed inequities. However, the remaining unexplained variance indicates that not all disparities can be attributed to additive effects alone. The interaction between characteristics creates unique configurations of disadvantage that go beyond the isolated influence of each variable. Thus, the results from Model 3 underscore the limitations of a purely additive approach and highlight the relevance of intersectional in understanding inequities in functional limitations.

As described in the methodology, the key statistical measures used to interpret the intersectional model are the VPC, PCV, and AUC; the OR values must be interpreted considering these measures. Accordingly, Model 3 presents the ORs for the main effects. The strata that contain more older adults from 70−79 (OR:2.17; 95%CI:1.86–2.55) and 80 and more years (OR:5.28; 95%CI:4.38–6.40), the ones with more people that report bad economic status (OR:2.46; 95%CI:2.05–2.96), less social engagement (OR:1.75; 95%CI:1.54–1.99), less education (OR:1.37; 95%CI:1.19–1.57) and with more women (OR:1.16; 95%CI:1.03–1.30) increase their probability of having functional limitations. Finally, all the AUC values show an acceptable discriminatory accuracy [39]. This indicates that a randomly chosen individual experiencing functional limitations is likelier to belong to a stratum with a higher predicted probability of such outcome than an individual without functional limitations.

Discussion

To our knowledge, this is the first research studying inequity in functional limitations in older Mexican adults using the intersectional framework and incorporating less frequently analyzed variables like health system and social engagement. Despite our findings aligning with classical epidemiological research, we provide key insights that enhance the understanding of disparities in functional limitations by demonstrating how the interaction between multiple variables contributes to the definition of inequities. So, this approach offers an integral, rather than partial, perspective. The main findings were: 1) age, social engagement, and economic status are the most important variables explaining the functional limitations in strata. 2) Contrary to prior evidence, sex didn’t have a relation with the outcome. 3) Access to health services didn’t relate to the outcome and 4) not all inequity is explained by additive effects. Therefore, the social context in which the individual lives plays a role in the differences in functional limitations.

Regarding the first finding, the VPC value in Model 1 indicates that the stratum where individuals are nested accounts for a significant variation in functional limitations, highlighting the contextual influence. This suggests the role of structural social determinants of health (SDH) in explaining disparities. Meanwhile, PCV values from Model 2 A-F reveal the contribution of specific variables to the between-intersection variance. Through the intersectional framework, we identified how these variables interact to create vulnerability patterns, highlighting their combined influence on functional limitations rather than considering them in isolation. As expected and reported previously, age is the main variable that reduces functionality [40–43], followed by social engagement, and economic status.

Functionality reduction is strongly related to age because of a progressive decline through time that can be slowed down with physical activity and healthy diets [44]. Thus, the link between age and functional limitation may reflect this process in our analysis. While age may have limited implications for functional inequities, other outcomes—such as abuse, mental health issues, cognitive decline, and poor quality of life—are significantly affected by the negative social perception of aging—ageism— [45]. Therefore, it’s important to include age as an intersection to completely understand the impact of ageism in shaping health differences [24,45–47].

As said before, SDHs are important in producing and maintaining inequities. Since 1790, poverty has been recognized as a root cause of diseases [48]. Subsequent scientific literature has supported this statement, indicating that lower economic status is associated with increased frequency of almost all diseases [49]. Regarding disability, it’s globally acknowledged that good economic status is linked to a lower likelihood of disability. Studies in Brazil [40], Spain [41], and Mexico [13] show a gradient where higher income is associated with a lower probability of disability. Education is another SDH closely related to inequity; globally, higher educational attainment is linked to lower disability rates [13,40,42,43,50–52]. Despite this, our findings show a PCV value for education of 9.70%, which means that a small percentage of differences between strata accounted for this variable. This may reflect the structural hierarchy of SDH [33]. Some authors argued that the only effective intervention to reduce health inequities is addressing the gaps in economic status, as social class significantly shapes inequity [53]. For example, a wealthy Black woman faces lower inequities than a poor Black woman, so this mix of privilege and oppression attributes may have a hierarchical influence on health outcomes [17].

However, economic status and education are difficult to modify to improve functionality as they require systemic, long-term efforts rather than direct interventions. We identified social engagement as a key modifiable factor, with a PCV value of 37.78%, which significantly explains disparities between strata. Social engagement is active participation in formal and non-formal community organizations. It reflects the integration of individuals in the diverse dynamics of society that imply interactions with others [54]. Research highlights its strong association with lower mortality [55], improved mental health [54], reduced chronic morbidity [56], and maintaining functionality. In the Netherlands, older adults with greater social networks and participation in activities had a lower disability risk compared to less engaged peers [50]. This is also true in Brazil [43], the United States [57], and Japan [58,59]; the last one shows a positive gradient between increased social activities performed and a lower probability of having functional limitations [60,61]. Additionally, in Taiwan and Sweden, this protective relationship demonstrated a long-term effect [52], and a positive annual improvement in physical function [62], respectively. Consequently, social engagement is the main variable that could be promptly intervened to maintain functionality.

The mechanism through which functionality is preserved via social participation is the active realization of BADLs and IADLs. Their consistent use strengthens musculoskeletal and neural systems, helping to prevent age-related physical decline. Additionally, social networks offer greater access to support and foster a sense of belonging, further contributing to maintaining functionality [57,61].

Social engagement has been extensively studied and promoted in high-income countries, where various initiatives encourage older adults to participate in cultural, sports, religious, and recreational activities. In contrast, Latin America has limited programs promoting older adults’ social participation and lacks sufficient research on the benefits of such activities [54]. Furthermore, in Latin America, the primary social network for older adults is the family, which provides crucial financial, healthcare, and social support. In Mexico, intergenerational households play an important role in socialization for older people, principally provided by the spouse and children [63]. This situation highlights the absence of programs for long-term care in Mexico, which places the burden of care on families, leading to increased costs and caregiver strain [64]. Additionally, this issue requires urgent attention, as demographic changes—such as smaller household sizes, the decline of nuclear families, and the rise in single-person households—further complicate socialization for older adults [63].

Although the Mexican government has promoted social initiatives to enhance the social participation of older adults, such as the PILARES para el Bienestar policy [65] and the National Institute for Older Adults (Instituto Nacional de las Personas Adultas Mayores, INAPAM) [66], there is a need to improve the availability and accessibility of these activities at the national level. This can be achieved by raising awareness about them, ensuring their geographical proximity, and providing adequate transportation options. This implies adapting urban infrastructure, including wider sidewalks, ramps, smooth pavements, and the installation of elevators. Additionally, these activities should be financially accessible and designed with an intergenerational approach [67].

Concerning the second finding, by analyzing the intersectionality statistics—VPC and PCV—we found that sex didn’t have a significant relationship with functional limitations, as previously reported. There’s evidence that demonstrated a substantial association between disability and being a woman in European countries [41,42], Brazil [43], India [51], the United States [68], Canada [69], and Mexico [8,13]. Moreover, more complex indicators, like disability-free life expectancy (DFLE), show that globally, women live longer but with a higher disability. This has been known as the male-female health-survival paradox that is explained by women’s protective biological factors—such as female hormones—, lower exposure to risk factors —like violent behaviors—, healthier lifestyles —including reduced smoking and alcohol consumption—, more use of healthcare systems compared to men, and a greater ability to communicate health problems to physicians [40,68,70–75]. However, in our results, the PVC value for sex was 1.70% which reflects that this variable has little impact on the disparities across strata.

Using intersectionality, our results suggest that having functional limitations is not simply linked to being a woman but to what it means to be a woman. In contemporary societies, older women are often stereotyped as less competent, more dependent, submissive, and less intellectually capable than men, which leads to a more complex discrimination process because of the intersection of sexism with other types of discrimination [46]. For example, older men have more socialization activities than older women [45], face less discrimination at work [61], and have a better socioeconomic status [51]. This perception leads to the undervaluation of older women’s significant contributions, including their roles in economic productivity, the transmission of knowledge and experiences, intergenerational caregiving, and providing essential family support. However, it must be acknowledged that multiple social determinants, such as employment status, pension, and health status, were not considered in this analysis, yet they may impact the differences in functionality between sexes.

Another significant observation is that health system access showed no relationship with functional limitations. Health systems are often regarded as intermediate SDH with the potential to reduce inequities [33]. For instance, in Brazil, affiliation with the health system lowers the likelihood of disability in older adults [40]. Our analysis contrasts with this finding. There may be organizational and contextual issues within the Mexican health system that hinder the goal of restoring functionality among older adults. For instance, despite the high percentage of health system affiliation among older adults in Mexico, healthcare services are underutilized due to structural challenges such as overcrowding, long waiting times, shortages of resources in the public sector [76], and ageism among healthcare professionals [24,46,77]. This finding is deeply concerning because the health system must provide adequate medical attention to improve the health status of older people. However, this result requires a more in-depth analysis due to the segmentation of the Mexican health system based on social security status [78]. This division has generated different health outcomes among the population, driven by variations in the quality of services and the availability of resources [79,80]. Moreover, recent transformations within the system, including the centralization of health services for the population without social security and modifications in the structure of health institutions providing health services, have led to significant changes in health system affiliation rates and access to healthcare services [78]. Therefore, a more precise classification of this variable is recommended.

Although our findings are consistent with classical epidemiological approaches, they also reveal that not all inequities can be explained solely by additive effects. This confirms that analyzing inequities without an intersectional perspective may oversimplify their complexity by focusing only on individual characteristics, overlooking the interconnected nature of social determinants [15,16].

Finally, we identified vulnerable groups with higher risk of health inequity that may be targeted for specific interventions. Being in the stratum defined as “older woman, with no education, low economic status, limited social engagement, and no access to the health system” confers more than nine times the probability of having functional limitations compared to “younger man, with education, good economic status, social engagement, and health system access” (S2 File). Intersectionality enables us to recognize the most and least advantaged groups and other at-risk populations that could benefit from tailored interventions. This is relevant as policy interventions that lack an intersectional perspective tend to intervene solely in the most disadvantaged groups and overlook others who might also be vulnerable. This can lead to “threshold effects”, where individuals just above the intervention threshold are excluded [15,16,21]. For this reason, intersectionality has been compared to precision medicine [18], as it allows for more precise identification of groups that could benefit from specific interventions [22]. As shown in Supporting Information 2, the groups requiring intervention are predominantly composed of women with no social engagement and fair or poor socioeconomic status.

This study presents several strengths. It’s the first to explore inequities among Mexican older adults using an intersectional framework. It also benefits from a representative sample and employs the gold standard for quantitative intersectional analysis. Furthermore, the study incorporates less frequently analyzed intersecting variables like health system and social engagement. However, certain limitations must be acknowledged. First, the outcome measure does not directly reflect disability. Further, the selection of variables was limited by the sample size [36], which may have prevented the inclusion of relevant factors associated with the outcome. Analyzing other variables could offer a more understanding of the complex interactions among determinants that drive health inequities. However, as the number of variables used to define strata increases, so does the number of resulting strata, which can lead to insufficient observations within some of them, thereby restricting the feasibility of applying MAIHDA. To overcome these limitations, fulfill the methodological requirements necessary for the robust application of MAIHDA, and incorporate a broader range of variables—such as ethnicity, migration status, lifestyle behaviors, and chronic health conditions— larger and more comprehensive datasets are needed. Also, longitudinal studies could also provide a clearer understanding of the progression from preserved functionality to functionality loss, offering insights into the development and persistence of inequities [14]. Although MHAS is a longitudinal study, the sample size was insufficient to perform MAIHDA longitudinally. Expanding the dataset would allow a deeper examination of the temporal and structural mechanisms driving disparities. A final limitation is that this information may not be generalizable to populations outside Mexico, as the interactions of these predictors could vary in other countries. Therefore, the results should be interpreted carefully when applied to other populations.

Conclusion

Adopting an intersectional approach allows for a more comprehensive understanding of health inequities by recognizing that individual characteristics are interconnected and shaped by context. This perspective moves beyond individual risk profiles and helps to explain the broader range of health disparities across populations. Our findings reveal that different groups experience varying functional limitations, enabling more targeted and effective interventions. Moreover, we demonstrate that enhancing social engagement represents a short-term strategy to improve functionality and promptly reduce inequities. Importantly, our study highlights that higher probabilities of disability are not inherently linked to being a woman but are shaped by societal perceptions and stereotypes about what it means to be a woman. Consequently, scientific findings on social determinants should be interpreted within the context of power dynamics and privilege. Regarding the health system, we recommend improving access for this age group and encouraging further research into its role in restoring and maintaining functionality. This will enable the provision of adequate medical attention and contribute to improving health outcomes of older populations.

Supporting information

Acknowledgments

The publication of this paper was supported by Instituto Nacional de Geriatría, Mexico.