A substantial body of research shows that higher educational levels correlate with better self-perceived health, lower comorbidity risks, and reduced mortality rates [e.g., (1, 2–6, 15)]. This relationship is typically explained by two factors: how education affects lifelong habits (diet, alcohol consumption, smoking, physical activity) and how it provides better structural and material conditions (6).

It is widely documented that older people who have attained higher levels of education generally enjoy a more favorable economic situation. This, in turn, enables them to reside in healthier living environments and to access superior quality medical care and a more balanced diet (5). Furthermore, personal behaviors can be influenced by education, insofar as greater schooling can engender a more profound understanding of the factors that influence their health, which in turn leads them to adopt preventive practices. Several studies have shown that individuals with higher educational attainment generally exhibit healthier behaviors and lifestyles, along with better access to health services, compared to those with lower educational qualifications (16, 17). In accordance with the cumulative advantage hypothesis, which posits that individuals accrue benefits over the course of their lives that positively impact their health, Alperin and Kyzyma (18) posit that individuals with higher levels of education demonstrate superior management of their health and engage in physical exercise with greater frequency in comparison to those with lower levels of education. Furthermore, individuals who were raised in families in which at least one parent had attained a high level of education demonstrated better health throughout their lives (18). This finding suggests that the health of the older adults is also significantly influenced by their earlier life stages, with education emerging as a prominent factor. This observation underscores the notion that inequalities experienced throughout the life course tend to accumulate in later years.

Recent studies conducted within the Italian context are consistent with international research on the links between education and health outcomes, including self-perceived health. Sarti and Spinola (8) highlight that socioeconomic factors, particularly education and occupational status, significantly influence health in both Italy and Argentina. Although their study does not focus exclusively on older adults, they find that education has a stronger impact on mental health among individuals from lower socioeconomic backgrounds in Italy compared to those from the working and middle classes. Similarly, Adjei et al. (40), using data from Germany, Spain, Italy, the United States, and the United Kingdom, confirm that educational attainment is consistently associated with better self-perceived health for both men and women across all countries considered, including Italy.

In a multilevel analysis of older Italians, Pirani and Salvini (3) identify several key determinants of self-perceived health: (1) education and income emerge as the primary drivers, with higher socioeconomic status associated with fewer chronic conditions, healthier behaviors, and more favorable health assessments; (2) women report worse health than men, although this gender gap diminishes after controlling for other variables, indicating that gender differences in health perception are partly compositional; (3) physical activity is positively associated with self-rated health, likely reflecting both its health benefits and the tendency of healthier individuals to remain active; (4) strong social ties contribute positively to perceived health; and (5) although regional differences are modest, more positive perceptions of local healthcare systems are associated with better self-assessed health.

Further supporting this perspective, Maniscalco et al. (7) examine the relationship between self-perceived health and quality of life among the older population in Spain, Greece, and Italy. Their findings indicate that while income is more closely tied to overall quality of life, education is a more critical determinant of self-perceived health, underscoring the distinct nature of these two constructs.

Taken together, these studies reinforce the idea that education plays a key role in shaping subjective health perceptions. However, both self-perceived and objective health outcomes are influenced by a broader array of factors. These include material and structural conditions, social capital, access to healthcare, lifestyle behaviors, social relationships, health literacy, economic inequality, and psychological well-being (8, 19). Accordingly, the determinants of health perceptions can be categorized into sociodemographic, behavioral, and psychological domains.

The literature on gender differences in health consistently reveals a well-documented paradox: men tend to have shorter life expectancies but report better health, whereas women live longer yet often experience poorer health conditions (5, 9, 10, 12, 20). In this context, gender is not merely a biological category but a social construct that shapes identities, roles, lifestyle habits, disease incidence, healthcare practices, and socioeconomic positioning (11).

Several authors (6, 12) have similarly demonstrated the effect of sex on health outcomes, highlighting that gender differences arise from a combination of biological factors and socially patterned behaviors, with women facing higher risks of comorbidities and disabilities partly due to higher obesity rates and lower levels of physical activity. Crimmins et al. (10) also challenge purely biological explanations for gender disparities in health, underscoring the role of contextual factors such as healthcare infrastructure, disease patterns, and public health systems. While men are more susceptible to severe diseases like heart conditions and diabetes, women more frequently report chronic but non-lethal conditions, including arthritis and depression. These conditions contribute to women’s lower self-perceived health despite their longer life expectancy (21). More recently, Osobsama et al. (20) show that women are more likely than men to use primary healthcare services, whereas men exhibit higher rates of hospitalization.

Focusing specifically on self-perceived health, the main outcome of interest in this study, Ross et al. (13) explore the joint influence of gender and education. Their findings indicate that although both factors independently shape health self-perception, the effect of education is stronger for women. Women with lower levels of education report significantly poorer health perceptions than their male counterparts. However, each additional year of education improves perceived health, and this educational gradient is steeper for women, such that the gender gap narrows considerably among those who have completed high school (usually 18–19 years old).

Crimmins et al. (22), in a comparative study including Italy, finds that women report poorer self-perceived health than men when controlling only for age. However, once the presence of illnesses is considered, the gender gap in health perception becomes statistically insignificant. This suggests that women’s more negative health reports may reflect not only their greater exposure to illness but also the types of conditions they experience, often those leading to disability rather than mortality.

Adjei et al. (40) also document persistent gender differences in health perception among older adults across five countries: Germany, the UK, Spain, Italy, and the United States. Their study shows that Italy displays the widest gender gap in perceived health among the older adults, with differences reaching up to 47%. The authors attribute these disparities to unequal participation in the paid workforce, differences in educational attainment, and varying distributions of leisure and domestic responsibilities between men and women.

These findings are consistent with Pirani and Salvini (3), who underscore the significant role of gender in shaping self-perceived health among Italy’s older population, with women consistently evaluating their health more negatively than men. At the broader European level, Dimitrova (23) also finds that women report worse health than men across nearly all countries included in her study, reinforcing the robustness of this gendered pattern.

Geographical differences represent a critical dimension in understanding health inequalities among the older population. A growing body of research highlights how economic, environmental, and cultural disparities across regions significantly influence population health (41, 42, 43). Curtis (44) and Fan et al. (41) underscore the role of unequal resource distribution in explaining regional health disparities, particularly with respect to access to healthcare, education, and income.

In the Asian context, several studies document marked spatial inequalities in older adults health. Fan et al. (41) identify substantial interprovincial disparities in China, with 2.4% of the variation in older adults health attributable to differences in economic, educational, and medical resources. These disparities are particularly pronounced in less developed rural areas, where individual income gaps further exacerbated inequality. Similarly, Cui et al. (42) observe both inter- and intra-regional health disparities among older Chinese adults, pointing to localized health and social conditions as key contributing factors. Crimmins et al. (10) argue that variations in health and mortality should not be ascribed solely to biological or hormonal differences; instead, they stress the importance of healthcare infrastructure, contextual risk exposures, and the influence of historical and systemic factors.

In Europe, research likewise confirms that geographic and social contexts shape health disparities among older adults. Oksuzyan et al. (21) identify multiple intersecting factors, such as region of residence, health behaviors, literacy, class, and education, that shape gendered health outcomes across countries. In Spain, Fernandez-Martinez et al. (43) report that older adults in the southern regions tend to rate their health more negatively than those in wealthier northern areas, linking these differences to socioeconomic and psychological factors. Poorer perceived health was associated with advanced age, chronic conditions, depressive symptoms, and reduced independence in daily activities. In Greece, Dermartis et al. (38) identify a clear gradient in perceived quality of life by settlement type: older residents in urban areas report the most positive health perceptions, followed by those in semi-urban and rural areas.

In the Italian context, although extensive research has examined regional disparities in healthcare organization and governance (24–27, 39), few studies have directly addressed regional variation in self-perceived health among the older population. Nonetheless, the influence of regional socioeconomic inequalities on health is well documented. Franzini and Giannoni (28) identify regional income disparities as a major driver of health outcomes, advancing two explanatory frameworks: material deprivation, whereby limited economic resources directly impair health, and social-psychological well-being, which emphasizes the role of social capital and the negative impact of social isolation. These dynamics are particularly salient in economically disadvantaged southern regions, where poverty, unemployment, and income inequality converge with reduced access to healthcare to worsen health outcomes. Mortality statistics further reflect these territorial disparities; Lenzi et al. (29) show that treatable mortality varies by gender and region, with men exhibiting higher rates overall, and southern Italy, except for Puglia, recording the highest levels of avoidable mortality.

Building on the extensive literature documenting a positive association between education and health in later life, the aim of this study is to clarify both the strength of this relationship and the mechanisms through which it operates among older adults in Italy. Our first objective is to estimate the association between educational attainment and self-perceived health for older individuals in Italy. Accordingly, our first research question is:

Beyond this primary association, our main contribution lies in attempting to open the “black box” of the relationship between education and self-perceived health by considering the role of intervening mechanisms related to health-related behaviors and material conditions. Education may indirectly affect health by reducing economic strain and by promoting the adoption of healthier lifestyles, such as engaging in physical activity and avoiding smoking and excessive drinking (30, 37). This leads to our second research question:

In the previous subsections, the role played by gender and territorial inequalities was highlighted, and it is therefore relevant to explore whether these mediating mechanisms operate differently across population subgroups. Geographical area of residence and gender may shape both access to resources and the adoption of a healthy lifestyle. In addition, differences across age groups cannot be excluded, as our sample consists of individuals aged 55 and older, and the size of this group may conceal significant heterogeneous effects. This motivates our third research question:

Figure 2 illustrates trends in self-reported health status from 2013 to 2019, stratified by different characteristics. The general trend (Panel A) shows a gradual improvement in perceived health. While there are slight fluctuations, the overall trajectory suggests a consistent rise in the proportion of individuals reporting good health. More precisely, the share of healthy older people increases from 40% in 2013 to almost 48% in 2021. Panel B highlights stark differences in health self-perception by on educational level. Individuals with tertiary education consistently report the highest levels of good health, followed by those with upper secondary and lower secondary education. Conversely, those with no educational qualifications exhibit the lowest levels of self-reported health. The gap between the lowest and highest educational groups remains significant but stable over time, underscoring the persistent impact of educational inequality on health outcomes.

Regional disparities are evident in Panel C. The North-West consistently reports the highest levels of good health, followed closely by the North-East and the Centre. In contrast, the South and Islands report lower health perceptions, with the Islands (Sicily and Sardinia) consistently showing the lowest levels. These regional differences may be linked to variations in healthcare access, economic opportunities, and socioeconomic determinants of health. While all regions show some improvement over time, the gap between the North and South persists, reflecting ongoing regional disparities.

Panel D, finally, reveals gender-based differences in perceived health. Men consistently report higher levels of good health compared to women throughout the observed period. Although both genders show gradual improvements over time, the gender gap remains stable, suggesting that factors such as differential exposure to health risks, societal roles, and access to healthcare shape these outcomes.

After presenting the descriptive data from our analysis, we proceed with a logistic regression analysis to examine how education level influences the self-reported health of the older population. Table 1 reports three models predicting self-perceived health. The first model controls sociodemographic characteristics, the survey year and employment status. Model 2 adds economic status, while Model 3 further includes three lifestyle variables (alcohol, smoking and physical activity). The primary aim is to compare how the coefficients vary relating to the educational qualification across the three models, to assess whether these mediators exert an intervening effect. Methodologically, the estimates presented in Table 1 are derived from logistic regression models and to facilitate the interpretation of the results, the parameters are presented in the form of average marginal effects. In this way, they can be interpreted as differences in terms of percentage points.

Educational attainment is positively associated with the probability of reporting good health across all models. In Model (1), individuals with primary education are 7.5 percentage points more likely to report good health compared to those with no formal education. This likelihood increases progressively with higher educational levels, reaching 21.1 percentage points for those with tertiary education. While the magnitude of these effects decreases slightly in Models (2) and (3) due to the inclusion of the mediators, the positive and significant relationship remains robust. This underscores the critical role of education in shaping health perceptions, as highlighted by the previous literature.

In Model (2), we introduce the first mediator: economic strain. It is negatively associated with self-perceived good health, as indicated by a significant coefficient of −0.095. This suggests that individuals experiencing economic difficulties are less likely to report being in good health, highlighting the adverse impact of financial stress on well-being.

Regarding lifestyle factors, we rely on three indicators: alcohol consumption, smoking habits and physical activity. Model (3) presents the results for these variables. The relationship between alcohol consumption and self-reported health perceptions is nonlinear. The coefficient for alcohol consumption is positive (0.038), while the squared term is negative (−0.005), indicating that moderate alcohol consumption may be associated with better health perceptions, but excessive consumption has a detrimental effect. Smoking is significantly negatively associated with the probability of reporting good health; the coefficient of −0.030 reflects the well-established health risks associated with smoking, reinforcing its negative impact on self-perceived well-being. Physical activity emerges as the strongest behavioral predictor of self-perceived good health (0.093), highlighting the significant health benefits of regular physical activity, which likely improves physical fitness, reduces stress, and enhances overall well-being.

The next step of the analysis is to estimate the mediation effects of economic strain and lifestyle factors. Tables 2, 3 quantify this mediation. In Table 2 we estimate four different models, each including a different mediator. For the overall population, the most influential mediators are: economic strain (15.63%) and recreational physical activity (11.49%), whereas smoking and alcohol consumption play a negligible role. The prominent role of economic strain aligns with the argument that material resources are pivotal in shaping health outcomes. Higher education often translates into better socioeconomic positions, providing individuals with greater financial stability and access to (private) healthcare and other resources. Reduced economic strain likely alleviates chronic stress and enables healthier lifestyle choices, indirectly improving health perceptions.

Among lifestyle factors, physical activity is the most significant mediator, explaining 12% of the total effect. Smoking (2.22%) and alcohol consumption (0.66%) play more minor roles. These findings suggest that education fosters healthier behaviors through increased awareness of health risks and benefits. Educated individuals are more likely to adopt regular physical activity, contributing to better health outcomes. Surprisingly, the mediating roles of smoking and alcohol consumption are very low. This result can be explained by the so-called survivor effect: among older people, those who engaged in heavy smoking or excessive alcohol consumption may already have experienced significant health consequences (i.e., died), potentially leading to selection bias. This means that those remaining in the sample are either moderate users or have successfully changed these behaviors, reducing their explanatory power as mediators. Older individuals may also reduce or cease risky behaviors such as smoking and excessive drinking, either due to age-related health concerns or medical advice.

Table 2 also presents the results of the mediation analysis broken down by age group. The overall picture is confirmed, but some variations emerge across age groups. More specifically, the role of economic strain is greater for the youngest age group (i.e., those aged between 55 and 64). This could be due to the fact that the oldest age group is better protected by pensions, which mitigate economic difficulties compared to younger cohorts. The mediation effect of alcohol consumption is minimal for all age groups, suggesting that its role as a mediator in the relationship between education and health is limited in this context. For smoking, the middle group (65–74) belongs to a cohort with historically higher smoking rates and is likely to have quit more recently. In contrast, the oldest group may have stopped smoking much earlier, or this may reflect some selection, as those who smoked heavily in the past may not have survived to the oldest age group. Regarding physical activity, the mediation effect increases with age. This is plausible for two reasons: firstly, cumulative effects over a lifetime likely play a greater role at older ages; secondly, physical activity helps to maintain or improve mobility, which is particularly important at ages when physical functioning is more vulnerable.

Table 3 considers a model including the two main mediators simultaneously and shows that approximately 27% of the total effect of education is explained by these two variables. The second column of Table 3 reports the weights of the individual mediators, which are consistent with what we saw in Table 2. Overall, our results underscore the crucial role of education in shaping self-reported health through access to resources, and healthier behaviors among more educated individuals. In fact, it is noteworthy to emphasize that a substantial part of the total effect (about 75%) is not explained by the mediators. The remaining 75% includes the direct effect of education as well as any other indirect mechanisms not accounted for in the current mediation models. The direct effect might reflect factors related to cognitive abilities and to stronger social networks. Cognitive abilities may encompass the ability to interpret symptoms, more effectively manage health, and access health information and medical resources (i.e., health literacy). Stronger social networks may provide emotional and practical support, further enhancing health outcomes.

The final step examines how the roles of the mediators can change according to geographical area of residence and gender. Table 4 shows the results broken down by North and Center-South, showing substantial differences in mediation percentages between regions. On average, the role of mediators is stronger in the North. Economic resources are particularly influential in the North, whereas physical activity plays a more prominent role in the South.

The greater mediation effect observed in the North may reflect better structural conditions that enhance the impact of economic strain reduction. More precisely, educational attainment may more effectively reduce economic strain, while in the South, structural barriers might limit its potential to improve financial stability and health outcomes. Physical activity appears to play a larger mediating role in the South, potentially due to cultural or lifestyle differences. In regions where economic opportunities are limited, educated individuals may rely more on personal health behaviors, such as physical activity, to maintain their health. In other words, the lack of structural support in the South could mean that health improvements rely more heavily on individual actions, rather than systemic factors such as reductions in economic strain.

Table 5 shows the mediation effects by gender. No significant differences were observed, although the mediation effect is slightly larger for males (29.5% vs. 25.3%). The absence of gender differences, at least in the mediation analysis, suggests that, for this older population, the pathways linking education, economic strain, and physical activity to health operate similarly for men and women.