An analytical study based on observational population data was conducted within the CARhES (CArdiovascular Risk factors for hEalth Services research) cohort. This is an open, dynamic, population-based cohort of subjects aged 16 and older diagnosed with hypertension, diabetes mellitus, and/or dyslipidaemia in the Spanish region of Aragon (14).

For the purposes of this study, we included patients from the CARhES cohort who experienced a first recorded episode of AMI between 2017 and 2022 identified using the International Classification of Diseases (ICD-10) code I21. We excluded subjects with a prior diagnosis of AMI at the onset of cohort follow-up, as well as those who died during the index event. In order to analyse health services utilization and CVRF monitoring, we only included subjects with complete clinical and administrative data available for the 365 days following the AMI event. A detailed flowchart illustrating selection criteria of the study population is provided in Figure 1.

The data used in this study were obtained from the CARhES cohort (14), a population-based, dynamic open cohort designed to analyse the impact of healthcare service use and pharmacological treatment on health outcomes in patients with CVRFs. The CARhES cohort is constructed using RWD, that is, routinely collected health information from clinical practice, including all individuals aged ≥16 years registered in the Aragón public health system (Spain) with a diagnosis of hypertension, diabetes mellitus, and/or dyslipidaemia since 2017. The cohort is subject to annual updates through data extractions.

This study constitutes a secondary analysis of anonymized RWD extracted from the BIGAN platform, which integrates multiple information systems of the Aragón Health Service (SALUD) for research and policy purposes. The integrated sources include the following: the user database (BDU) and adjusted morbidity groups (GMA) for demographic and clinical data; hospital discharge records (CMBD); specialist care data (CEX); primary care records (OMI-AP); emergency care data (PCH); and electronic pharmacy dispensing records (Receta Electrónica) for medication use. Collectively, these sources provide comprehensive information on patients' clinical profiles, socioeconomic conditions, and healthcare utilization. No additional instruments or patient-reported outcome measures were used; all data were obtained from routinely recorded clinical and administrative sources within the public health system.

In this study, the variables were grouped into two main categories: general patient characteristics at the time of the event (such as socio-demographic, anthropometric and clinical characteristics) and patient management variables, which included health service utilization, CVRF monitoring, and pharmacological treatment during the first year following AMI. The sex/gender variable was also included, as a key factor for analysing potential gender-based differences in care and outcomes.

Sociodemographic and baseline clinical characteristics of the subjects studied were described using counts and proportions for categorical variables and means with standard deviation (SD) for continuous variables. Bivariate analyses between men and women were performed using Pearson's Chi-squared test, and we compared means between groups using Student's T-test. Bivariate logistic regression analyses were conducted to investigate gender-based differences in patient management variables that had been previously defined. The threshold for statistical significance was set at P < 0.05. When statistically significant gender-based differences were identified, the Blinder-Oaxaca decomposition method was applied (18). A twofold decomposition was conducted using the Oaxaca R library and reference regression coefficients, which were calculated from a pooled regression model (19). Variables such as age, socioeconomic status, area of residence (urban or rural), and morbidity burden were examined to determine their contribution to the observed differences. This analytical approach enables the quantification of the extent to which observed gender differences can be attributed to measurable variables, and how much remains unexplained. Specifically, it decomposes the mean outcome differences between two groups into an explained fraction—linked to differences in observed characteristics—and an unexplained fraction, which may reflect the differential effects of those variables, unmeasured confounders, behavioural or structural inequalities, or potential discrimination (20). This method provides a nuanced understanding of the mechanisms underlying gender disparities in healthcare. All statistical analyses were performed using R 4.3.3. (R Core Team, 2022) a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/) (21) and JAMOVI (version 2.4) [Computer Software) Retrieved from https://www.jamovi.org (22).

This study is based on data from the CARhES cohort, whose protocol was approved by the Clinical Research Ethics Committee of Aragon (CEICA PI21/148). The research was conducted in accordance with local legislation and institutional requirements. As the study involved the retrospective analysis of anonymized, population-based data with no direct contact or interaction with participants, the requirement for written informed consent was waived by the Ethics Committee.

The study population included a total of 3,464 subjects who had experienced a first AMI during 2017–2022, with a minimum follow-up of one year. Of these, 980 (28.30%) were women, and 2,482 (71.70%) were men.

As presented in Table 1, women were significantly older at the time of the event (75.26 years) compared to men (67.23 years). The majority of the population were pensioners <18,000€, with significant differences (p < 0.001) observed between men (41.79%) and women (67.45%). 2,471 subjects (71.33%) lived in urban areas. A greater proportion of women (10.92%) were institutionalized in nursing homes compared to men (4.31%).

While dyslipidemia was the most prevalent CVRF overall, with a slightly higher prevalence among men, hypertension was significantly more common in women. Among the comorbidities analyzed, all conditions (except cirrhosis and chronic obstructive pulmonary disease) were more prevalent in women (p < 0.001). Women exhibited a significantly higher number of pathologies, more affected systems, and a greater overall morbidity burden compared to men (p < 0.001). Conversely, men had higher rates of overweight and obesity and had a higher prevalence of smoking (p < 0.001) (Table 1).

Nearly all participants (99.86%) had at least one primary care visit within the first year after AMI (Table 2). Women were more likely to visit nurse practitioners (93.67% vs. 91.77%) and specialists such as endocrinologists and ophthalmologists (p < 0.001),while men had significantly more visits to cardiologists (p < 0.001). Hospital admissions by all the causes were significantly higher in women (35.10% vs. 30.76%), while emergency room use was similar between genders (Table 2).

Men had higher rates of monitoring for several CVRFs (Table 2), including glycated hemoglobin (HbA1c), low-density lipoprotein cholesterol (LDL-c), and high-density lipoprotein cholesterol (HDL-c). Total cholesterol, triglycerides and waist circumference were also monitored more frequently among men, with smaller differences. Engagement with physical activity was reported more frequently by men (p < 0.001), while women were more likely to receive influenza vaccinations (p = 0.004). No significant differences were found in nutritional habits or treatment adherence (Table 2).

Men were more likely to receive the main guideline-recommended drugs, including beta-blockers, lipid modifying agents, and antiplatelet agents (p < 0.001) (Table 2). However, women were more frequently prescribed some concomitant medications, such as diuretics, CCBs, vitamin K antagonists, and direct Factor Xa inhibitors (p < 0.001), and showed a higher overall pharmacological burden (p < 0.001).

Logistic regression analyses, shown in Table 3, revealed significant gender differences. After age adjustment, logistic regression indicated that most gender differences in health services were no longer statistically significant, with the exception of endocrinologist visits, which remained higher for women (adjusted Odds Ratio: 1.41, 95% Confidence Interval: 1.10–1.80). Significant gender differences remained in CVRFs monitoring: women were less likely to achieve optimal blood pressure levels, HbA1c levels, and waist circumference measurements. They also had lower odds of reporting regular physical activity (adjusted OR: 0.67, 95% CI: 0.52–0.87). Differences in lipid profiles and influenza vaccination were narrowed after adjustment for age. After adjusting for age, women continued to have lower odds of receiving beta-blockers, lipid modifying agents, and antiplatelet agents. In contrast, they were significantly more likely to be prescribed diuretics.

The Oaxaca-Blinder decomposition analyses assessed the contribution of observed variables to gender inequalities. So, variables with negative values reduced the observed inequalities, while those with positive values increased them (Figuress 2, 3). Complete data from the OAXACA decomposition analyses, such as the explained fraction of the models and the contribution of each variable, can be found in the Supplementary Tables S1 and S2.

Regarding healthcare utilization, the OAXACA analysis revealed that only 19.74% of the observed gender inequalities in endocrinology visits were explained by the variables included in the model, with age and morbidity burden being the main contributing factors. The lower age of men mitigated these inequalities, while the lower morbidity burden of men with respect to women increased the existing differences.

Concerning physical activity, which was more reported in men, the lower morbidity burden of men was the main explanatory factor of these gender differences, along with their higher socioeconomic status. However, being a low-income pensioner, seemed to mitigate these differences, while age did not seem to influence gender differences.

In terms of CVRF monitoring, the explained fraction varied from 63.15% for HbA1c to 38.78% for physical activity. The explanatory variables in the model exhibited different effects on each of the CVRF studied. With regard to blood pressure, which was less monitored in women, the factors that contributed most to increasing gender inequalities were urban residence and high socioeconomic status, considering actives <18,000€ as reference. In contrast, the older age of women, being a low-income pensioner, and their higher morbidity burden in this study reduced these differences. A similar explanatory pattern was observed for capillary glycemia measurement, except that age did not appear to be associated with variations in gender differences observed. For HbA1c measurement, the factors more associated with increased gender inequalities were age, as the main contributing factor, and high socioeconomic status. In contrast, being a low-income pensioner and morbidity burden appeared to reduce these inequalities.

Regarding drug prescription, the explained fraction of observed gender differences varied across the pharmacological groups analyzed. The explanatory fraction of the model was high in diuretics and lipid modifying agents (more than 50%) while the explanatory factor of beta-blockers and antiplatelet agents was low, suggesting the influence of additional factors not considered in the model. As shown in Figuress 2, 3, age and morbidity burden were the main factors contributing to gender inequalities in the prescribing of guideline-recommended drugs. Regarding age, the older age of women was related with increasing gender differences in the prescription of antiplatelets, beta-blockers and lipid modifying agents. On the contrary, the younger age of men seemed to increase prescription differences in diuretics. Regarding socioeconomic status, none of the categories analysed showed a statistically significant effect on drug prescription, but it was possible to observe some associations. In the case of antiplatelets and beta-blockers, the high frequency of women that were pensioners with low income reduced the inequalities observed. In the case of diuretics, with a higher prescription in women, the lower frequency of men with low income increased the differences. Finally, the morbidity burden increased the observed differences in all the drugs considered. The lower morbidity burden of men increased the differences observed in men and women in the prescription of diuretics. On the other hand, the higher morbidity burden of women increased the differences in the prescription of antiplatelets, beta-blockers and lipid-modifying agents.

In our population following a first AMI, men and women had different socioeconomic and clinical characteristics. Women were older, more frequently institutionalized, and had higher morbidity burden and lower socioeconomic status. These factors are essential in order to comprehend the observed inequalities in healthcare utilization, CVRF monitoring, and treatment.

Nearly all patients had at least one primary care visit within the first year after AMI, as expected, since primary care is the principal setting for health care follow-up (23). However, women visited primary care nurses more frequently than men, as described in the literature (24–26) and may be due to differences in health-seeking behaviors or greater awareness of prevention (24). In this sense, a study by Vallejo-Torres et al. found that nurse utilization was associated with older age, female gender, the presence of more chronic conditions, and lower socioeconomic status (26) which is in line with our results.

In terms of specialist care, women were more likely to visit endocrinologists but had lower overall rates of visits to other specialists, particularly cardiologists, as the evidence shows (24,27). Our prior scoping review (28) suggested that this lower referral rate to cardiology may be related to an androgenic bias in cardiovascular care and under-recognition of the disease in women by both patients and healthcare providers (27,28), as well as, lower awareness of CVD, less social support, or lower socioeconomic status (29).

After adjusting for age, only endocrinologists' visits remained significant. Oaxaca decomposition attributed this to women's higher morbidity burden, which increased the observed differences.

This could be partly explained by the higher prevalence of endocrine disorders in women, particularly postmenopausal and elderly, such as thyroid disease, osteoporosis, or hormonal imbalances (30–32), which. often coexist with other comorbidities that increase cardiovascular risk and complicate disease management (33–36). In contrast, men's younger age helped reduce the differences observed. However, the association between age and referral patterns in older women is complex, and may be influenced by unmeasured factors such as race or marital status (37).

Hospital admissions were more frequent among women, however, became not significant after adjusting for age. These findings are consistent with previous studies (38,39), suggesting that while unadjusted data may show higher admission rates among women, factors such as age and other social determinants play a significant role in the observed differences.

Gender inequalities in CVRF monitoring were pronounced. Men had a higher frequency of recorded risk factor monitoring, including blood pressure, glycemic control (HbA1c), and cholesterol levels (8, 40–42). These findings align with existing literature suggesting a potential gender bias in management, with women receiving less frequent CVRF monitoring and more conservative treatments, despite being older and having a higher morbidity burden (43). Urban residence increased gender inequalities in CVRF management, which could be explained by differences in access to health care. Rural areas often have shorter waiting lists and greater availability of primary care services (44–47). The relationship between socioeconomic status and gender inequalities in health outcomes is somewhat controversial. We observed a reduction in gender inequalities in blood pressure, capillary glycemia, and HbA1C control associated with lower socioeconomic status in women. This could be related to the fact that women with lower socioeconomic status in our population were older and had a higher burden of disease, leading them to visit primary care more frequently. As a result, their CVRFs were monitored more regularly, narrowing the gap between men and women. Some authors support this explanation (48,49). Conversely, other studies suggest that lower socioeconomic status may limit women's CVRFs monitoring, thereby increasing gender inequalities (50,51). Also, other studies have associated higher socioeconomic status with better CVRF monitoring and greater access to health care, highlighting the complexity of this relationship (29,52,53).

For HbA1c measurement, age has an important role in increasing existing inequalities, which has been related with the older age of women (54–57). In terms of lifestyle practices, women reported greater adherence to healthy lifestyle recommendations, such as proper diet and immunizations. This is consistent with existing literature (58–63). For physical activity, consistent with what has been reported in the literature (60,61,64,65), women in our study were less likely to report practicing physical activity than men. Oaxaca analysis identified that morbidity burden and high socioeconomic status increased gender inequalities in physical activity practice, whereas lower income reduced inequalities. As has been widely described, gender, age and socioeconomic status are known to be important determinants of physical activity (66). While higher income has been associated with greater physical activity (67), in low-income settings, men and women have shown to face similar barriers to physical activity, reducing gender inequalities (68). In addition, chronic health conditions negatively impact physical activity levels, with women experiencing greater reductions, further widening inequalities (69).

Women were less likely to be prescribed guideline-recommended medications and were more likely to receive adjunctive therapies, as diuretics. These results are consistent with the literature (8,28,41,42,70,71). The Oaxaca analysis showed that older age and higher morbidity burden were the main factors contributing to this gender inequalities. This is in line with some studies that have found strong associations between age, morbidity burden and greater risk of exacerbations, leading to under-prescription of the pharmacological groups studied, increasing gender inequalities (24,48,72–74). Similarly, the large international PRAISE registry (75) reported lower prescription rates of guideline-recommended therapies in women following acute coronary syndromes. However, it also found that female gender was not associated with an increased risk of adverse outcomes, including bleeding, highlighting the need for more intensive, evidence-based treatment strategies in women based on their clinical profile rather than gender alone. The analysis of socioeconomic status reveals complex interactions with gender disparities in the prescribing patterns of guideline-recommended medications, as explored in our previous research (70). While higher socioeconomic status is generally associated with reduced gender disparities (29,52,76), our study found that lower income subjects, particularly older women with a higher morbidity burden, experienced a narrowing of these disparities in the prescription of antiplatelets and beta-blockers, although this association was not statistically significant. This suggests that women in our population may receive more clinical attention, leading to more equitable prescribing patterns of guideline-recommended therapies (48,50,77).

Regarding diuretic prescription, our findings align with existing research associating lower socioeconomic status with higher diuretic prescription rates among women, which could be associated with the management of multiple comorbidities (78). Studies indicate that women with low socioeconomic status are more susceptible to polypharmacy, with diuretics frequently prescribed as part of a prescribing cascade (79). Additionally, diuretic use tends to increase with the overall number of medications prescribed, which corresponds to the greater morbidity burden observed in our study population (80).

One of the major strengths of this study is the use of the CARhES cohort, a population cohort of RWD from the Aragon Health Service. This increases the internal validity of the study by ensuring a representative study population at the regional level.

In addition, a key strength of our study is the use of the Oaxaca-Blinder decomposition method to examine gender inequalities. This analytical approach allows us to separate observed gender differences into two components: an explained fraction, attributable to measurable variables, and an unexplained fraction, which may reflect differential effects of these characteristics, structural or behavioural inequalities, or potential discrimination (20). When we apply this method from a gender perspective, our analysis provides a more nuanced and comprehensive understanding of the mechanisms underlying inequalities in the use of services and management of CVD.

Nonetheless, certain limitations need to be acknowledged. A limitation of using registered diagnoses is the potential for diagnostic bias in CVD, as it excludes undiagnosed subjects. As highlighted in the literature (43,72), this issue could be particularly relevant for women because of potential underdiagnosis related to the nonrecognition of their symptoms in previous clinical guidelines. Another potential limitation of the study is the exclusion of patients who died during the index event, which may introduce survival bias. While the mortality rates were comparable between sexes, women represented a slightly higher proportion of the deceased and were, on average, older. No significant sex differences were observed in terms of morbidity burden or number of chronic conditions. However, the exclusion of these cases may have led to an underrepresentation of more severe female cases. Furthermore, the established follow-up period of one year may have further contributed to survival bias, as it excludes patients with shorter survival, potentially affecting older and more vulnerable subjects, particularly women. This decision was made to ensure a consistent observation period for all participants, thus minimising variability in follow-up duration and enabling reliable analysis of post-AMI care. Another limitation is the lack of information regarding the actual reason for the patient's visit. As a result, we cannot confirm whether consultations were directly related to the AMI episode. Finally, data on key lifestyle parameters, such as adequate diet, physical activity, smoking, or adherence to treatment, are underreported in electronic health records, as we have shown in our results. Nevertheless, we have chosen to present this information to highlight both their potential impact and the significant gaps in information registration.

It is important to assess whether factors such as age or comorbidities justify the reduced care observed or, on the contrary, reflect bias in clinical decisions.

Raising awareness of gender differences is essential to ensure equitable care. For example, better recording of CVRFs and lifestyle habits in health records is key to improving personalized care. Clinical practice guidelines must incorporate a gender-sensitive approach to ensure equitable and personalized care, as current recommendations do not fully address the differentiated healthcare needs of women, contributing to treatment inequities. Beyond clinical practice, these findings point to wider systemic challenges, including potential barriers that women face in accessing cardiovascular care as explored in our previous scoping review (28) and by Giordano et al. (11). Several social and cultural factors—like unrecognized symptoms, caregiving duties, fear of being a burden, low income, limited education, and reduced autonomy—can delay or prevent women with AMI from getting timely care. The establishment of follow-up protocols or structured monitoring programs, especially in primary care, can reduce the gender gap and improve health outcomes.

Finally, gender analysis is complex and closely linked to other social factors, making it hard to explain all the causes of disparities. This highlights the need for more research using an intersectional approach.