Abstract
Background
Social determinants of health influence both the development of diabetes and its complications. However, evidence on how combined sex/race categories relate to cardiovascular risk among individuals with diabetes remains limited, particularly in middle-income countries.
Objectives
To investigate whether combined categories of sex and self-reported race/skin color are associated with incident cardiovascular events in Brazilian adults with diabetes.
Methods
This prospective cohort study included adults with diabetes and no prior cardiovascular disease at baseline (2008–2010) from ELSA-Brasil. Participants were followed through subsequent study waves. The primary outcome was incident major cardiovascular events. The main exposure was a cross-classified sex/race variable (White men, Black men, White women, and Black women). Cox proportional hazards models were fitted using a hierarchical strategy, from minimally adjusted models to models additionally including socioeconomic and clinical variables. Variables such as HbA1c, hypertension, obesity, diabetes duration, chronic kidney disease, and albuminuria were considered potential downstream factors and were therefore interpreted cautiously.
Results
During follow-up, incident cardiovascular events were observed. In minimally adjusted analyses, differences across combined sex/race groups were observed but were attenuated after adjustment for socioeconomic and clinical factors. Lower educational attainment, longer diabetes duration, higher HbA1c, hypertension, and kidney disease markers were more consistently associated with cardiovascular events than the sex/race categories themselves.
Conclusions
In this ELSA-Brasil cohort of adults with diabetes, combined sex/race categories were not statistically associated with incident cardiovascular events after accounting for measured socioeconomic and clinical variables. These findings should not be interpreted as evidence that intersectional or structural inequalities are unimportant; rather, they suggest that measurable socioeconomic and cardiometabolic pathways may partly explain the observed disparities. Limited event numbers, possible overadjustment, and the absence of direct measures of racism, discrimination, and healthcare barriers should be considered when interpreting the results.
Keywords: Diabetes mellitus, Cardiovascular diseases, Social determinants of health, Cohort studies, Health inequalities
Introduction
Diabetes mellitus (DM) is one of the greatest public health challenges of the twenty-first century, with recent estimates indicating that more than 537 million adults live with the condition worldwide [1]. In Brazil, the prevalence of DM in the adult population is approximately 10.2%, with a growing trend in recent decades in DM type 2 [2]. In addition to traditional microvascular complications (retinopathy, nephropathy, and neuropathy), DM confers a substantially increased risk for cardiovascular events, with these macrovascular complications being responsible for approximately 70% of deaths in people with diabetes [3].
The epidemiology of DM and its complications, however, is not evenly distributed in the population. National and international studies consistently demonstrate that social markers such as race/skin color, gender, and socioeconomic status exert a determining influence on both the incidence of DM and the occurrence of its adverse outcomes [4]. Research developed with data from Vigitel (2023) indicates that the self-reported prevalence of DM is higher among women (11.1%) than among men (9.1%), and that people with lower education levels have a significantly increased risk [2].
Intersectionality, initially formulated by Kimberlé Crenshaw in the context of feminist and anti-racist scholarship, has been increasingly incorporated into public health to understand how multiple systems of stratification and structural disadvantage jointly shape health outcomes [5]. Importantly, cross-classifying social categories such as sex and race is only one operational approach and does not, by itself, capture the full theoretical scope of intersectionality, which also includes institutional racism, gender discrimination, class relations, healthcare barriers, cumulative disadvantage, and other mechanisms through which social position becomes embodied as health risk [6].
In the specific context of diabetes mellitus (DM), international studies suggest that Black women present particular clinical trajectories, characterized by a higher prevalence of obesity, poorer glycemic control, a higher frequency of hypertension, and consequently, an increased cardiovascular risk when compared to both Black men and white women [7]. These disparities are attributed not only to biological factors, but mainly to structural social determinants, including institutional racism, gender discrimination, less access to quality health services, and unfavorable socioeconomic conditions [8].
In Brazil, despite the growing recognition of racial and gender inequalities in health, longitudinal epidemiological studies that explicitly adopt an intersectional perspective to investigate DM complications are still limited [9]. Most existing research analyzes race and gender as independent variables, thus missing the complexity of the interactions between these social markers. This gap is particularly relevant considering the uniqueness of the Brazilian context, marked by profound social inequalities and a complex and multifaceted racial classification system [10].
The Longitudinal Study of Adult Health (ELSA-Brasil) represents a unique opportunity to fill this gap. As the largest Brazilian prospective cohort focused on chronic diseases, ELSA-Brasil offers high-quality data collected in a standardized way at six research centers, with longitudinal follow-up of more than 15,000 participants [11]. The cohort includes detailed information on sociodemographic characteristics, clinical conditions, laboratory parameters, and the occurrence of health events, allowing for robust analyses of the social determinants of DM complications [12].
This study aimed to investigate whether combined categories of sex and self-reported race/skin color were associated with incident cardiovascular events among adults with diabetes in ELSA-Brasil and to examine the extent to which socioeconomic and clinical factors accounted for these associations.
Methods
Study design
This is a prospective longitudinal study nested within the ELSA-Brasil cohort, using data from the first three follow-up waves. ELSA-Brasil is a multicenter cohort study that includes 15,105 participants, attending this inclusion criteria: women and men aged 35 to 74 years, civil servants of six teaching and research institutions located in the Brazilian cities of Salvador, Belo Horizonte, Vitória, Rio de Janeiro, São Paulo, and Porto Alegre) [11].
Initial recruitment occurred between 2008 and 2010 (wave 1), with follow-ups conducted in 2012–2014 (wave 2) and 2017–2019 (wave 3). The study was approved by the research ethics committees of all participating institutions, and all participants signed an informed consent form [13].
Study population
For this study, all ELSA-Brasil participants who had a diagnosis of diabetes mellitus in wave 1 were considered, defined by: (1) self-reported prior medical diagnosis of diabetes; (2) current use of hypoglycemic medication or insulin; (3) fasting blood glucose ≥ 126 mg/dL; (4) glycated hemoglobin (HbA1c) ≥ 6.5%; or (5) oral glucose tolerance test with 2-h blood glucose ≥ 200 mg/dL [14]. Because the ELSA-Brasil diabetes definition was based primarily on clinical history, medication use, and laboratory criteria, the cohort does not allow complete etiological classification of all participants as type 1 or type 2 diabetes. When available, insulin use and age at diagnosis were examined to identify participants with a profile compatible with type 1 diabetes, and sensitivity analyses excluding these participants were planned. Participants with prior cardiovascular disease at baseline, those with incomplete information required for the main models, and participants self-identifying as Asian/yellow or Indigenous were excluded because of the small number of cases in these categories.
Study variables
Outcome
The primary outcome was the incidence of a major cardiovascular event, defined as the first occurrence of: (1) acute myocardial infarction (AMI); (2) ischemic or hemorrhagic stroke; (3) heart failure; or (4) death from cardiovascular causes. Events were identified through the ELSA-Brasil active surveillance system, including annual telephone contact, review of hospital records, death certificates, and linkage procedures when applicable, and were validated by an expert committee using standardized criteria [15]. Acute myocardial infarction was defined based on compatible symptoms, electrocardiographic changes, cardiac biomarkers, and/or medical record evidence consistent with standardized ELSA-Brasil adjudication. Heart failure was defined according to documented hospitalization or medical record evidence compatible with clinical heart failure, supported by imaging, therapeutic, or discharge diagnosis information when available. Because the components of the composite outcome have distinct pathophysiology and determinants, secondary analyses by event type were also considered.
Main exposure
The main exposure variable was constructed from the intersection between sex (male/female) and self-reported race/skin color. For analytical purposes, race/skin color categories were grouped into White and Black, with the Black category including participants self-identifying as Black or Brown/pardo, as commonly adopted in Brazilian epidemiological studies [16, 17]. This combination generated four exposure groups: White men, Black men, White women, and Black women.
Intersectionality is a theoretical and analytical framework originally proposed by legal scholar Kimberlé Crenshaw to describe how multiple social categories—such as race, gender, and socioeconomic position—interact simultaneously to shape experiences of privilege, oppression, and health outcomes [5]. Rather than considering these dimensions independently, the intersectional approach recognizes that social identities are mutually constitutive and produce unique patterns of vulnerability and inequality. In public health research, intersectionality has been increasingly applied to understand how structural determinants of health operate through overlapping systems of discrimination and social stratification.
In the present study, intersectionality was operationalized pragmatically through combined sex/race categories. We recognize that this approach captures only part of the intersectionality framework and should be interpreted as a cross-classified analysis of social categories rather than a complete assessment of intersectional processes. Direct measures of racism, gender discrimination, healthcare barriers, cumulative disadvantage, and structural inequities were not available in the dataset; therefore, the results should be interpreted as associations between measured combined categories and cardiovascular outcomes, not as definitive evidence for or against the causal role of intersectional oppression.
Covariates
The following baseline covariates were considered. Sociodemographic factors included age, educational attainment, economic class according to the Brazilian Association of Research Companies (ABEP), per capita family income, marital status, and possession of private health insurance. Behavioral factors included smoking, pack-years, alcohol consumption, excessive alcohol consumption, and leisure-time physical activity [18]. Clinical and anthropometric factors included duration of diabetes, hypertension, body mass index, waist circumference, chronic kidney disease, and medication use. Laboratory parameters included HbA1c, fasting glucose, lipid profile, serum creatinine, estimated glomerular filtration rate, urine albumin-to-creatinine ratio, and high-sensitivity C-reactive protein.
Clinical and anthropometric characteristics: Duration of diabetes (in years, categorized as: 0–4, 5–9, 10–19, ≥ 20 years); Systemic arterial hypertension (defined as blood pressure ≥ 140/90 mmHg or use of antihypertensive drugs) [19]; Body mass index (BMI, continuous and categorized according to WHO classification) [20]; Waist circumference (in cm, continuous); Chronic kidney disease at baseline (defined as glomerular filtration rate < 60 mL/min/1.73m2 or albumin/creatinine ratio ≥ 30 mg/g) [21].
Medications in Use: Use of diabetes medications (yes/no); Use of antihypertensive medications (yes/no, subclassified for diuretics, angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers); Use of statins (yes/no).
Laboratory parameters: Glycated hemoglobin (HbA1c, in %); Fasting blood glucose (mg/dL); Lipid profile (total cholesterol, LDL-cholesterol, HDL-cholesterol, triglycerides); Serum creatinine (mg/dL); Estimated glomerular filtration rate (by the CKD-EPI equation, in mL/min/1.73m2); High-sensitivity C-reactive protein (mg/L).
Statistical analysis
Descriptive Analysis: The baseline characteristics of the participants were described according to the four intersectional sex/race groups. For continuous variables with a normal distribution, the mean and standard deviation were used; for variables with an asymmetrical distribution, the median and interquartile range were used. Comparisons between groups were performed using ANOVA (for normal continuous variables) or the Kruskal–Wallis test (for non-normal variables). For categorical variables, the chi-square test or Fisher’s exact test was used when appropriate.
Survival Analysis: Follow-up time was calculated from the date of the interview in wave 1 to the date of the first cardiovascular event, the date of the last interview (for censoring), or the end of the study period (wave 3), whichever occurred first. The cumulative probability of remaining free from cardiovascular events was estimated using the Kaplan–Meier method, with comparisons between groups performed using the log-rank test.
Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (95% CI) for the association between combined sex/race categories and incident cardiovascular events. A conceptual hierarchical approach was adopted. Age was treated as a potential confounder. Educational attainment, economic class, and health insurance were interpreted as socioeconomic indicators that may confound or partially mediate the association between race/sex social position and cardiovascular outcomes. Clinical factors such as HbA1c, hypertension, obesity, diabetes duration, chronic kidney disease, and albuminuria were interpreted as potential downstream pathways as well as predictors of cardiovascular events. Therefore, the fully adjusted model was considered an explanatory model rather than a model estimating the total effect of sex/race social position.
Model 1: Adjusted only for the sex/race intersectionality variable.
Model 2: Adjusted for sex/race intersectionality and sociodemographic factors (age, education, economic class).
Model 3: adjusted for the variables in Model 2 plus clinical and behavioral factors, including duration of diabetes, hypertension, BMI, smoking, medication use, HbA1c, lipid profile, chronic kidney disease, estimated glomerular filtration rate, and albuminuria when available. Because some of these variables may lie on the causal pathway from social disadvantage to cardiovascular disease, attenuation of the sex/race estimates in Model 3 was interpreted as possible mediation and/or overadjustment, rather than as absence of social inequality.
The proportional hazards assumption was assessed graphically and using Schoenfeld residuals. Interaction between sex and race was tested by including multiplicative interaction terms. Multicollinearity was assessed using variance inflation factors (VIF), with values > 5 indicating potentially problematic collinearity. Rates of censoring and losses to follow-up were described, and comparisons between included and excluded participants were performed to evaluate the potential impact of missing data.
Secondary and sensitivity analyses were performed or planned as follows: (1) excluding participants who developed cardiovascular events in the first two years of follow-up to minimize reverse causality bias; (2) using multiple imputation for missing data when appropriate; (3) analyzing individual components of the composite outcome (myocardial infarction, stroke, heart failure, and cardiovascular death) separately; (4) additionally adjusting for chronic kidney disease, estimated glomerular filtration rate, and albuminuria; and (5) excluding participants with a profile compatible with type 1 diabetes when information was available.
All analyses were performed using R software version 4.3.1, with survival, survminer, dplyr, and ggplot2 packages. The level of statistical significance was set at p < 0.05 (two-tailed).
Ethical aspects
The ELSA-Brasil protocol was approved by the Research Ethics Committee of the Federal University of Minas Gerais and by the committees of the other participating institutions (CONEP opinion 976/2006)13. This secondary study was evaluated and approved by the Research Ethics Committee of the University of Fortaleza (protocol nº 5.123.789). All participants signed a free and informed consent form.
Results
Baseline characteristics of the study population
Of the 15,105 participants in ELSA-Brasil, 1,215 (8.0%) had diabetes mellitus at baseline. Among the 1,215 participants with diabetes at baseline, 962 had complete information for all variables included in the primary analysis and no previous cardiovascular disease, constituting the final analytic sample. During follow-up, 186 participants developed incident cardiovascular events. The number excluded because of missing data and the percentage of missingness varied according to the analytic dataset: approximately 16% of initially eligible participants had at least one incomplete variable required for complete-case modelling, whereas missingness for each principal variable was generally low. The revised supplementary material provides the pattern of missing data and comparisons between included and excluded participants. Most participants were treated with oral hypoglycemic agents, a smaller proportion received insulin, and a subset had no pharmacological treatment at baseline. Table 1 presents baseline characteristics stratified by the combined sex/race groups.
Table 1.
Baseline characteristics according to the intersectionality of sex/race of diabetic patients who did not develop cardiovascular disease in waves B and C in ELSA-Brasil
| Did not develop CVD* in waves B and C | White men | Black men | White women | Black women | p-value2 |
|---|---|---|---|---|---|
| N = 2721 | N = 1181 | N = 2141 | N = 1721 | ||
| Age (wave A) | 56.1 ± 8.9 | 53.6 ± 7.7 | 57.2 ± 8.4 | 56.5 ± 8.0 | 0.003 |
| Sex (wave A) | < 0.001 | ||||
| Male | 272 (100.0%) | 118 (100.0%) | 0 (0.0%) | 0 (0.0%) | |
| Female | 0 (0.0%) | 0 (0.0%) | 214 (100.0%) | 172 (100.0%) | |
| Educational attainment (wave A) | < 0.001 | ||||
| Illiterate/Basic study | 22 (8.1%) | 31 (26.3%) | 24 (11.2%) | 37 (21.5%) | |
| High school | 77 (28.3%) | 54 (45.8%) | 71 (33.2%) | 93 (54.1%) | |
| University | 173 (63.6%) | 33 (28.0%) | 119 (55.6%) | 42 (24.4%) | |
| Health status perception (wave A) | < 0.001 | ||||
| Very good | 37 (13.6%) | 15 (12.7%) | 35 (16.4%) | 10 (5.8%) | |
| Good | 167 (61.4%) | 62 (52.5%) | 118 (55.1%) | 75 (43.9%) | |
| Regular | 58 (21.3%) | 37 (31.4%) | 52 (24.3%) | 73 (42.7%) | |
| Bad | 10 (3.7%) | 4 (3.4%) | 9 (4.2%) | 11 (6.4%) | |
| Very bad | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 2 (1.2%) | |
| Economic class (wave A) | 7.054.5 (3.734.5—7.884.5) | 2.904.5 (2.074.5—4.564.5) | 4.564.5 (2.904.5—7.884.5) | 2.904.5 (2.074.5—3.734.5) | < 0.001 |
| Per capita income (wave A) | 1.867.3 (1.078.9—2.628.2) | 933.6 (533.5—1.521.5) | 1.798.2 (1.037.3—2.628.2) | 950.9 (518.6—1.379.8) | < 0.001 |
| Health insurance (wave A) | 180 (66.2%) | 68 (57.6%) | 147 (69.0%) | 99 (57.6%) | 0.046 |
| Without health insurance | 92 (33.8%) | 50 (42.4%) | 66 (31.0%) | 73 (42.4%) | |
| With health insurance | 180 (66.2%) | 68 (57.6%) | 147 (69.0%) | 99 (57.6%) | |
| ABEP criterion** | < 0.001 | ||||
| A | 5 (1.9%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | |
| B | 135 (50.9%) | 19 (16.2%) | 83 (39.3%) | 23 (13.4%) | |
| C | 79 (29.8%) | 52 (44.4%) | 76 (36.0%) | 84 (48.8%) | |
| D | 46 (17.4%) | 46 (39.3%) | 52 (24.6%) | 65 (37.8%) | |
| Civil status (wave A) | < 0.001 | ||||
| Married | 227 (83.5%) | 94 (79.7%) | 107 (50.0%) | 73 (42.4%) | |
| Lives with partner | 25 (9.2%) | 11 (9.3%) | 35 (16.4%) | 27 (15.7%) | |
| Single | 14 (5.1%) | 5 (4.2%) | 34 (15.9%) | 40 (23.3%) | |
| Single / Divorced | 2 (0.7%) | 5 (4.2%) | 25 (11.7%) | 23 (13.4%) | |
| Widow | 4 (1.5%) | 3 (2.5%) | 13 (6.1%) | 9 (5.2%) | |
| Lifestyle | |||||
| Smoking (wave A) | < 0.001 | ||||
| Never smoked | 113 (41.5%) | 54 (45.8%) | 128 (59.8%) | 102 (59.3%) | |
| Ex-smoker | 132 (48.5%) | 50 (42.4%) | 60 (28.0%) | 52 (30.2%) | |
| Smoker | 27 (9.9%) | 14 (11.9%) | 26 (12.1%) | 18 (10.5%) | |
| Tobacco load (wave A) | 20.0 (10.0—20.0) | 10.0 (5.0—20.0) | 10.0 (5.0—20.0) | 9.0 (4.0—20.0) | < 0.001 |
| Excessive alcohol consumption (wave A) | 39 (14.3%) | 16 (13.6%) | 9 (4.2%) | 4 (2.3%) | < 0.001 |
| Current alcohol use (wave A) | < 0.001 | ||||
| Never drank | 10 (3.7%) | 3 (2.5%) | 38 (17.8%) | 34 (19.8%) | |
| Ex-alcohol use | 53 (19.5%) | 26 (22.0%) | 55 (25.8%) | 49 (28.5%) | |
| Currently drinks | 209 (76.8%) | 89 (75.4%) | 120 (56.3%) | 89 (51.7%) | |
| Level of physical activity (wave A) | 0.003 | ||||
| Sedentary | 189 (70.3%) | 98 (83.1%) | 163 (77.3%) | 145 (84.3%) | |
| Insufficiently active | 58 (21.6%) | 15 (12.7%) | 40 (19.0%) | 25 (14.5%) | |
| Active | 22 (8.2%) | 5 (4.2%) | 8 (3.8%) | 2 (1.2%) | |
| Anthropometric data | |||||
| Waist circumference (cm)—wave A | 104.0 ± 11.2 | 100.3 ± 11.1 | 96.7 ± 11.5 | 97.2 ± 13.0 | < 0.001 |
| BMI*** (continuous)—Onda A | 29.6 ± 4.6 | 28.7 ± 4.3 | 30.0 ± 5.0 | 30.3 ± 5.5 | 0.055 |
| BMI (WHO categories)—wave A | 0.002 | ||||
| Eutrophic | 33 (12.1%) | 23 (19.5%) | 37 (17.3%) | 29 (16.9%) | |
| Overweight | 130 (47.8%) | 57 (48.3%) | 69 (32.2%) | 60 (34.9%) | |
| Obesity I | 81 (29.8%) | 28 (23.7%) | 73 (34.1%) | 46 (26.7%) | |
| Obesity II | 21 (7.7%) | 9 (7.6%) | 29 (13.6%) | 28 (16.3%) | |
| Obesity III | 7 (2.6%) | 1 (0.8%) | 6 (2.8%) | 9 (5.2%) | |
| Obesity (wave A) | 109 (40.1%) | 38 (32.2%) | 108 (50.5%) | 83 (48.3%) | 0.004 |
| Clinical characteristics | |||||
| Time of diabetes (years) | 0.002 | ||||
| 0.102 | |||||
| 0–4 years | 198 (72.8%) | 70 (59.3%) | 132 (61.7%) | 108 (62.8%) | |
| 5–9 years | 32 (11.8%) | 19 (16.1%) | 29 (13.6%) | 29 (16.9%) | |
| 10–19 years | 34 (12.5%) | 21 (17.8%) | 36 (16.8%) | 23 (13.4%) | |
| ≥ 20 years | 8 (2.9%) | 8 (6.8%) | 17 (7.9%) | 12 (7.0%) | |
| Self-reported cardiovascular disease | 17 (6.5%) | 7 (6.2%) | 9 (4.5%) | 8 (4.8%) | 0.764 |
| Hypertension (wave A) | 161 (59.4%) | 73 (61.9%) | 119 (55.9%) | 124 (72.1%) | 0.009 |
| Kidney disease in waves B and C | 96 (37.4%) | 46 (40.0%) | 67 (34.2%) | 62 (36.5%) | 0.771 |
| Medications | |||||
| Use of diabetes medications | 209 (79.8%) | 85 (74.6%) | 170 (81.3%) | 135 (80.8%) | 0.505 |
| Use of other antihypertensive drugs | 5 (1.9%) | 1 (0.8%) | 5 (2.3%) | 10 (5.8%) | 0.034 |
| Use of antihypertensive drugs | 139 (51.5%) | 62 (52.5%) | 111 (51.9%) | 114 (66.3%) | 0.01 |
| Current use of antihypertensive drugs | 179 (65.8%) | 76 (64.4%) | 125 (58.4%) | 138 (80.7%) | < 0.001 |
| Use of diuretics | 55 (20.4%) | 21 (17.8%) | 69 (32.2%) | 65 (37.8%) | < 0.001 |
| Use of diuretic antihypertensive medication | 64 (23.5%) | 28 (23.7%) | 65 (30.4%) | 73 (42.7%) | < 0.001 |
| Use of angiotensin II blocker | 88 (32.4%) | 36 (30.5%) | 59 (27.6%) | 69 (40.4%) | 0.06 |
| Use of ACE inhibitor | 64 (23.7%) | 37 (31.4%) | 40 (18.7%) | 39 (22.7%) | 0.075 |
| Mean systolic blood pressure (wave A) | 124.8 (116.5—134.8) | 129.5 (120.5—145.0) | 123.3 (112.0—135.0) | 129.0 (117.0—142.0) | < 0.001 |
| Mean diastolic pressure (wave A) | 78.5 (72.5—85.5) | 80.8 (74.0—88.5) | 74.0 (69.0—81.5) | 79.5 (73.8—85.5) | < 0.001 |
| Framingham risk score (CHD LDL)—wave A | 0.1 (0.1—0.2) | 0.1 (0.1—0.2) | 0.1 (0.1—0.2) | 0.1 (0.1—0.2) | 0.123 |
| Cardiovascular risk score (Framingham)—wave A | 0.2 (0.2—0.4) | 0.2 (0.1—0.3) | 0.1 (0.1—0.2) | 0.1 (0.1—0.2) | < 0.001 |
| Laboratory variables | |||||
| Total Cholesterol (mg/dL)—wave A | 187.0 (163.5—213.0) | 199.0 (173.0—229.0) | 197.5 (175.0—224.0) | 197.0 (169.0—228.0) | 0.02 |
| HDL cholesterol (mg/dL)—wave A | 43.0 (38.0—50.0) | 47.0 (42.0—54.0) | 50.0 (44.0—57.0) | 51.0 (46.5—59.0) | < 0.001 |
| LDL cholesterol (mg/dL)—wave A | 103.0 (88.0—127.0) | 118.0 (92.0—142.0) | 108.0 (92.0—134.0) | 115.5 (87.0—140.0) | 0.004 |
| Triglycerides (mg/dL)—wave A | 153.0 (111.0—218.0) | 129.5 (86.0—205.0) | 133.0 (103.0—202.0) | 112.0 (89.0—176.0) | 0.2 |
| Fast glucose (mg/dL)—wave A | 132.5 (117.5—157.5) | 133.0 (118.0—165.0) | 128.0 (111.0—151.0) | 133.5 (112.0—160.5) | 0.461 |
| Glycated hemoglobin (HbA1c, %)—wave A | 6.0 (5.5—6.9) | 6.6 (5.9—7.6) | 6.1 (5.5—7.0) | 6.5 (6.0—7.4) | 0.002 |
| Creatinine (mg/dL)—Onda A | 1.0 (0.9—1.2) | 1.0 (1.0—1.2) | 0.8 (0.7—0.9) | 0.8 (0.8—0.9) | < 0.001 |
| Glomerular filtration rate (CKD-EPI, continuous)—wave A | 81.5 (70.3—93.7) | 83.4 (73.6—93.7) | 84.2 (72.1—95.2) | 81.7 (70.6—89.6) | 0.648 |
| GFR categories (KDIGO)—wave A | 0.436 | ||||
| G1: eGFR ≥ 90 | 90 (33.1%) | 39 (33.1%) | 73 (34.1%) | 41 (23.8%) | |
| G2: eGFR 60–89 | 166 (61.0%) | 69 (58.5%) | 125 (58.4%) | 117 (68.0%) | |
| G3a: eGFR 45–59 | 14 (5.1%) | 9 (7.6%) | 13 (6.1%) | 14 (8.1%) | |
| G3b: eGFR 30–44 | 2 (0.7%) | 0 (0.0%) | 1 (0.5%) | 0 (0.0%) | |
| G4: eGFR 15–29 | 0 (0.0%) | 1 (0.8%) | 1 (0.5%) | 0 (0.0%) | |
| G5: eGFR < 15 | 0 (0.0%) | 0 (0.0%) | 1 (0.5%) | 0 (0.0%) | |
| Urinary albumin/creatinine ratio—wave A | 0.469 | ||||
| Uric acid (mg/dL)—wave A | < 0.001 |
1 Categorical data expressed as absolute counts and percentages in parentheses. Quantitative data expressed as median and interquartile range in parentheses. 2 ANOVA or Kruskal–Wallis test for continuous data and chi-square test for categorical data
*CVD: cardiovascular disease
**ABEP criterion: Socioeconomic status assessed by the Brazilian Economic Classification Criterion
***BMI: body mass index
As shown in Table 1, significant differences were observed between the intersectional groups in several characteristics. White men had higher education levels and better socioeconomic status, while Black men and Black women were concentrated in lower socioeconomic strata. Women (especially Black women) had a higher prevalence of obesity and hypertension, while men (especially White men) reported a higher frequency of prior smoking. HbA1c values were significantly higher in the Black groups (both men and women) compared to the White groups.
Incidence of cardiovascular events
During follow-up, incident cardiovascular events occurred. The composite outcome included acute myocardial infarction, stroke, heart failure, and cardiovascular death. Because these components have distinct pathophysiological mechanisms, the revised supplementary analyses present event-specific estimates when the number of events allowed model convergence.
Black women had the highest incidence rate of cardiovascular events (21.6 per 1000 person-years), followed by white men (18.9 per 1000 person-years). White women had the lowest rate (12.6 per 1000 person-years). Regarding the specific types of events, acute myocardial infarction was observed to be more frequent in all groups, with a particularly high proportion among men (52–53% of events) (Table 2).
Table 2.
Baseline characteristics according to the intersectionality of sex/race of diabetic patients who developed cardiovascular disease in waves B and C in ELSA-Brasil
| Devevoloped CVD in waves B and C | White men | Black men | White women | Black women | p-value2 |
|---|---|---|---|---|---|
| N = 721 | N = 351 | N = 411 | N = 381 | ||
| Age (wave A) | 60.0 ± 8.4 | 57.1 ± 8.0 | 57.2 ± 7.5 | 59.3 ± 6.6 | 0.151 |
| Sex (wave A) | < 0.001 | ||||
| Male | 72 (100.0%) | 35 (100.0%) | 0 (0.0%) | 0 (0.0%) | |
| Female | 0 (0.0%) | 0 (0.0%) | 41 (100.0%) | 38 (100.0%) | |
| Educational attainment (wave A) | < 0.001 | ||||
| Illiterate/Basic study | 12 (16.7%) | 12 (34.3%) | 6 (14.6%) | 7 (18.4%) | |
| High school | 16 (22.2%) | 17 (48.6%) | 20 (48.8%) | 23 (60.5%) | |
| University | 44 (61.1%) | 6 (17.1%) | 15 (36.6%) | 8 (21.1%) | |
| Health status perception (wave A) | 0.286 | ||||
| Very good | 8 (11.3%) | 2 (5.7%) | 2 (4.9%) | 3 (7.9%) | |
| Good | 36 (50.7%) | 13 (37.1%) | 20 (48.8%) | 11 (28.9%) | |
| Regular | 23 (32.4%) | 16 (45.7%) | 15 (36.6%) | 19 (50.0%) | |
| Bad | 4 (5.6%) | 2 (5.7%) | 4 (9.8%) | 4 (10.5%) | |
| Very bad | 0 (0.0%) | 2 (5.7%) | 0 (0.0%) | 1 (2.6%) | |
| Economic class (wave A) | 7.884.5 (3.734.5—7.884.5) | 2.904.5 (2.074.5—4.564.5) | 3.734.5 (2.074.5—7.054.5) | 2.904.5 (2.074.5—3.734.5) | < 0.001 |
| Per capita income (wave A) | 2.351.5 (1.141.1—3.942.3) | 933.6 (580.9—1.452.3) | 1.410.9 (933.6—2.282.3) | 1.037.3 (622.3—1.971.1) | < 0.001 |
| Health insurance (wave A) | 47 (65.3%) | 20 (57.1%) | 23 (56.1%) | 27 (71.1%) | 0.463 |
| Without health insurance | 25 (34.7%) | 15 (42.9%) | 18 (43.9%) | 11 (28.9%) | |
| With health insurance | 47 (65.3%) | 20 (57.1%) | 23 (56.1%) | 27 (71.1%) | |
| ABEP criterion** | < 0.001 | ||||
| A | 1 (1.4%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | |
| B | 40 (58.0%) | 7 (20.0%) | 8 (20.5%) | 8 (21.1%) | |
| C | 15 (21.7%) | 9 (25.7%) | 17 (43.6%) | 16 (42.1%) | |
| D | 13 (18.8%) | 19 (54.3%) | 14 (35.9%) | 14 (36.8%) | |
| Civil status (wave A) | < 0.001 | ||||
| Married | 63 (87.5%) | 26 (74.3%) | 24 (58.5%) | 17 (44.7%) | |
| Lives with partner | 4 (5.6%) | 6 (17.1%) | 8 (19.5%) | 10 (26.3%) | |
| Single | 2 (2.8%) | 0 (0.0%) | 0 (0.0%) | 4 (10.5%) | |
| Single / Divorced | 1 (1.4%) | 2 (5.7%) | 7 (17.1%) | 4 (10.5%) | |
| Widow | 2 (2.8%) | 1 (2.9%) | 2 (4.9%) | 3 (7.9%) | |
| Lifestyle | |||||
| Smoking (wave A) | 0.002 | ||||
| Never smoked | 26 (36.1%) | 12 (34.3%) | 25 (61.0%) | 27 (71.1%) | |
| Ex-smoker | 40 (55.6%) | 19 (54.3%) | 11 (26.8%) | 8 (21.1%) | |
| Smoker | 6 (8.3%) | 4 (11.4%) | 5 (12.2%) | 3 (7.9%) | |
| Tobacco load (wave A) | 24.0 ± 14.6 | 23.2 ± 17.5 | 14.4 ± 10.9 | 7.8 ± 6.5 | 0.003 |
| Excessive alcohol consumption (wave A) | 5 (6.9%) | 3 (8.6%) | 2 (4.9%) | 1 (2.6%) | 0.706 |
| Current alcohol use (wave A) | 0.002 | ||||
| Never drank | 5 (6.9%) | 3 (8.6%) | 12 (29.3%) | 9 (23.7%) | |
| Ex-alcohol use | 18 (25.0%) | 15 (42.9%) | 10 (24.4%) | 15 (39.5%) | |
| Currently drinks | 49 (68.1%) | 17 (48.6%) | 19 (46.3%) | 14 (36.8%) | |
| Level of physical activity (wave A) | 0.242 | ||||
| Sedentary | 50 (71.4%) | 28 (80.0%) | 37 (90.2%) | 33 (86.8%) | |
| Insufficiently active | 16 (22.9%) | 5 (14.3%) | 3 (7.3%) | 3 (7.9%) | |
| Active | 4 (5.7%) | 2 (5.7%) | 1 (2.4%) | 2 (5.3%) | |
| Anthropometric data | |||||
| Waist circumference (cm)—wave A | 104.3 ± 8.3 | 101.8 ± 10.8 | 101.2 ± 12.0 | 99.4 ± 10.7 | 0.099 |
| BMI*** (continuous)—Onda A | 29.0 ± 3.6 | 29.4 ± 4.5 | 32.2 ± 5.5 | 31.0 ± 5.0 | 0.002 |
| BMI (WHO categories)—wave A | 0.019 | ||||
| Eutrophic | 11 (15.3%) | 5 (14.3%) | 1 (2.4%) | 3 (7.9%) | |
| Overweight | 33 (45.8%) | 20 (57.1%) | 14 (34.1%) | 19 (50.0%) | |
| Obesity I | 24 (33.3%) | 6 (17.1%) | 15 (36.6%) | 7 (18.4%) | |
| Obesity II | 3 (4.2%) | 3 (8.6%) | 6 (14.6%) | 7 (18.4%) | |
| Obesity III | 1 (1.4%) | 1 (2.9%) | 5 (12.2%) | 2 (5.3%) | |
| Obesity (wave A) | 28 (38.9%) | 10 (28.6%) | 26 (63.4%) | 16 (42.1%) | 0.015 |
| Clinical characteristics | |||||
| Time of diabetes (years) | 0.86 | ||||
| 0.307 | |||||
| 0–4 years | 40 (55.6%) | 19 (54.3%) | 25 (61.0%) | 22 (57.9%) | |
| 5–9 years | 11 (15.3%) | 4 (11.4%) | 10 (24.4%) | 9 (23.7%) | |
| 10–19 years | 14 (19.4%) | 11 (31.4%) | 4 (9.8%) | 5 (13.2%) | |
| ≥ 20 years | 7 (9.7%) | 1 (2.9%) | 2 (4.9%) | 2 (5.3%) | |
| Self-reported cardiovascular disease | 34 (48.6%) | 15 (42.9%) | 16 (40.0%) | 10 (27.0%) | 0.195 |
| Hypertension (wave A) | 56 (77.8%) | 29 (82.9%) | 31 (75.6%) | 32 (84.2%) | 0.735 |
| Kidney disease in waves B and C | 37 (52.9%) | 19 (54.3%) | 20 (51.3%) | 18 (50.0%) | 0.984 |
| Medicamentos | |||||
| Use of diabetes medications | 62 (87.3%) | 26 (83.9%) | 32 (78.0%) | 32 (86.5%) | 0.604 |
| Use of other antihypertensive drugs | 3 (4.2%) | 3 (8.6%) | 1 (2.4%) | 0 (0.0%) | 0.267 |
| Use of antihypertensive drugs | 54 (75.0%) | 27 (77.1%) | 31 (75.6%) | 31 (81.6%) | 0.884 |
| Current use of antihypertensive drugs | 65 (90.3%) | 31 (88.6%) | 33 (80.5%) | 34 (89.5%) | 0.467 |
| Use of diuretics | 19 (26.4%) | 14 (40.0%) | 16 (39.0%) | 19 (50.0%) | 0.091 |
| Use of diuretic antihypertensive medication | 22 (30.6%) | 15 (42.9%) | 19 (46.3%) | 20 (52.6%) | 0.116 |
| Use of angiotensin II blocker | 22 (30.6%) | 12 (34.3%) | 16 (39.0%) | 15 (39.5%) | 0.739 |
| Use of ACE inhibitor | 35 (48.6%) | 13 (37.1%) | 15 (36.6%) | 13 (34.2%) | 0.392 |
| Mean systolic blood pressure (wave A) | 127.8 (118.0—137.8) | 127.5 (115.0—150.0) | 128.0 (119.0—144.0) | 130.5 (118.0—143.5) | 0.698 |
| Mean diastolic pressure (wave A) | 76.0 (67.3—83.0) | 81.0 (71.0—92.0) | 79.0 (72.5—88.0) | 78.0 (72.0—85.5) | 0.062 |
| Framingham risk score (CHD LDL)—wave A | 0.1 (0.1—0.2) | 0.1 (0.1—0.2) | 0.1 (0.1—0.2) | 0.1 (0.1—0.2) | 0.581 |
| Cardiovascular risk score (Framingham)—wave A | 0.3 (0.2—0.4) | 0.3 (0.2—0.4) | 0.2 (0.1—0.3) | 0.2 (0.1—0.3) | < 0.001 |
| Laboratory variables | |||||
| Total Cholesterol (mg/dL)—wave A | 178.0 (143.0—197.5) | 186.0 (163.0—223.0) | 190.0 (162.0—227.0) | 204.5 (175.0—229.0) | 0.002 |
| HDL cholesterol (mg/dL)—wave A | 42.0 (37.0—47.0) | 46.0 (39.0—50.0) | 51.0 (43.0—57.0) | 50.5 (47.0—57.0) | < 0.001 |
| LDL cholesterol (mg/dL)—wave A | 98.5 (72.5—118.0) | 103.0 (88.0—136.0) | 101.0 (79.0—131.0) | 115.5 (95.0—141.0) | 0.007 |
| Triglycerides (mg/dL)—wave A | 148.5 (106.0—199.5) | 141.0 (90.0—225.0) | 160.0 (108.0—214.0) | 124.0 (86.0—169.0) | 0.062 |
| Fast glucose (mg/dL)—wave A | 144.0 (123.0—173.0) | 148.0 (128.0—221.0) | 128.0 (112.0—146.0) | 133.5 (117.0—179.0) | 0.065 |
| Glycated hemoglobin (HbA1c, %)—wave A | 6.3 (5.9—7.1) | 7.2 (5.8—9.1) | 6.0 (5.5—7.3) | 7.1 (6.2—7.9) | 0.016 |
| Creatinine (mg/dL)—Onda A | 1.0 (1.0—1.2) | 1.1 (1.0—1.2) | 0.9 (0.8—0.9) | 0.8 (0.7—0.9) | < 0.001 |
| Glomerular filtration rate (CKD-EPI, continuous)—wave A | 79.8 (68.3—89.6) | 75.7 (68.5—85.0) | 79.0 (68.2—91.9) | 83.3 (69.6—92.6) | 0.688 |
| GFR categories (KDIGO)—wave A | 0.716 | ||||
| G1: eGFR ≥ 90 | 17 (23.6%) | 7 (20.0%) | 11 (26.8%) | 12 (31.6%) | |
| G2: eGFR 60–89 | 49 (68.1%) | 23 (65.7%) | 26 (63.4%) | 22 (57.9%) | |
| G3a: eGFR 45–59 | 3 (4.2%) | 3 (8.6%) | 4 (9.8%) | 4 (10.5%) | |
| G3b: eGFR 30–44 | 2 (2.8%) | 2 (5.7%) | 0 (0.0%) | 0 (0.0%) | |
| G4: eGFR 15–29 | 1 (1.4%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | |
| G5: eGFR < 15 | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | |
| Urinary albumin/creatinine ratio—wave A | 5.6 (4.7—31.2) | 6.3 (4.4—33.6) | 8.2 (6.6—11.0) | 8.8 (6.8—22.8) | 0.762 |
| Uric acid (mg/dL)—wave A | 6.6 ± 1.4 | 6.5 ± 1.5 | 5.7 ± 1.5 | 5.7 ± 1.3 | < 0.001 |
1 Categorical data expressed as absolute counts and percentages in parentheses. Quantitative data expressed as median and interquartile range in parentheses. 2 ANOVA or Kruskal–Wallis test for continuous data and chi-square test for categorical data
*CVD: cardiovascular disease
**ABEP criterion: Socioeconomic status assessed by the Brazilian Economic Classification Criterion
***BMI: body mass index
Survival analysis
Figure 1 presents Kaplan–Meier curves for cardiovascular event-free survival according to combined sex/race categories. The log-rank test did not indicate statistically significant differences between curves. Figures 2, 3 and 4 show event-free survival according to educational attainment, economic class, and duration of diabetes, respectively, illustrating stronger crude separation according to education, socioeconomic position, and diabetes duration than according to combined sex/race categories.
Fig. 1.
Kaplan–Meier curves for cardiovascular event–free survival according to sex/race intersectionality in adults with diabetes from the ELSA-Brasil cohort. Participants were categorized into four groups: White men, Black men, White women, and Black women. No statistically significant differences in event-free survival were observed between groups (log-rank p = 0.432). Follow-up duration was up to 8 years
Fig. 2.
Kaplan–Meier curves for cardiovascular event–free survival according to educational attainment in adults with diabetes from the ELSA-Brasil cohort. Participants with lower educational levels showed reduced cardiovascular event–free survival compared with those with higher education, indicating a significant association between education and cardiovascular risk
Fig. 3.
Kaplan–Meier curves for cardiovascular event–free survival according to economic class in adults with diabetes from the ELSA-Brasil cohort. Economic class was categorized according to the Brazilian Economic Classification Criterion (ABEP). Differences in survival curves suggest a gradient in cardiovascular risk across socioeconomic strata
Fig. 4.
Kaplan–Meier curves for cardiovascular event–free survival according to duration of diabetes in adults from the ELSA-Brasil cohort. Longer diabetes duration was associated with a progressively lower probability of remaining free from cardiovascular events, highlighting duration of disease as a key predictor of cardiovascular risk
Cox regression models
Table 3 presents the results of the Cox regression models for the association between sex/race intersectionality and incidence of cardiovascular events.
Table 3.
Cox regression evaluating the relationship of sex/race intersectionality and other factors with the incidence of cardiovascular disease in waves B and C in ELSA-Brasil
| Univarate analysis (CVD*) | Model 1 | Model 2 | Model 3 | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| HR | 95CI | p | HR | 95CI | p | HR | 95CI | p | HR | 95CI | p | ||||
| Sex | |||||||||||||||
| White women | 1.000 | - | - | 1.000 | - | - | 1.000 | - | - | 1.000 | - | - | |||
| White men | 0.87 | 0.65 – 1.16 | 0.346 | 0.91 | 0.68 – 1.22 | 0.535 | 0.91 | 0.67 – 1.22 | 0.514 | 0.94 | 0,7 – 1,26 | 0,671 | |||
| Black men | 0.95 | 0.67 – 1.35 | 0.786 | 0.83 | 0.58 – 1.19 | 0.312 | 0.87 | 0.6 – 1.24 | 0.438 | 0.86 | 0,6 – 1,23 | 0,402 | |||
| Black women | 1.01 | 0.74 – 1.38 | 0.954 | 0.88 | 0.63 – 1.21 | 0.420 | 0.9 | 0.65 – 1.25 | 0.548 | 0.9 | 0,65 – 1,25 | 0,536 | |||
| Educational attainment | |||||||||||||||
| University | 1.000 | - | - | 1.000 | - | - | 1.000 | - | - | 1.000 | - | - | |||
| Illiterate/Basic | 1.83 | 1.41 – 2.37 | < 0.001 | 1.67 | 1.2 – 2.33 | 0.003 | 1.9 | 1.25 – 2.89 | 0.003 | 1.82 | 1,2 – 2,77 | 0,005 | |||
| High school | 1.57 | 1.26 – 1.96 | < 0.001 | 1.66 | 1.28 – 2.14 | < 0.001 | 1.9 | 1.35 – 2.68 | < 0.001 | 1.85 | 1,31 – 2,61 | < 0,001 | |||
| ABEP criterion** | |||||||||||||||
| A | 1.000 | - | - | 1.000 | - | - | 1.000 | - | - | ||||||
| B | 1.22 | 0.3 – 4.95 | 0.779 | 1.92 | 0.27 – 13.79 | 0.518 | 1.92 | 0,27 – 13,83 | 0,517 | ||||||
| C | 1.23 | 0.3 – 4.96 | 0.772 | 1.42 | 0.19 – 10.4 | 0.730 | 1.48 | 0,2 – 10,84 | 0,700 | ||||||
| D | 1.78 | 0.44 – 7.17 | 0.419 | 1.55 | 0.21 – 11.47 | 0.667 | 1.62 | 0,22 – 12 | 0,635 | ||||||
| Time of diabetes | |||||||||||||||
| 0—4 years | 1.000 | - | - | 1.000 | - | - | |||||||||
| 5–9 years | 1.49 | 1.16 – 1.92 | 0.002 | 1.33 | 0,98 – 1,81 | 0,068 | |||||||||
| 10–19 years | 1.37 | 1.06 – 1.78 | 0.017 | 1.28 | 0,94 – 1,75 | 0,117 | |||||||||
| ≥ 20 years | 1.66 | 1.14 – 2.43 | 0.009 | 1.53 | 0,99 – 2,39 | 0,058 | |||||||||
HR: hazard ratio. IC: Intervalo de confiança de 95%
*CVD: cardiovascular disease
**ABEP criterion: Socioeconomic status assessed by the Brazilian Economic Classification Criterion
In models evaluating combined sex/race categories, the estimates were attenuated after adjustment for socioeconomic and clinical factors. Therefore, the results are best interpreted as showing no statistically significant independent association between the measured combined sex/race categories and cardiovascular events after inclusion of the variables available in this cohort. This attenuation may reflect confounding, mediation through socioeconomic and cardiometabolic pathways, overadjustment, and/or limited statistical power.
Factors more consistently associated with increased cardiovascular risk in the adjusted models included lower educational attainment, longer diabetes duration, higher HbA1c, hypertension, and kidney disease markers. In response to reviewer comments, chronic kidney disease, estimated glomerular filtration rate, and albuminuria were considered in the expanded model because of their established relevance as predictors of cardiovascular outcomes in diabetes.
The interaction test between sex and race was not statistically significant. However, given the relatively limited number of cardiovascular events, the absence of statistical interaction should be interpreted cautiously.
Sensitivity analyses
Sensitivity analyses supported the direction of the main findings but should be interpreted in light of the limited number of outcome events and the risk of unstable estimates in highly adjusted models.
Excluding events occurring in the first 2 years of follow-up produced estimates broadly consistent with the primary analysis.
Multiple imputation analyses produced estimates similar to the complete-case analysis when applied to variables with missing information.
Analyses by individual cardiovascular outcome components showed broadly similar patterns, although precision was limited and confidence intervals were wide, particularly for heart failure and cardiovascular death.
Discussion
This longitudinal study investigated whether combined categories of sex and self-reported race/skin color were associated with incident cardiovascular events among adults with diabetes in ELSA-Brasil. The main findings were: (1) combined sex/race categories were not statistically associated with cardiovascular events after adjustment for measured socioeconomic and clinical variables; (2) educational attainment, diabetes duration, glycemic control, hypertension, and kidney disease markers were more consistent predictors of cardiovascular events; and (3) the attenuation of sex/race estimates after extensive adjustment should be interpreted cautiously because some adjusted variables may represent mediators of social inequality rather than true confounders.
Our findings are consistent with contemporary cohort studies showing that recurrent cardiovascular events remain strongly associated with cumulative cardiometabolic burden and that prediction models incorporating demographic and clinical characteristics improve cardiovascular risk stratification across race and sex groups, supporting the importance of integrated risk assessment in individuals with diabetes [22–24].
The lack of a statistically significant independent association between the measured sex/race categories and cardiovascular risk should not be interpreted as evidence that intersectional or structural inequalities are absent. First, the ELSA-Brasil cohort is composed of public employees, a group with greater employment stability and potentially better access to healthcare than the general Brazilian population. This characteristic may have attenuated disparities that would likely be more pronounced in less socially protected populations and is consistent with previous descriptions of the ELSA-Brasil cohort profile and diabetes burden in this population [25]. Second, the Brazilian Unified Health System (SUS) provides universal access to healthcare and medications for chronic diseases, which may reduce, although not eliminate, inequalities in diabetes and cardiovascular care [26].
Third, race/skin color classification in Brazil is complex and dynamic, and the grouping of Black and Brown ("pardo") participants may not fully capture the heterogeneity of racialized experiences [27]. Moreover, Brazilian epidemiological studies rely on nationally standardized classifications of race/skin color and socioeconomic position, which should be considered when comparing findings across different countries and healthcare systems [16, 17]. Beyond classification issues, the broader Latin American literature consistently demonstrates that health inequalities are deeply influenced by structural social determinants and racial inequities [26–29]. Furthermore, systematic reviews have shown that experiences of racial discrimination are consistently associated with adverse physical and mental health outcomes, supporting the interpretation that unmeasured structural determinants may contribute to cardiovascular disparities beyond those captured by the variables available in the present study [30, 31]. Fourth, direct measures of racism, discrimination, healthcare barriers, neighborhood deprivation, and cumulative social disadvantage were not available. These unmeasured factors are central to the intersectionality framework and may operate upstream of the socioeconomic and cardiometabolic variables included in our models [28].
Longer diabetes duration and poorer glycemic control were independently associated with higher cardiovascular risk, reinforcing the importance of early diagnosis, sustained glycemic management, and prevention of therapeutic inertia. These findings are consistent with landmark studies such as UKPDS and DCCT/EDIC, which demonstrated that intensive glycemic control during the early stages of diabetes confers long-term cardiovascular benefits through the phenomenon known as "metabolic memory” [29, 30]. More recent longitudinal evidence has further reinforced this concept by demonstrating a persistent legacy effect, whereby good glycemic control achieved early in the disease course continues to reduce future vascular complications even when glycemic control subsequently deteriorates [32]. Our findings are also consistent with contemporary cohort studies demonstrating that recurrent cardiovascular events remain strongly associated with cumulative cardiometabolic burden and that prediction models incorporating demographic and clinical characteristics improve cardiovascular risk stratification across race, sex, and age groups [22–24]. These observations are also aligned with the current ADA/EASD consensus recommendations, which emphasize individualized glycemic management together with comprehensive cardiovascular risk reduction as the cornerstone of contemporary diabetes care [33].
The association between lower educational attainment and higher cardiovascular risk, even after adjustment for economic class and other covariates, highlights education as a fundamental social determinant of health [31]. Education influences not only socioeconomic opportunities but also health literacy, self-management skills, treatment adherence, and the ability to navigate healthcare systems [32]. Low health literacy has consistently been associated with poorer glycemic control, lower adherence to treatment, increased hospitalization, and worse health outcomes among individuals with chronic diseases, reinforcing the importance of educational interventions as part of comprehensive diabetes management [34].
The persistence of hypertension as an independent predictor reinforces the need for an integrated approach to multiple cardiometabolic risk factors in diabetes. Multifactorial care programs simultaneously addressing glycemic control, blood pressure, lipid management, and lifestyle modification have demonstrated greater effectiveness in reducing cardiovascular events than interventions targeting a single risk factor [33]. In addition, large meta-analyses have confirmed that effective blood pressure reduction substantially decreases cardiovascular morbidity and mortality across different baseline blood pressure levels, supporting aggressive management of hypertension in high-risk individuals with diabetes [35].
The present findings suggest that cardiovascular prevention among individuals with diabetes should prioritize early diagnosis, sustained glycemic control, blood pressure optimization, kidney risk assessment including estimated glomerular filtration rate and albuminuria, and lifestyle interventions. These priorities are fully consistent with current international Standards of Care, which recommend comprehensive cardiovascular and renal risk assessment as essential components of diabetes management [14]. At the same time, these clinical strategies should be integrated into broader public health policies addressing health literacy, unequal access to healthcare, structural racism, gender inequities, and socioeconomic disadvantage.
Although combined sex/race categories were not statistically significant predictors after full adjustment, racial and gender inequalities should not be overlooked. Baseline differences in education, income, obesity, hypertension, and glycemic control suggest that socially patterned inequalities may exert their effects through intermediate socioeconomic and clinical pathways. Consequently, targeted interventions may still be required to reduce disparities in risk factor burden and improve equitable access to high-quality diabetes care. This interpretation is reinforced by recent analyses from the ELSA-Brasil cohort demonstrating poorer long-term glycemic control among Black women, highlighting the persistence of important disparities despite similar healthcare access [36].
This study has several limitations. First, the ELSA-Brasil cohort consists of public employees and is therefore not representative of the Brazilian population as a whole, potentially attenuating social gradients in cardiovascular risk. Second, self-reported race/skin color is subject to variability in self-classification, and the grouping of racial categories may obscure important heterogeneity. Third, the study did not directly measure racism, discrimination, healthcare barriers, neighborhood deprivation, or cumulative social adversity, all of which are central components of an intersectional framework. Fourth, the relatively limited number of cardiovascular events reduced statistical power for highly adjusted models and subtype-specific analyses, increasing the possibility of type II error and imprecise estimates, a recognized challenge in observational epidemiological studies [37]. Fifth, some variables included in the fully adjusted models may represent mediators rather than confounders, introducing the possibility of overadjustment. Sixth, diabetes type could not be completely classified for all participants, although sensitivity analyses excluding participants with characteristics compatible with type 1 diabetes were considered. Finally, missing data may have introduced selection bias, although missingness for the principal variables was generally low and sensitivity analyses yielded similar results.
The strengths of this study include its prospective longitudinal design, the use of data from ELSA-Brasil, one of the largest and most comprehensively characterized cohort studies conducted in Latin America, standardized data collection across multiple research centers, validated cardiovascular outcomes, detailed socioeconomic and clinical characterization, assessment of kidney disease markers, and explicit recognition that cross-classified sex/race categories represent an operational approximation rather than a complete measure of intersectionality [38].
Directions for future research
Future studies should investigate longitudinal trajectories of cardiovascular risk according to intersectional social positions, incorporate direct measures of racism, discrimination, chronic stress, resilience, and social support, evaluate more socially vulnerable populations with limited healthcare access, extend follow-up periods to improve statistical power for cardiovascular outcomes, and explore the biological mechanisms linking social disadvantage to cardiometabolic disease, including the potential contribution of chronic psychosocial stress and weathering processes [39, 40].
Conclusion
In this longitudinal analysis of Brazilian adults with diabetes, combined sex/race categories were not statistically associated with incident cardiovascular events after adjustment for measured socioeconomic and clinical variables. These findings should be interpreted cautiously because limited event numbers, potential overadjustment, residual confounding, and the absence of direct measures of racism, discrimination, and healthcare barriers may have influenced the results. The study supports the importance of integrated cardiovascular prevention in diabetes, including glycemic control, blood pressure management, kidney risk assessment, and educationally tailored care, while also reinforcing the need for research and policies that address the structural determinants underlying cardiometabolic inequalities.
Acknowledgements
We would like to thank all participants of the ELSA-Brasil study for their valuable voluntary contribution and the technical support from the employees of our institutions.
Author contributions
•Leonardo Fonseca Maia: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Writing – original draft. •Ane Karoline Medina Néri: Data curation; Investigation; Writing – review & editing. •Simon Fraser: Writing – review & editing. •Rosa Malena Fagundes Xavier: Investigation; Writing – review & editing. •Maria Teresita del Niño Jesus Fernandez Bendicho: Investigation; Writing – review & editing. •Jamile Costa Cerqueira: Data curation; Investigation; Writing – review & editing. •Maria da Conceição Chagas de Almeida: Investigation; Writing – review & editing. •Antonio Augusto Ferreira Carioca: Formal analysis; Methodology; Writing – review & editing. •Sandhi Barreto: Resources; Data curation; Supervision; Writing – review & editing. •Antonio Alberto Lopes: Supervision; Writing – review & editing. •Sheila Maria Alvim de Matos: Conceptualization; Methodology; Supervision; Writing – review & editing. •Geraldo Bezerra da Silva Junior: Conceptualization; Methodology; Supervision; Project administration; Writing – review & editing. All authors approved the final version of the manuscript and agree to be accountable for all aspects of the work.
Funding
This study received funding from the Edson Queiroz Foundation/University of Fortaleza. This study was supported by the Brazilian Ministry of Health (Department of Science and Technology) and the Brazilian Ministry of Science, Technology and Innovation/FINEP (Financiadora de Estudos e Projetos), grants #01 06 0010.00, #01 06 0212.00, #01 06 0300.00, #01 06 0278.00, #01 06 0115.00, and #01 06 0071.00. Additional support was provided by CNPq (National Council for Scientific and Technological Development) and CAPES (Brazilian Federal Agency for Support and Evaluation of Graduate Education). GBSJ received a CNPq productivity grant (#303453/2024–9). The funders had no role in study design, data analysis, interpretation, manuscript preparation, or the decision to submit the article for publication.
Data availability
No datasets were generated or analysed during the current study.
Declarations
Competing interests
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.International Diabetes Federation. IDF diabetes atlas. 10th ed. Brussels: International Diabetes Federation; 2021. [Google Scholar]
- 2.Brasil. Ministério da Saúde. Vigitel Brasil 2023: vigilância de fatores de risco e proteção para doenças crônicas por inquérito telefônico. Brasília: Ministério da Saúde; 2024. [Google Scholar]
- 3.Emerging Risk Factors Collaboration, Sarwar N, Gao P, Seshasai SR, Gobin R, Kaptoge S, Di Angelantonio E, et al. Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies. Lancet. 2010;375(9733):2215–22. [DOI] [PMC free article] [PubMed]
- 4.Hill-Briggs F, Adler NE, Berkowitz SA, Chin MH, Gary-Webb TL, Navas-Acien A, et al. Social determinants of health and diabetes: a scientific review. Diabetes Care. 2021;44(1):258–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Crenshaw K. Demarginalizing the intersection of race and sex: a black feminist critique of antidiscrimination doctrine, feminist theory and antiracist politics. Univ Chic Leg Forum. 1989;1989(1):139–67. [Google Scholar]
- 6.Bowleg L. The problem with the phrase women and minorities: intersectionality-an important theoretical framework for public health. Am J Public Health. 2012;102(7):1267–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Huebschmann AG, Huxley RR, Kohrt WM, Zeitler P, Regensteiner JG, Reusch JEB. Sex differences in the burden of type 2 diabetes and cardiovascular risk across the life course. Diabetologia. 2019;62(10):1761–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Williams DR, Lawrence JA, Davis BA. Racism and health: evidence and needed research. Annu Rev Public Health. 2019;40:105–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Chor D. Health inequalities in Brazil: race matters. Cad Saude Publica. 2013;29(7):1272–5. [DOI] [PubMed] [Google Scholar]
- 10.Travassos C, Williams DR. The concept and measurement of race and their relationship to public health: a review focused on Brazil and the United States. Cad Saude Publica. 2004;20(3):660–78. [DOI] [PubMed] [Google Scholar]
- 11.Aquino EM, Barreto SM, Bensenor IM, Carvalho MS, Chor D, Duncan BB, et al. Brazilian longitudinal study of adult health (ELSA-Brasil): objectives and design. Am J Epidemiol. 2012;175(4):315–24. [DOI] [PubMed] [Google Scholar]
- 12.Schmidt MI, Duncan BB, Mill JG, Lotufo PA, Chor D, Barreto SM, et al. Cohort profile: Longitudinal Study of Adult Health (ELSA-Brasil). Int J Epidemiol. 2015;44(1):68–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Bensenor IM, Griep RH, Pinto KA, Faria CP, Felisbino-Mendes M, Caetano EI, et al. Routines of organization of clinical tests and interviews in the ELSA-Brasil investigation center. Rev Saude Publica. 2013;47(Suppl 2):37–47. [DOI] [PubMed] [Google Scholar]
- 14.Committee ADAPP. Standards of medical care in diabetes - 2024. Diabetes Care. 2024;47(Suppl 1):S1–322. [Google Scholar]
- 15.Barreto SM, Ladeira RM, Duncan BB, Schmidt MI, Lopes AA, Bensenor IM, et al. Chronic kidney disease among adult participants of the ELSA-Brasil cohort: association with race and socioeconomic position. J Epidemiol Community Health. 2016;70(4):380–9. [DOI] [PubMed] [Google Scholar]
- 16.Instituto Brasileiro de Geografia e Estatística. Características étnico-raciais da população: um estudo das categorias de classificação de cor ou raça. Rio de Janeiro: IBGE; 2013. [Google Scholar]
- 17.Associação Brasileira de Empresas de Pesquisa. Critério Brasil 2020. São Paulo: ABEP; 2020. [Google Scholar]
- 18.World Health Organization. Global recommendations on physical activity for health. Geneva: WHO; 2010. [PubMed] [Google Scholar]
- 19.Williams B, Mancia G, Spiering W, AgabitiRosei E, Azizi M, Burnier M, et al. 2018 ESC/ESH Guidelines for the management of arterial hypertension. Eur Heart J. 2018;39(33):3021–104. [DOI] [PubMed] [Google Scholar]
- 20.World Health Organization. Obesity: preventing and managing the global epidemic. World Health Organ Tech Rep Ser 2000; 894: i-xii, 1–253. [PubMed]
- 21.Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group. KDIGO 2012 Clinical Practice Guidelines for the Evaluation and Management of Chronic Kidney Disease. Kidney Int Suppl. 2013;3(1):1–150.
- 22.Chi G, Lee JJ, Kazmi SHA, et al. Early and late recurrent cardiovascular events among high-risk patients with an acute coronary syndrome: meta-analysis of phase III studies and implications on trial design. Clin Cardiol. 2022;45(3):299–307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ulvestam A, Graipe A, Irewall AL, et al. Incidence and predictors of cardiovascular outcomes after acute coronary syndrome in a population-based cohort study. Sci Rep. 2023;13(1):3447. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hong C, Pencina MJ, Wojdyla DM, et al. Predictive accuracy of stroke risk prediction models across black and white race, sex, and age groups. JAMA. 2023;329(4):306–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Schmidt MI, Hoffman JF, de Fátima Sander Diniz M et al. High prevalence of diabetes and intermediate hyperglycemia - The Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Diabetol Metab Syndr 2014; 6: 123. [DOI] [PMC free article] [PubMed]
- 26.Paim J, Travassos C, Almeida C, Bahia L, Macinko J. The Brazilian health system: history, advances, and challenges. Lancet. 2011;377(9779):1778–97. [DOI] [PubMed] [Google Scholar]
- 27.Pólvora JB. Racismo à brasileira: uma nova perspectiva sociológica. Horizontes Antropológicos. 2004;10(21):320–3. [Google Scholar]
- 28.Krieger N. Measures of racism, sexism, heterosexism, and gender binarism for health equity research: from structural injustice to embodied harm - an ecosocial analysis. Annu Rev Public Health. 2020;41:37–62. [DOI] [PubMed] [Google Scholar]
- 29.Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med. 2008;359(15):1577–89. [DOI] [PubMed] [Google Scholar]
- 30.Bastos JL, Celeste RK, Faerstein E, Barros AJ. Racial discrimination and health: a systematic review of scales with a focus on their psychometric properties. Soc Sci Med. 2010;70(7):1091–9. [DOI] [PubMed] [Google Scholar]
- 31.Paradies Y, Ben J, Denson N, Elias A, Priest N, Pieterse A, et al. Racism as a determinant of health: a systematic review and meta-analysis. PLoS ONE. 2015;10(9):e0138511. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Laiteerapong N, Ham SA, Gao Y, Moffet HH, Liu JY, Huang ES, et al. The legacy effect in type 2 diabetes: impact of early glycemic control on future complications (The Diabetes & Aging Study). Diabetes Care. 2019;42(3):416–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Davies MJ, Aroda VR, Collins BS, Gabbay RA, Green J, Maruthur NM, et al. Management of hyperglycemia in type 2 diabetes, 2022. A consensus report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care. 2022;45(11):2753–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Berkman ND, Sheridan SL, Donahue KE, Halpern DJ, Crotty K. Low health literacy and health outcomes: an updated systematic review. Ann Intern Med. 2011;155(2):97–107. [DOI] [PubMed] [Google Scholar]
- 35.Brunström M, Carlberg B. Association of blood pressure lowering with mortality and cardiovascular disease across blood pressure levels: a systematic review and meta-analysis. JAMA Intern Med. 2018;178(1):28–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Garcia GAF, Giatti L, Camelo LV et al. Intersection of Gender-Race/Color and Glycemic Control in Diabetes in Nine Years of Follow-up of the ELSA-Brazil Cohort. J Racial Ethn Health Disparities 2025. 10.1007/s40615-025-02760-8. Online ahead of print. [DOI] [PubMed]
- 37.Woodward M. Epidemiology: study design and data analysis. 3rd ed. Boca Raton: CRC Press; 2013. p. 898p. [Google Scholar]
- 38.Lotufo PA. The Brazilian Longitudinal Study of Adult Health (ELSA-Brasil): the best science is providing health for all. Sao Paulo Med J. 2018;136(6):499–500. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Pascoe EA, Smart Richman L. Perceived discrimination and health: a meta-analytic review. Psychol Bull. 2009;135(4):531–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Geronimus AT, Hicken M, Keene D, Bound J. “Weathering” and age patterns of allostatic load scores among blacks and whites in the United States. Am J Public Health. 2006;96(5):826–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
No datasets were generated or analysed during the current study.