SciELO - Scientific Electronic Library Online

vol.62 número2Seguimiento de clonas de Staphylococcus aureus resistentes a meticilina en un hospital en el sur de MéxicoInfección experimental y competencia vectorial del virus Zika en mosquitos Aedes aegypti de Acapulco, Guerrero, México índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados




Links relacionados

  • No hay artículos similaresSimilares en SciELO


Salud Pública de México

versión impresa ISSN 0036-3634

Salud pública Méx vol.62 no.2 Cuernavaca mar./abr. 2020  Epub 28-Feb-2022 

Artículos originales

Linking socioeconomic inequalities and type 2 diabetes through obesity and lifestyle factors among Mexican adults: a structural equations modeling approach

Asociación entre las inequidades socioeconómicas y la diabetes tipo 2 mediada por la obesidad y factores del estilo de vida en adultos mexicanos: un enfoque de modelos de ecuaciones estructurales

Edgar Denova-Gutiérrez1 

Delfino Vargas-Chanes2 

Sheyla Hernández3 

Paloma Muñoz-Aguirre4 

David Napier5 

Simón Barquera1 

1Centro de Investigación en Nutrición y Salud, Instituto Nacional de Salud Pública. Cuernavaca, Mexico.

2Programa Universitario de Estudios del Desarrollo, Universidad Nacional Autónoma de México. Mexico City, Mexico.

3Facultad de Medicina, Universidad Nacional Autónoma de México. Mexico City, Mexico.

4Conacyt-Centro de Investigación en Salud Poblacional, Instituto Nacional de Salud Pública. Cuernavaca, Mexico.

5Department of Anthropology, University College London. London, United Kingdom.



To assess the association between type 2 diabetes (DM2) and socioeconomic inequalities, mediated by the contribution of body mass index (BMI), physical activity (PA), and diet (diet-DII).

Materials and methods:

We conducted a cross-sectional analysis using data of adults participating in the Diabetes Mellitus Survey of Mexico City. Socioeconomic and demographic characteristics as well as height and weight, dietary intake, leisure time activity and the presence of DM2 were measured. We fitted a structural equation model (SEM) with DM2 as the main outcome, and BMI, diet-DII and PA served as mediator variables between socioeconomic inequalities index (SII) and DM2.


The prevalence of DM2 was 13.6%. From the fitted SEM, each standard deviation increases in the SII was associated with increased scores of DM2 (β=0.174,P<0.001).


The results in the present study show how high scores in the index of SII may influence the presence of DM2.

Keywords: type 2 diabetes; socioeconomic inequalities; survey; Mexico



Evaluar la asociación entre diabetes tipo 2 y las inequidades socioeconómicas (IS), mediada por la contribución del índice de masa corporal (IMC), actividad física (AF) y dieta (dieta-DII).

Material y métodos:

Se realizó un análisis transversal utilizando datos de la Encuesta de Diabetes Mellitus de la Ciudad de México. Se midieron las características sociodemográficas, altura, peso, ingesta dietética, actividad de tiempo libre y presencia de diabetes. Se ajustó un modelo de ecuaciones estructurales (MEE) con diabetes como resultado principal, e IMC, dieta-DII y PA sirvieron como variables mediadoras entre el IS y la diabetes.


La prevalencia de diabetes fue de 13.6%. A partir del MEE ajustado, cada aumento de la desviación estándar en el IS se asoció con un aumento en las puntuaciones de diabetes (β=0.174,P<0.001).


Los resultados en el presente estudio muestran cómo los puntajes altos en las IS pueden influir en la presencia de diabetes.

Palabras clave: diabetes tipo 2; desigualdades socioeconómicas; encuesta; México


Worldwide, type 2 diabetes (DM2) prevalence has doubled in past decades,1representing one of the largest widespread epidemics faced by the world population. Due to increasing disease burden, DM2 has been recognized as one of the most important non-communicable diseases (NCDs).2According to the International Diabetes Federation (IDF), in 2017, there were 451 million people living with DM2 and this number is expected to increase to 693 million by 2045.3From an economic perspective, previous report4suggest that total DM2-related health expenditure in 2015 was 673 billion US dollars and it is projected that in 2040 it will cost 802 billion US dollars. The IDF shows that Mexico ranks among top five countries in the world for DM2,2with approximately 12 million people living with this condition. In addition, in 2016, in agreement with the Global Burden of Disease study, 65% of premature mortality and 88% of disability-adjusted life-years (DALYs) were caused by NCDs5in Mexico with DM2 among top five leading causes.6

Genetic as well as environmental factors, particularly lifestyle factors (i.e. poor diet, physical inactivity, sedentary lifestyle, smoking, etc.), joined by economic transition, urbanization, industrialization, and globalization have been defined as important components in the growing burden of NCDs.7In addition to environmental and genetic factors, primary determinants of disease, essentially social and economic in nature, have been linked with DM2 and also with changes in diet and other lifestyles.8In this sense, socioeconomic status (SES)-a construct determined through access to health-care and information, healthy foods and access to exercise venues, income level, education and occupational opportunities as well as individual lifestyle choices9 has been defined as a factor of the unequal distribution in the prevalence, incidence, and mortality of DM2.10Unequal distribution of risk factors for DM2 (i.e. obesity, unhealthy diet, physical inactivity) between levels of SES have been reported.11Unequal distribution of access to resources and opportunities necessary to achieve a healthy lifestyle has been defined as socioeconomic inequalities.12

Despite the fact that socioeconomic inequalities and DM2 have been widely studied in different populations, the causal pathway is not yet fully understood. Moreover, as previously reported,13there is a lack of comprehensive information about how some factors (i.e. body mass index, diet, physical activity, family history of DM2, etc.) mediate and potentially compound the impact of socioeconomic inequalities on DM2. Thus, the main objectives of this study are to evaluate, using a structural equation modeling approach, the relation between socioeconomic inequalities and DM2 and to examine body mass index, diet, and physical activity as potential mediators in this relationship in a group of adults participating in the Diabetes Mellitus Survey of Mexico City 2015 (DMS-MC 2015).

Materials and methods

Study design

The DMS-MC 2015, a probabilistic population-based survey, collected cross-sectional data from a sample representative of adult residents of Mexico City aged 20-69 years. The DMS-MC 2015 was conducted in 2015 by a research team from the National Institute of Public Health (INSP, by its Spanish acronym) to obtain information and measurements of NCDs, as well as socio-economic conditions, lifestyle risk factors (i.e. diet and physical activity), and other health indicators. The methodology and participant characteristics have been previously detailed.14

In total, 1 174 subjects were included in the present analysis. Altogether, 242 participants were excluded for the following reasons: 142 participants had missing information regarding biomarkers and other important covariates. We also excluded subjects with more than 10% missing items on their food frequency questionnaires, those who did not consume between 600 kcal and 5 500 kcal daily, and subjects with more than 12 months past their DM2 diagnosis date.

We executed the current study according to the Declaration of Helsinki guidelines. The Research, Ethics and Biosecurity Committee at INSP reviewed and approved the study protocol and informed consent forms. Written informed consent was obtained from each participant.

Dietary intake

A semi-quantitative food frequency questionnaire (SFFQ) previously validated in Mexican populations15was used to assess dietary intake over the past seven days prior to the interview. The SFFQ included information concerning the consumption of 140 foods items. For each food, a commonly used portion size was specified (i.e. one slice of bread or one cup of coffee). Frequency intake options were characterized by set categories ranging from “never” to “six or more times per day”. We converted frequency, which was originally measured as times per day, into portion size per day. Then, to estimate the nutrient and total energy (kcal/day) intake, we multiplied the frequency of consumption of each food by the estimated nutrient content with a food content database, compiled by INSP.16The administration and collection of the SFFQ was carried out by trained personnel using standardized data collection and entry procedures.

To compute the dietary inflammatory index (DII) scores for participants in the DMS-MC 2015, 27 food items and nutrients were linked to the world database that provided estimates of mean intake and standard deviation for each food parameter.17Then, these were expressed using Z-score. Finally, all the food parameter-specific DII scores were summed to create the overall DII score for every participant in the study. The methodology has previously been described.14


Total cholesterol (TC), high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), triglycerides (TG), glucose, and glycated hemoglobin A1c (HbA1c) were measured in a standardized laboratory. A fasting venous blood sample (fasting time was ≥ 8 hours) was collected from an antecubital vein from each participant.

Plasma TG was measured with a colorimetric method following enzymatic hydrolysis performed with the lipase technique. TC, HDL-c, and LDL-c were measured using the colorimetric method following enzymatic assay. Plasma glucose was measured with the enzymatic colorimetric methods by using oxidized glucose. In addition, the proportion of HbA1c was determined using the immunocolorimetric method.18

Type 2 diabetes

Previous DM2 diagnosis methods in the DMS-CM 2015 have been previously reported.14In short, first subjects were asked whether or not they had a diabetic condition and whether they had ever been diagnosed with DM2 by a physician. Additionally, subjects whose glucose concentration was ≥126 mg/dL and had levels of [HbA1c (%) ≥6.5] were defined as displaying fasting glucose and/or HbA1c values consistent with DM2 diagnostic criteria19and were also defined as participants with DM2. The manifest continuous variables glucose levels and HbA1c were included into the SEM model as a latent variable of DM2 for SEM analysis.

Anthropometric and blood pressure

Using standardized procedures, trained personnel measured participants’ height and weight. Body weight was assessed with a previously calibrated electronic (SECA 874) scale with a precision of 0.1 kg. Height was evaluated by using a conventional stadiometer (SECA 213) to the nearest 0.1 cm. Body mass index was estimated as weight in kilograms divided by the square of height in meters [BMI= weight (kg)/height (m2)]. We categorized BMI into three categories: normal <25.0 kg/m2, overweight ≥25.0 and <30.0 kg/m2, and obesity ≥30.0 kg/m2. For the structural equation model (SEM), BMI was included as a continuous variable. With a measuring tape, waist circumference was evaluated to the nearest 0.1 cm at the highest point of the iliac crest to the end of normal expiration, which was placed below any clothing, directly touching the participant’s skin. Central obesity was defined as a waist circumference of ≥90 cm in men and ≥80 cm in women.20Additionally, participants’ blood pressure was measured twice using an automatic medical grade monitor (OMROM HEM-907). The first measurement was taken after five minutes of rest, while participants were sitting with the dominant arm supported at heart level. The second measurement was taken in the same way, five minutes after the first.

Physical activity

A short version of the international physical activity questionnaire (s-IPAQ), previously validated in Mexican populations, was used to evaluate physical activity (PA).21The questionnaire includes nine items that assess time spent performing moderate-intense PA for at least 10 minutes of each activity over seven days. The data of the s-IPAQ was examined as follows: first, physical activity interval duration gathered in hours was converted into minutes; second, data which was described as a weekly frequency was transformed into an average daily time; and third, subjects whose responses were “do not know”, or “refused”, or had “missing data” for time duration or frequency were excluded. In the SEM analysis, PA was analyzed as a continuous variable.

Socioeconomic status

A household wealth index (HWI) was created using principal components analysis with household characteristics and family assets. In general, HWI was constructed by combining eight variables that assessed household characteristics, goods, and available services including: construction materials of the floor, ceiling, and walls; household goods (stove, microwave, washing machine, refrigerator and boiler); and electric goods (television, computer, radio and telephone). The index was divided into tertiles and used as a proxy for low, medium, and high SES, for the multivariate logistic regression models the highest tertile (high SES) was the category of reference.


Education level was based on a question, “What level of education do you have?”, and the responders were requested to state their maximum level of education. The response options ranged from “elementary school” to “master and doctoral degree”. For SEM analysis, we stratified subjects into five groups according to the highest level of education obtained: “elementary or less”, “secondary”, “high school”, “university”, and “master and doctoral degree”. For the multivariate logistic regression models, we stratified subjects into three categories according to the highest level of education obtained as follow: “elementary or less and secondary” “high school” and “Bachelor’s degree or higher”, being Bachelor’s degree or higher the category of reference.


Using a self-administered questionnaire (household level), income was determined using two questions, “What is your income for your actual job?”, and “Do you have other sources of income?”. Response categories ranged from “daily income” to “annual income.” With this information, we computed the monthly income of each person. In the SEM analysis, income was analyzed as a continuous variable. For the multivariate logistic regression models, income was divided in tertiles, being the highest tertile of income the category of reference.

Other participant characteristics

In general, two self-administered questionnaires (at the household and individual level) were completed for each participant. With these questionnaires, detailed information concerning their demographic characteristics (i.e. age, sex, marital status), self-perception of body weight, past medical history, family history of DM2, current medication use, lifestyle information (i.e., diet, physical activity, smoking status, etc.), depression symptoms, sleep quantity, and information about reproductive history (for females) were obtained.

Variables for the SEM

Index of socioeconomic inequalities

In order to establish a unidimensional measure of socioeconomic inequality, we incorporated sociodemographic variables (such as income, SES, and education) into a latent variable called the socioeconomic inequalities index (SII). The latent variable SII was included in the SEM analysis as an exogenous variable. The reliability of this latent variable is 0.70.

Type 2 diabetes

We used two manifest variables to assess DM2 as a latent variable, one is the glucose level and the other is the HbA1c. The reliability of this latent variable is 0.96. The rest of the variables in the SEM (PA, diet-DII., BMI, and family history of DM2) were measured as manifest variables, as described previously. Latent variables are indicated in the figures 1and 2 as ellipses and manifest variables as rectangles.

Statistical analysis

In all analyses, complex sampling design was considered using survey-related commands and specifying PSUs, strata, and weights. All P-values are 2-tailed andp<0.05 was considered significant. All analyses were performed using Stata software, version 13.0. For SEM analyses we used MPLUS 7.11.

A descriptive analysis of the main characteristics of interest, frequencies, means, and standard deviations, was conducted. One-way ANOVA was used to test for differences in general characteristics across socioeconomic inequalities tertiles, while, chi-square tests were used to evaluate the distribution of qualitative variables across SII tertiles. To evaluate the magnitude of the association between specific SII tertiles and DM2, we estimated multivariable adjusted odds ratios (OR) and 95% confidence intervals (95%CI) using logistic regression models. In all multivariate models, the first tertile of the SII score was used as the reference. The Mantel-Haenszel extension chi-square test was used to assess the overall trend of OR across increasing tertiles of SII scores. We also computed OR and 95%CI, using multivariate logistic regression models, to assess the association between education, SES, income (independent variables of the SII score), and DM2.

We fit an SEM model (using the DM2 latent variable as continuous outcome) to study direct effects, indirect effects, and gender effects. We included BMI, PA, and diet-DII as mediator variables between SII and DM2 (figure 1). The method of estimation was Maximum Likelihood based on the covariance matrix. Global goodness of SEM fit indices included the chi-square statistic that tested the null hypothesis: that the reproduced covariance matrix has the specified structure, or the model fits the data. In addition, the Comparative Fit Index (CFI) is presented, which ranges between 0 and 1 (model fits). Other goodness of fit statistics were also examined, and included the Root Mean Square Error of Approximation (RMSEA), and Standardized Root Mean Square Residual (SRMR).22

SES: socioeconomic status

DII: dietary inflammatory index

BMI: body mass index

HbA1c : glycated he moglobin A1c

DM2: type 2 dia betes

Figure 1 Conceptual framework of the association between diabetes socioeconomic inequality index, body mass index and lifestyle factors 

Finally, SEM was used to study the relationship between socioeconomic inequalities and DM2 onset, mediated by PA, diet-DII, BMI, controlled by family history of DM2. This model was replicated for men and women to study the possible changes in the estimates by gender.


The main characteristics of the participants in the DMS-MC 2015 are presented in table I. While DM2 prevalence was similar between men (14.0%) and women (13.8%), they differed in other characteristics. Particularly, women were more likely to be obese (37.9 vs. 30.6%), less educated (21.9 vs. 16.7%), and had lower SES (25.2 vs. 18.4%). In relation to family history of DM2 and the presence of hypertension, 41.2% of women had a family history of DM2 and approximately 19.0% had hypertension.

table I Main characteristics of adults participating in the Diabetes Mellitus Survey of Mexico 

Characteristic Men (n = 456) Women (n = 718)
% SE % SE
Age (years)* 39.1 0.67 40.7 0.65
Socioeconomic status, %
Low 18.4 2.0 25.2 2.0
Medium 37.1 3.0 35.6 3.0
High 44.4 4.0 39.1 3.0
Income (US dollar/month)* 277.5 28.4 126.9 11.6
Education, %
Elementary and secondary education 16.7 1.9 21.9 1.8
High school 27.8 2.6 26.6 2.3
Bachelor’s degree or higher 55.5 2.9 51.5 2.7
Socioeconomic inequality index -0.22 0.05 -0.11 0.04
Smoking status, %
Current 60.0 2.6 32.0 2.3
Past 11.5 1.8 11.7 1.4
Never 28.5 2.3 56.3 2.4
Physical activity, %
Low 28.6 2.5 26.7 2.1
Moderate/intense 71.4 3.1 73.3 2.2
Family history of DM2, (% Yes) 38.0 2.5 41.2 2.2
Hypertension, (% Yes)‡ 12.3 1.4 18.9 2.0
Body mass index (kg/m2)* 28.1 0.33 29.1 0.31
BMI, %
Normal 26.0 2.9 25.0 2.2
Overweight 43.4 3.0 37.1 1.8
Obesity 30.6 2.8 37.9 2.3
Glucose (mg/dL)* 108.7 2.5 106.9 2.6
Glycated hemoglobin (HbA1c)* 5.9 0.08 5.9 0.09
DM score 5.6 2.6 3.8 2.6
DM2, (% Yes) 14.0 1.8 13.8 1.6
Triglycerides (mg/dL)* 230.8 9.9 181.8 6.4
HDL-C (mg/dL)* 40.0 0.68 44.4 0.47
Dietary variables
Energy intake (kcal/day)* 2479.3 60.1 2007.4 53.1
Carbohydrates (% energy)* 54.4 0.54 56.3 0.39
Total fats (% energy)* 30.9 0.45 31.2 0.29
Saturated fats (% energy)* 12.7 0.14 13.2 0.14
MUFA (% energy)* 10.9 0.17 10.8 0.17
PUFA (% energy)* 7. 0 0.11 7.0 0.11
Fiber (g/day)* 29.4 1.0 26.2 0.83
Magnesium (mg/day)* 441.9 13.2 371.1 10.9
DII score* -0.35 0.013 -0.85 0.08

*Mean and SE (standard error)

≥ 140/90 mmHg

BMI: body mass index [normal (<25.0 kg/m2), overweight (≥ 25.0 to <30.0 kg/m2), obesity (≥ 30.0 kg/m2)]

MUFA: monounsaturated fatty acids

PUFA: polyunsaturated fatty acids

DII: dietary inflammatory index

HDL-c: high density lipoprotein cholesterol

DM: diabetes mellitus

According to SII score tertiles, when we compared participants in the highest level vs. participants in the lowest level of SII, individuals in the highest level were significantly older (42.5 vs. 38.5 years), with higher prevalence of obesity (37.6 vs. 31.1%) and a higher prevalence of DM2 (20.6 vs. 9.3%). Concerning DII, participants in the highest level of SII had higher DII scores (0.51, SE:0.16), which means that subjects with the highest level of SII were more likely to have a pro-inflammatory diet (table II).

table II Characteristics of participants according to tertiles of socioeconomic inequality index: The Diabetes Mellitus Survey of Mexico City 

Characteristic Socioeconomic inequality index (SII)
Low SII Medium SII High SII
% SE % SE % SE
Age (years)* 38.5 0.81 40.1 0.82 42.5 0.76
Sex (women), % 48.2 2.8 50.4 2.5 59.2 2.6
Socioeconomic status, %
Low 6.4 2.0 16.7 2.3 42.4 2.9
Medium 14.6 3.0 29.2 2.9 35.2 3.0
High 79.0 3.1 54.1 3.0 22.4 3.2
Income (US dollar/month) 364.4 41.3 194.4 12.6 148.5 11.5
Education, %
Elementary and secondary education 8.4 2.0 19.8 2.5 39.4 2.7
High school 17.8 3.0 34.2 3.0 33.4 2.9
Bachelor’s degree or higher 73.8 3.0 46.0 2.8 27.2 2.9
Smoking status, %
Current 46.7 3.2 45.9 3.0 43.1 3.8
Past 10.4 1.9 12.4 2.0 12.1 2.3
Never 42.9 2.7 41.7 3.2 44.8 3.4
Physical activity, %
Low 36.3 2.9 20.5 2.2 22.2 2.3
Moderate/intense 63.7 3.5 79.5 2.3 77.8 2.9
Family history of DM2, (% Yes) 41.6 2.9 43.1 3.2 41.1 2.9
Hypertension, (% Yes)‡ 14.8 1.9 15.7 2.0 17.8 2.2
Body mass index (kg/m2)* 28.1 0.31 28.9 0.39 29.1 0.35
BMI, %
Normal 26.0 2.8 25.8 2.5 24.0 3.0
Overweight 42.9 2.9 38.4 3.0 38.4 3.4
Obesity 31.1 2.9 35.8 2.8 37.6 3.3
Glucose (mg/dL)* 100.6 1.9 107.9 2.7 122.7 5.3
Glycated hemoglobin (HbA1c)* 5.6 0.08 5.9 0.10 6.5 0.15
DM2, (% Yes) 9.3 1.6 13.9 2.1 20.6 2.3
DM score -11.9 0.19 -5.6 0.27 10.7 0.53
Triglycerides (mg/dL)* 198.5 8.9 203.8 9.2 221.9 12.7
HDL-c (mg/dL)* 43.2 0.82 41.3 0.62 42.1 0.85
DII score* -0.91 0.18 0.13 0.14 0.51 0.16

*Mean and SE (standard error

≥ 140/90 mmHg

BMI: body mass index [normal (<25.0 kg/m2), overweight (≥ 25.0 to <30.0 kg/m2), obesity (≥ 30.0 kg/m2)]

DII: dietary inflammatory index

HDL-c: high density lipoprotein cholesterol

DM: diabetes mellitus

After adjustment for potential confounders, like: age, sex, tobacco use, family history of DM2, presence of hypertension, alcohol intake, PA, BMI, and diet-DII, DM2 was positively associated with SES, income, and education. For example, participants in the lowest category of SES were more likely to be diabetic (OR=1.99, 95%CI: 1.27-3.12), compared to subjects in the highest level of SES. On the other hand, participants in the lowest category of income had higher odds of DM2(OR= 1.45; 95%CI: 1.01-2.30) compared with subjects in the highest tertile of income.23

Figure 1 evaluates a theoretical model in which SII and DM2 were used to predict each other in a structured manner. Among the total population (figure 2a), pathways linking SII to DM2 included a pathway through diet-DII, BMI, and PA, in addition the major and significant contribution of indirect effects from SII to DM2 onset was via BMI (ß=0.071,p<0.01). In general, SII was negatively associated with PA (ß= -0.066,p<0.05), which, in turn, was inversely associated with DM2 (ß= -0.045,p<0.05). Finally, SII had a direct effect on DM2 (ß= 0.174,p<0.001).

Among women (figure 2b), the direct effect of SII on DM2 was the major pathway explaining its total effect (ß=0.233,p<0.001). Lastly, a direct effect of SII on DM2 was observed among men (figure 2c) (ß= 0.146,p<0.001). Statistical differences for the models by gender were tested using Wald Test, as indicated in table III.

Figure 2 Findings of estructural equation models: mechanism explaining socioeconomic inequalities (SII), diabetes through physical activity, dietary inflamatory index (DII), body mass index (BMI), and family history of DM2 

table III Parameter estimates for SEM and fit index, for the general model and by gender 

Coeficiente Overall Model Men Women Wald Test
β1 SII”Diabetes 0.194*** 0.146*** 0.233*** 0.2724
β2 SII “BMI 0.109** -0.024NS 0.146** 0.0174
β3 SII “PA -0.071* 0.072NS -0.085* 0.0601
β4 SII “Diet-DII -0.050NS -0.031NS -0.107* 0.3138
β5 BMI”Diabetes 0.091** 0.056NS 0.080* 0.8726
β6 PA”Diabetes -0.056* -0.085* -0.056* 0.9361
β7 Diet-DII “Diabetes 0.159*** 0.171*** 0.171*** 0.7794
β8 FHD”Diabetes 0.187*** 0.146*** 0.4109
Adjustment indexes
CFI 0.971 0.919
RMSEA 0.54 0.074
SRMR 0.34 0.046
BIC 52 726.69

SEM: structural equation model

SII: socioeconomic inequalities

DII: dietary inflammatory index

BMI: body mass index

PA: physical activity

FHD: family history of diabetes

SES: socioeconomic status

CFI: Comparative Fit Index

RMSEA: Root Mean Square Error of Approximation

SRMR: Standardized Root Mean Square Residual

BIC: Bayesian information criteria

NS: Not statistically significant


Using representative data collected in Mexico City, we assessed, through a structural equation model approach, the complex association between DM2 and socioeconomic inequalities, mediated by the contribution of body mass index, physical activity, and diet. Our results show that a higher score of socioeconomic inequalities was related to a higher score of DM2. Additionally, we found that this association was partially mediated by BMI, diet-DII, and PA--variables frequently considered as modifiable health behaviors or characteristics. The relation between DM2 and socioeconomic inequalities observed in the present study has been reported in other populations previously.24,25,26,27For example, in a representative sample of the European population, Espelt and colleagues24found that women with a lower socioeconomic position (characterized by low education, low income, and low socioeconomic status) had 2.2 (95%CI: 1.9-2.7) higher prevalence ratio of DM2, while for men it was 1.6 (95%CI: 1.4-1.9).

In this sense, multiple possible mechanisms have been described in the relation between socioeconomic inequalities and health. These mechanisms, among others, include: unhealthy behaviors (poor diet, physical inactivity, alcohol intake, etc.), restricted access to health care and inadequate resources, higher psycho-social stress levels, and inadequacies in material circumstances.28

Potential mediator variables (BMI, diet-DII, PA) examined in our study have been previously discussed.11,24,29,30We found that obesity was an important risk factor in the association between socioeconomic inequalities and DM2. Our data suggests that individuals with higher SII had higher BMI values (ß = 0.109,p<0.01). This correlation is consistent with a previous review31where subjects with lower socioeconomic status, lower educational level, and lower income had a higher prevalence of obesity. The relevance of BMI as a risk factor for DM2 has been amply demonstrated.32,33Several biological mechanisms, including at the molecular level, have been proposed to explain the relation between BMI and DM2. For example, high BMI has been associated with higher plasma levels of free fatty acids which promote insulin resistance, abnormal lipid levels, and increased inflammation, all of which are predictors of incident DM2.32,33

We also found that subjects with higher SII score had lower levels of physical activity (ß = -0.066,p<0.05). In agreement with our findings, Bird and colleagues27found that participants with lower income had higher levels of physical inactivity. This finding could relate to fewer and less safe environments, and/or lack of infrastructure (sidewalks and parks) that would decrease the practice of leisure physical activities.34Finally, our analysis indicates that individuals with higher SII had higher levels of diet-DII (ß= 0.037). Though in our study the relationship was not statistically significant, previous studies have demonstrated that people with lower socioeconomic status and income (and related time restriction), have less access to healthy foods and consume less diverse diets with important differences in micronutrient intake and status.12,35,36

Notwithstanding important findings, some potential limitations need to be considered in the interpretation of our results. The design is cross-sectional in nature; therefore, it is not possible for this study to address causal sequencing. Additionally, we cannot ignore the possibility of reverse causation. Some important elements related to socioeconomic inequalities (such as neighborhood characteristics) were not evaluated in the first cycle of the DMS-MC. However, using a robust statistical methodology, we developed a SII that includes variables such as socioeconomic status, education, and income, that reflect certain conditions of the individual’s living environment.12,35,36Additionally, the DMS-MC 2015 is a self-reported survey and consequently may be susceptible to measurement error and recall bias. Finally, the findings of the present study, while pertinent to Mexico City, may not be generalizable. Hence, we should use caution if drawing conclusions at the national level.

In spite of these limitations, our study has an important number of strengths:

First, the random stratified cluster design of the DMS-MC 2015 and the representation of the sample of adults in Mexico City, which means that the data collected possess a better picture of the relative characteristics of the general population included in our study.

Second, the use of validated questionnaires; in this case, using questionnaires that have been previously validated in the population of interest decreases the probability of measurement error, and therefore, our conclusions may be less biased.

Third, stratified analysis was carried out to identify gender differences in patterns of association.

And fourth, the use of SEM to analyze the complex relation between SII and DM2. SEM are very relevant in exploratory studies, particularly when variables of interest cannot be measured perfectly, as with those involving behavioral/social issues and having complex interrelationships between variables.

To the best of our knowledge, this is the first study that evaluates, through SEM, the complex relationship between socioeconomic inequalities and DM2 in the Mexican population. In summary, it was found that a higher socioeconomic inequalities score was related to a higher DM2 score and that this association is partially mediated by BMI, diet-DII, and PA.


In Mexico, the increasing prevalence of DM2 as well as the widening of the socioeconomic gap between individuals has become a pressing and important issue. Although multiple studies, mainly in high income countries, have evaluated the relation between socioeconomic inequalities and DM2; the complex mechanisms of this relationship are not completely documented. On the other hand, as reported previously,30,31the efforts to identify a primary or single risk factor or mechanism has not resulted in the development of effective strategies to reduce the increasing prevalence of DM2 globally. Thus, schemes for DM2 prevention and management must address social, political, and economic determinants and also individual-level risk factors. In this context, assessing and predicting the multiple risk factors related to the progression of the disease (DM2 and other chronic diseases), in order to control and prevent it, modeling complex relationships is a crucial task of epidemiology and public health. Therefore, the employment of innovative epidemiological methodologies and statistical techniques (i.e., structural equation models, machine learning algorithms, among others) commonly used in marketing or social studies should be widely applied in health sciences.

Lastly, our results emphasis the need for public health programs and policies that specifically target disadvantaged populations. In this sense, the promotion and access to healthy behaviors-like places to exercise and healthy foods, healthy living conditions, and access to and use of health care services-may help in diminishing the link between socioeconomic inequalities and DM2.


Funding information: Unrestricted research grant obtained by SB from Novo Nordisk (number Project: 1295. Cities Changing Diabetes: Mexico City Representative Diabetes survey) and a NIH-Fogarty grant (NIH-Fogarty RO3 TW009061) to the National Institute of Public Health.


1. Zimmet PZ, Magliano DJ, Herman WH, Shaw JE. Diabetes: A 21st century challenge. Lancet Diabetes Endocrinol. 2014;2(1):56-64. [ Links ]

2. World Health Organization. Global report on diabetes. Geneva: WHO, 2016 [cited June 4, 2019]. Available from:Available from: ]

3. International Diabetes Federation. IDF Diabetes Atlas. 8th ed. Brussels, Belgium: IDF, 2017 [cited June 4, 2019]. Available from:Available from:http://www.diabetesatlas.orgLinks ]

4. Zimmet P, Alberti KG, Magliano DJ, Bennett PH. Diabetes mellitus statistics on prevalence and mortality: facts and fallacies. Nat Rev Endocrinol. 2016;12(10):616-22. [ Links ]

5. GBD 2013 Risk Factors Collaborators, Forouzanfar MH, Alexander L, Anderson HR, Bachman VF, Biryukov S,et al. Global, regional, and national comparative risk assessment of 79 behavioural, environmental and occupational, and metabolic risks or clusters of risks, 1990-2015: a systematic analysis for the Global Burden of Disease Study 2015. Lancet. 2016;388:1659-724. [ Links ]

6. Gómez-Dantés H, Fullman N, Lamadrid-Figueroa H, Cahuana-Hurtado L, Darney B, Ávila-Burgos L,et al. Dissonant health transition in the states of Mexico, 1990-2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet. 2016;388(10058):2386-402. [ Links ]

7. Hills AP, Arena R, Khunti K, Yajnik CS, Jayawardena R, Henry CJ,et al. Epidemiology and determinants of type 2 diabetes in south Asia. Lancet Diabetes Endocrinol. 2018;6(12):966-78. [ Links ]

8. Hwang J, Shon C. Relationship between socioeconomic status and type 2 diabetes: results from Korean National Health and Nutrition Examination Survey (KNHANES) 2010-2012. BMJ Open. 2014;4(8):e005710. [ Links ]

9. Robbins JM, Vaccarino V, Zhang H, Kasl SV. Socioeconomic status and type 2 diabetes in African American and Non-Hispanic white women and men: Evidence from the third National Health and Nutrition Examination Survey. Am J Public Health. 2001;91(1):76-83. [ Links ]

10. Stringhini S, Tabak AG, Akbaraly TN, Sabia S, Shipley MJ, Marmot MG,et al. Contribution of modifiable risk factors to social inequalities in type 2 diabetes in the Whitehall II cohort study. BMJ. 2012;345:e5452. [ Links ]

11. Dowler E. Inequalities in diet and physical activity in Europe. Public Health Nutr. 2001 ;4(2b):701-9. 160 [ Links ]

12. Whitehead M. The concepts and principles of equity and health. Int J Health Serv. 1992;22(3):429-45. [ Links ]

13. Wang Y, Beydoun MA. The obesity epidemic in the United States- gender, age, socioeconomic, racial/ethnic, and geographic characteristics: a systematic review and meta-regression analysis. Epidemiol Rev. 2007;29:6-28. [ Links ]

14. Denova-Gutiérrez E, Muñoz-Aguirre P, Shivappa N, Hébert JR, Tolentino-Mayo L, Batis C,et al. Dietary inflammatory index and type 2 Diabetes Mellitus in adults: The Diabetes Mellitus Survey of Mexico City. Nutrients. 2018;10(4)pii:E385. [ Links ]

15. Denova-Gutiérrez E, Ramírez-Silva I, Rodríguez-Ramírez S, Jiménez-Aguilar A, Shamah-Levy T, Rivera-Dommarco JA. Validity of a food frequency questionnaire to assess food intake in Mexican adolescent and adult population. Salud Publica Mex. 2016;58(6):617-28. [ Links ]

16. Shivappa N, Steck SE, Hurley TG, Hussey JR, Hébert JR. Designing and developing a literature-derived, population-based dietary inflammatory index. Public Health Nutr. 2014;17(8):1689-96. [ Links ]

17. Hernández-Ávila JE, González-Avilés L, Rosales-Mendoza E. Manual de Usuario. SNUT Sistema de Evaluación de Hábitos Nutricionales y Consumo de Nutrimentos. Cuernavaca, México: Instituto Nacional de Salud Pública, 2003. [ Links ]

18. Twetz N. Textbook of Clinical Chemistry. Philadelphia, PA, USA: WB Saunders, 1999. [ Links ]

19. American Diabetes Association. Standards of medical care in diabetes-2014. Diabetes Care. 2014;37(suppl 1):S14-80. [ Links ]

20. Alberti KG, Zimmet P, Shaw J. The metabolic syndrome: A new worldwide definition. Lancet. 2005;366(9491):1059-62. [ Links ]

21. Martinez SM, Ainsworth BE, Elder JP. A review of physical activity measures used among US Latinos: Guidelines for developing culturally appropriate measures. Ann Behav Med. 2008;36(2):195-207. [ Links ]

22. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling. 1999;6(1):1-55. [ Links ]

23. Denova-Gutiérrez E, Vargas-Chanes D, Hernández S, Muñoz-Aguirre P, Napier D, Barquera S. Supplementary Table 1. Multivariate logistic regression analysis for socioeconomic inequity index, socioeconomic status, income, and education and type 2 diabetes in adults participating in the Diabetes Mellitus Survey of Mexico. Dryad, 2020. Dataset. [ Links ]

24. Espelt A, Borrell C, Roskam AJ, Rodríguez-Sanz M, Stirbu I, Dalmau-Bueno A,et al. Socioeconomic inequalities in diabetes mellitus across Europe at the beginning of the 21st century. Diabetologia. 2008;51(11):1971-9. [ Links ]

25. Espelt A, Arriola L, Borrell C, Larrañaga I, Sandín M, Escolar-Pujolar A. Socioeconomic position and type 2 diabetesmellitus in Europe 1999-2009: a panorama of inequalities. Curr Diabetes Rev. 2011;7(3):148-58. [ Links ]

26. Agardh E, Allebeck P, Hallqvist J, Moradi T, Sidorchuk A. Type 2 diabetes incidence and socio-economic position: a systematic review and meta-analysis. Int J Epidemiol. 2011;40(3):804-18. [ Links ]

27. Bird Y, Lemstra M, Rogers M, Moraros J. The relationship between socioeconomic status/income and prevalence of diabetes and associated conditions: A cross-sectional population-based study in Saskatchewan, Canada. Int J Equity Health. 2015;14:93. [ Links ]

28. MacIntyre S. The black report and beyond: what are the issues? Soc Sci Med. 1997;44(6):723-45. [ Links ]

29. Demakakos P, Marmot M, Steptoe A. Socioeconomic position and the incidence of type 2 diabetes: the ELSA study. Eur J Epidemiol. 2012;27(5):367-78. [ Links ]

30. Espelt A, Kunst AE, Palència L, Gnavi R, Borrell C. Twenty years of socio-economic inequalities in type 2 diabetes mellitus prevalence in Spain, 1987-2006. Eur J Public Health. 2012;22(6):765-71. [ Links ]

31. Dinsa GD, Goryakin Y, Fumagalli E, Suhrcke M. Obesity and socioeconomic status in developing countries: a systematic review. Obes Rev. 2012;13(11):1067-79. [ Links ]

32. Boden G. Effects of free fatty acids (FFA) on glucose metabolism: significance for insulin resistance and type 2 diabetes. Exp Clin Endocrinol Diabetes. 2003;111(3):121-4. [ Links ]

33. Pradhan AD, Manson JE, Rifai N, Buring JE, Ridker PM. C-reactive protein, interleukin 6, and risk of developing type 2 diabetes mellitus. JAMA. 2001;286(3):327-34. [ Links ]

34. Grant J, Mackay K, Manuel P, McHugh TL. Barriers to optimizing investments in the built environment to reduce youth obesity: Policy-maker perspective. Can J Public Health. 2010;101(3):237-40. [ Links ]

35. Caraher M, Dixon P, Lang T, Carr-Hill R. Access to healthy foods: part 1. Barriers to accessing healthy foods: differentials by gender, social class, income, and mode of transport. Health Educ J. 1998;57(3):191-201. [ Links ]

36. Donking AJ, Dowler EA, Stevenson SJ, Turner SA. Mapping access to food in a deprived area: the development of price and availability indices. Public Health Nutr. 2000;3(1):31-8. [ Links ]

Received: August 12, 2019; Accepted: November 19, 2019

Corresponding author: Dr. Simón Barquera. Centro de Investigación en Nutrición y Salud, Instituto Nacional de Salud Pública. Av. Universidad 655, col. Santa María Ahuacatitlán. 62100 Cuernavaca, Morelos, Mexico. email:

Declaration of conflict of interests. The authors declare that they have no conflict of interests.

Creative Commons License This is an open-access article distributed under the terms of the Creative Commons Attribution License