The Trend in the Prevalence of Diabetes Mellitus in the Mexican Indigenous Population From 2000 to 2018

HIGHLIGHTS • Mexican indigenous people are more likely to be diabetic than nonindigenous people.• Since 2006, indigenous people have experienced an increase in obesity.• Since 2006, indigenous people have experienced an increase in abdominal obesity.• Indigenous students had lower education scores than nonindigenous students.

T a g g e d P Diabetes is a global public health problem; it has been estimated that approximately 463 million (9.3%) adults aged 20−79 years were living with diabetes worldwide in 2019, and this number is expected to increase to approximately 578 million (10.2%) by 2030 and 700 million (10.9%) by 2045.An estimated 4.2 million deaths among adults in the same age group are attributable to diabetes.Globally, diabetes is estimated to contribute to 11.3% of deaths. 1 T a g g e d E n d T a g g e d P It was recently reported that 10.3% of Mexican adults had been diagnosed with type 2 diabetes (T2D). 2 A significant number of people living with diabetes do not know their condition; data from the Encuesta Nacional de Salud y Nutrici on 2006 showed that 7.07% of Mexicans did not know that they had diabetes. 3Diabetes has been associated with heart failure, 4 rheumatic diseases, 5 musculoskeletal disorders, 6 increased risk of fracture, 7 and colorectal cancer, 8 among many other negative health-related outcomes.T a g g e d E n d T a g g e d P The increase in energy density of Mexican eating patterns 9 and sedentary lifestyles 10 have been identified as the most important causes of diabetes.Both causes are often associated with the biological inability to adapt to aspects of urban culture. 11T a g g e d E n d T a g g e d P Mexico has seen an increase in the number of metropolitan areas in recent years. 12However, there is still a significant percentage of the population living in areas with <2,500 inhabitants. 13The traditional areas where the indigenous population is concentrated are located in the most rugged areas of the country, with the most difficult access and deficient communication systems.Nevertheless, the indigenous population is present in almost all municipalities and in almost all the country's entities. 14In 2020, there were 7,364,645 indigenous people aged ≥3 years speaking an indigenous language in Mexico, representing 6% of the total population. 15T a g g e d E n d T a g g e d P Indigenous people, representing 5% of the world's population and 10% of the world's poorest people, continue to be disproportionately poorer, less educated, and in worse health conditions than nonindigenous groups. 16Disparities in Mexico's indigenous population in terms of basic and crucial development indicators have been shown; although development indicators have improved for the indigenous population, when we compare indigenous with nonindigenous people, the gap in socioeconomic and development indicators persists. 17T a g g e d E n d T a g g e d P In addition, it has been reported that the prevalence of diabetes by previous medical diagnosis has been higher in urban localities than in rural localities; however, the difference has shortened over time, reported to be 9.5% vs 9.2% in 2016, respectively. 18It is noteworthy that a high rural component is located in the south of Mexico and is the region where many indigenous language speakers are concentrated. 19T a g g e d E n d T a g g e d P This study aims to estimate the trends in the prevalence of diabetes from 2000 to 2018 in the Mexican indigenous language−speaking population and to analyze the main sociodemographic and clinical characteristics of this population group.We hypothesized that in recent years, the change in the prevalence of diagnosed diabetes in the indigenous population is greater than in the nonindigenous population.T a g g e d E n d  2012, and 2018.All of these are national, cross-sectional, and population-based surveys.For this report, we included the information of people aged ≥20 years with a complete questionnaire and biochemical analyses.All surveys were stratified, and probability sampling was used with national representation.We used the linguistic perspective as a proxy for indigenism.We classified an individual as indigenous if he or she responded affirmatively to the question, (NOMBRE) habla alguna lengua indígena (dialecto)?Does she/he speak any indigenous language?, making a distinction between people who speak an indigenous language and those who speak only Spanish (no/yes).T a g g e d E n d T a g g e d H 2 MeasuresT a g g e d E n d T a g g e d P We considered that an individual had a previous diagnosis of diabetes if he/she answered yes to the questionnaire question, Has any doctor told you that you have diabetes?
We considered an individual to have undiagnosed diabetes if they answered NO to the previous question and had a fasting serum glucose ≥126 mg/dL with at least 8 hours of fasting or had HbA1c levels ≥6.5%.All measurements were performed under standard procedures.T a g g e d E n d T a g g e d P and South (Campeche, Chiapas, Guerrero, Hidalgo, Oaxaca, Puebla, Quintana Roo, Tabasco, Tlaxcala, Veracruz, Yucat an).Finally, regarding the urbanicity of the area of residence, 2 categories were considered: rural (<2,500 inhabitants) and urban (≥ 2,500 inhabitants). 13T a g g e d E n d T a g g e d P A descriptive analysis was performed to present an overview of the disease in this population, including people with diabetes (both previously diagnosed by a physician and those who were not).Table 2 shows the distribution (frequencies and 95% CIs) of diabetes by health characteristics stratified by the indigeneity group for each survey.We estimated the age at diabetes onset by subtracting the time of diabetes diagnosis from the current age.The former information was obtained from the questionnaire question, How long ago were you told you had diabetes?T a g g e d E n d T a g g e d P The mean weight in kilograms, stature in meters, and waist circumference in centimeters by sex were reported.BMI was calculated as a person's weight in kilograms divided by the square of the person's height in meters (kg/ m 2 ).Then, it was categorized using the WHO criteria (BMI: underweight, <18.5 kg/m 2 ; normal, 18.5−24.9kg/ m 2 ; overweight, 25−29.9kg/m 2 ; and obese, ≥30 kg/m 2 ). 20nderweight and normal weight categories were grouped for analysis.Frequency for each BMI category was reported.T a g g e d E n d T a g g e d P We used the cut off point of waist circumference >102 cm for men and >88 cm for women to define abdominal obesity and then estimated its proportion.To estimate the frequency of hypertriglyceridemia, hypertension, chronic kidney disease, previous stroke, chronic heart failure, cerebral vascular disease, and depression, we considered the questionnaire question that inquired about the presence of this condition in the past or present, depending on the event asked.T a g g e d E n d T a g g e d P Finally, participants were considered current smokers if they answered affirmatively to one of the questions in the questionnaire: do you currently smoke tobacco every day?Do you currently smoke tobacco some days?They were considered current drinkers if they answered affirmatively to the question, do you currently drink?T a g g e d E n d T a g g e d H 2 Statistical AnalysisT a g g e d E n d T a g g e d P To describe the study population, nonindigenous and indigenous, according to socioeconomic characteristics, we obtained means and percentages with their CIs, considering the variable type (quantitative or qualitative).Change in the prevalence of diabetes (i.e., diagnosed, undiagnosed, and total diabetes) by indigenous group, national level, and year of the survey was obtained with the coefficient from a logistic regression model.We also calculated the prevalence adjusted for age using the world population in 2010 by the direct method. 21T a g g e d E n d T a g g e d P In addition, we presented health conditions related to diabetes and lifestyle habits with means and percentages with CIs by the indigenous/nonindigenous group.Finally, to estimate diabetes prevalence trends from 2000 to 2018 for each indigenous/nonindigenous group, we used multiple logistic regression models to adjust the risk of diabetes by year of the survey.Both models were adjusted for sex, age, waist circumference, and urbanicity of the area of residence.In the same way, we presented a graph with the predicted probabilities of having diabetes by indigenous group.Previously, bivariate models were constructed; the diagnosis of diabetes and the year of the survey were fixed, and then each variable was included one by one in the models to identify possible confounding factors.The variables considered were those with p<0.20 in the bivariate analysis.T a g g e d E n d T a g g e d P The goodness of fit of the models was assessed with the Hosmer−Lemeshow test.We considered a significance level of p<0.05 and a 95% CI in all cases.For all analyses, we considered the effect of the complex sample design, using the survey module ( svy T a g g e d P This study included 73,744 participants (45,021; 5,829;  9,732; and 13,162 for 2000, 2006, 2012, and 2018  National Surveys considered, respectively).This sample represented 247.2 million people aged ≥20 years for the whole period.The indigenous stratum had 6,576 participants, representing 16.0 million Mexican adults.T a g g e d E n d T a g g e d P The nonindigenous group had a lower frequency of adults aged ≥60 years than the indigenous group for all surveys (12.1% vs 14.2%, 14.7% vs 19.2%, 15.0% vs  19.1%, and 19.8% vs 23.6% for 2000, 2006, 2012, and  2018 surveys, respectively) (Table 1).T a g g e d E n d T a g g e d P Regarding educational level, for both the nonindigenous and indigenous groups, the category of primary or less showed the highest frequencies; however, there was a wide difference between them (44.9% vs 77.3%, 47.8% vs 82.1%, 39.9% vs 75.5%, and 29.3% vs 65.0% for 2000,  2006, 2012, and 2018 surveys, respectively).T a g g e d E n d T a g g e d P For the nonindigenous group, the third tertile of socioeconomic level reported the highest frequency T a g g e d E n d  1).T a g g e d E n d T a g g e d P Table 2 reports the change in the prevalence of diagnosed diabetes, undiagnosed diabetes, and total diabetes for the nonindigenous, indigenous, and the entire population.A positive and significant prevalence change was found for diagnosed diabetes for the whole population and for the nonindigenous and indigenous groups (3.5%, 3.3%, and 6.4%, respectively).In the case of undiagnosed diabetes, a significant negative percentage change was observed for the nonindigenous group and for the total population (−2.8% and −2.2%, respectively).In contrast, for the indigenous group, it was significantly positive (7.7%).Data from the National Health Survey 2000 survey were not available to estimate the prevalence of undiagnosed diabetes.T a g g e d E n d T a g g e d P In Table 3, a total of 8,992 participants living with diabetes were considered: 2,940 (2,778+162), 2,252 (2,067 +185), 1,415 (1,258+157), and 2,385 (2,155+230) for the 2000, 2006, 2012, and 2018 National Surveys, respectively.This represented 28.4 million people with diabetes aged ≥20 years for the whole period.The indigenous stratum had 734 participants in the period, representing 1.9 million indigenous adults with diabetes.T a g g e d E n d T a g g e d P In the case of participants with previously diagnosed diabetes, the age of diabetes onset was <50 years in both groups.The mean time of diabetes diagnosis was slightly greater than 8 years.For the remaining variables, total diabetes was considered.The obesity category showed a significant increasing trend for both groups; however, in the indigenous group, this was higher (13.2%, 40.2%, 37.8%, and 52.1% for 2000, 2006, 2012, and 2018 surveys,  respectively).A similar result was found for abdominal obesity (53.25%, 54.8%, 66.9%, and 65.7% for 2000, 2006, 2012, and 2018 surveys, respectively) (Table 3).T a g g e d E n d T a g g e d P We found ORs for diabetes trends adjusted for each sociodemographic characteristic or anthropometrics stratified by indigenous group (Appendix, available online); these trends (ORs) were similar when they were simultaneously adjusted for sex, age, waist circumference, and urbanicity.In the indigenous group, we found an increased risk of a person living with diabetes in recent years compared with in the 2006 survey year (OR=1.77,95% CI=1.03, 3.03 for 2012 and OR=2.22 95% CI=1.35, 3.66 for the 2018 survey years).In contrast, for the nonindigenous group, we found lower risk (OR=0.7395% CI=0.65, 0.88 for 2012 and OR=0.82 95% CI=0.70, 0.97 for 2018) (Table 4).In addition, the probability of having diabetes from 2006 to 2018 by the indigenous group was observed; a positive trend was found in the indigenous group, whereas a flat trend was observed in the nonindigenous group (Figure 1).T a g g e d E n d T a g g e d P The term diabetes describes a group of metabolic disorders characterized and identified by the presence of hyperglycemia in the absence of treatment.The long-term specific effects of diabetes include retinopathy, nephropathy, and neuropathy, among other complications. 26,27Since the beginning of this century, mortality from this cause has been among the top 3 in Mexico; for example, in 2020, the leading cause of death was heart disease, with 220,000 deaths, whereas there were just over 150,000 deaths from diabetes (the third leading cause). 28T a g g e d E n d T a g g e d P Although the total prevalence of diabetes has remained stable during the last few years, it is possible to observe that undiagnosed diabetes has decreased during the study period.Both diagnosed and undiagnosed diabetes continue to represent a challenge for the health system.T a g g e d E n d T a g g e d P This study is the first to analyze the trend in the prevalence of diabetes among the Mexican indigenous population.We found an increased risk of having diabetes in this group as the survey year increased.In contrast, among the nonindigenous group, we did not find an increased risk, which indicates the differential behavior of the disease between indigenous and nonindigenous groups.T2D is associated with several risk factors, including genetics and the interaction between genetics and environmental factors.On the basis of the findings of previous studies, some of the variants associated with the presentation of T2D in the Mexican mestizo population have been identified.Some of these are related to insulin resistance and insulin secretion.However, the variants found do not explain the high prevalence in the population.S anchez-Pozos and Menjívar 29 recommend carrying out additional studies to find new genetic markers for predicting diseases as well as new therapeutic targets.In an analysis of samples from 11 indigenous T a g g e d E n d groups, Granados-Silvestre et al. 30 identified the T130I polymorphism in the HNF4A gene.This polymorphism is associated with the presence of elevated levels of triglycerides and with the early onset of T2D in mestizo Mexicans.In addition, the high frequency of the T130I polymorphism in Mexican indigenous populations reveals a diabetogenic background, which contributes to the high prevalence of T2D in the Mexican population. 30T a g g e d E n d T a g g e d P In a systematic review, Esparza-Romero and colleagues 31 reported the prevalence of diabetes mellitus in different indigenous groups in Mexico.The highest prevalence was reported among Mixtecs from Baja California (26.2%) and Yaquis from Sonora (18.3%), and the lowest was reported in Tepehuanos (0.83%) and Mazatecs (2.0%). 31Other researchers have obtained a high prevalence of diabetes in indigenous populations of Mexico, such as in San Quintin, Baja California (21.8%) (diagnosed and undiagnosed diabetes included). 32T a g g e d E n d T a g g e d P Between November 2012 and October 2017, 2,596 Mexican Amerindians from 73 indigenous communities of 60 different ethnic groups participated in a cross-sectional study at Instituto Nacional de Medicina Gen omica.This research only included people who identified themselves as indigenous, had parents and grandparents born in the same community, and spoke the native language.Mendoza-Caamal et al. 33 found that a prevalence of T2D previously diagnosed was 13.7% (95% CI=12.4,15.1), was 14.3% (95% CI=12.7,16) in females, and was 12.3% (95% CI=10.1, 14.8) in males.The overall prevalence of elevated fasting glucose (≥100 mg/dL or previous diagnosis of T2D) was 27.9% (95% CI=26.1, 29.6%), with an increase with age-up to 40% in adults aged from 41 years to 70 years-and no difference by sex. 33The differences obtained in the previously listed studies were also due to the definition used to determine a person as indigenous-self-identified as indigenous or speaker of an indigenous languageand the definition of diabetes, with diagnosis and without a diagnosis, that is, by the blood glucose result from a blood sample, which were not included by all the studies.T a g g e d E n d T a g g e d P By contrast, the correlation between education and health is well established and well studied; some studies have even estimated the causal effect of education on health. 34Because of this, we highlighted that in our study, the indigenous group continues to have the lowest education levels.Although this percentage has decreased by about 10 percentage points during the study period, more is T a g g e d E n d  36 However, in our analysis, this association could not be established.We report a lower prevalence of diabetes in the early years of the study among the poorest people, which is the indigenous group.Interestingly, for the last year of the survey, the prevalence in this group was higher than that of the nonindigenous group, suggesting that factors other than SES could affect the prevalence of diabetes more.T a g g e d E n d T a g g e d P According to our results, the trend of obesity among the indigenous group in Mexico has significantly increased from 13.2% in 2000 to 52.1% in 2018; this difference represents about 40 percentage points in 18 years.Although the trend increased in the nonindigenous group, the difference in the same period is about 10 percentage points, going from 34.7% in 2000 to 44.3% in 2018.This trend is consistent with the annualized average of 2.3 percentage points, suggesting that the distribution has shifted to the right with a total increase of 42.2% over the 2000−2018 period. 37The percentage change obtained between surveys for diabetes prevalence for the indigenous group was 6.7% (2.6, 11.0) and 0.1% (−1.1, 1.4) for the nonindigenous group.Nationally, the percentage change was 0.5% (−0.7, 1.8).Only the percentage for the indigenous group was statistically significant.T a g g e d E n d T a g g e d P By contrast, more than half of the nonindigenous group presented abdominal obesity during the entire study period.This result was superior to that of Villalta et al. 38 in a cross-sectional study of 195 indigenous women aged over 45 years; they found that the most frequent component of metabolic syndrome (30%) was abdominal circumference.On the basis of the analysis of another national survey, the first wave (2002) of the Mexican Family Life Survey, a longitudinal study of Mexican households and communities, Stoddard and colleagues 39 found that indigenous adults had significantly lower odds of obesity (OR=0.58,95% CI=0.49, 0.68) and diabetes (OR=0.59,95% CI=0.40, 0.68) than nonindigenous adults, when indigenous status was defined as the ability to speak an indigenous language, after adjusting for lifestyle indicators.In this case, diabetes was defined only for diagnosed diabetes. 39T a g g e d E n d T a g g e d P Some authors stated that acculturated societies are entering a dangerous era with rapidly increasing rates of obesity in populations leading to rising levels of T2D 40,41 ; in this sense, Mexico has experienced a change in eating habits in recent decades.Groenendael stated that the most critical change had been the shift from a traditional rural diet of corn and beans to a highly commercialized and industrialized diet under the influence of changing economic models.This change is also observed in rural areas of Mexico,42 where the indigenous population is mainly found.Consumption of ultraprocessed food and sugarsweetened beverages has been linked to an increased risk of obesity and other metabolic disorders. 43,44In a study conducted in communities in Mexico with a predominantly indigenous population, the high consumption of tortillas (15−20 tortillas per day) and the daily, social, and sumptuary consumption of soft drinks and beer have been associated with excessive caloric intake. 45T a g g e d E n d T a g g e d P Differences in the health lifestyles of indigenous people may be associated with a higher prevalence of obesity; therefore, it is possible to suggest that this change in body composition could be related to the increased prevalence of diabetes.Soto-Estrada et al. 9 suggested that the steady increase in the mortality rate from T2D from 1990 to 2015 coincides with the increase in the energy density of Mexican dietary patterns from 1961 to 2013.Different dietary patterns are one of several factors involved, some of which could be related to lifestyle changes.T a g g e d E n d T a g g e d P Although this study did not examine diabetes control, Cruz-S anchez and Cruz-Arceo's findings deserve attention.In an indigenous community in Tabasco, they conducted qualitative research about diabetes.The researchers found that women have access to the health service once a month to receive health education and to control chronic diseases to maintain social inclusion programs.They are diagnosed with diabetes when they already have symptoms or complications.In some cases, peritoneal dialysis or insulin was applied when the damage had progressed to a very advanced level.The abandonment of medical treatment can result in blindness or death in the indigenous population.There is a greater trust in herbal medicine among this group.Diabetes selfcare requires daily glucose monitoring, preferably in the morning, as part of self-care.There was no equipment available to any of the study participants.Every month, Study PopulationT a g g e d E n d T a g g e d P We used data set from the National Health Survey 2000 and the National Health and Nutrition Surveys 2006, ) of the Stata software, Version 15 (Stata Corp, College Station, Texas).T a g g e d E n d T a g g e d H 2 Ethical ConsiderationsT a g g e d E n d T a g g e d P All surveys considered in this report were approved by the Ethics, Research, and Biosafety Commissions of the National Institute of Public Health.Participants signed an informed consent form after the survey procedures were explained to them. 22−25 T a g g e d E n d

Table 1
Table 1.Population According to Sociodemographic Characteristics in Mexican Adults From 2000 to 2018, by Indigenous Group

Table 3 .
Health Characteristics in Mexican Adults With Diabetes by Indigenous Group and Year of Survey T a g g e d E n d Castro-Porras et al / AJPM Focus 2023;2(2):100087

Table 4 .
The Adjusted Risk of Diabetes by Year of the Survey Shown are models for risk of diabetes by year of survey, stratified by indigenous group and adjusted for sex, age, waist circumference, and urbanicity.An inverse association between SES and type 2 prevalence had previously been claimed.
Note:T a g g e d F i g u r e Figure 1.Predicted probabilities by indigenous/nonindigenous group.The models were adjusted for sex, age, waist circumference, and urbanicity.