Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Relationship between marital status and incidence of type 2 diabetes mellitus in a Brazilian rural population: The Baependi Heart Study

  • Camila Maciel de Oliveira ,

    Contributed equally to this work with: Camila Maciel de Oliveira, Luciane Viater Tureck

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration

    cmaciel@mit.edu (CMO); alexandre.pereira@incor.usp.br (ACP)

    Affiliations Laboratory of Genetics and Molecular Cardiology, Heart Institute (InCor), University of São Paulo Medical School, São Paulo, Brazil, Department of Integrative Medicine, Federal University of Parana, Curitiba, Brazil, Global CoCreation Lab, Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Cambridge, MA, United States of America

  • Luciane Viater Tureck ,

    Contributed equally to this work with: Camila Maciel de Oliveira, Luciane Viater Tureck

    Roles Formal analysis, Writing – original draft, Writing – review & editing

    Affiliation Department of Genetics, Federal University of Parana, Curitiba, Brazil

  • Danilo Alvares,

    Roles Data curation, Formal analysis, Writing – original draft

    Affiliation Department of Statistics, Pontificia Universidad Católica de Chile, Santiago, Chile

  • Chunyu Liu,

    Roles Formal analysis, Writing – original draft, Writing – review & editing

    Affiliations Framingham Heart Study, Framingham, MA, United States of America, Department of Biostatistics, Boston University, Boston, MA, United States of America

  • Andrea Roseli Vançan Russo Horimoto,

    Roles Conceptualization, Data curation, Investigation

    Affiliation Laboratory of Genetics and Molecular Cardiology, Heart Institute (InCor), University of São Paulo Medical School, São Paulo, Brazil

  • Mercedes Balcells,

    Roles Writing – review & editing

    Affiliations Global CoCreation Lab, Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Cambridge, MA, United States of America, Bioengineering Department, Institut Quimic de Sarria, Ramon Llull Univ, Barcelona, Spain

  • Rafael de Oliveira Alvim,

    Roles Conceptualization, Data curation, Writing – original draft

    Affiliations Laboratory of Genetics and Molecular Cardiology, Heart Institute (InCor), University of São Paulo Medical School, São Paulo, Brazil, Department of Physiological Sciences, Federal University of Amazonas, Manaus, Brazil

  • José Eduardo Krieger,

    Roles Conceptualization, Data curation, Funding acquisition, Project administration

    Affiliation Laboratory of Genetics and Molecular Cardiology, Heart Institute (InCor), University of São Paulo Medical School, São Paulo, Brazil

  • Alexandre Costa Pereira

    Roles Conceptualization, Data curation, Funding acquisition, Methodology, Project administration, Writing – original draft

    cmaciel@mit.edu (CMO); alexandre.pereira@incor.usp.br (ACP)

    Affiliation Laboratory of Genetics and Molecular Cardiology, Heart Institute (InCor), University of São Paulo Medical School, São Paulo, Brazil

Abstract

Many factors influence the incidence of type 2 diabetes mellitus (T2DM). Here, we investigated the associations between socio-demographic characteristics and familial history with the 5-year incidence of T2DM in a family-based study conducted in Brazil. T2DM was defined as baseline fasting blood glucose ≥ 126 mg/dL or the use of any hypoglycaemic drug. We excluded individuals with T2DM at baseline or if they did not attend two examination cycles. After exclusions, we evaluated a sample of 1,125 participants, part of the Baependi Heart Study (BHS). Mixed-effects logistic regression models were used to assess T2DM incident given different characteristics. At the 5-year follow-up, the incidence of T2DM was 6.7% (7.2% men and 6.3% women). After adjusting for age, sex, and education status, the model that combined marital and occupation status, skin color, and familial history of T2DM provided the best prediction for T2DM incidence. Only marital status was independently associated with T2DM incidence. Individuals that remained married, despite having significantly increased their weight, were significantly less likely to develop diabetes than their divorced counterparts.

Introduction

Type 2 diabetes mellitus (T2DM) is a multifactorial metabolic disease characterized by the development of insulin resistance and, subsequently, the loss of b-cell function. The worldwide prevalence was 30 million 70 years ago and 108 million 35 years ago [1,2]. It is known that T2DM is rising faster in low-income and middle-income than in high-income countries [1,2]. Brazil has risen from seventh in 1980 to fourth in 2014 in the worldwide country rank of diabetes prevalence (from 2.7 to 11.7 million adults with diabetes) [1].

There are important limitations in generalizing determinants of T2DM incidence from different populations [17]. This fact is partially explained by differences in obesity rates, lifestyle, health system resources, and access to medications for preventing the disease [3,8]. The association between marital status and various diseases has been investigated. Especially for T2DM, while some results have highlighted the beneficial effect of marriage [911], a poor marital quality may be a unique risk factor in men [12] or being widowed has been associated with a lower risk in women [13]. Moreover, marriage patterns have changed in the last years: people get married later and less often than in the past, there are more divorces and gender roles in a marriage have changed [14], all of which could modify these relationships.

This study aimed to identify the relative importance of socio-demographic variables, in particular marriage status, associated with T2DM incidence in a Brazilian sample from a rural area, after a 5-year follow-up period.

Materials and methods

Study population

The Baependi Heart Study (BHS) is a Brazilian cohort that seeks to investigate cardiovascular risk factors and other non-communicable diseases, including both genders aged 18 years old or above. At baseline (cycle 1 from 2005 to 2006), 1,695 individuals in 95 families were recruited in Baependi (19,117 inhabitants, 752 km2) located in Minas Gerais State, Brazil [15]. Five years later (cycle 2 from 2010 to 2013), 2,495 individuals from 125 families were evaluated [16]. At each examination cycle, socio-demographic, behaviour, medical history, and physical characteristics were assessed by a standardized protocol. A trained staff collected socioeconomic and clinical data, and all participants were examined in the same research center [15,16].

Of those 2,495 individuals at cycle 2, 1,341 individuals were the same assessed at cycle 1; thus, 354 participants were lost during the follow-up period or died, and 800 were new participants assessed only at cycle 2.

For this study, we carried out the analysis in individuals who attended both examination cycles (n = 1,341). Participants who had some missing data (n = 84, cycle 1; n = 45, cycle 2) were excluded. Individuals with fasting blood glucose ≥ 126 mg/dL or individuals that used hypoglycaemic medications in cycle 1 (n = 87) were also excluded. After exclusions, data on 1,125 diabetes-free individuals in cycle 1 were used to access T2DM incident in cycle 2.

The study protocol was approved by the ethics committee of the Hospital das Clínicas (SDC: 3485/10/074), University of São Paulo, Brazil, and each individual provided informed written consent before participation.

Sample characteristics

Socio-demographic characteristics included education, marital and occupation status, income, and skin color/race. Those were assessed via interviews using a standardized questionnaire. Education status included four categories: 1) illiterate or never attended school despite reading and writing or attended school for 1 to 4 years; 2) attended school for 5 to 8 years (incomplete or completed primary schooling); 3) attended school for 9 to 11 (incomplete or completed secondary schooling); 4) attended school for more than 11 years or finished university. For analysis, we grouped education into low (categories 1 and 2) or high (categories 3 and 4) levels. Marital status was defined as 1) married, 2) single, and 3) divorced/widower. Occupation status was categorized as 1) employed or retired and 2) unemployed. Since income was very homogeneous in this sample (about 80% of the sample were in the same range of 250–500 dollars/month), we only included occupation status in our analysis. Skin color/race was self-reported (white, brown, black, and indigenous) and stratified into white and non-white for the current analysis.

Social behaviour was also assessed. Smoking status was dichotomized into current/former smokers or never smokers. Alcohol consumption was defined as never drinkers versus current or former drinkers.

Clinical and laboratorial characteristics

Body mass index (BMI) was calculated as body weight (kg) divided by height squared (m2). BMI was categorised as normal weight (< 25kg/m2), overweight (25 kg/m2 to 29.9 kg/m2) and obesity (≥ 30 kg/m2). Waist circumference was measured half-way between the lowest rib and the iliac crest while the subject was at minimal respiration. Blood pressure (BP) was measured using a standard digital sphygmomanometer (OMRON, model HEM-741CINT) on the left-arm after 5 minutes of rest in the sitting position. Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were calculated from the mean value of three readings.

Hypertension status was defined by the presence of SBP ≥ 140mmHg or DBP ≥ 90mmHg or by the use of antihypertensive medications. Dyslipidaemia treatment was defined by the use of lipid-lowering drugs. Diabetes mellitus was defined as fasting blood glucose ≥ 126 mg/dL or use of hypoglycaemic medications.

Blood collection was standardized, and laboratory assays were conducted in the same clinical chemistry laboratory. The fasting status was declared by the participants at the time of blood collection and the duration of 12 hours was requested.

Statistical analysis

The incidence of T2DM was assessed after a 5-year follow-up of individuals free of the disease at baseline (n = 1,125 participants).

For the descriptive analysis, categorical variables are presented as percentages and only age is summarised as the mean ± standard deviation (SD). The comparisons of categorical covariates were performed by the Chi-square test, and the means (age) were compared by the Student’s t-test.

Mixed-effects logistic regression models were used to assess the incidence of T2DM adjusting for different characteristics and family (as a cluster variable). The choice to use logistic regression models instead of Cox proportional hazards models was based on the fact that our study included only two visits with the same time interval for all participants [17]. All analyses were corrected for age and sex. Exploratory analyses—sensitivity analyses and a model for diabetes incidence adjusted for BMI change, were conducted post-hoc after identification of marital status as the main socio-demographic predictor for diabetes incidence, in order to search for changes in marital status during a 5-year follow-up and the interaction between sex and BMI changes. All analyses were performed using R version 3.4.2.

Results

General characteristics in the Baependi Heart Study

From Table 1, we can see that 57% of participants were women. Approximately 76% of all individuals reported themselves as white and 30% had a familial history of T2DM. More men than women were single (31% vs 25%), smoker (20% vs 12%) and had an occupation with own income (87% vs 57%). At baseline, the mean age was similar for both sexes (41 ± 15 years for women and 43 ± 17 years for men).

thumbnail
Table 1. Socio-demographic and clinical characteristics at baseline and 5-year follow-up in the Baependi Heart Study cohort.

https://doi.org/10.1371/journal.pone.0236869.t001

Obesity and altered waist circumference increased over time in both sexes. Dyslipidaemia medication use increased approximately 3-fold in both sexes, and hypertension medication use increased by almost 100% in men (15% to 28%).

Type 2 diabetes mellitus status according to socio-demographic characteristics

The incidence of T2DM was 6.7% in the general sample (75 T2DM in 1125 participants) over 5 years, and there was no significant difference based on sex (7.2% for men and 6.3% for women) (p = 0.63). Based on age groups, the T2DM incidence was 6% (< 29 years); 5.1% (30 to 39 years); 6.4% (40 to 49 years); 6.8% (50 to 59 years); 14% (60 to 69 years); 7% (> 70 years). In the BHS sample, the rate of undiagnosed cases was 30%.

The incidence of T2DM was also analysed according to socio-demographic variables (Table 2). T2DM was more frequent in individuals with high education status, divorced or widower.

thumbnail
Table 2. Socio-demographic characteristics according to T2DM status after the 5-year follow-up in the Baependi Heart Study cohort.

https://doi.org/10.1371/journal.pone.0236869.t002

The only socio-demographic variable independently associated with increased odds of presenting diabetes was marital status (Table 3). In our sample, 13% of divorced, 6% of married and 6% of single individuals developed T2DM. After adjusting these estimates for age and sex, being married was associated with a 0.39 odds of developing diabetes; being single was associated with an odds of 0.33 of developing diabetes.

Further investigating this relationship, at baseline, there was no difference in the glucose levels between these three marital status groups, as well as between married and divorced/widower regarding BMI (p value = 0.86), nor between single and divorced/widower (p value = 0.12). In addition, adding baseline BMI to a model predicting diabetes incidence did not significantly change the estimated effect size of marital status suggesting that the observed association is not being mediated by baseline BMI (for being married odds changed from 0.39 to 0.38; for being single odds remained in 0.33).

Nonetheless, the observed association could be mediated by changes in the marital status between baseline and 5-years follow-up. Comparing baseline and 5-years marital status, 63% of individuals remained in their baseline marital status. From the 37% that changed their marital status, the majority of changes occurred in single individuals that have married in the last 5 years (38% of those who changed marital status).

A sensitivity analysis using only individuals that remained in their baseline marital status showed that the estimated effect sizes for being married and being single did not change.

Another analysis, based on a model for diabetes incidence, adjusted for BMI change in addition to all previous potential confounders, showed that the BMI change was highly associated with increased odds of developing diabetes (p value = 0.01), however, its addition did not change the estimated effect size of baseline marital status (OR of 0.39 for being married and 0.31 for being single). Only those that married (p value = 0.0001) or remained married (p value = 0.01) presented significant changes towards increased BMI after 5 years. Despite the increased weight gain, individuals from these groups were still significantly less likely to develop diabetes than divorced/widower individuals (Fig 1).

thumbnail
Fig 1. Predicted values of BMI changes and predicted probabilities of T2DM incidence.

BMI: Body Mass Index; T2DM: Diabetes Mellitus type 2.

https://doi.org/10.1371/journal.pone.0236869.g001

Discussion

This is one of the first studies describing T2DM incidence in a sample from a rural city in Brazil. The associations between socioeconomic factors and T2DM occurrence were investigated, and it was possible to identify, among sociodemographic variables, the independent effect of marital status on T2DM incidence.

Baependi is a small city, whose economy is based on family farming. The role of women is predominantly linked to family care. The general characteristic of this sample reflects a typical rural Brazilian population: men had a higher frequency of being single, smoking and a higher rate of employment. It is known that smoking rates are similar between women and men in high-income countries, but the sex difference increases as the country’s income becomes lower [18], as demonstrated in our sample.

We observed a higher prevalence of overweight/obesity among women at baseline, followed by a marked increase in men over the next five years. This seems to have an important relationship with the lifestyle of the population, where in general, women execute the more sedentary activities compared to men, leading to the early onset of overweight/obesity in women. In contrast, men showed a marked decline in their metabolic health later, possibly when they approached the age to retire from rural work. Although this observation is somewhat predictable, it shows the particularities that should be considered in this kind of study, and it may indicate practices for obesity and hypertension prevention, aimed to specific sex and age ranges, that will be more effective since our results are quite different from others who investigated these relationships in urban populations.

Regarding diabetes incidence, in a cross-sectional study, Iser and collaborators found a 6.3% prevalence of self-reported diabetes (5.9% men vs 6.6% women) for the combined population of capitals of Brazil [19]. However, to the best of our knowledge, incidence data for T2DM is still missing for the Brazilian rural population.

The Framingham Heart Study examined T2DM incidence over 8 years within three distinct periods [20]. The age-adjusted 8-year incidence rate of diabetes was higher among men in the 1970s (3.4% vs 2.6%), 1980s (3.6% vs 3.0%) and 1990s (5.8% vs 3.7%) [16]. In the BHS sample, although, there was no statistical difference, the incidence rate of diabetes was also higher in men (7.2% vs 6.3%).

Previous studies have assessed the association between T2DM and socioeconomical factors in the Brazilian population. In an expressive Brazilian sample, in which the prevalence of self-reported T2DM was 7.5%, after adjustments, diabetes remained associated with age (≥ 40 years), education (< 8 years of study), marital status (non-married), obesity, sedentary lifestyle and comorbidity, such as hypertension and hypercholesterolemia [21]. In a specific Brazilian sample, assessed to verify the low adherence to anti-diabetic treatment, including only diabetic patients aged over 20 years, age, female sex and lower income status were associated to T2DM [22]. Other findings have shown that some Brazilian States with greater poverty and lower levels of education had higher rates of T2DM or hyperglycaemia as well [22]. However, these were all based on prevalent cases and mostly self-reported. Here, we add data on predictors for the T2DM incidence rate.

In addition to well-known risk factors for diabetes, such as diet and physical activity, the socioeconomic and sociodemographic factors have shown great importance in this context. The socioeconomic position–measured by educational levels, occupation or income is frequently inversely associated with diabetes [23,24]. Smoking, especially for people with low socioeconomic status, was also identified as a mediator for diabetes development [25]. Although these factors were investigated in our study, the only socio-demographic factor that seemed to have greater importance in predicting the 5-year T2DM incidence in the Baependi population was marital status.

The relationship between marriage and improved health outcome has been previously suggested [26]. Some studies have shown a lower incidence of diabetes [27] and improved adherence to diabetes treatment [28] in partnered patients, since marital relationship influences health behaviours and socioeconomic status. Despite there being no difference in the incidence of T2DM between men and women in our study, the influence of marital status on T2DM seems to be modulated by gender. In a recent study which investigated the diabetes mortality in a large Spanish sample, the highest mortality was observed in divorced/widower women, while single men showed highest mortality [29]. Considering the T2DM incidence, another study found that widowed women compared to married women showed lower risk of T2DM development [13]. In our study, the influence of marital status seemed to be independent of sex.

Our results suggest that, only those who remained married or married during the 5-years follow-up have had a significant weight gain, which was associated with an increased risk of developing T2DM. However, the risk associated to marital status did not change, even after this adjustment. In fact, individuals that remained married, despite having significantly increased their weight, were significantly less likely to develop diabetes than their divorced counterparts.

There are two primary theories that can explain the beneficial effect of marriage on health. The first one is regarding the “selection”: healthier individuals tend to get married and remain married. The second hypothesis corresponds to post-marriage effect: reduction of stress, adoption of healthy behaviours [3033]. In our study, it is not possible to verify which hypothesis was more coherent, however, probably, both have had an effect on DMT2 development.

In this context, Cornelis and collaborators conducted an important study with a large number of men for ≤22 years and, after various models of adjustment, including lifestyle, BMI, family history, and other variables, widowhood was associated with an increased risk for T2DM in a robust way [34]. In this study, widower and divorced/separated were analysed separately, which is important, as the widowhood and divorce can have different stressful effects [35]. It was reported that the alcohol consumption increased between men who became widower, while both widower and divorced/separated men showed decreased in their BMI and vegetables consumption [35]. Since these factors have influence on T2DM, more studies are necessary to clarify the possible differences regarding the relationship between marital status and T2DM risk.

Some limitations were important in our study context. The lack of adjustment for physical activity as the potential/residual confounding factor is one of them. Additionally, classification of occupation can be mentioned, which may not have been effective in distinguishing participants (lack of distinction for domestic and part-time employees, for example), as well as the marital status, since it was not possible to distinguish widower and divorced/separated individuals.

Conclusions

In summary, lifestyle influences sex-specific metabolic changes over time. Marital status appears to be a predictor of T2DM incidence and the underlying factors for this association should be further characterised for they may provide important information in the better design and implementation of preventive programs.

Acknowledgments

We are grateful to the population of Baependi for their participation in the Baependi Heart Study. Additionally, we are pleased to engange an incredible group working on the data collection.

References

  1. 1. Ogurtsova K, da Rocha Fernandes JD, Huang Y, Linnenkamp U, Guariguata L, Cho NH, et al. IDF Diabetes Atlas: Global estimates for the prevalence of diabetes for 2015 and 2040. Diabetes Res Clin Pract. 2017; 128: 40–50. pmid:28437734
  2. 2. NCD Risk Factor Collaboration (NCD-RisC) (Collaborators 514). Worldwide trends in diabetes since 1980: a pooled analysis of 751 population-based studies with 4.4 million participants. Lancet. 2016; 387: 1513–1530. pmid:27061677
  3. 3. Camelo LV, Giatti L, Duncan BB, Chor D, Griep RH, Schmidt MI, et al. Gender differences in cumulative life-course socioeconomic position and social mobility in relation to new onset diabetes in adults-the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Ann Epidemiol. 2016; 26: 858–864. pmid:27894568
  4. 4. Danaei G, Finucane MM, Lu Y, Singh GM, Cowan MJ, Paciorek CJ, et al.; Global Burden of Metabolic Risk Factors of Chronic Diseases Collaborating Group (Blood Glucose). National, regional, and global trends in fasting plasma glucose and diabetes prevalence since 1980: systematic analysis of health examination surveys and epidemiological studies with 370 country-years and 2.7 million participants. Lancet. 2011; 378: 31–40. pmid:21705069
  5. 5. Tamayo T, Brinks R, Hoyer A, Kuß OS, Rathmann W. The Prevalence and Incidence of Diabetes in Germany. Dtsch Arztebl Int. 2016; 113: 177–182. pmid:27118665
  6. 6. Vega T., Gil M., and Lozano J. Age and sex differences in the incidence of diabetes mellitus in a population-based Spanish cohort. J Diabetes. 2015; 7: 411–417. pmid:24981073
  7. 7. Boehme MW, Buechele G, Frankenhauser-Mannuss J, Mueller J, Lump D, Boehm BO, et al. Prevalence, incidence and concomitant co-morbidities of type 2 diabetes mellitus in South Western Germany—a retrospective cohort and case control study in claims data of a large statutory health insurance. BMC Public Health. 2015; 15: 855. pmid:26334523
  8. 8. Emmerick ICM, Campos MR, Luiza VL, Chaves LA, Bertoldi AD, Ross-Degnan D. Retrospective interrupted time series examining hypertension and diabetes medicines usage following changes in patient cost sharing in the 'Farmacia Popular' programme in Brazil. BMJ Open. 2017; 7: e017308. pmid:29101135
  9. 9. Metsä-Simola N, Martikainen P. The short-term and long-term effects of divorce on mortality risk in a large Finnish cohort, 1990–2003. Popul Stud (Camb). 2013; 67(1): 97–110.
  10. 10. Manzoli L, Villari P, M Pirone G, Boccia A. Marital status and mortality in the elderly: a systematic review and meta-analysis. Soc Sci Med. 2007 Jan; 64(1): 77–94. pmid:17011690
  11. 11. Tuoyire DA, Ayetey H, Gender differences in the association between marital status and hypertension in Ghana. J Biosoc Sci. 2019 May; 51(3): 313–334. pmid:29781417
  12. 12. Whisman MA, Li A, Sbarra DA, Raison CL. Marital quality and diabetes: results from the Health and Retirement Study. Health Psychol. 2014 Aug; 33(8): 832–840. pmid:25068454
  13. 13. Ramezankhani A1, Azizi F2, Hadaegh F. Associations of marital status with diabetes, hypertension, cardiovascular disease and all-cause mortality: A long term follow-up study. PLoS One. 2019 Apr 22;14(4): e0215593. pmid:31009512
  14. 14. Lundquist JH, Anderton DL, Yaukey D. Demography: the study of human population. Waveland Press, 2014.
  15. 15. Oliveira CM, Pereira AC, de Andrade M, Soler JM, Krieger JE. Heritability of cardiovascular risk factors in a Brazilian population: Baependi Heart Study. BMC Med Genet. 2008; 9: 32. pmid:18430212
  16. 16. Alvim RO, Horimoto AR, Oliveira CM, Bortolotto LA, Krieger JE, Pereira AC. Heritability of arterial stiffness in a Brazilian population: Baependi Heart Study. J Hypertens. 2017; 35: 105–110. pmid:27763993
  17. 17. Annesi I, Moreau T, Lellouch J. Efficiency of the logistic regression and Cox proportional hazards models in longitudinal studies. Stat.Med. 1989; 8: 1515–21. pmid:2616941
  18. 18. Peters S.A., Huxley R.R., and Woodward M. Do smoking habits differ between women and men in contemporary Western populations? Evidence from half a million people in the UK Biobank study. BMJ Open. 2014; 4:e005663. pmid:25550291
  19. 19. Iser BP, Malta DC, Duncan BB, de Moura L, Vigo A, Schmidt MI. Prevalence, correlates, and description of self-reported diabetes in brazilian capitals—results from a telephone survey. PLoS One. 2014; 9: e108044. pmid:25255096
  20. 20. Fox CS, Pencina MJ, Meigs JB, Vasan RS, Levitzky YS, D'Agostino RB Sr. Trends in the incidence of type 2 diabetes mellitus from the 1970s to the 1990s: the Framingham Heart Study. Circulation. 2006; 113: 2914–2918. pmid:16785337
  21. 21. Flor L.S. and Campos M.R., The prevalence of diabetes mellitus and its associated factors in the Brazilian adult population: evidence from a population-based survey. Rev Bras Epidemiol. 2017; 20: 16–29. pmid:28513791
  22. 22. Meiners MMMA Tavares NUL, Guimarães LSP Bertoldi AD, Pizzol TDSD Luiza VL, et al. Access and adherence to medication among people with diabetes in Brazil: evidences from PNAUM. Rev Bras Epidemiol. 2017; 20: 445–459. pmid:29160437
  23. 23. 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: 804–18. pmid:21335614
  24. 24. Sommer I, Griebler U, Mahlknecht P, Thaler K, Bouskill K, Gartlehner G, et al. Socioeconomic inequalities in non-communicable diseases and their risk factors: an overview of systematic reviews. BMC Public Health. 2015;18:914.
  25. 25. Williams ED, Tapp RJ, Magliano DJ, Shaw JE, Zimmet PZ, Oldenburg BF. Health behaviours, socioeconomic status and diabetes incidence: the Australian Diabetes Obesity and Lifestyle Study (AusDiab). Diabetologia. 2010; 53:2538–45. pmid:20740271
  26. 26. Kiecolt-Glaser J.K. and Newton T.L. Marriage and health: his and hers. Psychol Bull. 2001; 127: 472–503. pmid:11439708
  27. 27. Johnson NJ, Backlund E, Sorlie PD, Loveless CA. Marital status and mortality: the national longitudinal mortality study. Ann Epidemiol. 2000; 10: 224–38. pmid:10854957
  28. 28. Haines L, Coppa N, Harris Y, Wisnivesky JP, Lin JJ. The Impact of Partnership Status on Diabetes Control and Self-Management Behaviors. Health Educ Behav. 2018; 45: 668–671. pmid:29361845
  29. 29. Escolar-Pujolar A, Córdoba Doña JA, Goicolea Julían I, Rodríguez GJ, Santos Sánchez V, Mayoral Sánchez E, et al. The effect of marital status on social and gender inequalities in diabetes mortality in Andalusia. Endocrinol Diabetes Nutr. 2018; 65: 21–29. pmid:29233513
  30. 30. Umberson D. Gender, marital status and the social control of health behavior. Soc Sci Med. 1992; 34: 907–917. pmid:1604380
  31. 31. Wyke S, Ford G. Competing explanations for associations between marital status and health. Soc Sci Med. 1992; 34(5):523–352. pmid:1604359
  32. 32. Murray JE. Marital protection and marital selection: evidence from a historical-prospective sample of American men. Demography. 2000; 37: 511–52. pmid:11086576
  33. 33. Espinosa J, Evans WN. Heightened mortality after the death of a spouse: marriage protection or marriage selection? J Health Econ. 2008; 27: 1326–1342. pmid:18513810
  34. 34. Cornelis MC, Chiuve SE, Glymour MM, Chang S-C, Tchetgen EJT, Liang L, et al. Bachelors, divorcees, and widowers: Does marriage protect men from type 2 diabetes? PLoS One. 2014; 9(9):e106720. pmid:25229473
  35. 35. Eng PM, Kawachi I, Fitzmaurice G, Rimm EB. Effects of marital transitions on changes in dietary and other health behaviours in US male health professionals. J Epidemiol Community Health. 2005; 59: 56–62. pmid:15598728