Introduction

Depressive disorders have a negative impact on the mental health of many people around the world. According to the systematic analysis and update of The Global Burden of Diseases, Injuries and Risk Factors Study 2017 (GBD 2017)¹, since 2007 depressive disorders became the third leading cause of non-fatal health loss, measured in years lived with disability (YLD), which in 2017 amounted to 14.3% (13.1 - 15.6%). Based on data from GBD 2002, Mathers and Loncar projected that by 2030, in the baseline scenario, unipolar depressive disorders will rank second in disability-adjusted life years (DALYs), defined as the sum of years of potential life lost due to mortality plus the years of productive life lost due to disability².

Several studies have estimated the prevalence of depression in México. In the National Survey of Psychiatric Epidemiology (2001-2002), the WHO World Mental Health Survey Initiative version of the Composite International Diagnostic Interview was applied to 5826 adults (54% women) aged 18-65, living in communities with >2500 inhabitants. It was found that major depressive disorder had a lifetime prevalence of 3.3%, and 1.5% in the previous year³. In another study performed during 1999 and 2001, 2509 adults (63.9% women) aged 18-92, residing in four major cities in different regions of México were interviewed, and major depression had a 12.8% lifetime prevalence, 6.1% of them in the previous year⁴. In the 2002-2003 National Performance Evaluation Survey based on the DSM-IV diagnostic criteria for depression, an adult aged ≥18 years was interviewed in each of 38700 households sampled in cities classified as rural, urban or metropolitan (with <2500, 2500-100000, or > 100000 inhabitants, respectively). A total prevalence of depression of 4.5% (95% CI=4.1-4.9) was estimated⁵.

Projections of global mortality and burden of disease predict that, by 2030, unipolar depressive disorder will represent 6.7% of all DALYs in middle-income countries², which currently includes México. To avoid this increase and the excessive burden on health systems, preventive measures are necessary, such as identifying and reducing modifiable factors that increase the risk of depression.

Strong evidence indicates that suffering type 2 diabetes mellitus (T2DM) increases the risk of developing depression^6,7,8,9,10. A systematic review with meta-analysis of 24 high quality longitudinal cohort studies showed that people with T2DM had a greater risk of developing depression (pooled odds ratio 1.33; 95% CI, 1.18-1.51)¹⁰. The comorbidity of diabetes and depression acts in synergy to further reduce the quality of life (QoL) of afflicted persons, increasing the loss of productivity and health- care expenditure^6,8,9. This interaction of both maladies is bidirectional, since depression in diabetic patients favors negligence in treatment adherence^6,8. Thus, many studies have documented that depression in patients with T2DM is associated with poor glycemic control, evidenced by higher levels of fasting blood glucose^11,12,13,14, or glycated hemoglobin A (HbA1c) levels^15,16. In a meta-analysis it was estimated that reducing the burden of T2DM by 10-25% could likely prevent between 0.93 and 2.34 x 10⁶ cases of depression globally¹⁰. The above effect would be of utmost importance since in the GBD 2017, T2DM emerged as the fourth leading cause of age-standardized YLD worldwide, and the leading cause of YLD in Mexico¹. Another reason for implementing measures to reduce the incidence of diabetes is that its comorbidity with depression multiplies the risk of mortality^6,8,10.

In a study carried out in Mexicans over 40 years old, it was found that the prevalence of depression (DSM-IV) in diabetics with at least 5 years of evolution was 52%, while in non-diabetics it was 18% (χ2 = 12.7; p<0.05)⁷. Moreover, many studies have shown a high prevalence of depression in Mexican patients with T2DM, ranging from 27 to 73%, depending on the geographical area, and the type of instrument used^{7,11,12,13,14,17,18,19}. These values are higher than the reported prevalence among the general population (3.3- 6.1%)^3,4,5.

The prevalence of depression in Mexican patients with T2DM has been mainly based on subjects from urban populations^{7,11,12,13,17,18,19}, rather than from rural communities. In one study, comprising 4555 outpatients who attended 6 rural clinics in the state of Chiapas during 2014, a 7.9% global annual prevalence of depression was found (10.1% in women, 3.9% in men), but the association with T2DM was not significant (p=0.06)²⁰. In another study, carried out between 2017 and 2018, the Beck Depression Inventory (BDI) was applied to 112 patients with T2DM attending a rural clinic in the state of Yucatán. The prevalence of depressive symptoms was 62%, and the capillary blood glucose level was directly proportional to depression severity¹⁴.

Since in Mexico there is a great diversity of ethnic groups that live in rural communities, it is important to assess the relationship between depressive symptoms and diabetes control in these populations. Here we report the results of a cross-sectional study in a small sample of T2DM outpatients treated only with oral antidiabetic drugs and living in a rural community of predominantly Mayan ethnicity. The aim was to evaluate the possible correlation between the glycemic control assessed by HbA1c levels during the previous three months, and the depressive symptoms measured with the CES-D questionnaire, as well as with the QoL assessed with the D-39 instrument.

Materials and methods

Study population. T2DM patients from the rural community of Cuncunul, Yucatan, and attending the local health center for their follow-up appointments, were invited to participate in this study. Data were collected between March 2, 2018 and June 19, 2018. Only subjects ≥18 years-old and taking oral glucose-lowering drugs (metformin and/or glibenclamide) were included. Subjects on insulin regimen were excluded since in this type of diabetic patients the association between depressive symptoms and poor glycemic control has already been established¹⁶. Likewise, patients taking antidepressant medication or presenting T2DM complications (diabetic ketoacidosis, hyperglycemic hyperosmolar syndrome, retinopathy, nephropathy, or coronary artery disease), were also excluded.

Ethical considerations. This study was approved by the ethics board of the CIR-UADY, being classified as ‘risk- free’ research (article 17- I: application of surveys, physical exams, or routine tests) according to the guidelines of the Regulation of the General Health Law on Health Research Topics in Mexico (last amendment in March 28, 2014). Subjects received information regarding the aim, procedures, likely benefits of the study, and the guarantee of data confidentiality, and those willing to participate signed a letter of informed consent (articles 20-22).

General procedures. The questionnaire used to assess the presence or absence of depressive symptoms was applied first, and then the one concerning QoL. All interviews (≤40 min) were performed by the same person (LASD), who was performing his social service in medicine in the community of Cuncunul at the time of the study. Since most patients were not proficient in reading because of their low education level, the interviewer read aloud all the questions from both questionnaires, explaining their meaning in words that the subjects could better understand. Afterwards, the interviewer asked them to indicate the answer that best approximated how they had felt and recorded it. Body weight and height were measured at this time. Diagnosis of arterial hypertension was obtained from the clinical record (blood pressure cut-off points: systolic ≥140 mmHg and/or diastolic ≥90 mmHg)²¹. Having at least one Mayan surname was the criterion for Mayan ethnicity. In Mexico, two last names are used (from the father’s side and from the mother’s side).

CES-D questionnaire. The Center for Epidemiological Studies - Depression (CES-D) questionnaire (20 questions) was developed as a screening test to identify people at risk of clinical depression by assessing the subjective perception of mood and the level of daily functioning^22,23. Each response is scored from zero to three on a scale of frequency of occurrence of the symptom during the previous week. Four of the items are worded considering positive affect conditions and are scored inversely, so that their interpretation is the same as for the other questions. The resulting 20 values are added to obtain the total score (max. 60). Values ≥16 indicate clinically relevant depressive symptoms²². A meta- analysis reported that with the above criterion CES-D has a sensitivity of 0.87, a specificity of 0.70, and a diagnostic odds ratio of 16.2²³.

Diabetes-39 questionnaire. The Diabetes-39 (D-39) questionnaire was used since it was specifically designed to evaluate QoL in patients with type 1 and type 2 diabetes²⁴. A Spanish version has been validated in Mexican subjects with T2DM, searching for associations between QoL and glycemic indicators, such as HbA1c and fasting plasma glucose, resulting in high internal consistency and construct validity²⁴. The 39 queries are grouped into five dimensions: energy-mobility (EM), diabetes control (DC), anxiety-worry (AW), social burden (SB), and sexual functioning (SF). Patients indicate how much their QoL was affected during the previous month by the action described in each question. Results for each section are converted into a 0 to 100 scale using a linear transformation formula. The total score is obtained from the five sections and reflects the impact of diabetes on QoL as a whole (better QoL for lower scores). It also contains two final questions that rate the patient’s perception of his or her overall QoL and diabetes severity (DS)²⁴.

Body mass index. Body weight (kg) and height (m) were measured with a mechanical weighing scale with stadiometer. Body mass index (BMI) was calculated with the formula: body weight / (height squared) and expressed in kg/ m²; then subjects were classified as normal (20-24.9 kg/m²), overweight (25-29.9 kg/m²) or obese (≥30 kg/m²)^13,16.

Glycosylated hemoglobin. The levels of HbA1c determined during the previous three months were obtained from the medical records (n=9). In the absence of such measurement, the lab test was ordered during the visit (n=12). If the patient could not go to the lab, the measurement was made with a portable glucometer (AC1Now®⁺, PTS Diagnostics) at the end of the visit (n=12). No significant differences were found between HbA1c values (mean ± SD) from clinical records (9.9±2.5%), those requested to the lab (8.9±2.7%), and those measured with glucometer (9.4±2.3%) (One-way ANOVA, F2,30=0.44, p=0.65). HbA1c <8% was chosen as the criterion for good glycemic control, according to the guidelines of the American College of Physicians²⁵.

Statistical analysis. Reported values are means ± SD, unless otherwise noted. HbA1c and BMI data sets (n=33) had a Normal distribution (D’Agostino-Pearson omnibus test), and unpaired comparisons were made with Student’s t-test (two-tailed). For questionnaire scores, some were normally distributed (CES-D, and the EM, DC, AW and QoL dimensions of D-39), while others were not (SB, SF, and DS dimensions, and total D-39 score). Hence, the Mann- Whitney U-test (two-tailed) was deemed appropriate to compare CES-D and D-39 scores from patients having good (HbA1c <8%, n=11) or poor (HbA1c ≥8%, n=22) glycemic control, or for having depression (CES-D ≥16, n=12) or not (CES-D <16, n=21). For some questionnaire scores that had significant differences, box plots were made to illustrate the dispersion of values in quartiles. Fisher’s exact test was applied to ascertain associations between some variables, such as depression, D-39 dimensions, glycemic control, BMI or hypertension. Pearson determination coefficient (two- tailed) was obtained to establish the possible covariation between normally distributed variables. A p value < 0.05 was considered significant. Graphs and statistical analyses were made using Prism Ver. 8.0.2 (GraphPad Software Inc., San Diego, CA, USA).

Results

A total of 33 patients with T2DM were studied, with a mean age of 58.5 ± 11.5 years (range: 38-80 years), and most were of Mayan descent (≈83%). Only a third of the patients had good glycemic control (HbA1c <8%), slightly more than one third had depression (CES-D ≥16), the majority (≈88%) was overweight or obese, and about half (54.5%) had hypertension (Table 1). No significant differences were found in the proportion of women and men within the categories of these variables (Fisher’s exact test).

In this sample, 69.7% were women and 30.3% men. No significant differences were found between women and men with respect to age, HbA1c levels or BMI (Student’s t-test), nor in the CES-D or D-39 total scores (Mann-Whitney U-test) (Table 2). Thus, the data from both genders were pooled for subsequent analyses.

Relationship between HbA1c levels and CES-D questionnaire scores

Fifty percent (11/22) of T2DM patients with poor glycemic control (HbA1c ≥8%) had CES-D scores ≥16, while only 9.1% (1/11) of those with good glycemic control (HbA1c<8%) surpassed this cutoff value (p=0.027, Fisher’s exact test). In diabetics with poor glycemic control (n=22) CES-D scores were significantly higher than in those with good control (n=11) (Figure 1a). Similarly, HbA1c levels in patients with depression (10.7±2.5%, n=12) were higher than in those without depression (8.6±2.2%, n=21; p=0.015, Student’s t-test). When CES-D scores were plotted against HbA1c levels, a significant positive correlation was found between the two variables (Figure 1b).

Figure 1 CES-D scores as a function of glycemic control 

a) Box-plots showing CES-D score distributions in T2DM patients grouped according to their glycemic control (good or poor). Open symbols represent individual CES-D scores. Values on or above the cut-off value of 16 (horizontal dashed line) indicate depression. The dotted line marks a more stringent cut-off value (19) for depression. The p value is from the Mann-Whitney U-test (U=52.5). b) Scatterplot of CES-D scores and HbA1c % values measured in each diabetic patient. Points to the right of the cut-off value of 8 (vertical dashed line) indicate poor glycemic control. Triangles represent patients with depression; circles, patients without depression. Solid symbols are women; open symbols, men. r² is the Pearson coefficient of determination.

Source: design and elaboration by the authors with Prism software, using CES-D scores, and HbA1c values from laboratory tests.

Relationship between D-39 questionnaire dimensions and CES-D scores

Although the D-39 questionnaire does not specify cut-off points for good or poor QoL, some authors have used the median of the values to sort them into these two categories²⁴. In our sample, only the anxiety-worry (AW) dimension scores (median =20), had significant associations with other variables. Nineteen (57.6%) of the 33 diabetic patients scored ≥20 on the AW dimension. Among those classified as depressed (CES-D ≥16, n=12), a significant proportion (83.3%) had AW scores ≥20 (p=0.033, Fisher’s exact test). Indeed, after grouping T2DM subjects according to the absence or presence of depression, the AW dimension scores were significantly higher in those with CES-D scores ≥16 than in those with lower values (Figure 2a). Noticeably, among the ten patients with AW scores ≥20 and CES-D scores ≥16, nine were women. When AW scores were plotted against CES-D scores, a significant positive correlation was found (Figure 2b). None of the other D-39 dimensions correlated with CES-D scores.

Figure 2 D-39 anxiety-worry scores as a function of CES-D scores 

Discussion

Here we describe the characteristics of a sample of patients with T2DM, analyzing the relationship between their glycemic control (assessed through HbA1c levels), depressive symptoms (CES-D scores) and quality of life (D- 39 scores). A notorious finding was the significant difference in CES-D scores between patients with HbA1c blood levels <8% and those having higher values, which defines good or poor glycemic control, respectively, according to the Clinical Guidelines of the American College of Physicians²⁵. In addition, in T2DM patients, a significant positive correlation was found between the magnitudes of the intensity of depressive symptoms and HbA1c levels, suggesting that they are directly related. Regarding QoL, assessed with the D-39 questionnaire, a significant difference was found only in the scores of the anxiety-worry dimension between patients with and without depression²². In this case, a significant positive correlation was also found, which suggests that the two variables influence each other.

Conclusions

From the results of this study, it is concluded that T2DM subjects classified as depressed according to the CES-D criteria have higher levels of HbA1c than those without depression. The finding that HbA1c levels were positively correlated with CES-D scores suggests that, as the glycemic control in T2DM patients worsens, their risk of suffering from depression increases. This is supported by the observation that depression occurred in 33.3% of the diabetics with poor glycemic control (HbA1c ≥8%), but only in 3.0% of those with good control (HbA1c <8%). The fact that D-39 “Anxiety-Worry” (AW) dimension scores were higher in diabetics with depression than in those without depression, and that both variables were correlated, suggests that T2DM subjects with depression have a higher risk of suffering anxiety disorders. Considering that depression and anxiety have been associated with poor self-care to achieve good glycemic control, we propose the concomitant use of CES-D and D-39, which are validated and easy-to-apply instruments, as screening tests to detect depression and anxiety in DM2 patients residing in rural communities. Therefore, if patients test positive on one or both instruments, they can be referred to a psychiatrist to confirm the diagnosis and provide appropriate therapy. This would help promote adherence to diabetes control measures and to improve their quality of life