We undertook a cross-sectional analytical study. Secondary data were taken from the Demographic and Health Survey of Peru (ENDES, in Spanish), anually executed by Instituto Nacional de Estadística e Informática-Peru (freely available in: https://iinei.inei.gob.pe/microdatos/) covering the years 2018 to 2021, and analyzed. The STROBE guidelines (Strengthening the Reporting of Observational Studies in Epidemiology) were followed for the present study. ⁸

The ENDES is a nationally representative survey with a two-stage sampling design (Instituto Nacional de Estadística e Informática, 2015). The sample was characterized by being probabilistic of a balanced, stratified, and independent type, at the departmental level and in urban and rural areas. For our research, only the data of the respondents of both sexes and that had the main variables of interest were analyzed.

The outcome of interest was the presence of depressive symptoms, which was assessed with the Patient Health Questionnaire-9 (PHQ-9) in the Peruvian survey. The questionnaire consists of nine items formulated to assess and monitor the severity of depression for patients in primary care and other environments. It was designed to be self-administered, and collects information on depressive symptoms over the course of the previous 2 weeks. Each item has a score ranging from 0 to 3, with a total maximum of 27 points. ^{9 , 10} Individuals with a score of 15 or more are considered to have depression. ¹⁰ The PHQ-9 was previously validated in the Peruvian population ¹¹ and its psychometric properties have been recognized as adequate for different population groups. ¹²

The exposure variable was the presence of obesity, which was assessed by BMI and AC. Anthropometric measurements (weight, height) of all participants were evaluated by trained personnel following standardized techniques based on the WHO and the Instituto Nacional de Salud from Peru. ¹³ Obesity according to BMI was defined with a cut-off point of BMI ≥ 30 kg/m ², while abdominal obesity was defined if there was AC ≥ 104 cm in men and ≥ 88 cm in women, measurements recommended by the Adult Treatment Panel III (CA-ATP-III).

The factors evaluated were gender (man vs. woman); categorized age (15-34, 35-60, 61-69, and ≥70 years); educational level (primary, secondary and higher); the wealth index (poor, medium, rich and richest), the natural region (Metropolitan Lima, rest of the coast, Andean and jungle), daily tobacco use (yes vs no), physical disability (yes vs no), the self-reported alcohol consumption in the previous 12 months (yes vs no), history of hypertension (yes vs no) and Diabetes Mellitus type 2 (DM2) (yes vs no).

We used STATA 17 software for analysis and the prevalence of depressive and obesity symptoms was estimated. Bivariate analysis through a Chi-square test was used to analyze each possible factor associated with depression. Finally, the crude and adjusted odds ratio (cOR and aOR respectively) were calculated using logistic regression. Each marker was independently adjusted for sex, categorized age, natural region, educational level, wealth index, daily smoking, alcohol consumption, physical disability, history of hypertension, and history of DM2.

All analyses were performed considering complex samples. It was considered statistically significant if the p value was <0.05. Both prevalence and association measures were presented with 95% confidence intervals.

This study was developed with an analysis of survey data sets that are openly published and available online (at http://iinei.inei.gob.pe/microdatos/). In the ENDES survey, run by trained interviewers, informed consent was obtained from all participants. Additionally, in order to ensure data privacy, the responses were anonymized through coding.

In the present study, no significant correlation was found between obesity measured by BMI and depressive symptoms. BMI is a regular and easy tool for the assessment of excess adiposity, but it has restrictions, that include the inability to distinguish between adipose tissue distribution and lean body mass ¹⁴ and also, there are significant differences in the performance of BMI between ethnic groups. ^{15 , 16} These limitations may be greater in men due to their greater muscle mass compared to women. This may explain the fact that several studies have reported that BMI is an important predictor of depressive symptoms in women but no in men, ^{17 , 18} although when the studies are carried out prospectively, the BMI is related to depression. ¹⁹ According to our findings, Guedes et al. ²⁰ suggested that particularly, the body fat percentage and not BMI was related to a greater severity of depressive symptoms.

However, other studies that only included BMI as an anthropometric variable reported an association between BMI and depressive symptoms, such as the study by De Godín et al., ²¹ which documented that a high BMI is considered a risk factor for manifestation of depressive symptoms among older adult subjects in France, compared to normal BMI. Furthermore, Sachs-Ericsson et al. ²² reported that BMI was a predictor of depression in old age, and that its effect was stronger in African-Americans than in the white population, regardless of sex. However, differences have also been found in relation to sex, since Anderson et al. ²³ carried out a prospective longitudinal study to evaluate the association between depression and weight variation in a study carried out from the early years to adulthood, and found out that depression was associated with elevated BMI in women but not in men. Similarly, the systematic review by Luppino et al. ¹⁹ showed that obesity defined by BMI was related to a major risk of depression in American subjects compared to Europeans, and being overweight was associated with a higher risk of depression in adult populations but not in young ones. Another aspect to consider is that the association between BMI and depression varies depending on the different subtypes of depression, as reported in a recent meta-analysis. ⁵ The differences found between the different studies can be explained by the methodological variation, including population, follow-up, cut-off point in diagnostic tools, and criteria for obesity and depression.

The evidence seems to indicate that some anthropometric markers have a better explanatory value in regard the association between obesity and depression. An example of this is the study by Zhao et al., which found that abdominal obesity among obese and overweight people was strongly associated with an increase in depressive symptoms. ²⁴ Likewise, other works such as that of Hadi et al., ²⁵ in which different anthropometric indicators of obesity were studied, concluded that those related to abdominal adiposity have a better association with depression, compared to BMI. On the other hand, Lee et al. reported that depressed mood in overweight premenopausal women is associated with visceral fat, but not subcutaneous fat. ²⁶ The follow-up study by Herva et al. ²⁷ argued that in both men and women, abdominal obesity may be closely associated with depression.

A Swiss cohort study by Lasserre et al. ²⁸ reported that depressive disorder was an important risk factor for obesity as AC increased in both sexes. Ma and Xiao ²⁹ reported that higher waist circumference was associated with depression, regardless of BMI. While Williams et al. ³⁰ found out that women with antecedents of depressive issues tended to have higher BMI, weight, waist circumference, and body fat than those without antecedents of mental issues.

Abdominal adipose tissue induces the activation of the immune system, the release of regulatory molecules and citokines that, in turn, unchain inflammatory signaling pathways. ^{31 , 32} Symptoms of depression can be aggravated by systemic inflammation, so it is relevant to focus on central adiposity when discussing the association between obesity and depression. ³³ In addition, some mechanisms have been suggested for the linkage between obesity and depression, which includes the hypothalamic-pituitary-adrenocortical axis dysregulation resulting from reduced glucocorticoid receptors and excessive cortisol secretion. ³⁴

Among the limitations of this study, it is important to mention two. First, due to the cross-sectional nature of this work, we were unable to determine the direction of causality between anthropometric measures and depressive symptoms. Secondly, it is not possible to talk about the diagnosis of depression itself, since what was assessed was the presence of depressive symptoms. Likewise, subtypes of depression could not be established.

Despite the limitations, this study has the strength of having benefited from a nationally representative sample, as well as from the methodology used to obtain it. Finally, we emphasize that the present study gives us a first impression about the importance of appropriately selecting the anthropometric marker of obesity used to assess its association with depressive symptoms among the Peruvian population.