Behavioral risk factors for noncommunicable diseases associated with depression and suicide risk in adolescence

Noncommunicable diseases (NCDs) and mental disorders cooccur in adult-hood, which is why their determinants and common risk factors should be addressed at an early age. Therefore, we estimated the association of the major risk factors for NCDs with depression and suicide risk by structural equation modeling considering pathways triggered by social vulnerability or mediated by obesity. This population-based study included 2,515 Brazilian adolescents. The following exposures were the major risk factors for NCDs: substance use behaviors (variable deduced from alcohol, tobacco, and drug use), physical inactivity, and components of unhealthy eating markers (added sugar and saturated fat). Obesity was assessed using the fat mass index. The outcomes were depression and suicide risk. Depression was associated with substance use behaviors (SC = 0.304; p < 0.001), added sugar (SC = 0.094; p = 0.005), and females (SC = 0.310; p < 0.001). Suicide risk was also associated with substance use behaviors (SC = 0.356; p < 0.001), added sugar (SC = 0.100; p = 0.012


Introduction
Mental disorders often coexist with noncommunicable diseases (NCDs).Depression tends to be classified as a mental NCD 1 and have a bidirectional association with other NCDs 2 .Social determinants of poverty act primarily on depression and increase the future risk of other NCDs 3 whereas side effects of antidepressants may cause weight gain 2 .Moreover, NCDs can increase morbidity and drugs used for their treatment could aggravate psychiatric symptoms 4 , especially in adulthood.
Considering the bidirectional relationships between depression and other NCDs, obesity emerges as a metabolic risk factor for NCDs of physical scope 5 which is consistently associated with depression 6 .Common risk factors could therefore be involved in the complex relationships between obesity, depression, and NCDs in adolescents.
According to the World Health Organization (WHO), NCDs are mostly driven by four major risk factors: unhealthy diet, physical inactivity, smoking, and the harmful use of alcohol 7 .These factors increase metabolic risk and are associated with mental disorders, depression, and suicide risk 8 in individuals, including adolescents 9 .They can also induce inflammation, which is an underlying mechanism of all of these diseases 10 , and thus triggering or aggravating depression and suicide risk 11,12 .
The cooccurence of risk factors for NCDs characterizes an unhealthy lifestyle 8,13,14 .The crosssectional, nationwide, school-based study found that 5.7% of adolescents smoked 15 , 21% had consumed alcohol on at least one occasion in the last 30 days 16 , and 52.4% were physically inactive 17 .Data from the Brazilian Household Budget Survey (POF) 2017/2018 18 show that adolescents have lower nutritional quality diet markers per capita and a higher consumption of sandwich biscuit, soft drinks, dairy beverages, chips, sandwiches, and pizzas.
Risk factors associated with depression and suicide risk in complex relationships must be further investigated.Regarding limitations, previous studies often selected only one risk factor as exposure without adjusting for the other risk factors for NCDs.Furthermore, many studies based on selfreports or symptoms and not on diagnostic measures could make inconsistent associations between major risk factors for NCDs and depression or suicide risk 8,13 .
Depression is a known risk factor for suicide risk and the second leading cause of death among young people aged from 15 to 29 years worldwide 19 .Death rates are increasing among adolescents and suicide has become a major preventable public health problem.Adolescents aged 10 to 19 years in Brazil have a 8.25% 20 prevalence of suicide and 17% prevalence of depression 21 .We hypothesize that the major risk factors for NCDs are associated with depression and suicide risk in adolescence, which could cause clusters of NCDs and mental health disorders in adulthood.Therefore, this study estimated the association of the major risk factors for NCDs with depression and suicide risk by structural equation modeling (SEM) considering pathways triggered by social vulnerability or mediated by obesity.

Study design
This cross-sectional study was conducted using a population-based sample derived from the RPS Brazilian Birth Cohorts Consortium (Ribeirão Preto, Pelotas and São Luís) follow-up at 18-19 years old.This study used data from the 1997 São Luís Birth Cohort obtained during the follow-up of adolescents (18-19 years old) from January to November 2016.
The sample included all participants from the original cohort who participated in the follow-up (n = 687).At that time, more adolescents born in São Luís (Maranhão State) in 1997 were included to increase sample power and prevent future losses (n = 1,828).These adolescents were selected from a random sample (n = 1,133) using the Brazilian Information System on Live Births (SINASC) database or among volunteers identified in schools and universities (n = 695), as long as they were in the SINASC database.This population-based study included 2,515 Brazilian adolescents.

Data collection procedures
The following information was collected using questionnaires: adolescent's sex, schooling level, economic class, monthly household income, cigarette smoking, illicit drug use, consumption of alcoholic beverages, physical activity, food intake, and diagnosis of depression and suicide risk, and schooling level of the head of household.
Sex was dichotomized into male and female.The schooling level of the head of household and of the adolescent was categorized into elementary school, high school, and higher education.The economic class was categorized into A, B, C, and D/E based on the Brazilian Economic Classification Criteria (CCEB) 22 .Monthly household income was based on the Brazilian minimum wage in 2016 (USD 270.76).
Cigarette smoking was dichotomized into no and yes.Substance use (never used, previously used, or currently using) and consumption of alcoholic beverages were evaluated using the Alcohol Use Disorders Identification Test (AUDIT) and divided into low and high risk 23 .
Physical activity was assessed by the Self-Administered Physical Activity Checklist (SAPAC) 24 .The adolescents were classified as physically active (≥ 150 minutes of physical activity/week) or physically inactive (< 150 minutes of physical activity/week).
Food intake was obtained with a validated food frequency questionnaire (FFQ) containing 106 items 25 .Details of the FFQ methodology were published by Sousa et al. 26 .Unhealthy eating markers were added sugar and saturated fat consumption.Added sugar intake was the percentage of calories from sugar of the total daily energy intake, found in soft drinks, fruit-flavored juice, chocolate drinks, energy drinks, and in a wide range of food groups, including dairy products, bread, cookies, breakfast cereals, desserts, chocolate, mayonnaise, salty snacks, and cold cuts.Saturated fat intake was also estimated as the percentage of calories from saturated fat of total daily energy intake, found in fast food, salty snacks, sweets, meat, chicken, fish, eggs, and milk and dairy products.Added sugar and saturated fat intake were analyzed as continuous variables.
Fat mass was determined by dual-energy X-ray absorptiometry (DXA) and height was measured with a stadiometer (Altura Exata; http://www.alturexata.com.br/).The fat mass index (FMI), corresponding to fat mass (kg) calculated by DXA and divided by the height in square meters (m 2 ), was used to measure obesity.
Depression and suicide risk were diagnosed by the Mini-International Neuropsychiatric Interview (MINI) 27 .Participants were diagnosed with depression based on the occurrence of major depressive and recurrent major depressive episodes.Suicide risk was categorized as low (score ranging 1-5) and high (score ≥ 6).

Latent variables
The following latent variables were used: Socioeconomic status: deduced from the shared variance of the head of household's and the adolescent's schooling level, economic class according to CCEB criterion, and monthly household income; Substance use behaviors: deduced from the shared variance of cigarette smoking, illicit drug use, and consumption of alcoholic beverages.

Proposed theoretical model
The latent variable socioeconomic status was considered as a more distal determinant that affects other variables of the model.Sex also affects socioeconomic status, substance use behaviors, physical inactivity, added sugar, saturated fat, FMI, depression, and suicide risk.The following exposures of interest were the major risk factors for NCDs: substance use behaviors, physical inactivity, and unhealthy eating markers.Obesity -represented by FMI -was considered a mediator since it could be chain linking the risk factors for NCDs to depression or suicide risk (Figure 1).

Statistical analysis
The descriptive data were analyzed using the Stata software, version 14.0 (https://www.stata.com).Categorical variables were described by absolute and relative frequencies.Depression and suicide risk were assessed in separate models by SEM.Factor loadings > 0.5 were considered as an indication that the factor influences the variable in the confirmatory factor analysis.The root mean square error of approximation (RMSEA) with an upper limit of the 90% confidence interval (90%CI) < 0.08 and a comparative fit index (CFI) and Tucker-Lewis index (TLI) > 0.95 were considered as indicators of adequate model fit.The weighted least squares means and variance adjusted (WLSMV) estimator and theta parameterization were used.The goodness-of-fit of the structural equation model was assessed based on the same estimates as described for the confirmatory factor analysis 28 using the Mplus 7.0 software (https://www.statmodel.com/).
Firstly, to represent the unhealthy diet, added sugar and saturated fat consumption was analyzed, considering that these are the main components of the investigated diet associated with mental disorders 8,14,29,30 .Then, sugar was analyzed as a separate component of diet because of its addictive potential in the central nervous system 31
The unhealthy eating markers were correlated with each other in the depression (SC = 0.408; p ≤ 0.001) and suicide risk model (SC = 0.408; p < 0.001).
During consistency analyses, we regressed the model using the healthy diet and unhealthy diet latent variables and found that unhealthy diet was associated with depression (SC = 0.142; p = 0.004) and the risk of suicide (SC = 0.207; p = 0.010) (Supplementary Material; http://cadernos.ensp.fiocruz.br/static//arquivo/suppl-e00055621_6499.pdf).Then, we analyzed each component of unhealthy diet (fast food, salty snacks, sweet foods, sugar-sweetened beverages) in the proposed model, replacing added sugar and saturated fat, and found that they were all associated with depression (p < 0.05) and suicide risk (p < 0.05).However, when fast food and salty snacks were assessed in the model using a correlation command to sugar consumption, neither were associated to depression and suicide risk (Supplementary Material; http://cadernos.ensp.fiocruz.br/static//arquivo/suppl-e00055621_6499.pdf).

Discussion
The major risk factors for NCDs (substance use behaviors and added sugar consumption) were associated with both depression and suicide risk in adolescents.Regarding unhealthy eating markers, only added sugar consumption was associated with mental disorders.The female sex was also associated with depression and suicide risk.As a more distal determinant, a better socioeconomic status protected individuals against both depression and suicide risk.
Substance use behaviors composed a latent variable deduced from the shared variance of cigarette smoking, illicit drug use, and consumption of alcoholic beverages.This variable was strongly associated with both depression and suicide risk in adolescents.The early use of psychoactive substances Cad.Saúde Pública 2022; 38(2):e00055621 has also been shown to be related to depression and suicide risk 13,32,33 .One explanatory mechanism is that smoking leads to dysregulation of the hypothalamic-pituitary-adrenal axis and to hypersecretion of cortisol, which can induce depression 34 .However, we cannot rule out a reverse causality, in which depressed individuals are at higher risk of cigarette smoking and psychoactive substance use 35 .
Our study found that added sugar consumption was associated with depression and suicide risk, corroborating with previous reports 8,9,36,37 .Consuming added sugars induces addiction and is correlated with other addictive behaviors, including alcoholism, smoking, and illicit drug use 38 ; the exposure to these addictive substances is, in turn, associated with depression and suicide risk 32,33,35 .
A meta-analysis indicated that the higher the frequency of sugar-sweetened beverages consumption, the higher the risk of having depression; however, this analysis found inconsistencies in the diagnosis of depression and emphasized the need to better adjust for confounders 8 .A population-based study involving adolescents found an association between a higher frequency of sugarsweetened beverages consumption and suicidal behavior 9 .Our study corroborates with the topic since depression and suicide risk outcomes were measured using a validated diagnostic instrument MINI 27 .Moreover, SEM allowed adjusting for confounders, including other diet components and direct path analysis.
Added sugar consumption could be associated with depression and suicide risk in adolescents because the fructose in these beverages affects the hypothalamus-pituitary-adrenal axis 39 .Furthermore, sugar-rich diets can increase inflammation and neurodegeneration 40 .Persistent hyperglycemia can lead to the synthesis of reactive oxygen species that exert neurotoxic effects on the brain 41 .The consumption of added sugar can cause insulin resistance, which is also a known risk factor for depression 42 and suicidal behavior 43 .
Saturated fats, fast foods, and salty snacks were not associated with depression or suicide risk when the model adjustment included added sugar consumption.These findings are contrary to previous studies, which show that the consumption of food rich in trans and saturated fats are associated with symptoms of depression 29 and suicide risk 30 .One study grouped foods to reduce the correlation among unhealthy items (sweets, cake/cookies, snacks, and fast food) in multiple logistic regression models to depression 29 .Another study included soft drink adjustment in the association between foods rich in trans/saturated fats with suicide risk 30 , without considering the correlation among these components.A high correlation between added sugar consumption and saturated fat consumption was observed in the depression model (SC = 0.408; p < 0.001) and in the suicide risk models (SC = 0.408; p < 0.001) (data not shown).Altogether, these findings allowed us to speculate that added sugar consumption is the main component of an unhealthy diet associated with depression and suicide risk.
Studies have shown that physical activity is a protective factor against depression 44 and suicide risk in youngsters 45 which adjusts to smoking but not to other behavioral risks, including unhealthy diet and alcohol.In depressed individuals, physical activity can increase neurogenesis and hippocampal volume 46 and balance anti-inflammatory and proinflammatory markers 47 .However, physical inactivity in adolescents was not associated with depression and suicide risk in our study.This could be because of the model adjustment to components of unhealthy lifestyle and the correlation among them in adolescence.A study which showed that physical activity protects against the risk of suicide 45 did not adjust the model to unhealthy diet, alcohol consumption, and smoking.
Higher body fat levels were not associated with depression or suicide risk in adolescents.Other studies showed that obesity was associated with depression 48 and suicide risk 49 since visceral fat increases the levels of inflammation 11,12 and because obese adolescents with depression had insulin resistance and lower hippocampal volume 50 .However, common causes of insulin resistance, including added sugar consumption and alcohol abuse, were consistently associated with depression and suicide risk in our results and could precede the association between obesity and mental disorders.
A better household socioeconomic status protected individuals against depression and suicide risk.The effect of socioeconomic status was direct and not mediated by the major risk factors for NCDs.Other untested paths could explain the association between social vulnerability and increased depression and suicide risk, including a greater exposure to violence and epigenetic inflammatory mechanisms that can occur in deprived environments throughout life.This study has some limitations.The cross-sectional design of the study does not allow us to affirm temporality between the major risk factors for NCDs and mental disorders nor to rule out reverse causality or bidirectionality in the associations found.However, to the best of our knowledge, this is the first study to analyze all the major behavior risk factors for NCDs according to WHO, showing that substance use behaviors and added sugar intake already affect depression and increase suicide risk in adolescents, regardless of the social vulnerability.These findings could help explain the clustering of mental health disorders and NCDs in adulthood.A strong point of our study is the use of fat mass index, and not body mass index, to measure obesity.
The major risk factors for NCDs, including substance use behaviors and added sugar consumption, were associated with depression and suicide risk behavior in adolescents.Our results indicate that coordinated actions are essential to surveil, to prevent, and to control NCDs and mental disorders according to the WHO Cluster of Noncommunicable Diseases and Mental Health (NMH) 51 , including the behavior risk factors for NCDs and beginning at the earliest possible age.Further studies must explore the extent to which modifications on the risk factors for NCD can affect mental disorders.

Figure 1
Figure 1Theoretical model proposed to analyze the association between risk factors for noncommunicable diseases (NCDs) and mental disorders in adolescents.São Luís, Maranhão State, Brazil.
Cad. Saúde Pública 2022; 38(2):e00055621 . Furthermore, latent variables of healthy diet components (fiber [g], fruits [g], and vegetables [g]) and unhealthy diet components (fast food [g], salty snacks [g], sweet foods [g], and sugar-sweetened beverages [g]) were tested as analyses of consistency, replacing added sugar and saturated fat consumption in the theoretical model associated with depression and suicide risk, and analyzed by SEM (Supplementary Material; http://cadernos.ensp.fiocruz.br/static//arquivo/suppl-e00055621_6499.pdf).This project was approved by the Research Ethics Committee of the University Hospital of the Federal University of Maranhão (HU-UFMA; approval n. 1,302,489, on October 29, 2015) and all participants signed an informed consent form.

Table 3
Factor loading, standard error, and p-values of the indicators of latent variables (household socioeconomic status and substance use behaviors).São Luís, Maranhão State, Brazil, 2021.
SC: standardized coefficient; SE: standard error.* Household socioeconomic status: latent variable defined by the schooling level of the head of household, economic class, adolescent's schooling level, and household income; ** Substance use behaviors: defined by cigarette, illicit drug, and alcohol use.

Table 4
Standardized coefficient (SC), standard error (SE), and p-values of the total and direct effects of the explanatory variables on multiple outcomes.São Luís, Maranhão State, Brazil, 2021.
* Household socioeconomic status: latent variable defined by the schooling level of the head of household and the adolescent's economic class, schooling level, and household income; ** Substance use behaviors: defined by cigarette, illicit drug, and alcohol use.