Obesity and depressive symptoms in mid-life: a population-based cohort study

Obesity and depression are both highly prevalent public health disorders that affect many age groups and communities [1, 2]. Globally, more than 600 million people are living with obesity while depression affects over 350 million people [1, 3]. Obesity and depression are interrelated; both are known risk factors for cardiovascular disorders and diabetes [4‐6], and relate to negative health and lifestyles factors such as disturbed sleep patterns, sedentary behaviours and dysregulation in appetite and food intake [7]. These interrelations are clearly complex as use of antidepressant medication often leads to weight gain, as suggested from a review of ten clinical studies that explored the effect of antidepressants on body weight [8]. Furthermore, as evidenced by clinical trials conducted on women [9], dietary restriction targeting weight reduction in obesity may in turn exacerbate depression.

Previous observational studies on the association between obesity and depressive symptoms have provided mixed findings [10‐17], with some suggesting a positive association [10], while others have reported an inverse association (obesity associated with lower risk of depression) [12], no association [13], or a u-shaped association (higher risk of depressive symptoms among underweight and obese individuals) [14]. These inconsistencies can arise from differences in population characteristics (for example, ethnicity and age) [12, 15, 18], residual confounding [19, 20], or potentially, differences in measures used to define obesity [21]. For example, a study of 3666 individuals from different ethnic groups in USA suggested an association among White but not Black or Mexican Americans [18]. A meta-analysis of 13 cross-sectional obesity-depression association studies (median n = 1215), indicated differences in point estimates and statistical precision based on the number of confounders included in the adjustment [20]. Furthermore, most previous studies consider “general” obesity, as measured by body mass index (BMI) [22, 23], while notably less attention has been paid to potential differences in risk based on body fat distribution. While BMI is widely used, it has well-known limitations and may reflect differences in muscle rather than adipose tissue [21]. Some studies have associated central obesity, as measured by waist circumference (WC) or waist-hip ratio (WHR), with depression risk [24]. Adipose tissue is an active endocrine organ associated with increased inflammatory response which relates to depression [25]. White adipose tissue which accumulates typically around the waist is a source of a range of inflammatory markers, which can pass the brain-blood barrier and thereby affect depression risk [26]. Another possibility is that dysregulation of the hypothalamus-pituitary-adrenal (HPA) axis could increase the accumulation of fat around the abdomen as well as leading to alterations in mood [27], possibly indicating a greater importance of central rather than general obesity with respect to depression risk.

This study used data from a large-scale population-based cohort to establish whether depressive symptoms in mid-adulthood are associated with concurrent general or central obesity and related lifestyle factors, and further, whether obesity in turn predicts the longer term risk of depressive symptoms after accounting for potential sociodemographic and lifestyle confounders and the presence of depressive symptoms at baseline.

At age 45 years, around a quarter of participants were classified as having general obesity and 35% with central obesity. The prevalence of depressive symptoms was 8.4% at 45 years and 12.4% at 50 years. Sociodemographic and lifestyle factors associated with greater prevalence of general and central obesity were mostly also associated with higher prevalence of depressive symptoms (Table 1).

Table 2 shows the cross-sectional associations of depressive symptoms with obesity and lifestyle factors at 45 years. The odds of obesity were higher for those reporting ≥2 symptoms compared to individuals with no depressive symptoms (OR 1.31, 95%CI 1.11 to 1.55 and OR 1.26, 95%CI 1.08 to 1.47, for general and central obesity, respectively). The cross-sectional associations between depressive symptoms and general or central obesity did not vary by sex (P _{sex-interaction} > 0.33). Number of depressive symptoms was also related to sedentary behaviour, smoking and low fruit consumption (P < 0.001 for all comparisons after adjustment for sex, region, social class at birth and in adulthood). The association of depressive symptoms with cigarette smoking and fruit consumption varied by sex (P_{sex-interaction} < 0.02 for both comparisons), and both associations were observed in women but not men. Compared to women without symptoms, those with symptoms at baseline had an OR 1.82 (95% CI 1.38 to 2.19; men OR 1.05, 95% CI 0.78 to 1.40) for smoking and OR 0.51 (95% CI 0.40 to 0.67; men OR 0.76, 95% CI 0.57 to 1.00) for fruit consumption.

Both general and central obesity at 45 years were associated with higher odds of depressive symptoms at age 50 years in women but not in men (Table 3, P_{sex-interaction} < 0.01 for all comparisons). In women, the observed associations between BMI and WC with depressive symptoms were somewhat attenuated by adjustment for sociodemographic and lifestyle factors, but for both obesity indicators the associations remained even after accounting for baseline symptoms, sociodemographic and lifestyle factors (BMI: OR per 10 kg/m² 1.36, 95%CI 1.15 to 1.62; WC: OR per 5 cm 1.07, 95%CI 1.03 to 1.11). In men the fully adjusted OR for BMI was 0.83 (95%CI 0.64 to 1.10, P_trend = 0.20) and for WC 0.99 (95%CI 0.94 to 1.04, P_trend = 0.67). In a model including all covariates and mutually adjusted for BMI and WC, the association in women for BMI was abolished (OR 0.96, 95%CI 0.77 to 1.18) while the estimate for WC remained significant (OR 1.08, 95%CI 1.02 to 1.15). Again, the corresponding mutually adjusted model among men did not suggest an association either for WC or BMI (P > 0.16 for all comparisons).

To further understand the association between obesity and depressive symptoms, we explored the prospective associations between lifestyle factors and depressive symptoms. Alcohol consumption, smoking, sedentary behaviour and fruit consumption were all associated with depressive symptoms at 50 years after adjustment for depressive symptoms at 45 years, sex, sociodemographic and other lifestyle factors (Table 4). The presence of depressive symptoms at age 45 years was associated with an over four-fold greater odds of depressive symptoms at 50 years, attenuating only slightly after adjustment for obesity and lifestyle factors (OR reduced from 4.43 (95%CI 3.61 to 5.43) to 4.11 (95%CI 3.34 to 5.06)), suggesting further unmeasured influences on the tracking of depressive symptoms.

Tackling the current obesity epidemic both in terms of addressing its causes and consequences remains a key public health priority. Mental wellbeing has broad influences on health and as shown in our large-scale study, the presence of depressive symptoms is strongly associated both with obesity and a broad range of related lifestyle factors. However, as shown in our analyses, even after controlling for these concurrent associations, obesity several years earlier appears to affect subsequent odds of depressive symptoms among women but not in men, providing an important target group for implementation of preventative strategies. Furthermore, this study provided some support for the hypothesis that central rather than general adiposity may have a key role in contributing to the association between obesity and the risk of developing depressive symptoms, highlighting the need of further studies to establish related mechanisms and pathways.

Our work builds on an earlier analysis in the 1958 cohort, which suggested that when looking at categories of BMI from childhood to adulthood (age 7 onwards) obesity was associated with the subsequent odds of depressive symptoms in women but not in men [15]. Inclusion here of central obesity is important because BMI has known weaknesses, especially for men, due to the strong correlation with muscle mass [21]. The result from our previous work [15] is consistent with the current study, and both the analysis using BMI (or general obesity) and WC (or central obesity) suggest that obesity is affecting the likelihood of developing depressive symptoms in women only.

Our findings for higher odds of depressive symptoms by general and central obesity are consistent with several other previous studies [20, 22]. For example, a meta-analysis including eight cross-sectional studies reported the odds of depression to be 35% higher in people with obesity compared to normal weight individuals [42]. A meta-analysis reporting prospective associations also included eight studies, six of them assessed to be of poor quality by the authors, suggesting an overall 55% higher odds of depression for those with obesity compared to normal weight, with no evidence for sub-group differences by sex [22]. Another meta-analysis including 15 studies showed 38% higher odds of depression in individuals with central obesity compared to those without, with higher estimates among women than men [20]. Importantly, unlike these previous studies [20, 22], our study included an adjustment for depressive symptoms at baseline. Furthermore, the weaker effect estimate in our study compared to previous meta-analyses may be due to better control for sociodemographic or other confounding factors as most previous studies included in the two meta-analyses [20, 22, 42] were either unadjusted or adjusted for limited set of confounders.

We observed a dose-dependent increase in the odds of depressive symptoms by categories of BMI, which is consistent with most previous studies [10, 13, 43]. However, in some studies, the association between general obesity and depressive symptoms has favoured the “Fat and Jolly” hypothesis, suggesting that obesity may protect against depression [44] while others show a higher odds both for the very thin and obese individuals compared to those with normal weight [14, 16, 45]. It is possible that an increased odds of depression for those who are thin or underweight, can reflect sub-optimal health status [46]. These inconsistencies among studies could be further explained by contextual, and psychosocial influences contributing to the obesity – depression relationship. Individuals who experience stigma and discrimination due to their weight may have low self-esteem and high degree body image dissatisfaction that further aggravate psychological stress and lead to depressive symptoms [47]. Indeed, the possible stigma and discrimination due to excessive body weight is likely to have a greater impact on women than men which may explain the observed gender difference [47].

Mechanisms underlying the association between obesity and depression remains unclear, however, molecular and clinical studies have provided some evidence for the involvement of HPA axis, inflammatory pathways and insulin sensitivity [48], with dysregulation of HPA axis, elevation of certain inflammatory markers and insulin resistance being observed in both individuals with obesity and in individuals with depression [49‐51]. Disturbance in any of these systems may affect the secretion or metabolism of neurotransmitters, such as serotonin, norepinephrine and dopamine, in the brain and consequently influence mood [48, 52]. Findings regarding inflammatory pathways and insulin sensitivity are particularly interesting: the observed elevated inflammatory markers, including leptin, adiponectin and IL-6 and TNF-α are mainly secreted from white adipose tissue found on the abdomen [26]; abdominal obesity, in particular, places people at higher risk for developing insulin resistance [53]. All these observations resonate with our findings of an independent association for WC and highlight the particular importance of central obesity on the risk of developing depressive symptoms.

Our study also suggests that the association between obesity and depressive symptoms is in part related to socioeconomic and lifestyle factors, with obesity associations attenuating by about 30% after allowing for such factors. This attenuation was expected considering that most studies show lower socioeconomic status and unhealthy lifestyle factors to be associated with obesity and depression [54, 55] although some factors may be on the causal pathway from obesity to depressive symptoms. For example, obesity may lead to less active lifestyles as well as inactivity increasing the risk of obesity [56]. Even so, given the links shown here between lifestyle factors and depressive symptoms, our study suggests that these behaviours may be additional targets for interventions to reduce obesity and depression.

Depressive symptoms are associated with obesity and related lifestyle factors in mid-life. Our study suggests that obesity affects subsequent risk of developing depressive symptoms among women independently of these concurrent associations, providing an important target group for implementation of preventative strategies. In the management of depressive symptoms, one of the alternative approaches could be targeting and managing obesity and unfavourable lifestyle factors.