Introduction

Common mental disorder—comprising anxiety and depression—and obesity exact a considerable public health burden.[1,2] The secular rise in the prevalence of both in recent years has, in part, prompted speculation about their inter-relationship. In observational studies, there is growing evidence to suggest that people with common mental disorders have a greater risk of obesity than those who are apparently free of such mental health problems, although these findings are not entirely concordant. [3–5]There are also some studies that have explored the converse hypothesis: obesity as a risk factor for future mental disorders. Again, the results from these studies have been contradictory, with evidence of positive,[6,7] null,[8,9] and even an inverse association.[10]

Several mechanisms may explain these apparent bidirectional associations. These may be context specific and include attitudes towards body size and mental ill health in the population under study.[11] For example, in societies where obesity is stigmatised, being overweight or obese may lead to increased risk of anxiety and depression, whereas the reverse effect may be seen in communities where a large body size is prized. Additionally, commonly used treatments for depression have known side effects that result in weight gain.[12,13] Given the public health importance of increasing rates of both obesity and common mental disorder, understanding the nature of the association between these two conditions is crucial as it could inform prevention and treatment. This is particularly true in Western societies as the United Kingdom, where the prevalence of obesity almost doubled in the last three decades.[14,15]

Prospective studies with repeated measurements of both common mental disorder and different forms of obesity, such as general and abdominal adiposity, offer an important opportunity for an in-depth analysis of the inter-relationship between obesity and mental health. However, such studies are scarce.[16,17] The Health and Lifestyle Survey, which is a representative sample of the UK population, had an assessment of adiposity and mental health both at baseline and re-survey seven years later. These data afford us the opportunity to examine the potential for a prospective reciprocal association between common mental disorder and general or abdominal obesity.

Methods

Re-survey (HALS-2)

In 1991, seven years after the first survey, a follow-up study (re-survey) was carried out in order to identify changes that occurred in health and lifestyle among the original respondents.[18] Fig 1 shows the response at tracing and each level of interview. Of the original 9 003 participants of the baseline survey, a total of 6 626 respondents were traced, alive at the time of the re-survey and 81% (5 352) were successfully re-interviewed. From these, 869 participants were not measured for anthropometric variables and 612 participants did not return the GHQ questionnaire. The response proportion was higher in the younger age groups, and lowest among those over 80 years of age. A higher proportion of interviews were achieved in non-manual than manual occupational social classes. The response proportion for those undergoing the measurements by the nurse was 94.2% of those who had been measured at the baseline survey. The proportion of those returning the self-completion questionnaire that also returned it at the baseline survey was 84.1%. The analytical sample consisted of 3,388 participants with complete data on obesity and mental health at baseline and re-survey, and the covariates. All the data used for the present analyses are available upon request at http://www.esds.ac.uk/findingData/halsTitles.asp.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Participants selection flow in the HALS.

*These subjects with missing data for any of the following covariates: age, general health questionnaire, marital status, occupational social class, alcohol consumption, smoking status, height, weight and waist circumference.

https://doi.org/10.1371/journal.pone.0119970.g001

Results

Compared to participants excluded from the analyses (n = 3375), there was no statistically significant difference for gender (p = 0.42), alcohol consumption (p = 0.39) and BMI (p = 0.64). However, excluded participants were one year older (p = 0.002), smoked more and had a one centimetre higher waist circumference (S1 Table).

Characteristics of the study population are shown in Table 1. At baseline, men had significantly higher BMI (p < 0.001) and waist circumference (p < 0.001), but the prevalences of global obesity (p < 0.009) and severe abdominal obesity (p < 0.065) were higher in women. As expected, common mental disorder was more frequent in women than in men (p < 0.001). At follow-up, the prevalences of global and severe abdominal obesity were significantly higher among women (both p values < 0.001, see Table 1). Moreover, the development of severe abdominal obesity (p < 0.001) was more common in women. The prevalence of common mental disorder was more frequent in women (all p < 0.001) at baseline and at the re-survey studies.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Study participants’ characteristics at baseline (1984) and re-survey (1991) according to gender.

https://doi.org/10.1371/journal.pone.0119970.t001

Baseline common mental disorders as a risk factor for re-survey global and abdominal obesity

In Table 2 we show the relation between common mental disorder at baseline and the development of general and abdominal obesity at the re-survey. Relative to those who were free of common mental disorder, men with the condition experienced a doubling of the risk of developing general obesity by re-survey; there was essentially no suggestion of an effect for global overweight. There was no relation between common mental disorder and waist circumference in men. In women, there was an association of common mental disorder at baseline with later overweight (p = 0.02). Also, the odd of developing common mental disorder was higher in women having baseline moderate (p = 0.001) or severe (p = 0.01) abdominal obesity compared to those with normal waist circumference. Throughout these analyses, controlling for a range of covariates had essentially no impact on the effect estimates.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Odds ratios and 95% CIs for the relation of between common mental disorder among non-obese subjects at baseline (1984) and the development of global or abdominal obesity at re-survey (1991).

https://doi.org/10.1371/journal.pone.0119970.t002

In Table 3 we present the relation of changes in common mental disorder status between baseline and re-survey on the development of general or abdominal obesity at re-survey. Compared to study members free of common mental disorder at both surveys, men with common mental disorder at baseline and with normal mental health at re-survey had higher risk of obesity but not of overweight. The prevalence of neither moderate nor severe abdominal obesity in men differed across the different classes of exposition to common mental disorder. The prevalences of overweight, obesity and abdominal obesity were 61% to 200% significantly higher in women with common mental disorder at both surveys in comparison to those with normal mental health at both surveys. For both genders, these differences persisted when all the other studied covariates were included in the models.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Odds ratios and 95% CIs for the relation between changes in common mental disorder status between baseline and re-survey on the development of general or abdominal obesity at re-survey.

https://doi.org/10.1371/journal.pone.0119970.t003

General and abdominal obesity as a risk factor for the development of common mental disorders at re-survey

In Table 4 are presented the relationship between general and abdominal obesity at baseline and common mental disorder at re-survey. Neither baseline BMI categories nor baseline waist circumference categories was associated with common mental disorder at re-survey in either women or men. Compared to women having a BMI < 30 kg/m² at both surveys (data not shown), women having a BMI > 30 kg/m² at baseline but < 30 kg/m² at re-survey study had higher risk of common mental disorder when adjustments were made for age (1.74, 1.04–2.91) or in a fully-adjusted model (1.81, 1.07–2.09).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Odds ratios and 95% CIs for the relation between general or abdominal obesity among subjects free of common mental disorders at baseline (1984) and the development of common mental disorders at the re-survey study (1991).

https://doi.org/10.1371/journal.pone.0119970.t004

Discussion

Findings from this well characterized population-based cohort of British men and women provide evidence that common mental disorders may be a risk factor for future obesity. This relationship was found for measures of obesity based on BMI (men and women) and waist circumference (women only). Similar patterns of associations were found in women when BMI and waist circumference as a continuous, rather than categorical, variables were used as markers of anthropometry. We found no evidence of a bidirectional or reciprocal association in which obesity also predicts future risk of common mental disorder.

Despite high response at the successive data collection phases, differential drop out rates between the two surveys due to death, refusal and non-tracing affected the sample distribution. The loss to follow-up in the HALS is similar to other studies. This attrition has implications for estimating the prevalence of a given risk factor or disease. However, as we have also demonstrated with these data, using the exemplar of CVD risk, this attrition has essentially no impact on the estimation of risk factor-disease association. Thus, in relating risk factors to CVD mortality, the same magnitude of association was evident in people who participated in the baseline survey but not the resurvey relative to the group that took part in both the baseline survey and re-survey (p-value for interaction: 0.108).[26] Nevertheless, the distributions of the HALS-2 population compared reasonably well with that of the 1991 British census data respectively.[19,20,27] Common mental disorder was assessed using the self-administered 30-item GHQ, which focuses on self-reported symptoms of anxiety and depression, and associated psychosocial dysfunction.[21] This device is designed as a screening instrument for use in community settings. It has been validated against standardized clinical interviews and has shown high reliability.[28] However, the GHQ is not a measure of clinically recognised psychiatric disorder.[21] Although the symptom scale is reliable, we cannot be certain that our findings would be directly transferable to individuals meeting the DSM–IV[29]or ICD–10[30] criteria for specific mental disorders such as major depressive disorder or anxiety disorders.

Anti-depressive drugs intake was not recorded during the HALS-1 Study. This would have helped in better characterizing subjects at high risk of developing obesity by the time of the re-survey study. Anti-depressant use in the general population is rare as found during the re-survey study (2% of men and 4% of women). Few participants were present in the higher categories of obesity. This could lead to unstable findings. However, significant results in the presence of a low statistical power usually indicate a strong association.

Among previous studies that evaluated the association between mental health and obesity, only two used the GHQ to characterize common mental disorder.[17,31] Our findings confirm the results previously described [17,31] such as common mental disorder might be a risk factor of future general obesity. None of these studies examined the association with both general and abdominal adiposity. Other longitudinal studies have evaluated various markers of mental health in predicting general obesity, and found a positive[32,33] or no association.[8,9,34] Pan et al [16] reported a bidirectional association between depression and obesity in older women. Luppino et al [7] reported in the most recent meta-analysis of longitudinal studies in this field that baseline depression (symptoms and disorder) was predictive of general obesity over time (OR: 1.58, 1.33–1.87). The specific role of abdominal adiposity was not evaluated in this report.

In agreement with a previous study using repeat measures of GHQ to characterize mental health,[17] we found no robust association between obesity and subsequent common mental disorder. However, using various surrogates of mental health, others authors have found a positive [6,35] or no association.[8,9] It has been suggested that the association between common mental disorder and general obesity might be influenced by gender, mean age of study participants at baseline, follow-up duration and ethnic groups.[4,7,36] However, we performed statistical analyses separately for men and women, had little variation in follow-up periods between participants and the findings were robust to adjustment for age. This suggests that major bias is unlikely.

Although evidence of biological links between mental health and obesity remain complex and not definitive,[3,10,37] it seems relevant to highlight the most current hypotheses explaining the possibility of a biological pathway. First, common mental disorders are associated with eating disorders, which could influence future changes in adiposity.[38,39] Physical inactivity, a major contributing factor to obesity, is more prevalent among people with mental health problems.[40] Furthermore, commonly used pharmacological treatments for mental health problems have known side effects that may result in weight gain, weight loss or both.[13,41] The fact that mental health disorders induce an increase of weight over time may also be related to neuroendocrine disturbances, such as long-term activation of the hypothalamic–pituitary–adrenocortical (HPA) axis in the context of an increased visceral fat.[42] Cortisol, in the presence of insulin, inhibits lipid-mobilizing enzymes, a process mediated by glucocorticoid receptors that are found in fat depots and especially in intra-abdominal visceral fat.[43]

The association of obesity with future depression may be related to diagnosis of a chronic disease (or the treatment in terms of drugs and dietary restrictions) rather than to obesity. This possibility is consistent with our findings suggesting no real association between obesity and subsequent common mental disorder. Obesity can be seen as an inflammatory state, as weight gain has been shown to activate inflammatory pathways[44] and inflammation in turn has been associated with depression.[45]

In conclusion, our findings of a longitudinal association between common mental disorders and obesity may have important implications for clinical practice. Because weight gain appears to be a late consequence of common mental disorders, care providers should be aware that within patients showing symptoms of psychological distress or depression, weight should be monitored carefully. This awareness should lead to prevention, early detection, and co treatment of patients at risk, which could ultimately reduce the burden of common mental disorders. Longitudinal epidemiological research is especially warranted to further establish the plausible mechanisms underlying the link between common mental disorders and the onset of obesity, such medication use, physical activity and dietary patterns.

Supporting Information

Acknowledgments

We are grateful to the original creator and principal investigator of the HALS-1 and HALS-2 studies, Prof. Cox D.B, from the school of clinical medicine, university of Cambridge. We also thank the UK Health Promotion Trust who funded the HALS-1 and HALS-2 studies. The UK Data Archive provided free of charge to us the HALS-1 and HALS-2 databases. The original data creators, depositors or copyright holders, the funders of the Data Collections and the UK Data Archive bear no responsibility for data analyses or interpretation.