RandallcHath Depression, Emotional Eating and Long-term Weight Changes Paper Depression, emotional eating and long-term weight changes: a population-based

RandallcHath Depression, Emotional Eating and Long-term Weight Changes Paper Depression, emotional eating and long-term weight changes: a population-based prospective study Konttinen et al. International Journal of Behavioral Nutrition and Physical Activity
(2019) 16:28
https://doi.org/10.1186/s12966-019-0791-8
RESEARCH
Open Access
Depression, emotional eating and longterm weight changes: a population-based
prospective study
Hanna Konttinen1,2* , Tatjana van Strien3,4, Satu Männistö5, Pekka Jousilahti5 and Ari Haukkala2
Abstract
Background: Emotional eating (i.e. eating in response to negative emotions) has been suggested to be one
mechanism linking depression and subsequent development of obesity. However, studies have rarely examined this
mediation effect in a prospective setting and its dependence on other factors linked to stress and its management. We
used a population-based prospective cohort of adults and aimed to examine 1) whether emotional eating mediated
the associations between depression and 7-year change in body mass index (BMI) and waist circumference (WC), and
2) whether gender, age, night sleep duration or physical activity moderated these associations.
Methods: Participants were Finnish 25- to 74-year-olds who attended the DILGOM study at baseline in 2007 and
follow-up in 2014. At baseline (n = 5024), height, weight and WC were measured in a health examination. At follow-up
(n = 3735), height, weight and WC were based on measured or self-reported information. Depression (Center for
Epidemiological Studies – Depression Scale), emotional eating (Three-Factor Eating Questionnaire-R18), physical activity
and night sleep duration were self-reported. Age- and gender-adjusted structural equation models with full
information maximum likelihood estimator were used in the analyses.
Results: Depression and emotional eating were positively associated and they both predicted higher 7-year increase in
BMI (R2 = 0.048) and WC (R2 = 0.045). The effects of depression on change in BMI and WC were mediated by emotional
eating. Night sleep duration moderated the associations of emotional eating, while age moderated the associations of
depression. More specifically, emotional eating predicted higher BMI (P = 0.007 for the interaction) and WC
(P = 0.026, respectively) gain in shorter sleepers (7 h or less), but not in longer sleepers (9 h or more).
Depression predicted higher BMI (P < 0.001 for the interaction) and WC (P = 0.065, respectively) increase in younger participants, but not in older participants. Conclusions: Our findings offer support for the hypothesis that emotional eating is one behavioural mechanism between depression and development of obesity and abdominal obesity. Moreover, adults with a combination of shorter night sleep duration and higher emotional eating may be particularly vulnerable to weight gain. Future research should examine the clinical significance of our observations by tailoring weight management programs according to these characteristics. Keywords: Depressive symptoms, Emotional eating, Three-factor eating questionnaire, Physical activity, Sleep, Weight gain, Obesity, Longitudinal study, Moderation, Mediation * Correspondence: hanna.konttinen@helsinki.fi 1 Department of Food and Nutrition, P.O. Box 66, 00014 University of Helsinki, Helsinki, Finland 2 Faculty of Social Sciences, P.O. Box 54, 00014 University of Helsinki, Helsinki, Finland Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Konttinen et al. International Journal of Behavioral Nutrition and Physical Activity Background It has been estimated that worldwide over 300 million people suffer from depression and over 650 million are affected by obesity [1, 2]. The consequences of these conditions in terms of lost health, functioning and quality of life are huge – depression and obesity are both related to an elevated risk of developing several chronic diseases and depression is a major contributor to suicide deaths [1, 2]. There is thus a critical need to develop interventions that are effective in reducing the occurrence of both conditions. Numerous studies have demonstrated that depression and obesity often occur together and are bi-directionally associated over time [3, 4]. In an exploration of possible underlying mechanisms linking depression and obesity, a population-based cross-sectional study showed the link to be mediated by emotional eating [5, 6]. Emotional eating refers to a tendency to eat in response to negative emotions (e.g. depression, anxiety, stress) with the chosen foods being primarily energy-dense and palatable ones [6–8]. It can be caused by various mechanisms, such as using eating to cope with negative emotions or confusing internal states of hunger and satiety with physiological changes associated with emotions [9]. Using the 7-year follow-up data of the same population-based sample, the present study assessed whether emotional eating also acts as a mediator between depression and subsequent weight gain, and whether such a mediation effect is dependent on other factors, including gender, night sleep duration and physical activity. A more detailed knowledge of these factors may point out novel targets for improved obesity and depression interventions to decrease the global burden of disease and increase individual well-being. Depression (depression-melancholia) is typically characterized by loss of appetite and subsequent weight loss, but there exists also a depression subtype that is characterized by the a-typical vegetative symptom of increased appetite and weight gain [10–12]. Emotional eating has been considered a marker of this a-typical depression subtype, because it shares with this depression subtype the a-typical feature of increased appetite in response to distress [13, 14]. The depression – obesity link may therefore be mediated by emotional eating, for which there was indeed support in various cross-sectional studies for both genders [5, 6, 15, 16] and for women [17]. To date, studies have rarely examined the links between depression, emotional eating and weight gain in a prospective setting. As an exception, a 5-year study in Dutch parents [18] and an 18-year study in mid-life US adults [19] demonstrated that emotional eating acted as a mediator between depression and BMI gain or obesity development particularly in women. With the evidence from the above studies regarding gender being partly mixed, it remains inconclusive whether the mediation (2019) 16:28 Page 2 of 11 effect of emotional eating varies across women and men. Gender was therefore one of the moderators tested in the present prospective study. The mediation effect of emotional eating between depression and weight gain may also depend on physical activity and sleep duration, though to our best knowledge their moderating effects in this context have not been directly tested before. Both factors have been linked to stress and its management, with exercise being a treatment for depression and anxiety disorders [20–22] and short sleep duration being associated with psychological stress [23, 24]. Higher physical activity has also been associated with lower emotional eating [25, 26]. Accordingly, it has been proposed that increasing physical activity could be a viable strategy to reduce excessive intake of high-fat and -sugar foods under negative emotional states [27] and extending sleep duration could have comparable effects [28]. Exercise could thus attenuate the effects of depression and emotional eating on weight gain via improvements in emotion regulation. In contrast, short sleep duration might strengthen their effects on weight gain – i.e. reduced sleep can be seen as a stressor itself and a marker of perceived stress [29, 30] and evidence is emerging that it interferes with emotion regulation [31]. In support of this, findings from a laboratory study of 64 women suggested that short sleep duration (less than 7 h per night) may act as a stressor and lead to elevated snack intake in those prone to emotional eating [32]. A few observational studies have also found that sleep duration and physical activity moderated the emotional eating – weight gain association. Van Strien and Koenders [29] studied a sample of Dutch employees and observed that women with a combination of short sleep duration and high emotional eating experienced the greatest increases in body mass index (BMI) over 2 years. A similar pattern of findings was reported by Chaput et al. [33] in a sample of French Canadian adults with 6-year follow-up and information on disinhibited eating behaviour (i.e. tendency to overeat in response to food or emotional cues). Moreover, emotional eating was less strongly associated with BMI and its gain in participants with high physical activity than in those with low physical activity in the Dutch employee sample [34] and in a Swiss population survey [26]. However, it is important to explore whether these findings can be replicated and extended using an independent sample of adults with long-term follow-up as well as information on symptoms of depression and change in abdominal obesity. In the present study, we used a large population-based 7-year prospective cohort of adults to increase our knowledge on the interplay between depression, emotional eating and weight changes in the context of gender, night sleep duration and physical activity patterns. Konttinen et al. International Journal of Behavioral Nutrition and Physical Activity (2019) 16:28 Page 3 of 11 Because of the large age range (between 25 and 74 years at baseline) in this sample, we were also interested in the possible moderating effects of age. More specifically, our aims were to examine 1) whether emotional eating mediated the associations between symptoms of depression and 7-year change in BMI and waist circumference (WC), and 2) whether gender, age, night sleep duration or physical activity moderated these associations. The research protocols of the DILGOM baseline and follow-up studies were designed and conducted in accordance with the guidelines of the Declaration of Helsinki and have been approved by the Ethics Committee of Helsinki and Uusimaa Hospital District (decision numbers 229/E0/2006 and 332/13/03/00/2013, respectively). In addition, written informed consent was obtained from all participants. Methods Outcome variables BMI and WC Participants and procedure Participants were 25- to 74-year-old Finnish men and women who attended the baseline (n = 5024) and follow-up (n = 3735) phases of the DIetary, Lifestyle and Genetic determinants of Obesity and Metabolic syndrome (DILGOM) study (for a participant flow chart, see [35]). The baseline phase was conducted in 2007 as a part of the FINRISK 2007 study in which a random sample of 10,000 people, stratified by 10-year age groups and gender, was drawn from the Finnish population register in five large study areas [36]. All participants who attended the FINRISK 2007 study (n = 6258, response rate = 63%) in January–March were invited to the DILGOM 2007 study (n = 5024, response rate = 80%) conducted in April–June. The baseline phase contained a health examination (including measurements on height, weight and WC) at a study center and several self-administered questionnaires completed either during the visit or at home. All baseline participants alive at the end of the year 2013 received an invitation to the follow-up phase conducted in April–June 2014 (n = 3735, response rate = 82%). The data collection was carried out in two groups: 1) participants who lived in the areas of Turku and Loimaa and in the cities of Helsinki and Vantaa were invited to a similar health examination to the one at baseline (n = 1312); 2) participants who lived in the other three study areas (North Karelia, North Savo, Oulu) received a mail-back questionnaire and self-reported their current weight and height (n = 2423). They also measured their WC themselves, with a measurement tape that was sent to them together with detailed measurement instructions. Participants who did not attend the follow-up phase were more often men (?2 = 7.22, df = 1, P = 0.007) and tended to be younger (F(1, 5022) = 13.83, P < 0.001, ?2 = 0.003) and have higher BMI and WC (F(1, 5015) = 26.56, P < 0.001, ?2 = 0.005 and F(1, 4992) = 30.88, P < 0.001, ?2 = 0.006, respectively) at baseline than follow-up participants, but these mean differences were small in size. There were no statistically significant differences between these two groups of participants in terms of baseline education (F(1, 4983) = 3.68, P = 0.055, ?2 = 0.001), depression (F(1, 4727) = 3.70, P = 0.055, ?2 = 0.001) or emotional eating (F(1, 4853) = 0.60, P = 0.438, ?2 = 0.000). Trained research nurses measured participant’s height, weight and WC by using standardized international protocols [37] at baseline and follow-up. Weight was measured to the nearest 0.1 kg, height to the nearest 0.1 cm and WC to the nearest 0.5 cm. All measurements were made in a standing position in light clothing and without shoes. WC was measured at a level midway between the lower rib margin and iliac crest. At baseline, weight and height measurements were available for 5017 (99.9%) participants to calculate BMI (kg/m2), while WC measurement was available for 4994 (99.4%) participants. At follow-up, BMI and WC were based on measured (n = 1310 and 1305, respectively) or self-reported (n = 2352 and 2288, respectively) information. In a recent validation study conducted in a subset of DILGOM participants, the mean differences between self-reported and nurse-measured height, weight and WC were small and the intra-class correlations were 0.95 or greater in both genders [38]. Respondents with measured and self-reported anthropometric data at follow-up were therefore included in this study. Predictor variables Depression The 20-item Center for Epidemiological Studies - Depression (CES-D) Scale [39] was used to measure depressive symptoms at baseline. The scale is designed to measure depressive symptomatology in the general population, and it has been found to be adequately related to clinical ratings of depression [40]. For each item, respondents indicate how often they have felt in the described way during the past week using a four-point scale (from 0 “rarely or none of the time” to 3 “almost all of the time”). A meta-analysis of 28 studies examining the structure of the CES-D scale concluded that the proposed four-factor structure (negative affect, somatic and retarded activity, lack of positive affect, interpersonal difficulties) best described the scale [41]. In line with this and our previous cross-sectional study [5], we modelled depression as a latent factor with four indicators where each indicator was the mean of the items belonging to the respective original factor. It is noteworthy that the CES-D scale contains one item on loss of appetite (“I did not feel like eating; my Konttinen et al. International Journal of Behavioral Nutrition and Physical Activity appetite was poor”), while there is no corresponding item on increased appetite. We decided to exclude the loss of appetite item from the present analyses, because it represents an unbalanced measurement of appetite change with potentially biasing the measurement towards depression subtype characterized by decreased appetite and weight loss. Thus, somatic and retarded activity indicator variable was calculated based on 6 items instead of 7 items. Emotional eating Emotional eating at baseline was assessed by using the emotional eating scale of the 18-item Three-Factor Eating Questionnaire (TFEQ-R18) [42]. Karlsson et al. [42] developed the TFEQ-R18 on the basis of a factor analysis of the original 51-item TFEQ in the Swedish Obese Subjects study and it has been found to be valid in the general population [43, 44]. The scale contains three items that are all rated on a four-point scale (from 1? does not describe me at all? to 4? describes me exactly?): 1) When I feel anxious, I find myself eating, 2) When I feel blue I often overeat, and 3) When I feel lonely, I console myself by eating. In line with our previous cross-sectional study [5], emotional eating was modelled as a latent factor with the three items as indicators. Moderators and covariates Night sleep duration and physical activity Night sleep duration at baseline was assessed with the following question “How many hours per night do you usually sleep?”. The item was treated as a continuous variable in the analyses. Physical activity at baseline was measured with the International Physical Activity Questionnaire - Short Form (IPAQ-SF) [45]. It asks respondents to report their physical activity during the past 7 days across a comprehensive set of domains (leisure time, work, transport, domestic work and gardening) and three intensity levels (vigorous activities, moderate activities and walking). The data were scored according to the IPAQ manual and a combined total physical activity score (minutes per week) was used on a continuous scale in the main analyses. We repeated the analyses with vigorous physical activity score (minutes per week), but it should be noted that 41.6% of participants had not engaged in any vigorous activities during the past week. Age and gender Baseline age was treated as a continuous variable (years) and gender as a dichotomous variable (1 = men, 2 = women) in the analyses. Statistical methods We used structural equation modelling (SEM) to test the hypothesized mediation models between depression, (2019) 16:28 Page 4 of 11 emotional eating and 7-year change in adiposity indicators. Depression and emotional eating were modelled as latent factors because ignoring measurement error in predictors can lead to biased regression coefficients and latent variables allow taking measurement error into account [46]. The analyses were conducted in three steps. Firstly, confirmatory factor analysis with two latent factors (depression and emotional eating) was used to test whether the four depression indicators and the three emotional eating indicators loaded on separate factors. Secondly, the hypothesized mediation models with baseline age and gender as covariates were estimated separately for change in BMI and WC – change modelled by regressing the measurement at follow-up on the baseline measurement. The absence of an interaction between exposure (i.e. depression latent factor) and mediator (i.e. emotional eating latent factor) in both models allowed us to apply the SEM approach to mediation analysis (? = 0.12, SE = 0.07, P = 0.080 and ? = 0.04, SE = 0.07, P = 0.585 for the interaction in the model for BMI and WC, respectively) [46, 47]. The results were reported as the total, direct and indirect effects (i.e. regression coefficients and bias-corrected bootstrap 95% confidence intervals) of depression and emotional eating. The reported indirect effect reflects how much of the association between depression and change in adiposity indicator is explained by emotional eating [48]. The total effect represents the relationship between depression and change in adiposity indicator before adjustment for emotional eating. Thirdly, the moderator effects of gender, age, night sleep duration and physical activity were examined in a separate set of models by adding a moderator (in the case of sleep duration and physical activity) and interaction terms (moderator × emotiona... Purchase answer to see full attachment