Emotional Eating Interventions for Adults Living with Overweight or Obesity: A Systematic Review and Meta-Analysis

Background: Emotional eating (EE) may be defined as a tendency to eat in response to negative emotions and energy-dense and palatable foods, and is common amongst adults with overweight or obesity. There is limited evidence regarding the effectiveness of interventions that address EE. Objectives: To synthesize evidence on the effectiveness of EE interventions for weight loss and EE in adults living with overweight or obesity. Methods: This is a systematic review and meta-analysis. Adhering to the PRISMA guidance, a comprehensive electronic search was completed up to February 2022. Random effects meta-analysis was carried out to determine the percentage change in weight and EE scores. Results: Thirty-four studies were included. The combined effect size for percentage weight change was −1.08% (95% CI: −1.66 to −0.49, I2 = 64.65%, n = 37), once adjusted for publication bias. Similarly, the combined effect size for percentage change in EE was −2.37%, (95% CI: −3.76 to −0.99, I2 = 87.77%, n = 46). Cognitive Behavioural Therapy showed the most promise for reducing weight and improving EE. Conclusions: Interventions to address EE showed promise in reducing EE and promoted a small amount of weight loss in adults living with overweight or obesity.


Introduction
In the UK, 33% of adults are overweight and 28% have obesity [1]. Overweight and obesity are defined as "abnormal or excessive fat accumulation that may impair health" [2]. In adults, Body Mass Index (BMI) is used as an estimate of body fat, with a BMI of 25 to 29.9 kg/m 2 representing overweight, and a BMI of 30 kg/m 2 and above being used to define obesity [3]. Living with obesity increases the risk of premature mortality, comorbidities, and reduces wellbeing [4]. Randomised Control Trials (RCTs) report that behavioural weight loss interventions targeting energy intake and physical activity produce modest weight losses of 7-10% in interventions lasting up to 30 weeks [5]. However, individuals may regain more than 80% of the weight lost within 5 years [6]. Behavioural Weight Loss (BWL) interventions are reported to be ineffective for individuals living with overweight/obesity [7]. These interventions do not address the psychological factors associated with eating [8] (such as emotional eating (EE) and/or binge eating behaviours) that are associated with overweight, obesity, and poor mental wellbeing [9].
There is no ubiquitous definition of EE, but EE is commonly defined as responding to negative feelings (e.g., stress, upset, or furiousness) for temporary comfort by overconsuming energy-dense and palatable foods [10][11][12][13]. Theories suggest such behaviours may become coping strategies and foreshadow unhealthy eating habits, causing weight gain [12,13]. Other theories suggest that reported positive emotions (e.g., joy or excitement) may also lead to EE and are commonly followed by negative feelings such as shame [14]. Macht (2008) proposed a five-way model of how emotions affect eating, including (1) emotional control of food choice, (2) emotional suppression of food intake, (3) impairment of 1.
Whether interventions that address EE are effective for achieving weight loss and/or improving EE in adults living with overweight or obesity. 2.
Which psychological approach appears most effective for weight loss and/or improving EE for adults living with overweight or obesity.

Materials and Methods
The reporting of this review follows the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [26]. The PRISMA checklist [26] is included in Supplementary Table S1. The systematic review protocol is registered on the PROSPERO international prospective register of systematic reviews (CRD 42022302799).

Eligibility Criteria, Search Strategy and Selection of Studies
All studies that evaluated interventions with components that specifically addressed EE in adults with overweight or obesity were included. All primary research methodologies were included for consideration. Studies performing secondary data analysis on included RCTs were excluded. There were no date or geographical limits; however, searches were limited to papers written in English.
The inclusion criteria are reported in line with the PICOS guidance [27]. Participants: There were no limits on gender, ethnicity, or country of origin. However, studies had to be available in the English language. Studies were only included if >70% of participants did not meet criteria for BED or a clinically diagnosed eating disorders without EE as defined below. Similarly, >70% of the sample had to have a BMI of ≥25 kg/m 2 . Studies involving subjects under the age of 18 or animal studies were excluded. Interventions: Studies involving any psychological interventions with components specifically targeted at addressing EE were included. The definition of EE used was "the tendency to eat energydense and palatable foods, in response to negative emotions and/or eating to cope with unclear states of hunger and satiety with physiological changes and/or emotion regulation difficulties and/or various indicators of psychopathology including symptoms of anxiety and depression, negative self-concept, overeating". Following discussion with the research team, studies involving medical interventions or medical devices were excluded. Furthermore, interventions that included components of EE post-bariatric surgery were excluded (pre-surgery were included). Pharmacological interventions and psychological therapies for weight loss that were not targeted at addressing EE were excluded, as were studies that more generally targeted food cravings. Comparison: Treatment as usual, waiting list control or no comparison group. Outcome: Studies reporting pre-and post-intervention data for weight and/or a measure of EE were included. Study design: Only studies with primary data were included. Systematic reviews, umbrella reviews, and scoping reviews were excluded [28], as were follow-up papers, protocol papers, non-peer reviewed studies and grey literature sources without scientific credibility (studies without robust designs or high risk of bias).
Keywords were generated by reviewing the literature in the field and consulting a leading Academic Psychologist in the field of Disordered Eating (Table 1). On the advice of the Academic Psychologist, a list of specific search terms was created for EE to ensure that appropriate citations were identified (Table 2). The following databases were searched: CINAHL, PsychInfo, MEDLINE, and EM-BASE. Reference lists of the included papers were searched for any additional articles [28]. A grey literature search was also conducted of the first 100 hits on Google and MEDNAR (using "Emotional Eating in adults living with overweight or obesity" as the search term). All dates were searched. The searches were conducted from database inception to February 2022, ensuring that all eligible studies were included prior to data analysis. The PRISMA flow diagram [26] is shown in Figure 1. The searches were performed by X.Q.A. and uploaded into Rayyan, a web and app-based tool for systematic reviews [29]. Eligible studies were screened by two reviewers (X.Q.A. and J.S.). Both reviewers screened 100% of the titles and abstracts. In the second stage, all full-text articles were reviewed by X.Q.A., J.S., E.L.G., and G.T.T., with conflicts resolved through consensus. Approximately 10% of full-text articles required a consensus discussion.

Data Extraction
Data were extracted from eligible studies and recorded in an Excel spreadsheet to obtain relevant data regarding: the aim of the study; study population characteristics; study design; sample population; intervention approach; duration of intervention; main findings; author conclusions; and author contact details. Two reviewers (X.Q.A. and J.S.) independently extracted the data for completeness and accuracy, and any conflicts were resolved through consensus. A third reviewer (G.T.T.) checked the final data extraction.

Critical Appraisal
Two reviewers (X.Q.A., and E.L.G.) independently assessed the study quality using the CASP Critical Appraisal Tools [61] for RCTs and JBI Checklist for Quasi-Experimental Studies [62] for single group design studies. Studies meeting at least 75% of the appraisal criteria were considered to be of high methodological quality; those meeting 50-74.9% were considered medium quality; and those meeting less than 50% were considered low quality.

Synthesis of Results and Analytical Strategy
The quantitative data were reviewed and were deemed sufficient for meta-analysis of weight outcomes and EE outcomes. They were converted into percentage change due to the variation in tools to measure EE and the wide variation in baseline data. The primary outcome of this review was mean weight change pre-and post-intervention, as opposed to BMI which has a high specificity but low sensitivity for assessing obesity [63]. The secondary outcome was change in mean EE score pre-and post-intervention.
Where data were missing, the authors were emailed to request this data [21,23,32,38,42,46,48,54,57,64]. The Meta-Essentials Workbooks for Meta-analysis version 1.5 were used for the meta-analysis [65]. The results of individual studies are outlined in a table of frequencies in Supplementary Table S2.

Data Extraction
Data were extracted from eligible studies and recorded in an Excel spreadsheet to obtain relevant data regarding: the aim of the study; study population characteristics; study design; sample population; intervention approach; duration of intervention; main findings; author conclusions; and author contact details. Two reviewers (X.Q.A. and J.S.) independently extracted the data for completeness and accuracy, and any conflicts were resolved through consensus. A third reviewer (G.T.T.) checked the final data extraction.

Critical Appraisal
Two reviewers (X.Q.A., and E.L.G.) independently assessed the study quality using the CASP Critical Appraisal Tools [61] for RCTs and JBI Checklist for Quasi-Experimental Studies [62] for single group design studies. Studies meeting at least 75% of the appraisal criteria were considered to be of high methodological quality; those meeting 50-74.9%

Results
After removal of duplicates, 3220 citations were identified. Following the title and abstract screen, 287 full text papers were evaluated against the eligibility criteria and 253 papers were excluded. The reasons for exclusion are summarized in Figure 1, and fully reported in the Hierarchy of Exclusion in Supplementary Table S3. After screening the grey literature, no further scientifically credible studies were identified, therefore the PRISMA diagram [26] (Figure 1) includes databases and registers only.

Weight and BMI
Twenty-five studies reported pre-and post-intervention means for weight. Most studies included participants with a BMI over 25 kg/m 2 , with no upper BMI limit for inclusion. However, the study by Lillis et al. (2016) excluded participants with a BMI over 50 kg/m 2 [47].

Interventions
Several studies examined more than one psychological intervention. The most utilised psychological interventions were Behavioural Weight Loss (BWL) and Mindfulness. Table 2 provides full details of the interventions used in each study. The main intervention categories are outlined below.

Mindfulness
Mindfulness was used in 10 studies [31,[35][36][37]39,43,45,50,56,57]. Mindfulness interventions in this review included general Mindfulness techniques drawn from Mindfulness-based stress reduction, along with specific mindful eating training derived from Mindfulness-Based Eating Awareness Training [78]. The Mindful Eating techniques involved identifying and responding adaptively to food cravings, and providing skills for emotion regulation that would allow individuals to sit with, rather than trigger EE [79].

Acceptance-Based Interventions
Seven studies involved Acceptance-based interventions [23,30,33,38,40,47,54]. Acceptancebased interventions build upon the behavioural skills used in BWL programs by adding components derived from ACT. The three principal components of Acceptance-based interventions are distress tolerance, mindfulness, and commitment enhancement [80].

Behavioural Therapy
Five studies used Behavioural Therapies [24,38,41,47,53] such as Standard Behavioural Therapy (SBT) (e.g., counselling on diet and physical activity), a BWL intervention generally focus on increasing awareness of one's eating behaviour through dietary self-monitoring and goal setting [5]. This category also included Enhanced Behavioural Treatment (EBT) which integrates SBT components with techniques that specifically target EE [24].

Cognitive Behavioural Therapy
Three studies involved CBT [48,55,59]. The European Society of Physical and Rehabilitation Medicine [81] guidelines strongly recommend CBT as the gold standard psychological intervention for obesity.

Dialectical Behavioural Therapy
One study used Dialectical Behavioural Therapy (DBT) [21]. The principal components for DBT are mindfulness, emotion regulation, and distress tolerance, which required coaching with a professional [82].

Weight
Twenty-five studies reported pre-and post-intervention mean weight or provided this data upon request. Twenty-five studies, reporting on 37 different interventions were therefore included in the weight meta-analysis (some studies reported on multiple interventions). The combined effect size for percentage weight change across all interventions was −2.59% (95% CI: −3.59 to −1.58, z = −5.22, p < 0.0001, n = 37). The heterogeneity of the studies was moderate (I 2 = 49.34%). A forest plot for the combined effect size of all interventions is shown in Figure 2.

Emotional Eating Score
Thirty studies reported pre-and post-intervention mean EE scores or provided this data upon request. Thirty studies, reporting on 46 different interventions were in-cluded in the meta-analysis (some studies reported on multiple interventions). The combined effect size for percentage change in EE scores across all interventions was −22.74% (95% CI: −27.49 to −17.98, z = −9.62, p < 0.0001, n = 46). The heterogeneity of the studies was considerable (I 2 = 82.52%). A forest plot for the combined effect size of all interventions is shown in Figure 4.

Discussion
This systematic review and meta-analysis aimed to investigate whether interventions to address EE are effective for achieving weight loss and/or reducing EE in adults living with overweight or obesity. In summary, 34 studies were included in this review, with half being published in the USA. The combined effect size for all interventions on percentage weight change was −1.08%, once adjusted for publication bias. The combined effect size for percentage change in EE was −2.37%, once adjusted for publication bias. CBT

Discussion
This systematic review and meta-analysis aimed to investigate whether interventions to address EE are effective for achieving weight loss and/or reducing EE in adults living with overweight or obesity. In summary, 34 studies were included in this review, with half being published in the USA. The combined effect size for all interventions on percentage weight change was −1.08%, once adjusted for publication bias. The combined effect size for percentage change in EE was −2.37%, once adjusted for publication bias. CBT showed the most promise for weight loss and improving EE in the sub analyses. Both effects sizes were small, which was expected due to the short-term nature of the interventions included in the studies. Furthermore, the authors anticipated a larger effect size for the change in EE, as a change in eating behaviour would be an expected precursor to longer term weight loss.
There was significant variability in the 34 included studies in terms of research design, methodology, outcome measures, measurement tools, and intervention delivery/fidelity. This review concurred with Chew et al. (2022) [25] in that there is inconsistency in use of standard protocols for psychotherapeutic interventions. Our review identified a wide variety of self-reported tools have been developed and validated to measure EE (n = 11), with the Dutch Eating Behaviour Questionnaire (DEBQ) [66] being the most commonly used tool. Furthermore, there was wide variation in the length of intervention (1 day to 24 months); the setting (in person, groups, telephone, internet, self-help, and inpatient admissions); amount of support the participants received (self-help to 1:1 weekly support); and the amount of focus on EE within the intervention. Future research therefore needs to explore some of these factors and better understand the active behaviour change components of effective interventions.
The searches in this review identified 3220 records (after duplicates were removed), which is higher than the review by Chew et al. (2022) (n = 528) [25]. More studies were included in our meta-analysis (n = 25 for weight, and 30 for EE score) compared to Chew et al. (n = 19). This is likely due to the broader search strategy as described in Section 1.

Intervention Effects on Weight
Although health education is important, the psychological triggers and context of eating need to be considered in obesity treatment [51]. Several psychological interventions such as Mindfulness, ACT, CBT, and DBT have shown promising results in reducing EE and facilitating weight loss [84]. Our meta-analysis reported a small effect on weight (−1.08% from pre-to post-intervention). However, a meta-analysis by Chew et al. (2022) [25] reported no significant effect of EE interventions on weight outcomes (p = 0.12). Our review found that CBT showed the most promise for weight loss (−9%), followed by combined interventions (−6%). Previous studies without a focus on EE have largely found that CBT is effective for weight loss [80,85]. The average length of CBT interventions in this review lasted between six weeks and six months, which is longer than some of the other interventions used. This may explain why CBT (and combined interventions that include elements of CBT) showed more promise for weight loss compared to other interventions. Chew et al. [25] reported that purely Mindfulness-based interventions showed a higher interventional effect size for weight compared to other interventions. This differs from the findings of our review where Mindfulness-based interventions showed the least promise for percentage weight change. Furthermore, Chew et al.'s review found that combined interventions had counter-productive effects on weight whereas our review found they showed a degree of promise. This may be due to the methodological differences between the reviews, as Chew used Hedges g to assess the effect size (kg) pre-and post-intervention, whereas our review converted weight changes into percentages to enable us to include more studies and aid interpretation of results (there was a wide variation in mean baseline weights reported between the included studies).

Interventions for Improving Emotional Eating
The meta-analysis showed a higher percentage change in EE than weight (−2.37%) following intervention. This is encouraging as a reduction in EE is an anticipated precursor to weight loss and improved mental wellbeing. CBT showed the most promise in reducing EE (−38%), followed by Acceptance-based interventions (−25%). A possible intervention approach for adults living with obesity and EE is to educate individuals to understand and recognise EE, instead of restricting dietary intake which often exacerbates the issue [84]. Both CBT and Acceptance-based interventions may reduce EE episodes by recognizing stressors and resultant emotions and replacing the urge to eat with alternative positive actions [38,86]. CBT is a common approach to support change in EE by keeping a food and mood diary to identify the triggers and situational context of eating [87,88]. Glisenti and Strodl [89] suggested that DBT may be more effective than CBT in reducing EE as DBT focusses on both emotion regulation and Mindfulness [89], but there was only one study identified in this review that used DBT. Forman et al. [90] suggested Acceptancebased strategies may eliminate and reduce EE compared to BWL interventions. One study indicated a 1-day ACT workshop can be beneficial in reducing EE, however acceptance and value clarification skills may require more time to incarnate and develop as a longer term intervention [23]. Numerous studies concluded that changing eating behaviour may lead to improvements in both physical and mental health outcomes [91][92][93]. Mindfulness was the most common intervention used by authors. This has been shown in existing research to reduce EE by regulation of emotions [94][95][96]. Studies have shown mixed results for Mindfulness on eating behaviour [40,97]. When considering only Mindfulness-based interventions, individuals seeking to lose weight may benefit more from Mindfulness skills emphasizing increased awareness, acceptance, and overriding of hedonic drives to eat, rather than those promoting reliance on homeostatic cues to reduce consumption [71]. One study in this review [34] examined Mindfulness plus/minus professional contact and resulted a positive improvement in EE; however, the authors reported an increase in weight in participants without professional contact. Chew et al. (2022) [25] reported a small to medium interventional effect for Mindfulness interventions on EE post-intervention (p = 0.01). They also reported that pure Mindfulness interventions showed a higher interventional effect size for EE, when compared to other interventions or combinations of interventions. Mindfulness is a key component in Acceptance-based interventions and DBT [94]. Studies reported combined interventions such as CBT and Mindfulness may be advantageous as they enable individuals to explore behaviour change techniques whilst also learning acceptance strategies to reduce EE [24,95]. Greater awareness of emotions and stress enabled individuals to improve regulation of eating and make healthier choices, which leads to fewer EE episodes [45].

Intervention Setting and Mode of Delivery
The majority of participant contact in the studies was face-to-face, in a group setting, and was delivered by trained professionals. Comparator groups included personal contact with trained professionals, and manual-based treatment with telephone support. Three studies compared personal contact to remote delivery/self-help interventions [31,46,55]. However, there were no consistent findings between the modes of delivery. Further research is therefore required to determine the impact of the setting and mode of delivery on weight and EE in adults with overweight or obesity. Existing evidence from studies reported that personal contact with a health professional increases weight loss success [96,97] and motivation [98]. However, online CBT studies do report that remote delivery was useful for people living in certain locations or those who were unable to attend regularly face-toface sessions [99,100]. Considering the recent COVID-19 pandemic, this online platform is likely to become more popular with service users [101] and therefore all methods of delivery should be considered in clinical practice. Whilst CBT showed the most promise for reducing weight and improving EE, it is important to acknowledge that implementing CBT in treatment would require longer intervention times and requires specialist healthcare professionals with extensive training. It is therefore vital that further work is undertaken to explore practical ways in which healthcare professionals supporting adults with overweight or obesity might implement CBT approaches to address EE. One example would be through brief, manualised, or online self-help approaches (guided self-help).
Six of the interventions in this review were multi-component and therefore future research should aim to identify the active components that are most effective for improving EE and achieving weight loss. An important first step in weight management practice would be the development of an agreed pathway for EE.

Screening Tools for Emotional Eating
This review identified a wide variety in the available tools to measure EE (n = 11). The most commonly used tool was the Dutch Eating Behaviour Questionnaire (DEBQ) [66]. Different validated tools may allow for the comparison of the predictive validity of EE in relation to weight outcomes to see which construct has a greater effect on weight loss [84]. However, the use of different tools in the included studies highlights the need for some consensus amongst health professionals and researchers regarding the most appropriate tool to use in weight management practice [102]. Further research is therefore required to scope the available tools; their reliability and validity; their strengths and weaknesses; and their feasibility and acceptability for both healthcare professionals and service users.

Demographic Differences
In this review, participants were 32% Caucasian and 66% female. This is supported by existing research that reports the demand for healthcare services is greater in women living with overweight and obesity [103]. Women tend to report higher EES than men [104]. However, a further study found that men with EE were almost three times more likely be overweight than women [105]. Black adults are at higher risk of developing obesity than Caucasians, and Chinese adults are at lower risk of developing obesity [106]. Whilst there were a mix of genders and ethnicities within this review, future research should focus on a wider demographic population to increase the generalizability of the findings of this review. 4.6. Strengths and Limitations 4.6.1. Strengths The strengths of this review included broad and clear inclusion criteria that were designed through public and patient involvement and engagement. The authors did not limit the searches to RCTs in the way that Chew et al. (2022) [25] designed their review. This allowed for the inclusion of non-randomised intervention studies and single group designs. Furthermore, the changes in weight and EE scores were converted to percentages due to the range in baseline means, and the wide variety of tools used to measure EE. This allowed the authors to include additional studies and improved the interpretation of results. Furthermore, there were a mix of participant ages, genders, and ethnicities in the included participants.

Limitations
There were several limitations to this systematic review and meta-analysis. First, there was heterogeneity in measurement tools and interventions. The studies used multiple tools to measure EE (n = 11). Determining the validity of EE questionnaires is vital, given that these measures are more feasible for settings that do not allow observation of food intake and there is known measurement invariance in EE measures (namely DBEQ) by gender, age and BMI; therefore, it may be useful to segregate these groups in future analyses [107]. There was also wide variation in included interventions, and many were multi-component interventions, making it was difficult to fully understand the active components. Second, the statistical heterogeneity of the studies included in the meta-analysis and the significant publication bias in both meta-analyses reduce the validity of the overall effect size for weight and EE. The effect sizes were therefore adjusted using the Trim and Fill procedure to estimate an unbiased effect size. Third, the primary focus of several studies was on weight or BMI which can take longer to show significant changes over a relatively short intervention. Finally, whilst we reported on the risk of bias of individual studies and considered the impact of heterogeneity and publication bias in the meta-analysis, a formal GRADE [108] evaluation was not undertaken for this review. Future research should therefore consider a GRADE evaluation prior to making practice recommendations.

Conclusions
This comprehensive systematic review and meta-analysis found that psychological interventions to address EE showed some promise in reducing EE and promoting a small amount of weight loss in adults living with overweight or obesity. CBT showed the most promise for reducing weight and improving EE in the sub analyses. The review identified a wide range of available psychological interventions to address EE in this population, and wide variation in the available tools to assess for EE. A consensus is therefore required amongst healthcare professionals and academics regarding the most feasible and acceptable tool to identify EE in weight management practice settings. An important first step in weight management practice would be the development of an agreed pathway for EE in adults with overweight or obesity. Further research is required involving participants from a wide range of ethnic backgrounds, genders, and/or socio-economic statuses. Future research should focus on the long-term effectiveness and acceptability of interventions that address EE in adults with overweight or obesity and to better understand the active components and mechanisms of change in effective interventions.