Food Addiction Therapy For Weight Loss Treatment
Overview
Food addiction (FA) may serve as a potential predictor of outcomes in weight loss interventions. This systematic review and meta-analysis aimed to evaluate (1) the impact of FA diagnosis and symptom severity on weight loss outcomes in individuals with overweight or obesity, and (2) the influence of factors such as intervention type, study characteristics, and participant demographics (e.g., age, gender, ethnicity) on these associations. A comprehensive search of PubMed, PsycINFO, and Web of Science identified studies examining the relationship between FA, assessed via the Yale Food Addiction Scale, and weight loss outcomes in individuals with overweight or obesity without clinically diagnosed eating disorders.
The review included 25 studies involving 4,904 participants (71% women, average age 41 years, average BMI 40.82 kg/m²). Meta-analyses synthesized data from 18 studies on weight loss correlations with FA symptom count and 21 studies comparing weight loss outcomes between FA diagnosis groups. Results showed limited evidence supporting a negative impact of FA symptom count and diagnosis on weight loss outcomes. However, associations appeared stronger in behavioral weight loss programs and in ethnically diverse populations. The findings highlight the need for further investigation into how FA interacts with mental health conditions and environmental factors to affect weight loss success.
Introduction
The conceptual overlap between addiction and obesity has fueled the proposal of “food addiction” (FA), a pattern of compulsive eating behavior marked by a lack of control over food consumption, reliance on high-calorie foods to cope with emotional distress, and persistent cravings for highly palatable foods. Despite its potential relevance to the development and management of obesity (OB), FA remains a contentious subject. Key debates include whether it aligns more closely with substance-based or behavioral addiction, its distinction from established eating disorders, and its broader clinical significance.
To clarify these uncertainties, a systematic review and meta-analysis was conducted to assess whether FA influences the outcomes of weight loss interventions in individuals with OB. The Yale Food Addiction Scale (YFAS) is the primary tool used to measure FA. This instrument, derived from the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) criteria for substance dependence, evaluates symptoms such as loss of control, failed attempts to quit, and withdrawal-like effects. The original YFAS includes 25 items, and its variants include adaptations for children (YFAS-C), a shorter nine-item version (mYFAS), and the updated YFAS 2.0. These tools offer both symptom count and diagnostic scoring, with the latter expanded under DSM-5 criteria to include stratified severity levels.
Studies indicate that FA prevalence among individuals with OB ranges from 7.7% to 47%, with a weighted average of approximately 30% based on meta-analytic data. Higher rates are observed in populations with eating disorders (55–84%) and among women, though the influence of age and other sociodemographic factors requires further investigation. FA has been linked to worse psychological outcomes, such as increased depressive and anxiety symptoms, diminished quality of life, and difficulties in weight management, potentially positioning FA as a significant factor in the prognosis of OB treatment.
Existing evidence highlights considerable variability in weight loss outcomes from OB treatments, which include lifestyle modifications, bariatric surgery, pharmacotherapy, and cognitive-behavioral therapies. Only a subset of patients—20% to 30%—achieve sustained weight loss. While some research suggests that pre-treatment FA diagnosis or higher FA symptom counts are associated with poorer weight loss outcomes, other studies report no significant relationship. The heterogeneity in findings likely reflects differences in study designs, intervention types, and follow-up durations.
To date, only one systematic review (2017) has explored the impact of pre-treatment FA on post-bariatric surgery outcomes, finding no significant predictive relationship within the limited data available. However, more recent studies have examined FA’s role in various weight loss interventions, warranting an updated and comprehensive analysis.
This systematic review and meta-analysis aimed to quantify the predictive impact of FA diagnosis and symptom severity on post-intervention weight loss in individuals with OB. Additionally, it sought to identify sources of variability in these associations, considering factors such as intervention type, study methodology, and participant demographics. These findings aim to advance understanding of FA’s role in obesity treatment and guide future clinical strategies.
Method
This systematic review and meta-analysis protocol was registered prospectively in the PROSPERO database (CRD42022384703) and adheres to PRISMA guidelines for systematic reviews and meta-analyses. A comprehensive search strategy was implemented using keywords and Medical Subject Headings (MeSH) across PubMed, PsycINFO, and Web of Science. The initial search occurred on January 26, 2023, and was updated on June 3, 2024. Additionally, reference lists of included studies were manually reviewed to identify other relevant research.
The search focused on three main domains: (1) food addiction (e.g., food addiction, compulsive eating, YFAS), (2) patient weight status (e.g., overweight, obesity), and (3) weight loss interventions (e.g., bariatric surgery, physical exercise, or pharmacological approaches like orlistat). Keywords and search terms were applied systematically, with the complete list provided as supplementary material. Covidence software was used to automatically remove duplicate entries.
A team of five reviewers (L.S., G.H., M.P., C.R., Z.A.) independently screened titles and abstracts, while three reviewers (L.S., G.H., M.P.) conducted full-text screenings. Each study required two decisions for inclusion or exclusion, and disagreements were resolved through consensus meetings, with a final decision made by G.H.
Three reviewers independently extracted data, ensuring duplicate validation. Key data included publication details, study design, participant demographics, treatment types, outcomes, and follow-up intervals. Statistical data for effect size calculations, such as correlations, regression coefficients, and p-values, were also collected.
The quality of included studies was assessed using the Cochrane ROBINS-I tool, focusing on seven bias domains, such as confounding factors, participant selection, and missing data. Each study was categorized into risk levels, ranging from low to critical. Numerical scores were assigned to quantify bias, with higher scores reflecting greater risk.
Two random-effects meta-analyses were conducted using restricted maximum likelihood estimation. The first meta-analysis examined correlations between pre-intervention YFAS symptom counts and weight loss outcomes, applying Fisher’s r-to-z transformation for variance stabilization. The second analysis evaluated differences in weight loss between individuals with and without a pre-intervention food addiction diagnosis, using standardized mean differences (SMDs).
Subgroup analyses explored variations in weight loss outcomes based on intervention types (e.g., bariatric surgery vs. behavioral treatments) and participant demographics (e.g., gender, age, ethnicity). Dependent effect sizes were managed using cluster-robust variance estimation, accounting for multiple follow-ups or groups within studies.
Heterogeneity was quantified using the I² statistic, with thresholds set for low (≤40%) and high (≥75%) heterogeneity. Publication bias was assessed through funnel plots and Egger’s test, examining the symmetry of effect sizes relative to standard errors.
All analyses were performed using R software (version 4.3.3) with the metafor package (version 4.4.0).
Inclusion Criteria
This study focused on peer-reviewed publications available in English, German, Greek, or Spanish that met specific inclusion criteria. Eligible studies had to: (1) examine weight outcomes measured multiple times in (2) individuals with overweight or obesity (BMI ≥ 25 kg/m²) who did not have a medically diagnosed eating disorder, (3) assess the effectiveness of a weight loss intervention, (4) evaluate food addiction (FA) using any form of the Yale Food Addiction Scale (YFAS) prior to the intervention, and (5) analyze the relationship between pre-intervention YFAS symptom counts and weight loss outcomes, or compare weight loss results between individuals with and without a pre-intervention FA diagnosis.
Exclusion Criteria
Studies were excluded if they: (1) focused solely on participants with diagnosed eating disorders, (2) used an ineligible design, such as reviews, editorials, case reports, or cross-sectional studies, or (3) did not provide adequate data for inclusion in the meta-analysis.
Result
After excluding 1,602 duplicate entries, 5,646 titles and abstracts were screened, achieving a 94% agreement rate among reviewers. Following consensus discussions, 88 full-text articles were assessed for eligibility, resulting in the inclusion of 25 studies.
The studies used, were published between 2013 and 2024. Geographically, 13 studies originated from the Americas, 8 from Europe, and 4 from Asia. Among these, 11 focused on bariatric surgery (e.g., sleeve gastrectomy, Roux-en-Y bypass, or combinations including adjustable gastric banding), 1 examined pharmacological intervention (semaglutide), and 13 explored behavioral weight loss interventions (e.g., dietary modifications, physical activity promotion, or interventions addressing eating behaviors, including those targeting addiction-like patterns).
The studies’ average follow-up interval was 12.53 months (SD = 7.58), with an average of 1.72 follow-ups per study (SD = 0.84). The primary outcomes included percent total weight loss (%TWL, reported in 14 studies), raw BMI or weight changes (5 studies), and percent excess weight loss (%EWL, 3 studies). In children, age-adjusted BMI standard deviation scores were reported in 3 studies.
Regarding food addiction (FA), different versions of the Yale Food Addiction Scale (YFAS) were used across studies, with the most common being YFAS 3.0 (18 studies). The 25 studies collectively included 4,904 participants (3,475 women and 1,429 men), with 3 studies focused on children (n = 299) and the remainder on adults (n = 4,605). Four studies exclusively involved women. The mean age across all participants was 41.4 years (adults: 45.4 years, children: 13.3 years). Adults had an average BMI of 40.82 (SD = 6.04), while children had an average BMI-SDS of 2.12 (SD = 0.14). FA prevalence averaged 31%, with a mean symptom count of 2.58 (SD = 0.46).
Using the ROBINS-I tool, 9 studies were rated as having low risk of bias, another 9 had moderate risk, and 6 were classified as serious risk due to inadequate confounding controls. One study lacked sufficient data for a complete risk assessment. Risk-of-bias scores were used as predictors in subsequent meta-regressions.
Across 13 studies, pre-intervention YFAS symptom count showed a non-significant negative correlation with weight loss (z = 0.05, 95% CI [0.11; 0.02], p = 0.13). Funnel plot inspection indicated some asymmetry, but Egger’s test suggested minimal publication bias (p = 0.14). Effect size heterogeneity was moderate (I² = 35.41%). While the type of intervention significantly influenced correlations (p = 0.03), higher symptom counts predicted lower weight loss only in behavioral interventions (p = 0.009), not in surgical contexts (p = 0.99).
Comparisons between FA-positive and FA-negative individuals across 14 studies revealed non-significant differences in weight loss outcomes (SMD = 0.07, 95% CI [0.27; 0.14], p = 0.48). Ethnicity was the only significant moderator (p = 0.04), with homogenous White samples showing a larger difference favoring FA-negative individuals (SMD = 0.47, p = 0.04), compared to more diverse samples.
This analysis underscores the complex relationships between FA symptoms, intervention types, and weight loss outcomes, highlighting the influence of study design and population characteristics on findings.
In conclusion, this systematic review and meta-analysis assessed the prognostic influence of food addiction (FA) diagnosis and symptom severity on outcomes of weight loss interventions among individuals with overweight or obesity (OB) without medically diagnosed eating disorders. Across 25 studies, the prevalence of FA before interventions aligned with rates observed in earlier research. Neither FA diagnosis nor symptom count directly correlated with weight loss following bariatric surgery or pharmacological treatments. However, a higher FA symptom count was linked to reduced weight loss in behavioral interventions, while FA diagnosis showed relevance in specific subgroups.
The findings suggest that severe FA symptoms may impair adherence to behavioral interventions, which require sustained lifestyle changes, such as dietary adjustments and exercise. These symptoms, characterized by loss of control over eating and cravings for highly palatable foods, could hinder long-term compliance. However, the absence of a significant relationship between FA diagnosis and weight loss in these interventions warrants further investigation, particularly using updated criteria like the Yale Food Addiction Scale 2.0 (YFAS 2.0), which includes severity grading.
Ethnicity emerged as a potential moderator in secondary analyses, with diverse populations showing lower weight loss in FA-positive individuals. This may reflect sociocultural and environmental factors, such as differential access to healthy foods. Although the YFAS scale has been validated across diverse populations, interpreting these findings requires caution due to possible cultural influences on FA expression.
FA appears to be a frequent comorbidity in OB, potentially linked to hedonic dysregulation in weight control. Its overlap with subthreshold binge-eating disorder may also explain the lack of consistent associations with intervention outcomes. Interestingly, some OB interventions may indirectly reduce FA symptoms, though longitudinal tracking of these changes was insufficient across the studies.
The review excluded individuals with medically diagnosed eating disorders, limiting its scope to FA-specific contributions. Variation in study inclusion criteria, reliance on DSM-IV-based YFAS criteria, and lack of data on intervention adherence and dropout rates were notable limitations. Additionally, FA’s relationship with mental health issues, which are associated with poorer weight loss outcomes, remains underexplored. Future research should examine whether addressing FA could enhance treatment efficacy and whether addiction-specific interventions might benefit individuals with severe FA symptoms.
This is the first systematic review and meta-analysis exploring FA’s prognostic role in weight loss interventions. While evidence of FA’s negative impact on outcomes was limited, it may predict poorer results in behavioral interventions and among ethnically diverse populations. Future studies should investigate FA’s indirect effects, its interaction with mental health challenges, and the potential benefits of targeted interventions for individuals with higher FA severity.
