Effects of mindfulness‐based interventions on obesogenic eating behaviors: A systematic review and meta‐analysis

Summary

This systematic review and meta‐analysis examined the effects of mindful‐based interventions (MBIs) on changes in obesogenic eating behaviors. Seven databases (CINAHL, PubMed, PsycINFO, Cochrane, Web of Science Core Collection, Embase, Sociological Abstracts) were searched. Random‐effects models were performed to estimate the pooled effects, and mixed‐effects models were used to explore potential moderators of MBIs on eating behavioral changes. The significant effects on mindless eating habits included controlled eating (Hedge's g = 0.23, p = 0.005), external eating (g = −0.62, p = 0.001), fullness awareness (g = 0.64, p < 0.001), hunger eating (g = −0.69, p = 0.032), energy intake (g = −0.60, p = 0.003), sweet intake (g = −0.39, p < 0.001), and impulsive food choice (g = −0.43, p = 0.002). However, small and insignificant effects were noted for stress‐related eating habits like emotional eating (g = −0.27; p = 0.070) and binge eating (g = −0.35, p = 0.136). The long‐term effects were significantly sustained on hunger eating (g = −0.50, p = 0.007) but insignificant on emotional eating (g = −0.22, p = 0.809). MBIs delivered in clinical settings were more effective for decreasing emotional eating compared with those in school settings. Our findings support the effectiveness of MBIs. The pooled effects on improving mindless eating habits were stronger than the modification of stress‐related eating habits.

1. INTRODUCTION

Using mindfulness‐based interventions (MBIs) to target obesity‐related eating behaviors has demonstrated some promising results and been increasingly implemented to prevent or treat overweight or obesity. ¹ , ² The significant effects of MBIs on anthropometric outcomes (BMI: g = −0.36, waist circumference: g = −1.20) have been delineated in our previous meta‐analysis, ³ but the effects of MBIs on eating behavioral outcomes remain relatively obscure. Mindless and autopilot eating habits/styles (↓control, ↓fullness awareness, ↑external eating, ↑hunger, ↑energy intake, ↑sweet intake, ↑unhealthy food choice) ⁴ and stress‐related habits (emotional and binge eating) can position individuals at a greater risk for overweight and obesity. ¹ , ⁵ , ⁶ Thus, this systematic review and meta‐analysis examined the effects of MBIs on changes in various obesogenic eating behaviors including controlled, external, hunger, binge, and emotional eating as well as fullness awareness, sweet and energy intake, and impulsive food choice.

1.1. MBIs

Over the past few decades, MBIs for obesity prevention or treatment have been used with promising results. ¹ , ² , ³ MBIs typically are designed to improve individuals' psychological adaptation through enhanced awareness of their own emotions, acceptance, and cognitive processes. MBI techniques have been shown to support individuals' nonjudgmental self‐motivation and self‐awareness. ⁷ Mindful eating training, in particular, is designed to heighten individuals' awareness of the present moment, with a focus on their body's sensations when eating, as well as their thoughts and feelings about food. This approach has been frequently employed to overcome obesogenic eating behaviors such as mindless and stress‐related eating styles/habits. ⁸ , ⁹ Existing relevant reviews and meta‐analyses on the effects of MBIs mostly targeted populations suffering from overweight/obesity ¹⁰ , ¹¹ or eating disorders ¹² and assessed anthropometric outcomes. ³ , ¹¹ Of the few systematic reviews that investigated the effects of MBIs on eating behaviors, many of them aggregated various eating behavioral outcomes to reflect the improvements in healthy eating (r = 0.14), ¹³ overall eating behavior (g = 1.08), ¹¹ or eating attitude (g = 0.57). ¹¹ Various aspects of mindful eating habits (↑eating control or disinhibition, ¹⁴ ↑fullness awareness, ² ↓external eating [↓reactive to external cues], ¹⁵ ↓physical hunger, ⁹ ↓energy ¹⁶ and ↓sweet intake, ¹⁷ , ¹⁸ or ↓impulsive food choices ⁶ ) were not explicitly examined. Understanding specific impact of MBIs on various obesogenic eating habits can shed lights on how to better tailor MBIs and more effectively prevent or treat overweight or obesity. Currently, stress‐related eating habits (↑emotional or ↑binge eating) often were assessed when MBIs were implemented with populations diagnosed with eating disorders. ¹¹ , ¹⁹ There is scarce understanding about how MBIs may have an impact on mindless ⁴ or stress‐related eating habits ²⁰ that can contribute to what is known about obesogenic eating behaviors, particularly among individuals who do not have an eating disorder. Bridging this knowledge gap in literature has potential to broaden the implementation of MBIs.

1.2. Obesogenic behaviors: Mindless eating habits

Mindless (or autopilot) eating refers to the absence of conscious awareness when consuming snacks and meals and often leads to overeating behaviors because it can reduce individuals' ability to respond adequately to internal/external cues and/or physical hunger and to monitor the amount (or quality) of food intake. ⁴ , ¹⁵ In fact, individuals whose behavior includes mindless eating may not be able to discern the internal/external triggers to eat, the source of their hunger, and/or the level of fullness while eating. Subsequently, mindless eating can result in overweight and obesity, which puts them at higher risk for type 2 diabetes, cardiovascular diseases, and cancer. ²¹ Because MBIs are designed to increase individuals' awareness of their actions by bringing their attention to the present moment, mindful eating training can help turn off individuals' autopilot eating mode by training them to really savor foods and establish a positive and healthy relationship with food. In addition, MBIs have provided strategies to improve physical movement and stress management and potentially can be used to mitigate stress‐related eating behaviors (e.g., emotional and binge eating).

Currently, few systematic reviews or meta‐analysis have comprehensively and explicitly examined the effects of MBIs on these mindless eating habits including enhancing awareness of fullness (the recognition of the unpleasant full sensation after eating), ² physical hunger, ⁶ and minimizing external eating (in response to external cues rather than internal cues of fullness). ¹⁵ MBIs, particularly mindfulness training, can help decrease individuals' energy intake, ¹⁶ sweet intake (as a coping mechanism), ¹⁷ , ¹⁸ and impulsive food choices. ⁶

1.3. Obesogenic behavior: Stress‐related eating habits

Many individuals eat as a reaction to stress. In fact, increased perceived stress is a known obesogenic risk factor for emotional or binge eating behaviors and indirectly impacts individuals' weight status. ²² Emotional eating is defined as a tendency to overeat in response to negative emotions, such as anxiety or irritability, ²³ while binge eating refers to eating a large amount of food within a short amount of time and is associated with a sense of loss of control. ²⁴ Emotional eating, for example, is highly prevalent among individuals who struggle with their weight; this may progress to binge eating without timely and effective interventions. MBIs are recognized as a promising approach to reduce stress‐related eating behaviors, ² , ⁵ , ²⁵ , ²⁶ and improve mental health outcomes. ²⁷ Currently, the benefits of MBIs on emotional and binge eating are better understood among individuals who have suffered from mental health problems such as anxiety/depression ²⁷ and eating disorders. ¹² However, the results of these studies may not be applicable to emotional and binge eating behaviors manifested by people without mental health problems or eating disorders, and the benefits of MBIs on emotional and binge eating among the general population remain relatively unclear. Understanding the effects of MBIs on emotional and binge eating behaviors is essential because many individuals use eating to cope with excessive stress. ²⁰ , ²²

The aim of this review and meta‐analysis was to separate obesogenic eating behaviors into mindless or stress‐related eating habits to better understand the effect of MBIs on each domain. The result of this review will help develop interventions tailored toward improving obesogenic eating habits. Furthermore, the present review seeks to provide insights about potential moderators (e.g., MBI type and focus, sessions, duration, setting, age, treatment fidelity) on the effects of MBIs on various eating habits. The moderation analysis will delineate the role of potential confounders on the pooled effects of MBI with various eating habits.

2. METHODS

2.1. Literature search

The current systematic review was undertaken to identify experimental studies that investigated the effects of MBIs on obesogenic eating behaviors. The study report was developed according to the Preferred Reporting Items for Systematic Reviews and Meta‐analyses statement (PRISMA). ²⁸ Figure 1 illustrates our search process. With the help of the university librarian, studies were identified from the following databases: CINAHL Plus with Full Text, PubMed, PsycINFO, Cochrane, Web of Science Core Collection, Embase, and Sociological Abstracts. The search also included gray literature, such as ClinicalTrials.gov, MedRxiv, and a manual search of reference lists. The following keywords and phrases were used in this search: (Mindful OR yoga OR meditate OR Mindfulness) AND (Eat OR eating OR eats OR food OR “eating behavior” OR “feeding behavior” OR “feeding behaviors” OR “food habit” OR “food habits” OR diet). To some extent, the search strings were modified for each database to enhance the search and meet some of the databases' formatting requirements.

FIGURE 1.

PRISMA flow diagram. Source: Adopted from: Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021; 372:n71. doi: 10.1136/bmj.n71

2.2. Eligibility criteria

The inclusion criteria were experimental studies with general participants, had a comparison group, included any of the eating behavioral measurements (i.e., hunger, fullness awareness, external eating, controlled [or disinhibited] eating, emotional eating, binge eating, energy and sweet intake, and impulsive food choice), and the intervention that included components of mindful eating, mindful movement, or mindful meditation. Exclusion criteria were studies that explicitly targeted participants with acute or chronic medical conditions, who had eating disorders or mental health problems, or those were not published in English.

2.3. Data screening and extraction

A two‐step screening approach was used to code and extract data. In the initial step, the first author created the inclusion/exclusion criteria for data extraction and evaluation. Two of the authors (MA and AA) independently performed the eligibility assessment in a standardized manner using Covidence software. After removing duplicates, these authors reviewed the title and abstract for all studies to determine eligibility. Then, a full‐text review was conducted on all potentially eligible studies. Finally, those studies that met the inclusion and exclusion criteria were included in the review and meta‐analysis. Any disagreements between the authors were resolved by consensus. The search was completed in June 2023. The data extraction included the author's name, year of publication, country, design, sample size, participants' characteristics, type and focus of MBIs, duration, type of eating habits (outcomes), and key findings. The second step included multiple discussions among the two independent coders during the evaluation; consensus then was reached after resolving any discrepancies with the first author.

2.4. Quality appraisal and risk of bias

The two independent raters (MA and AA) scored eligible articles using the Alberta Heritage Foundation of Medical Research tool. ²⁹ This tool has 14 items that address five domains of quantitative study design, including study design (five items, sampling selection and data analysis), selection bias (two items), randomization procedure (one item), concealment procedure (two items), and reporting/overall synthesis (four items, internal/external validity). Each item was scored as one of the following: 0 (Not met), 1 (Partially met), and 2 (Met). A quality appraisal score (QA range = 0 to 1) for each article was calculated and averaged by using the total sum score of the 14 items divided by 28 (Table S2). Following the tool guideline, a quality score lower than 0.55 was considered low quality and consequently excluded from this review. Inter‐rater consistency was evaluated using Cohen's kappa before averaging the quality score between these two raters.

2.5. Statistical data analysis

The Comprehensive Meta‐Analysis program (version 3.0) was used to perform all meta‐analyses. To calculate the pooled effect size for each study outcome, we used the mean, standard deviation (SD), and the sample size in the experimental and control groups at each data assessment (baseline, post‐intervention, and follow‐up). If the mean or SD was not reported in the article, we contacted the corresponding author twice (2–3 weeks apart). If we did not receive a response within 6 weeks, we calculated the mean or SD from the reported median, sample size, interquartile range, and confidence intervals if applicable. ²⁶ , ²⁷ Influential outliers were defined as having residual >2.56 and I ² being decreased by ≥10% after removing each potential outlier. ²⁸ Hedge's g was calculated using random‐effects models as the effect size in order to examine each study's sample size and was interpreted as small = 0.20, medium = 0.50, or large = 0.80. ³⁰ A negative Hedge's g indicated that the intervention group had a greater decrease in the targeted eating habits than the control group. The pre‐ and post‐intervention correlation was fixed at 0.50. ³¹ The effects of long‐term sustainability from baseline to follow‐up and post‐intervention to follow‐up assessment were also evaluated for the articles with follow‐up data. Subgroup and meta regression analyses with mixed‐effects models were applied to conduct exploratory moderation analyses. Heterogeneity among studies was evaluated using the Q test and I ² statistics: low heterogeneity = 25%, moderate heterogeneity = 50%, and high heterogeneity = 75%. There was no strong evidence of publication bias when the results from the Begg and Mazumdar rank correlation test and the Egger's regression asymmetry test were nonsignificant, and the funnel plot ³² was symmetric. When publication bias existed, the Duval and Tweedie's trim and fill method were applied to adjust the effect size. Sensitivity analyses were performed to examine the robustness of the results according to each study's quality score.

3. RESULTS

3.1. Study characteristics

As illustrated in Table S1, of 32 included studies, 31 (97%) ¹⁵ , ¹⁶ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸ were randomized controlled trials (RCT); 11 (34%) ¹⁶ , ³³ , ³⁵ , ³⁶ , ³⁷ , ³⁹ , ⁴¹ , ⁴⁵ , ⁵⁵ , ⁵⁷ , ⁵⁸ were conducted in the United States, followed by the United Kingdom (k = 8, 25%), ³⁸ , ⁴⁰ , ⁴³ , ⁴⁶ , ⁵² , ⁵³ , ⁵⁴ , ⁵⁶ the Netherlands (k = 4, 13%), ³¹ , ³² , ³⁴ , ⁴⁴ Spain (k = 3, 9%), ¹⁵ , ⁴⁷ , ⁵⁹ and Brazil (k = 2, 6%), ⁴⁹ , ⁵⁰ as well as one each from Austria, ⁵¹ Denmark, ⁴² Greece, ³⁰ and Portugal. ⁴⁸ A total of 2431 participants (78% female) were included, with ages ranging from 9 ⁵⁹ to 58 ⁴⁷  years (overall M _age  = 36.72). Sample sizes ranged from 19 ³² to 257, ³⁴ with three studies (9%) ³³ , ⁵⁴ , ⁵⁹ targeting children (M _age  = 11.96).

3.2. Intervention characteristics

Interventions were mostly delivered at a clinic (k = 27; 84%) ¹⁵ , ¹⁶ , ³⁰ , ³¹ , ³² , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ , ⁵⁴ , ⁵⁶ , ⁵⁷ , ⁵⁸ followed by school settings (k = 5; 16%). ³³ , ³⁹ , ⁴⁴ , ⁵⁵ , ⁵⁹ Mean intervention length was 8.2 weeks (range: 1–24 weeks), with 20 studies (63%) having a minimal length of 7 weeks and nine (28%) having a maximum length of 3 weeks. About 47% (k = 15) of studies targeted participants with BMIs greater than 25. ³⁰ , ³² , ³⁵ , ³⁶ , ⁴² , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵⁵ , ⁵⁶ , ⁵⁸ , ⁶⁰ Although mindful eating was introduced in all included studies to various degrees, the mindfulness components can be demarcated into mindful eating [k = 20], mindful movement [k = 4], and mindfulness‐based stress reduction [k = 8]. About 59% of studies (k = 19) used mindfulness strategies as the primary intervention component, while the remaining 41% (k = 13) ³⁰ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁵⁰ , ⁵¹ , ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁸ , ⁶⁰ incorporated other health promotion programs to support healthy eating behaviors.

Of the 32 selected studies, nine (28%) utilized certified mindfulness practitioners as intervenors, while the rest trained their own intervenors. Most control conditions were waitlist or no‐intervention control (k = 11, 34%), while others applied attention control including standardized weight‐loss program (k = 4), existing care program (k = 3), reading (k = 3), PA (k = 2), PA/diet (k = 1), nutrition (k = 2), education (k = 2); and non‐mindful eating (k = 1). Furthermore, 27 studies (84%) reported immediate post‐intervention effects and six (19%) had follow‐up assessments (T3) ¹⁵ , ⁴⁵ , ⁴⁷ , ⁵¹ , ⁵⁴ , ⁵⁵ to evaluate long‐term effects. Treatment fidelity was described in two studies (6%). ⁴¹ , ⁵⁴ While most studies (k = 31) delivered the intervention through in‐person contacts, one study ³⁵ delivered mindfulness training remotely (using phone calls).

3.3. Risk of bias

Of the 32 included studies (see Table S2), no study was identified as having a high risk of bias (QA ≤0.55), while 15 studies (47%) had a moderate risk (QA = 0.56–0.74), ³⁰ , ³¹ , ³² , ³³ , ³⁶ , ³⁹ , ⁴² , ⁴⁶ , ⁴⁸ , ⁵¹ , ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁷ and 17 studies (52%) had a low risk of bias (QA ≥0.75). The moderate risks of bias were mostly due to selection bias (random sequence or allocation concealment), performance bias (blind outcome assessment), and attrition bias (incomplete data). No study was excluded as a result of high risk of bias. The inter‐rater agreement between the two raters was acceptable (Cohen's Kappa = 0.83).

3.4. Intervention pooled effects

3.4.1. Mindless eating habits

Controlled eating

Among the 17 comparisons, one influential outlier ³² was identified (z = 3.70), and I ² decreased from 58.70% (p = 0.001) to 29.23% (p = 0.131) after removing the outlier. Similarly, the pooled effects on increasing controlled eating decreased from 0.30 to 0.23 (see Figure 2A). Results from the Begg and Mazumdar rank correlation test (Tau = 0.12, p = 0.528), the Egger's regression intercept test (p = 0.304), and the funnel plot (see Figure 2B) showed no evidence of publication bias.

FIGURE 2.

Mindless‐related eating habits: controlled eating. (A) Forest plot for controlled eating. (B) Funnel plot for the controlled eating. Note. Begg and Mazumdar rank correlation test (Tau = 0.12, p = 0.528), the Egger's regression intercept test (β = 1.26, p = 0.304).

Moderators

As shown in Table S3, MM type was a significant moderator for the immediate intervention effects (p = 0.010). Programs that incorporated mindful movements and stress reduction interventions resulted in better effects than those that used mindful eating alone (g = 0.50, 0.28 vs. −0.01). Although not significant, interventions delivered by trained research assistants had greater effects than those delivered by a certified intervenor (g = 0.32 vs. 0.10). Moreover, MM as an add‐on had smaller effects than MM‐focused interventions (g = 0.16 vs. 0.34). Interventions delivered remotely that focused on older children or that were community‐based tended to have better effects on improving controlled eating.

Impulsive food choices

One significant outlier was identified (z = 3.32). ⁵⁵ After removing the outlier, the heterogeneity decreased from 83.40% (p < 0.001) to 71.05% (p < 0.001). The pooled effects on decreasing impulsive food choices increased from −0.28 to −0.43 (see Figure 3A). There was no strong evidence of publication bias based on the results from the Begg and Mazumdar rank correlation test (p = 0.055) and the Egger's regression intercept test (p = 0.072) as well as the funnel plot (see Figure 3B).

FIGURE 3.

Mindless‐related eating habits: impulsive food choices. (A) Forest plot for impulsive food choices. (B) Funnel plot for the impulsive food choices. Note. Begg and Mazumdar rank correlation test (Tau = −0.42, p = 0.055), the Egger's regression intercept test (β = −2.28, p = 0.072).

Moderators

As demonstrated in Table S4, no significant moderator was identified. When comparing the different effect sizes, the interventions had better effects when focusing on mindful eating (g = −0.53), with female participants (g = −0.63), not assessing treatment fidelity (g = −0.53), being an add‐on component to another program (g = −0.81) or targeting OW/OB participants (g = −0.81).

External eating

After removing one influential outlier (z = −5.32), ⁶⁰ the heterogeneity decreased from 93.13% (p < 0.001) to 61.66% (p = 0.011). The pooled effect size decreased from −1.29 to −0.62 (see Figure S1a). Results from the Begg and Mazumdar rank correlation test (p = 0.138), the Egger's regression intercept test (p = 0.554), and the funnel plot (see Figure S1b) showed no evidence of publication bias.

Fullness awareness and hunger

The heterogeneity was small for fullness awareness (k = 4, I ²  = 5.52%; Q = 3.18, p = 0.365) and high for hunger (k = 4, I ²  = 78.42%, Q = 13.90, p = 0.003). The pooled intervention effects were 0.64 on improving fullness awareness (see Figure S2a) and −0.69 on reducing hunger (see Figure S3a). Long‐term sustained effects from post‐intervention to follow‐up were noted (k = 2; g = −0.50, 95%CI: −0.87, −0.13; p = 0.007) on hunger. As shown in Figures S2b and S3b, there was no strong evidence of publication bias. This was supported by the results from the Begg and Mazumdar rank correlation test (Tau = 1.00, p = 0.042; Tau = 0, p = 1.00) and the Egger's regression intercept test (β = 5.70, p = 0.295; β = 16.96, p = 0.342).

Energy and sweet intake

The heterogeneity was moderate for energy intake (I ²  = 55.56%; Q = 13.50, p = 0.036) and small for sweet intake (I ²  = 0%; Q = 1.94, p = 0.585). The pooled intervention effects were −0.60 and −0.39 on reducing energy and sweet intake, respectively (see Figure S4a, S5a). Results from the Begg and Mazumdar rank correlation test (Tau = −0.05, p = 0.881; Tau = −0.33, p = 0.497), the Egger's regression intercept test (β = 16.96, p = 0.342; β = 0.17, p = 0.970), and the funnel plot (see Figure S4b, S5b) showed no evidence of publication bias.

3.4.2. Stress‐related eating habits

Emotional eating

The pooled immediate intervention effects on reducing emotional eating were −0.27 (see Figure S6a). The long‐term sustained effects from post‐intervention to follow‐up were −0.22 (95%CI: −2.04, 1.59; p = 0.809). Results from both the Begg and Mazumdar rank correlation test (Tau = −0.28, p = 0.053) and the Egger's regression intercept test (β = −3.68, p = 0.123) as well as the funnel plot (see Figure S6b) indicated no evidence of publication bias.

Moderators

Community‐based interventions had significantly better effects than school‐based interventions (g = −0.32 vs. 0.96, p = 0.038; see Table S5). In addition, interventions that focused on stress reduction (g = −0.61), among females (g = −0.41), with young children (g = −0.32), and were delivered in‐person (g = −0.31) or by trained research assistants (g = −0.37) resulted in more positive effects, but the results were not significant.

Binge eating

The heterogeneity was moderate (Q = 14.12, p = 0.007, I ²  = 71.68%). The pooled effects on reducing binge eating were −0.35 (Figure S7a). Results from both the Begg and Mazumdar rank correlation test (Tau = 0.60, p = 0.142) and the Egger's regression intercept test (β = 11.72, p = 0.200) indicated no evidence of publication bias. In addition, the funnel plot of standard error by Hedges' g was symmetric (Figure S7b).

3.5. Sensitivity analysis results

Studies with a moderate quality had significantly greater effects on decreasing impulsive food choices than those with a high quality (g = −0.81 vs. −0.20, p = 0.012). Other outcomes were not significantly related to study quality. However, high‐quality studies were more likely to have larger effects on decreasing binge eating (g = −0.51 vs. −0.09, p = 0.434), energy intake (g = −0.70 vs. −0.51, p = 0.663), and increasing controlled eating (g = 0.29 vs. 0.13, p = 0.348) than moderate‐quality studies. In contrast, studies with a moderate quality tended to have greater effects on decreasing emotional eating (g = −0.35 vs. −0.20, p = 0.603) than those that were of high quality.

4. DISCUSSION

To the best of our knowledge, this is the first systematic review and meta‐analysis that has explicitly evaluated the effects of MBIs for improving eating behaviors. Obesogenic eating habits can increase energy intake and contribute to elevated adiposity and obeisty. ⁶¹ This systematic review summarized the range of MBIs designed for preventing and treating obesogenic eating habits as well as the effectiveness of MBIs on various mindless and stress‐related eating habits. Results from this meta‐analysis revealed significant small‐to‐moderate pooled effects (g = 0.23 to −0.69) on improving mindless eating habits. These results are comparable to the effect sizes displayed conjointly in other studies, such as healthy eating (r = 0.14), ¹³ eating attitudes (g = 0.57), and overall eating behaviors (g = 0.53). ¹¹ Subsequently, MBIs can be utilized to increase individuals' awareness with eating (↑awareness of fullness, physical hunger, and external/internal cues) and help them realize why, what, and how much they should eat. Additionally, mindful eating training can be used to successfully mitigate an autopilot eating style and help to re‐establish a healthy and satisfactory relationship with food (↓impulsive food choices). Our results endorse the possibility of increasing participants' physical senses (satiety) to the eating experience and enable them to enjoy the food choices they have made (↑gratitude).

In addition, the significant increases in controlled eating and fullness awareness as well as the decreases in external eating and physical hunger are unique. It is possible that MBIs can be used to significantly connect individuals' sense of awareness with food and repurpose the meaning of eating. Mindful eating training might be a better approach for weight reduction than traditional dieting (dietary restriction for weight loss purpose) because MBIs help individuals purposefully pay attention to their food without judgment. As a result, MBI practitioners are more connected to their physical senses (hunger, fullness) and more appreciative of their food (satiety, gratitude, and enjoyment). Consequently, mindful eating practitioners are less likely to be concerned about restricting their food intake. Since mindfulness is a nonjudgmental approach, mindfulness practitioners are encouraged to choose what and how much they want to consume and are more likely to select foods that are beneficial to their health. For example, our findings in decreased energy/sweet intakes and impulsive food choices further authenticate the promise of using MBIs to tackle awareness‐related mindless eating habits. Moreover, the significant long‐term sustained effects on physical hunger further validate the benefits of using MBIs to improve mindless eating habits.

In terms of the effects on stress‐related eating habits (binge and emotional eating), the effect sizes were small (−0.27 and −0.35, respectively) and insignificant. Our effect size on binge eating is much lower than outdated meta‐analyses (>6 years) conducted with overweight/obese adults (g = −0.90) ¹⁰ or with adults who needed psychological interventions to reduce their problematic eating behaviors (g = −0.70). ¹² For the more recent meta‐analysis (conducted in 2023) targeting adults living with overweight or obesity, MBIs had a combined effect size (percentage change) of −2.37% on emotional eating. ⁶² In another recent systematic review focused on adults with problem eating behaviors, the positive effects of MBIs were concluded, but no pooled effect size was available for comparison. ⁶³ Overall, the effects of MBIs on emotional or binge eating habits remain relatively obscure, particularly among the general population (those people with or without weight concerns). However, our present findings in improving emotional eating do provide some insights into how MBIs may be utilized to handle stress‐related eating habits. Most importantly, the small effects on stress‐related eating habits have some clinical significance, particularly when delivered in clinical settings, because our moderation analysis showed that the effects were significantly stronger in clinical settings than in school settings. This finding suggests that clinically tailored MBIs may be more useful for meeting participants' psychological needs and consequently have greater potential for improving participants' emotional and binge eating habits. Unfortunately. our study did not find a significant long‐term sustained effect on reducing emotional eating (g = −0.22). Additional efforts are needed to understand how MBIs can be modified to improve stress‐related eating habits over short‐ and long‐term periods, particularly among those who do not yet have severe eating or mental health problems.

Only two studies (6.25%) in this review reported on treatment fidelity, which may affect studies' reproducibility. Assessing, monitoring, and reporting treatment fidelity (i.e., the degree to which an intervention has been implemented as intended) is essential because it is impossible to know the actual effects of the intervention if the intervention is not consistently delivered. This gap can seriously compromise the evolution of practical applications for MBIs (e.g., scaling up in the real world). ⁶⁴ This concern is also reflected in the result of our sensitivity analysis in which study quality (moderate and high QA score) seemed to play a role on the effects of decreasing impulsive food choices. Future MBI studies should consistently and rigorously evaluate treatment fidelity to enhance intervention programs' replicability and scalability.

4.1. Limitations

There were some limitations noted. First, there was high heterogeneity (78.42%) in hunger eating habits, which could impact the reliability of the result. Second, because different measurement tools were utilized to assess eating habits, not all tools were equally validated for people of all genders, ages, and BMIs. Moreover, only two studies (6.25%) in this review assessed treatment fidelity, which may have an impact on interventions' reproducibility. Finally, while we reported on the risk of bias of all included studies and considered the impact of heterogeneity and publication bias during meta‐analysis, a formal GRADE ⁶⁵ evaluation was not performed for this review. Future research should consider a GRADE evaluation prior to making any practice related recommendations.

5. CONCLUSION

This meta‐analysis found that utilizing MBIs to address mindless eating habits resulted in significant improvements on controlled eating, fullness awareness, external eating, hunger, energy and sweet intake, and impulsive food choices. In terms of stress‐related eating, MBIs showed some promise in reducing emotional and binge eating habits. Very few studies investigated the long‐term, sustained effects of MBIs. Thus, long‐term effectiveness on maintaining obesogenic eating behaviors warrants further investigation, particularly among the general population with or without weight‐related issues. Such studies also should strive to better understand the active components and mechanisms of changes in effective MBIs on obesity prevention.

AUTHOR CONTRIBUTIONS

All authors contributed to the design of the study and writing of the manuscript. All authors have approved the final manuscript.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Figure S1. Mindless‐related Eating Habits: External Eating.

Figure S1.a. Forest plot for external eating.

Figure S1.b. Funnel plot for the external eating. Note. Begg and Mazumdar rank correlation test (Tau = −0.43, p = 0.138), the Egger's regression intercept test (β = −2.54, p = 0.554).

Figure S2. Mindless‐related Eating Habits: Fullness Awareness.

Figure S2.a. Forest plot for fullness awareness.

Figure S2.b. Funnel plot for the fullness awareness. Note. Begg and Mazumdar rank correlation test (Tau = 1.00, p = 0.042), the Egger's regression intercept test (β = 5.70, p = 0.295)

Figure S3. Mindless‐related Eating Habits: Hunger.

Figure S3.a. Forest plot for hunger.

Figure S3.b. Funnel plot for the hunger. Note. Begg and Mazumdar rank correlation test (Tau = 0, p = 1.0), the Egger's regression intercept test (β = 16.96, p = 0.342)

Figure S4. Mindless‐related Eating Habits: Energy Intake.

Figure S4.a. Forest plot for energy intake.

Figure S4.b. Funnel plot for the energy intake. Note. Begg and Mazumdar rank correlation test (Tau = −0.0.05, p = 0.881), the Egger's regression intercept test (β = 0.17, p = 0.970)

Figure S5. Mindless‐related Eating Habits: Sweet Intake.

Figure S5.a. Forest plot for sweet intake.

Figure S5.b. Funnel plot for the sweet intake. Note. Begg and Mazumdar rank correlation test (Tau = −0.0.33, p = 0.497), the Egger's regression intercept test (β = −0.67, p = 0.684)

Figure S6. Stress‐related Eating Habits: Emotional Eating.

Figure S6.a. Forest plot for emotional eating.

Figure S6.b. Funnel plot for the emotional eating. Note. Begg and Mazumdar rank correlation test (Tau = −0.0.28, p = 0.123), the Egger's regression intercept test (β = −3.68, p = 0.123)

Figure S7. Stress‐related Eating Habits: Binge Eating.

Figure S7.a. Forest plot for binge eating.

Figure S7.b. Funnel plot for the binge eating. Note. Begg and Mazumdar rank correlation test (Tau = 0.0.60, p = 0.142), the Egger's regression intercept test (β = 11.72, p = 0.200)

Table S1. Article Summary. Note. Measurements: BASC = Basic Assessment System for Children; BES = Binge Eating Scale; DEBQ = Dutch Eating Behavior Questionnaire; DTS = The Dichotomous Thinking Scale; Eat‐26 = The Eating Attitudes Test for eating disorder; EBI = Eating Behavior Inventory; EI = Eating Inventory‐51; EEQ = Emotional Eating Questionnaire; EES = Emotional Eating Scale; FCQ = Food Craving Questionnaire; FFMQ = Five Facet Mindfulness Questionnaire; FFQ = Food Frequency Questionnaire; G‐FCQ‐T = General Food Craving Questionnaire Trait; GHQ = General Health Questionnaire; IES‐2 = Overall Intuitive Eating Scale‐2; KIM‐E = Kentucky Inventory Mindfulness Skills extended; MCD = Mindful Construal Diaries; MAAS = Mindfulness Awareness Scale; MetaCheck™ = an open‐circuit calorimeter; MEQ = Mindfulness Eating Questionnaire; ORWELL = Obesity Related Well‐being Questionnaire; PANAS = Positive and Negative Affect Schedule; PFS = Power of Food Scale; PSRS = Perceived Self‐Regulatory Success in dieting; RST‐PQ = The Reinforcement Sensitivity Theory Personality Questionnaire; SFVQ = Short Fruit and Vegetable Questionnaire; SMS = State Mindfulness Scale: SCS = Self‐Compassion Scale; TFEQ = Three‐Factor Eating Questionnaire; TSST = Trier Social Stress Test; TWT = Toy Wait Task: VAS = visual analogue scale for hunger and satiety; WEB‐SG = Weight‐ and Body‐Related Shame and Guilt scale; WSSQ = Weight Self‐stigma Questionnaire. Abbreviations: ABMT = Attention Bias Modification Training; b/t = between; Ed = Education; hrs = hours; MM = Mindfulness Meditation; ME = Mindfulness Eating; MT = Mindfulness Training; NQ = Nutritional Questionnaires; TAU = treatment as usual; Nut = Nutrition; PA = physical activity; sig = significant; Std = standard; Tx = treatment; YD = Yoga Dance.

Table S2. Quality Appraisal: Assessment Questions, Domains, and Quality Appraisal Scores. Notes. 0 = not met; 1 = partially met; 2 = met: QA = total score/28

Table S3. Moderators for the intervention effects on controlled eating. Note. OW = overweight; OB = obese; PA = physical activity; TX = treatment

Table S4. Moderators for the intervention effects on impulsive food choices. Note. OW = overweight; OB = obese; PA = physical activity; TX = treatment

Table S5. Moderators for the intervention effects on emotional eating. Note. OW = overweight; OB = obese; PA = physical activity; TX = treatment

Kao T‐SA, Ling J, Alanazi M, Atwa A, Liu S. Effects of mindfulness‐based interventions on obesogenic eating behaviors: A systematic review and meta‐analysis. Obesity Reviews. 2025;26(3):e13860. doi: 10.1111/obr.13860