Based on developmental and media influence theories, this study proposes a conceptual framework in which watching mukbang is associated with adolescent anxiety both directly and indirectly through dietary behaviors. Adolescents are particularly sensitive to rewarding and emotionally salient media stimuli due to ongoing neurodevelopment and heightened reward responsiveness [14,15]. Frequent exposure to mukbang content is expected to increase high-caffeine beverages (Hypothesis 2) and sugar-sweetened beverages consumption (Hypothesis 3) through observational learning and cue-reactivity mechanisms [5,7,16]. In turn, higher intake of high-caffeine beverages and sugar-sweetened beverages is hypothesized to elevate anxiety levels among adolescents, thereby mediating the association between mukbang viewing and anxiety (Hypothesis 4) [10,11,17]. In addition to these indirect pathways, mukbang viewing is expected to have a direct positive association with anxiety, independent of dietary behaviors (Hypothesis 1), potentially through emotional arousal, parasocial interaction, and social comparison processes [18]. Fig. 1 illustrates the conceptual framework guiding the present study.

Adolescence is a critical developmental period characterized by heightened emotional sensitivity, increased reward responsiveness, and incomplete maturation of cognitive control systems [14,15]. These developmental characteristics increase adolescents’ vulnerability to external influences, including digital media exposure. Growing evidence suggests that food-related digital content can influence both dietary behaviors and mental health outcomes among adolescents [3,4,5].

According to social cognitive theory, adolescents are likely to model behaviors displayed by media figures when those behaviors appear enjoyable or socially reinforced [8]. Mukbang content frequently includes the consumption of coffee and energy drinks, often portrayed as enhancing pleasure or endurance during eating. Empirical studies have shown that exposure to food- and beverage-related media increases adolescents’ intake of the featured products [16,17].

Visual exposure to sugar-rich, highly palatable foods can activate reward-related neural pathways and increase cravings, particularly during adolescence [4]. Prior research on digital food marketing and influencer-based eating content indicates that such exposure promotes greater consumption of sugar-sweetened beverages and snacks among youth [6,7]. Mukbang videos often emphasize large portions of sweet foods and beverages, reinforcing these consumption patterns.

Caffeine consumption has been consistently associated with increased physiological arousal, sleep disturbance, and anxiety symptoms among adolescents [9,10]. Adolescents are especially sensitive to caffeine’s anxiogenic effects due to heightened stress reactivity [11]. Similarly, excessive intake of sugar-sweetened foods and beverages has been linked to emotional dysregulation and anxiety in youth populations [12,13]. Given that mukbang viewing may increase both high-caffeine beverages and sugar-sweetened beverages, these dietary behaviors are hypothesized to mediate the association between mukbang exposure and anxiety.

This study is a secondary analysis study conducted to prepare countermeasures to prevent adolescent mental health by analyzing original data from the seventeenth Korea Youth Risk Behavior Web-based Survey (KYRBS) in 2022. The sample size of the KYRBS is 400 middle schools and 400 high schools. The sample number of schools is allocated based on the city/province, city size (metropolitan, small, and medium-sized city and county area), group, male/female for middle schools, and male/female and general high school and specialized high school for high schools by applying the proportional allocation method. Two-stage cluster sampling was used for sampling, and the sample school was selected by a systematic sampling method by estimating the sampling interval after sorting the school list of the population by strata for the primary sampling. For the secondary sampling, one class was randomly selected by grade from the selected sample school. Subjects of the survey ranged from first-year middle school students to third-year high school students in 800 sample schools nationwide, and the response rate of the KYRBS in 2022 was 92.9%.

Watching mukbang was assessed by asking participants how frequently they had watched mukbang content during the past 12 months. Responses were originally measured on a 7-point ordinal scale with the following anchors: 1 = “not at all,” 2 = “less than once a month,” 3 = “once to three times a month,” 4 = “once to twice a week,” 5 = “three to four times a week,” 6 = “five to six times a week,” and 7 = “every day.” For analytic purposes, the scale was linearly transformed to a 0–6 metric by subtracting one from each response, such that higher values indicated more frequent mukbang viewing. This transformation did not alter the relative spacing or interpretation of the original response categories. Anxiety symptoms were assessed using the Generalized Anxiety Disorder-7 (GAD-7), which measures the frequency of anxiety-related symptoms over the past two weeks on a 4-point scale ranging from 0 (“not at all”) to 3 (“nearly every day”). The GAD-7 has demonstrated good internal consistency in prior research (Cronbach’s α = 0.89) [19], and reliability was similarly high in the present study (Cronbach’s α = 0.91).

High-caffeine beverage consumption was measured using a single item assessing intake frequency during the past 7 days, with response options ranging from 1 (“did not consume during the past 7 days”) to 7 (“three or more times per day”). Participants were asked, “During the past 7 days, how often did you consume high-caffeine beverages?” Sugar-sweetened beverage consumption was assessed using an identical 7-day recall period and response scale. Different reference periods were used to reflect the conceptual nature of each construct. Mukbang viewing was measured over 12 months to capture habitual exposure, anxiety symptoms were assessed over the past two weeks following the GAD-7 guidelines, and dietary behaviors were measured using a 7-day recall period to reduce recall bias.

Demographic covariates included gender, grade, academic record, and household economic level. Respondents were divided into ‘males’ and ‘females’. In order to see the difference according to the school grade, it was classified into two categories, middle and high schools, and their academic record and household economic level were measured on the 5-point Likert scale of high (5), mid-high (4), middle (3), mid-low (2), and low (1).

Data were analyzed using SPSS Statistics version 21.0 (IBM Corp., Armonk, NY, USA). In this study, the survey employed a stratified, clustered, and weighted sampling design. Individual sampling weights, stratification variables, and primary sampling unit (PSU) cluster variables provided in the raw data were applied in the computation of descriptive statistics to reflect the complex survey design. In contrast, inferential analyses were conducted using standard statistical procedures that did not fully incorporate the complex sampling structure. In the raw data, response options such as “not applicable,” “don’t know,” and “no response” were treated as non-informative and recoded as missing values prior to analysis. The total sample comprised 51,850 respondents, with negligible missing data for variables. Analyses were conducted using valid responses, and the effective sample size remained 51,850.

Survey-weighted descriptive statistics were computed to summarize the characteristics of the study variables. Pearson’s correlation coefficients were calculated to examine the associations among watching mukbang, high-caffeine beverage consumption, sugar-sweetened beverage consumption, and anxiety. The internal consistency of the anxiety measure (GAD-7) was assessed using Cronbach’s α coefficient. Mediation analyses were conducted to test whether high-caffeine beverage consumption and sugar-sweetened beverage consumption mediated the relationship between watching mukbang and anxiety. Parallel mediation analyses were performed using PROCESS Model 4 with 5000 bootstrap samples, adjusting for demographic covariates including academic record, household economic level, grade, and gender. Unstandardized coefficients (B), standard errors, and 95% confidence intervals were reported.

Table 1 presents the descriptive statistics for all study variables. Participants reported an average anxiety score of 4.20 (SD = 4.65), with scores ranging from 0 to 21. The mean frequency of watching mukbang content was 2.22 (SD = 1.91), ranging from 0 to 6. Caffeine consumption showed a mean of 1.89 (SD = 1.24), and sugar-sweetened beverage consumption had a mean of 1.19 (SD = 0.85), each ranging from 1 to 7. Regarding the distribution of mukbang viewing frequency, 14,781 participants (29.4%) reported not watching mukbang at all, 5326 (10.3%) reported watching less than once a month, 9580 (18.3%) reported watching once to three times a month, 9195 (17.5%) reported watching once to twice a week, 5957 (11.3%) reported watching three to four times a week, 2575 (4.8%) reported watching five to six times a week, and 4418 (8.3%) reported watching mukbang every day. The mean academic record score was 3.12 (SE = 0.01), and the mean household economic level was 3.43 (SE = 0.01). The average grade level of participants was 3.47 (SE = 0.24). In terms of gender, 51.6% of the sample were male and 48.4% were female.

Table 2 displays the correlations among the main study variables. Anxiety scores exhibited a small but significant positive correlation with watching mukbang content (r = 0.07, p < 0.01), high-caffeine beverages consumption (r = 0.13, p < 0.01), and sugar-sweetened beverages consumption (r = 0.09, p < 0.01). Watching mukbang content was also positively correlated with high-caffeine beverages consumption (r = 0.06, p < 0.01) and sugar-sweetened beverages consumption (r = 0.09, p < 0.01). In addition, high-caffeine beverages consumption was moderately correlated with sugar-sweetened beverages (r = 0.28, p < 0.01). These results indicate that higher engagement in mukbang watching and greater intake of high-caffeine beverages and sugar-sweetened beverages are each associated with higher levels of self-reported anxiety.

A parallel multiple mediation analysis was conducted using PROCESS Model 4 to examine indirect associations between mukbang viewing and anxiety through high-caffeine beverage intake and sugar-sweetened beverage intake. Gender, academic record, household economic level, and grade were included as covariates (Table 3 and Table 4).

Mukbang viewing was positively associated with higher reported consumption of high-caffeine beverages (b = 0.036, β = 0.15, p < 0.001) and sugar-sweetened beverages (b = 0.066, β = 0.14, p < 0.001). In the model predicting anxiety, both high-caffeine beverage intake (b = 0.406, β = 0.18, p < 0.001) and sugar-sweetened beverage intake (b = 0.246, β = 0.12, p < 0.001) were positively associated with anxiety scores. Mukbang viewing remained significantly associated with anxiety after accounting for the mediators (b = 0.074, β = 0.13, p < 0.001), and the total association between mukbang viewing and anxiety was also significant (b = 0.105, β = 0.19, p < 0.001).

The indirect association via high-caffeine beverage intake was statistically significant (B = 0.0148, 95% CI for B [0.0120, 0.0175]), as was the indirect association via sugar-sweetened beverage intake (B = 0.0163, 95% CI for B [0.0136, 0.0192]). The total indirect association was also statistically significant (B = 0.0311, 95% CI for B [0.0271, 0.0352]). Because all 95% confidence intervals for the unstandardized coefficients (B) did not include zero, both indirect pathways were statistically supported. Overall, the models explained a modest proportion of variance. Mukbang viewing and covariates accounted for 2.9% of the variance in high-caffeine beverage consumption and 2.1% of the variance in sugar-sweetened beverage consumption. In the mediation model, the predictors explained 5.6% of the variance in anxiety, indicating small effect sizes despite statistical significance. Accordingly, these findings reflect modest statistical associations rather than strong practical or causal effects, and no causal inferences can be drawn from these cross-sectional data.