Nonlinear association between body mass index, perceived discrimination, and psychological sub-health problems among adolescents: risks of undernutrition versus overnutrition Authors: Muyuan Liu, Yan Xiong, Yuan Yao, Xinqi Du, Han Yang, Xiaoyun Liang, Qiyuan You, Yang Liu, Hong Li, Feifei Gao Published: 2026 ABSTRACT AbstractIntroductionPrevious studies have largely emphasized the adverse effects of overnutrition on mental health. However, undernourished adolescents are also likely to experience social prejudice and exclusion, which may negatively correlate to psychological well-being. However, evidence showing associations between undernutrition and mental health remains limited. In this study, we examine the relationship between nutritional status and psychological sub-health (PSH) among Chinese adolescents and investigate a mediating role for perceived discrimination.MethodsIn this cross-sectional study, a total of 2,699 adolescents from three provinces in China participated in this questionnaire-based study. Analysis of variance, logistic regression, and restricted cubic spline (RCS) analyses were conducted to examine nutritional status and perceived discrimination associations with PSH. Structural equation modeling was used to examine pathways linking body mass index (BMI), perceived discrimination, and PSH across different nutritional status subgroups, while some covariates (such as gender, sleep duration, and alcohol consumption) were adjusted.ResultsOverall, 461 adolescents (17.08%) exhibited PSH. Both undernutrition (Odds ratio (OR) = 1.57, 95% Confidence interval (CI) = 1.08–2.27, P = 0.017) and overnutrition (OR = 1.67, 95% CI = 1.32–2.12, P < 0.001) were associated with an increased risk of PSH, indicating a U-shaped nonlinear dose–response relationship (χ² = 15.37, P for nonlinearity < 0.001). Perceived discrimination significantly mediated the association between BMI and PSH in undernourished (β = −0.069, P = 0.012) and overnourished (β = 0.037, P = 0.010) adolescents. Additionally, overnutrition was directly associated with a higher PSH risk among male adolescents (β = 0.118, P = 0.007).ConclusionsA significant nonlinear association between BMI and PSH was observed in Chinese adolescents, with perceived discrimination playing an important mediating factor in both undernourished and overnourished groups. These findings provide clues to the causes of psychological sub-health problems among adolescents with different nutritional statuses and also point out the direction for future research on developing targeted mental health promotion strategies tailored to adolescents with different nutritional statuses.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40359-026-04965-w.Keywords: Weight-related stigma, Psychological sub-health, Perceived discrimination, Body mass index, Nutritional status, Adolescents Introduction Puberty is a critical developmental stage characterized by the gradual maturation of multiple organs and physiological systems [1]. During this period, adolescents’ mental and emotional states are highly plastic and particularly susceptible to self-evaluation and external judgments [2]. Globally, an estimated 10%–20% of adolescents experience mental health problems at some point during adolescence [3]. Common conditions include depression, anxiety, and sleep disturbances [4–6], which may adversely affect academic performance and social functioning [7, 8] and can persist into adulthood or even throughout life [9]. According to the Statistical Report on Youth Development in China (2020), approximately 30 million adolescents under 17 years old experience emotional or behavioral problems, underscoring a major public health challenge [10]. Psychological sub-health (PSH) refers to a persistent negative psychological state that lies between mental health and clinically diagnosed mental disorders [11]. The state is characterized by symptoms such as unexplained fatigue, emotional dysregulation, anxiety, and memory decline [11]. When compared with overt mental disorders, PSH differs in severity, clinical presentation, duration, reversibility, and functional impact, reflecting a gradual transition from quantitative to qualitative impairment [12]. Owing to its concealed nature and tendency to be overlooked, PSH carries a high progression risk to clinically diagnosable mental illness [13]. Although PSH often represents a precursor to more severe mental health conditions, it also presents a critical window for early identification and intervention [14]. Compared with the health hazards, the limited accessibility of treatment services and the intensive resources consumption of mental disorders [15], early management of PSH has diversified intervention paths and options, and has significant cost-benefit advantages and public health values [16–18], due to the reversibility natural of PSH. Therefore, identifying PSH risk factors and implementing preventive strategies in adolescents is of substantial public health importance, yet targeted mental health promotion efforts remain insufficient. Over the past three decades in China, rapid economic development and improved living standards have contributed to a marked increase in overweight and obesity prevalence among children and adolescents, making overnutrition an increasingly serious public health issue [19, 20]. Overnutrition poses a dual threat to adolescents’ physical and mental health development [21]. In addition to increasing the risk of multisystem disorders, including hypertension, hyperlipidemia, hyperuricemia, and respiratory diseases [22], overnutrition is also strongly associated with psychological problems [23]. Adolescents with obesity often experience reduced self-esteem, persistent inferiority feelings, and diminished well-being and quality of life [19, 21]. These adverse psychological outcomes are further exacerbated by weight-related stigma and body dissatisfaction issues [24]. While considerable research efforts have examined the relationship between overnutrition and adolescent mental health [2, 20], the potential impact of undernutrition should not be overlooked. Malnutrition prevalence among Chinese adolescents is estimated at approximately 5.9%, but the true burden may be underestimated [25]. Chronic undernutrition can impair brain development and normal physical growth, weaken immune function, increase infectious disease susceptibility, and negatively affect mental health [26]. Previous studies have reported a higher likelihood of PSH among malnourished adolescents [27, 28]. Considering the reality that undernutrition and overnutrition coexist in Chinese adolescents, and both overnutrition and undernutrition may be associated with adverse mental health outcomes, it is necessary to simultaneously investigate the relationship between nutritional status and PSH issues, as well as the potential pathways of their influence. Weight stigma often reflects social devaluation in individuals who are overweight or obese [29]. Within the context of prevailing aesthetic norms and social values in China, being thin is commonly equated with health, attractiveness, and social worth [30], whereas obese individuals are often stereotyped as lazy, unmotivated, or lacking self-discipline [29]. Such phenomena, termed weight stigma in sociological research, can influence adolescents’ social interactions and exert substantial negative effects on their mental health [31]. Although the evidence suggests that weight stigma is detrimental to the psychological well-being of young people, the underlying psychological mechanisms in adolescents remain poorly understood [2, 32]. Previous studies have predominantly focused on weight-related discrimination among overweight or obese adolescents [33, 34], with relatively little attention given to discrimination experienced by undernourished adolescents. Owing to their low body weight or thin physique, undernourished adolescents may be stigmatized as “sickly” or “unattractive,” which can lead to social exclusion, and in turn, impair self-esteem and mental health [35]. Social exclusion theory posits that devaluation or exclusion based on specific personal attributes, such as body weight, can diminish feelings of belonging and self-worth, ultimately contributing to adverse mental health outcomes [36]. Taken together, both overnutrition and undernutrition adolescents may have experienced high weight-related discrimination and social exclusion, potentially exhibiting a nonlinear dose–response relationship between nutritional status and psychological well-being. However, this relationship remains poorly understood in adolescents within the social and economic context in China. Importantly, the translation of such weight stigma and social exclusion into psychological harm, however, operates through how individuals perceive and interpret these experiences [37]. A longitudinal research demonstrates that perceived discrimination serves as a critical proximal mechanism linking social disadvantage to mental health outcomes, which is not merely a correlate of social position, but appears to be a crucial mechanism through which social adversity affects mental health [38]. For adolescents across different nutritional profiles, objective weight status (both undernutrition and overnutrition) exposes them to stigmatizing environments. This exposure fosters the internalization of negative self-perceptions (i.e., perceived discrimination) [34], which in turn directly undermines psychological well-being and may indirectly do so through its adverse effects on health behaviors [38]. From this theoretical and empirical perspective, perceived discrimination may represent an important risk factor and mediating pathway linking nutritional status to PSH among adolescents with different nutritional profiles, a hypothesis that requires further empirical validation in Chinese adolescents. Additionally, societal expectations regarding the ideal body shape differ by gender: girls are generally encouraged to be slender, whereas boys are more often expected to be physically robust [39]. Consequently, gender differences may underlie associations between body mass index (BMI) and mental health, and also have a mediating role in perceived discrimination. Clarifying these differences may help the development of gender-specific mental health promotion strategies among Chinese adolescents. Accordingly, grounded in social exclusion and weight stigma theory, we analyzed a large, multi-regional, Chinese adolescent sample and examined the association between nutritional status and PSH. We further evaluated the mediating role of perceived discrimination in the relationship between BMI and PSH across different nutritional status groups, as well as potential gender differences in these associations. Our hypotheses were as follows: first, a nonlinear relationship exists between nutritional status and mental health among Chinese adolescents; second, perceived discrimination mediates an association between BMI and PSH in different nutritional status subgroups; and third, gender differences occur in the relationship between BMI, perceived discrimination, and PSH. Our findings may clarify the pathways linking nutritional status to adolescent mental health and inform the development of targeted mental health promotion strategies for Chinese adolescents with different nutritional statuses, thereby contributing to “Healthy China” initiatives/goals. Methods Participants Adolescents (10–18 years old) enrolled in middle or high schools were recruited between April 2024 and January 2025. Inclusion criteria: aged between 10 and 18 years; of Chinese nationality; enrolled in middle or high schools; voluntary participation; and written informed consent from a parent/legal guardian. Exclusion criteria included the absence from school during the survey period; inability to participate due to serious disease or mental disorders; failure to provide written informed consent; or determination by school principals/teachers that participation was inappropriate for administrative or other reasons. A stratified cluster random sampling method was used to select 1–2 middle or high schools from each of three provinces (Jiangxi, Henan, and Liaoning). All eligible adolescents in selected schools were invited to participate in this questionnaire survey. In total, 2,881 adolescents were invited. After excluding 182 incomplete (> 20% missing data) or blank questionnaires, 2,699 valid questionnaires were included in analyses. The school level intraclass correlation coefficient (ICC) for the main variables in the present study was 0.003 < 0.05, indicating that the variance attributable to the school level was minimal [40]. Research instruments Sociodemographic variables and lifestyles Referring to professional knowledge and existing literature [41, 42], the following sociodemographic characteristics and lifestyles variables that may be related to PSH were collected: age (years); gender (male or female); place of residence (urban or rural); school level (middle or high school); primary guardian (parents, grandparents, or others); physical activity time (< 2 h/day or ≥ 2 h/day); screen time (< 2 h/day or ≥ 2 h/day); sleep duration (< 7 h/day, 7–8 h/day, or ≥ 8 h/day); alcohol consumption (yes/no); and smoking status (yes/no). Mental health Mental health status was assessed using the Brief Instrument on Psychological Health of Youths [43], which consists of 15 items across three dimensions: emotional problems, character problems, and social adaptation difficulties. Items indicating sub-health symptoms persisting for > 1 month (score > 3) were considered positive. Higher scores reflected a longer duration of sub-health symptoms. Participants with ≥ 7 positive items are classified with PSH [43]. The original scale has previously demonstrated excellent internal consistency (Cronbach’s α = 0.928) [43]. Perceived discrimination Perceived discrimination was assessed using the Perceived Discrimination Scale [44], which comprises six items covering two dimensions: individual discrimination and group discrimination. Each item is rated on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The mean score across the six items was used to represent the overall level of perceived discrimination, with higher scores indicating greater perceived discrimination. The original instrument has previously demonstrated good reliability, with Cronbach’s α coefficients = 0.810 and 0.820 for individual and group discrimination subscales, respectively [44]. Nutritional status Nutritional status was evaluated using BMI (kg/m²), calculated as weight (kg) divided by the height (m) squared. Height and weight data were obtained from the most recent physical examination records (local education bureaus) and supplied to the research team by the principals of participating schools after informed consent was obtained from adolescents. Nutritional status was categorized into three groups—undernutrition, normal nutrition, and overnutrition—according to Chinese health industry standards; Screening for Malnutrition in School-aged Children and Adolescents (WS/T 456–2014) [45] and Screening for Overweight and Obesity in School-aged Children and Adolescents (WS/T 586–2018) [46]. Specific BMI cutoff values are shown in Supplementary Table 1. Procedures Permission and cooperation were obtained from principals and educational administrations at participating schools. Prior to data collection, parents/legal guardians received detailed information on the study’s purpose, significance, and procedures. Oral informed consent was obtained from parents/legal guardians of all students, and written informed consent was collected from adolescents. Before questionnaire administration, trained researchers provided detailed instructions on completion requirements and precautions, and emphasized confidentiality, anonymity, and voluntary participation principles. The survey was conducted by class unit; students simultaneously completed self-administered questionnaires in their respective classrooms. Students were encouraged to seek clarification if questions/uncertainties arose during completion. Most students required approximately 10–15 min to complete questionnaires. Strict quality control measures were implemented to ensure data reliability. Logic-check items were embedded in questionnaires to minimize systematic and random errors. Prior to data collection, investigators received standardized training to ensure a thorough understanding of study objectives and adherence to standardized procedures. Additionally, questionnaires were completed voluntarily and anonymously to reduce potential information bias. To ensure consistency and standardization, data collection was monitored throughout the study in strictly accordance with predefined protocols. Ethics This study was conducted in accordance with Declaration of Helsinki principles. The study protocol was reviewed and approved by the Ethics Committee of Jinzhou Medical University (approval no. JZMULL2025235). Data analysis An electronic database was established using Epi Data (v.3.1), with double data entry and verification independently performed by two researchers. Missing data from eligible questionnaires were handled using the random forest imputation method, which can handle multivariate data consisting of continuous and categorical variables simultaneously [47]. This non-parametric approach does not require assumptions about distributional aspects of the data and provides excellent imputation results especially in data settings where complex interactions and non-linear relations are suspected [48]. Statistical analyses were conducted using SPSS (v.27.0), AMOS (v.24.0), and R (v.4.4.1). A two-sided significance level (α = 0.05) was applied. Continuous variables were presented as the mean ± standard deviations (mean ± SD), whereas categorical variables were expressed as frequencies and percentages. One-way analysis of variance (ANOVA) and multivariate analysis of variance (MANOVA) were used to examine PSH differences and perceived discrimination across nutritional status groups. Logistic regression analyses were used to identify factors associated with PSH, with effect sizes reported as odds ratios (ORs) and 95% confidence intervals (CIs). Interaction effects between sociodemographic characteristics and nutritional status or perceived discrimination toward PSH were also assessed in logistic regression models. After adjusted potential confounding factors that are statistically associated with psychological sub-health (according to the results of the logistic regression analysis), restricted cubic spline (RCS) analyses were conducted to evaluate dose–response relationships between BMI, perceived discrimination, and PSH risk. Structural equation modeling (SEM) was used to examine pathways linking BMI and PSH across different nutritional status groups, with perceived discrimination tested as a potential mediator. Potential confounding factors that are statistically associated with PSH were also controlled in SEM models. To assess gender differences in association between BMI and PSH, multigroup analyses stratified by gender were conducted in each nutritional status subgroup. Path coefficient differences were considered statistically significant when the critical ratio (Z) exceeded 1.96 (P < 0.05). Results Basic adolescent characteristics Among the 2,699 eligible adolescents, 1,444 (53.50%) were male and 1,255 (46.50%) were female, with a mean age of 14.55 ± 1.23 years. The majority were middle school students (74.06%, n = 1,999), while 25.94% (n = 700) were enrolled in high schools. In total, 1,630 adolescents (60.39%) resided in rural areas and 1,069 (39.61%) in urban areas. Most adolescents lived with their parents (79.88%, n = 2,156), whereas 484 (17.93%) were primarily cared for by grandparents and 59 (2.19%) by other guardians. Most adolescents spend less than 2 h per day on physical activities (85.88%, n = 2,318) and screen exposure (77.03%, n = 2,079). There were 1,298 adolescents (48.09%) slept 7–8 h a day, whereas 876 (32.45%) slept less than 7 h and 525 (19.45%) more than 8 h. A small proportion of adolescents smoked (1.82%, n = 49) or drunk alcohol (6.82%, n = 184). Detailed demographic characteristics by province are shown Table 1. Table 1Characteristics of participants in different provincesVariablesLiaoning Province(n = 1,270)Henan Province(n = 479)Jiangxi Province(n = 950)Total(n = 2,669)Age (years), mean ± SD14.07 ± 0.9114.46 ± 0.9515.24 ± 1.4114.55 ± 1.23BMI (kg/m2), mean ± SD21.44 ± 4.4819.83 ± 4.1620.63 ± 3.7320.87 ± 4.22Gender, n (%) Male657 (45.50)263 (18.21)524 (36.29)1,444 (53.50) Female613 (48.84)216 (17.21)426 (33.94)1,255 (46.50)Residence, n (%) Urban605 (56.59)7 (0.65)457 (42.75)1,069 (39.61) Rural665 (40.80)472 (28.96)493 (30.25)1,630 (60.39)School level, n (%) Middle school1,270 (63.53)471 (23.56)258 (12.91)1,999 (74.06)Primary guardian, n (%) Parents1,134 (52.60)347 (16.09)675 (31.31)2,156 (79.88) Grandparents122 (25.21)127 (26.24)235 (48.55)484 (17.93) Others14 (23.73)5 (8.47)40 (67.80)59 (2.19)Physical activity time, n (%) <2 h per day1,150 (49.61)323 (13.93)845 (36.45)2,318 (85.88) ≥ 2 h per day120 (31.50)156 (40.94)105 (27.56)381 (14.12)Screen time, n (%) <2 h per day1,122 (53.97)166 (7.98)791 (38.05)2,079 (77.03) ≥ 2 h per day148 (23.87)313 (50.48)159 (25.65)620 (22.97)Sleep duration, n (%) <7 h per day362 (41.32)78 (8.90)436 (49.77)876 (32.45) 7–8 h per day680 (52.39)192 (14.79)426 (32.82)1,298 (48.09) ≥ 8 h per day228 (43.43)209 (39.81)88 (16.76)525 (19.45)Drinking alcohol, n (%) No1,200 (47.71)437 (17.38)878(34.91)2,515 (93.18) Yes70 (38.04)42 (22.83)72(39.13)184 (6.82)Smoking, n (%) No1,257 (47.43)451 (17.02)942 (35.55)2,650 (98.18) Yes13 (26.53)28 (57.14)8 (16.33)49 (1.82) Characteristics of participants in different provinces Nutritional status, PSH, and perceived discrimination The mean BMI of the 2,699 adolescents was 20.87 ± 4.22 kg/m². Of these, 1,718 (63.65%) were classified as having normal nutrition, 223 (8.26%) as undernourished, and 758 (28.08%) as overnourished. The proportion of normal nutrition in girls (75.14%) is higher than that of boys (53.67%), whereas both the proportion of undernourished and overnourished in boys (9.49%; 36.84%) were higher than that of girls (6.85%; 18.01%), there is a statistically difference (χ² = 139.07, P < 0.001). There was no significant difference in nutritional status across other sociodemographic characteristics (P > 0.05). Student sociodemographic characteristics across nutritional status groups are shown in Supplementary Table 2. Cronbach’s α coefficients of both the Brief Instrument on Psychological Health of Youths and the Perceived Discrimination Scale were greater than the recommended threshold of 0.70. Besides, confirmatory factor analysis was applied to evaluate the validity of the scales. The results showed that both composite reliabilities were greater than 0.70 and average variance extracted greater than 0.50, indicating the good reliability and validity of the scales in the current sample (Supplementary Table 3). The mean PSH score was 2.00 ± 1.19. Both overnourished and undernourished adolescents exhibited significantly higher PSH levels than those with normal nutrition (MANOVA (F = 4.15, P < 0.001) and ANOVA (F = 5.06, P = 0.007)) (Table 2). Overall, 461 students (17.08%) met PSH criteria. Chi-square analysis demonstrated significant differences in PSH prevalence across nutritional status groups (χ² = 17.63, P < 0.001). The mean perceived discrimination score among all adolescents was 1.80 ± 0.90. Perceived discrimination levels were also significantly higher among overnourished and undernourished adolescents when compared to those with normal nutrition (MANOVA (F = 16.67, P < 0.001) and ANOVA (F = 4.73, P = 0.009)) (Table 2). Table 2The levels of psychological sub-health and perceived discrimination in different nutritional subgroupsVariablesNutritional statusANOVAMANOVAUndernutritionNormal nutritionOvernutritionF P F P Psychological sub-health problems2.17 ± 1.301.95 ± 1.132.07 ± 1.285.060.0074.15<0.001Emotional problems2.27 ± 1.502.00 ± 1.262.10 ± 1.394.430.012Character problem1.87 ± 1.311.71 ± 1.151.91 ± 1.376.960.001Social adaptation difficulties2.28 ± 1.412.09 ± 1.362.19 ± 1.432.720.066Perceived discrimination1.92 ± 1.021.76 ± 0.861.86 ± 0.954.730.00916.67<0.001Individual discrimination2.12 ± 1.111.95 ± 0.991.92 ± 1.023.350.035Group-based discrimination1.72 ± 1.111.57 ± 0.901.80 ± 1.0513.90<0.001Note. Psychological sub-health: undernutrition vs. normal nutrition (F = 3.07, P = 0.027), normal nutrition vs. overnutrition (F = 5.53, P<0.001), undernutrition vs. overnutrition (F = 2.98, P = 0.031); Perceived discrimination: undernutrition vs. normal nutrition (F = 3.19, P = 0.041), normal nutrition vs. overnutrition (F = 30.88, P<0.001), undernutrition vs. overnutrition (F = 9.64, P<0.001) The levels of psychological sub-health and perceived discrimination in different nutritional subgroups Note. Psychological sub-health: undernutrition vs. normal nutrition (F = 3.07, P = 0.027), normal nutrition vs. overnutrition (F = 5.53, P<0.001), undernutrition vs. overnutrition (F = 2.98, P = 0.031); Perceived discrimination: undernutrition vs. normal nutrition (F = 3.19, P = 0.041), normal nutrition vs. overnutrition (F = 30.88, P<0.001), undernutrition vs. overnutrition (F = 9.64, P<0.001) Factors associated with PSH among adolescents Logistic regression analyses were performed to identify factors associated with PSH (yes = 1, no = 0) among adolescents. Both undernutrition (OR = 1.57, 95% CI = 1.08–2.27, P = 0.017) and overnutrition (OR = 1.67, 95% CI = 1.32–2.12, P < 0.001) were significantly associated with an increased PSH risk. Higher perceived discrimination levels were also strongly associated with PSH (OR = 1.79, 95% CI = 1.61–1.98, P < 0.001). In addition, female adolescents had a higher PSH risk than males (OR = 1.62, 95% CI = 1.30–2.02, P < 0.001). Short sleep duration (< 7 h/day) and alcohol consumption were further associated with an elevated PSH risk (Fig. 1). Analysis of influencing factors of psychological sub-health among adolescents. (Psychological sub-health: yes = 1, no = 0) Sex-stratified analyses revealed that undernourished male adolescents had a significantly higher PSH risk when compared with those with normal nutrition (OR = 1.91, 95% CI = 1.21–3.04, P = 0.006), whereas this association was not observed among undernourished girls (OR = 1.45, 95% CI = 0.84–2.52, P = 0.187). Conversely, overnourished girls exhibited a markedly higher PSH risk (OR = 2.46, 95% CI = 1.76–3.42, P < 0.001), while the association among overnourished boys did not reach statistical significance (OR = 1.34, 95% CI = 0.98–1.83, P = 0.066). A significant interaction between gender and nutritional status in relation to PSH risk was also observed (P for interaction = 0.011). No significant interactions were identified between nutritional status and sleep duration or alcohol consumption in relation to PSH. Similarly, no significant interactions were identified between perceived discrimination and gender, sleep duration or drinking alcohol (P > 0.05) (Supplementary Table 4). Dose–response relationships between BMI, PSH, and perceived discrimination After adjusting for gender, sleep duration, and alcohol consumption, which statistically associated with PSH, RCS analyses were performed to examine dose–response relationships between BMI, perceived discrimination, and PSH risk (Fig. 2). A significant nonlinear association between BMI and PSH risk was observed, characterized by a U-shaped curve (χ² = 15.37, P for nonlinearity < 0.001). The BMI associated with the lowest PSH risk was approximately 19.00 kg/m². When BMI values were < 19.00 kg/m², the PSH risk increased progressively as BMI decreased. Conversely, when the BMI exceeded 19.00 kg/m², the PSH risk increased with increasing BMI (Fig. 2A). A similar nonlinear dose–response relationship was identified between BMI and PSH levels (χ² = 12.67, P for nonlinearity < 0.001; Fig. 2B). In addition, significant nonlinear associations were observed between BMI and perceived discrimination (χ² = 12.70, P for nonlinearity < 0.001; Fig. 2C) and between perceived discrimination and PSH levels (χ² = 85.84, P for nonlinearity < 0.001; Fig. 2D). The dose-response relationship between BMI, perceived discrimination and psychological sub-health (A: The dose-response relationship between BMI and the risk of psychological sub-health; B: The dose-response relationship between BMI and the level of psychological sub-health; C: The dose-response relationship between BMI and perceived discrimination; D: The dose-response relationship between perceived discrimination and the level of psychological sub-health) Pathway analysis between BMI, perceived discrimination, and PSH Given the observed nonlinear association between BMI and PSH, SEM was separately conducted in different nutritional status subgroups to examine pathways linking BMI and perceived discrimination to PSH. In each model, BMI was specified as the independent variable, PSH as the dependent latent variable, and perceived discrimination as the mediating variable. Gender, sleep duration, and alcohol consumption were included as covariates in all SEM models. All models demonstrated acceptable fit indices according to recommended criteria (Supplementary Table 5). Among undernourished adolescents, BMI was negatively associated with perceived discrimination (β = −0.174, P = 0.012), whereas perceived discrimination was positively associated with PSH (β = 0.400, P < 0.001). BMI showed a significant negative indirect association with PSH through perceived discrimination (β = −0.069, P = 0.012), while a direct association between BMI and PSH was not statistically significant (β = −0.086, P = 0.212) (Fig. 3A). Thus, in undernourished adolescents, lower BMI was associated with higher perceived discrimination levels, which in turn were associated with greater PSH, suggesting full mediation by perceived discrimination. In adolescents with normal nutritional status, perceived discrimination was also positively associated with PSH (β = 0.364, P < 0.001). However, neither a direct association between BMI and PSH (β = 0.013, P = 0.616) nor an indirect association via perceived discrimination (β = 0.009, P = 0.343) reached statistical significance (Fig. 3B), indicating no mediating effect in this group. Among overnourished adolescents, BMI was positively associated with perceived discrimination (β = 0.114, P = 0.004), and perceived discrimination was positively associated with PSH (β = 0.324, P < 0.001). A significant positive indirect association between BMI and PSH through perceived discrimination was observed (β = 0.037, P = 0.010), whereas a direct association was not significant (β = 0.063, P = 0.105) (Fig. 3C). Thus, among overnourished adolescents, a higher BMI was associated with greater perceived discrimination, which subsequently contributed to higher PSH levels, with perceived discrimination acting as a full mediator. The effect paths of BMI and perceived discrimination on psychological sub-health of adolescents in different nutrition groups. A Undernutrition group; B Normal nutrition group; C Overnutrition group. Covariates: Gender, Sleep duration and Drinking. *P < 0.05, **P < 0.01, ***P < 0.001 Comparison test results for path coefficients between BMI and PSH across nutritional status subgroups are shown in Table 3. Significant differences were observed in path coefficients linking BMI to perceived discrimination across nutritional groups, whereas no statistically significant differences were identified for the other paths. Specifically, the path coefficient from BMI to perceived discrimination among undernourished adolescents differed significantly from those observed among overnourished (Z = 2.919, P < 0.05) and adolescents with normal nutritional status (Z = 2.659, P < 0.05). Table 3The difference test of the path coefficients between BMI and psychological sub-health of adolescents across different nutritional subgroupsPathsCritical ratio of difference (Z)Undernutrition vs. Normal nutritionUndernutrition vs. OvernutritionOvernutrition vs. Normal nutritionBody mass index →Perceived discrimination2.659*2.919*1.016Body mass index →Psychological sub-health1.3081.5000.809Perceived discrimination →Psychological sub-health-0.591-0.1960.595Note. *P < 0.05 The difference test of the path coefficients between BMI and psychological sub-health of adolescents across different nutritional subgroups Note. *P < 0.05 Gender differences in associations between BMI and PSH To examine gender differences in associations between BMI and PSH, multigroup analyses stratified by gender were conducted in each nutritional status subgroup (Table 4). Among boys with undernutrition and normal nutrition, BMI was indirectly associated with PSH through perceived discrimination. Among overnourished boys, BMI was not only indirectly associated with PSH via perceived discrimination but also directly associated with PSH (β = 0.118, P = 0.007), with the direct pathway accounting for 71.41% of the total effect. The standardized path coefficient linking BMI to perceived discrimination among undernourished boys differed significantly from those observed among boys with normal nutrition (Z = 2.932, P < 0.05) and overnourished boys (Z = − 2.878, P < 0.05). In contrast, among girls, BMI showed neither direct nor indirect associations with PSH across all nutritional status groups. However, perceived discrimination was positively associated with PSH in girls with normal nutrition and overnutrition statuses, with no significant differences in the strength of these associations between groups (Z = 0.842, P > 0.05). Thus, in boys, BMI influenced PSH primarily through indirect pathways, and additionally through a direct pathway in those with overnutrition, whereas among girls, PSH was unrelated to BMI perse, and was instead, associated with perceived discrimination when nutritional status was normal or excessive. Table 4Subgroup analysis of gender in different nutritional adolescentsPathsGender (Standard path coefficient)MaleFemaleUndernutritionNormal nutritionOvernutritionUndernutritionNormal nutritionOvernutritionBody mass index →Perceived discrimination-0.219*0.084*0.144**-0.156-0.0240.039Body mass index →Psychological sub-health-0.0990.0370.118**-0.0500.016-0.058Perceived discrimination →Psychological sub-health0.480***0.395***0.328***0.2400.350***0.367**Critical ratio of difference (Z)Undernutrition vs Normal nutritionUndernutrition vs OvernutritionNormal nutrition vs OvernutritionUndernutrition vs Normal nutritionUndernutrition vs OvernutritionNormal nutrition vs OvernutritionBody mass index →Perceived discrimination2.932*2.878*-0.4031.2211.4530.872Body mass index →Psychological sub-health1.3741.5510.6700.5030.246-0.971Perceived discrimination →Psychological sub-health-0.789-0.6960.1000.2200.7290.842Critical ratio of difference (Z)Undernutrition (Male) vs Undernutrition (Female)Normal nutrition (Male) vs Normal nutrition (Female)Overnutrition (Male) vs Overnutrition (Female)Body mass index →Perceived discrimination0.987-2.237*-1.180Body mass index →Psychological sub-health0.418-0.453-1.982*Perceived discrimination →Psychological sub-health-0.611-0.1820.733Note. *P < 0.05, **P < 0.01, ***P < 0.001 Subgroup analysis of gender in different nutritional adolescents Note. *P < 0.05, **P < 0.01, ***P < 0.001 Further analyses revealed no significant gender differences in any of the BMI–PSH pathways among undernourished adolescents (P > 0.05). However, among adolescents with normal nutrition, BMI was positively associated with perceived discrimination in boys but not in girls, resulting in a significant gender difference in this pathway (Z = − 2.237, P < 0.05). Moreover, among overnourished adolescents, a higher BMI was associated with increased PSH both directly and indirectly in boys, whereas no such associations were observed in girls. A significant gender difference was identified in the direct effect of BMI on PSH between overnourished boys and girls (Z = − 1.982, P < 0.05). Overall, these findings indicated that the pathways linking BMI to PSH varied by gender, with gender moderating the relationship between BMI and PSH across different nutritional status groups. Discussion We analyzed data from 2,699 adolescents across three provinces in China to examine associations between BMI, perceived discrimination, and PSH. Adolescents with undernutrition and overnutrition had 1.57- and 1.67-fold higher odds of PSH, respectively, when compared with those with normal nutritional status. A clear U-shaped nonlinear dose–response relationship was identified between BMI and PSH risk. Perceived discrimination played a significant mediating role in the association between BMI and PSH, and notable gender differences were observed in pathways linking BMI to PSH across different nutritional status groups. Consistent with previous studies [49, 50], both undernourished and overnourished adolescents were at higher PSH risk than their normally nourished peers. Adolescents with overnutrition, whose body size deviated from prevailing social and aesthetic norms, are more likely to experience weight stigma, bullying, or social marginalization in school settings [33]. During childhood and adolescence, weight-related stigmatization may manifest as direct behaviors, such as verbal abuse or physical aggression, or as indirect behaviors, such as social exclusion, all of which are associated with adverse mental health outcomes [24]. In contrast, undernutrition is associated with impaired brain development, reduced cognitive function, poorer academic performance, and behavioral problems [51]. Undernourished adolescents may also experience body image dissatisfaction and heightened sensitivity to pressures from family, peers, and society, further increasing vulnerability to psychological problems [52]. The observed threshold effect supports the hypothesis that deviations from an optimal BMI range—either above or below—are associated with a significantly increased PSH risk, underscoring the psychosocial consequences of both undernutrition and overnutrition. The current study, which integrates adolescents with overnutrition, undernutrition, and normal nutrition, not only depicts the full picture of how the risk of PSH changes with BMI among adolescents, but also can help to identify vulnerable populations for adolescent mental health education and specify the target range for weight management in adolescents. Moreover, it can also provide empirical evidence to establish a novel framework of “a healthy weight range rather than simply weight loss” in the health promotion of adolescents. However, given the cross-sectional nature of our study, the causal relationship between BMI and PSH still cannot be determined. Meanwhile, previous studies have suggested that some adolescents with psychological problems are prone to obesity, due to emotional eating and their brains instinctively stimulate the secretion of serotonin through the intake of high-sugar and high-fat foods to copy with psychological stress [53, 54]. It is possible that weight-related problems may lead to high risk of mental health problems, poor psychological status in turn promote the further development of nutritional status. The two may be mutually causal, mutually reinforcing, and form a vicious cycle [55], which still need to be verified through further longitudinal research that involves knowledge from multiple disciplines. We observed that perceived discrimination significantly mediated the association between BMI and PSH among overnourished and undernourished adolescents. Adolescents in these groups are particularly vulnerable to social stigma and discrimination related to body weight [56]. Such discrimination often originates from evaluations by peers and the surrounding social environment [57]. Prolonged exposure to these negative evaluations may lead to their internalization, resulting in reduced self-esteem, increased social anxiety, and a higher risk of mental health problems [58]. Additionally, our path analysis indicated that BMI was significantly associated with perceived discrimination among adolescents with undernutrition and overnutrition, but not among those with normal nutritional status. This finding may be explained by the fact that normally nourished adolescents experience different forms of discrimination-related psychological stress when compared with undernourished and overnourished peers [59]. Adolescents with normal nutrition might be exposed to other prominent stressors that is not related to weight, including academic [8], family [60], and broader social pressures [61], which may attenuate or overshadow the specific influence of BMI on mental health outcomes. Notably, the association between perceived discrimination and PSH was significant across all three nutritional status groups, indicating that discrimination is a common experience in adolescents’ daily lives and exerts a broad and substantial association with adverse mental health outcomes [62, 63]. This finding underscores the importance of further research examining how discrimination interacts with adolescent mental health, and highlights potential directions for future research and interventions. Although adolescents with different nutritional statuses may encounter different discrimination sources, both undernourished and overnourished adolescents exhibited elevated PSH risk through perceived discrimination. This observation further supports a U-shaped nonlinear relationship between BMI and PSH, suggesting that deviation from socially accepted body norms—whether toward underweight or overweight—can increase exposure to discrimination, and consequently, PSH risk. Unlike previous studies, which have primarily examined weight-related discrimination among individuals with overnutrition [29, 64], we have extended the theoretical weight stigma framework by demonstrating similar discriminatory psychological mechanisms among undernourished adolescents. This finding highlights the limitations of traditional linear models in evaluating the mental health associations with extreme BMI values, and suggests that a U-shaped model more accurately reflects the complex interplay between physical status and psychological well-being. Weight-related stigma operates not solely as a phenomenon tied to overweight, but a bidirectional process. Deviation from the normal range, no matter undernutrition or overnutrition, have similar psychosocial vulnerabilities. Our study broadens the theoretical scope of the weight stigma framework, by uncovering this shared mechanism across both ends of the nutritional spectrum. Addressing weight-related stigmatization is of particular importance for undernourished and overnourished adolescents. Beyond the traditional weight stigma framework, the current study also found that the perception of discrimination is generally harmful to the mental health of adolescents, regardless of the types and reasons for discrimination (such as body shape, academic performance, family background, etc.). Once an individual perceives that they are not respected or treated maliciously, it will activate the brain’s stress response. Long-term psychological stress will further lead to the imbalance of physiological and psychological systems, resulting in mental health problems [65, 66]. To some extent, the findings of this study support the biopsychosocial mechanism of mental health problems in adolescents, and suggest that anti-discrimination intervention might be a generalizable strategy to promote the mental health of adolescents. Meanwhile, it should not be overlooked that people with certain psychological issues tend to be more sensitive and pay closer attention to or care more about others’ evaluations or attitudes [67], thereby heightening the level of perceived discrimination. This often results in a situation where mental health problems and perceived discrimination become mutually reinforcing and exacerbate each other in a vicious cycle. Therefore, efforts reducing discrimination, strengthening psychological resilience, and enhancing social support are required to improve mental health outcomes among adolescents, particularly for those who are nutritionally vulnerable and to foster a more inclusive social environment. Consistent with previous studies [68, 69], a significant interaction association between nutritional status and gender was observed, with undernourished males and overnourished females showing a higher likelihood of developing PSH. Our gender-stratified mediation analyses further revealed distinct association patterns between BMI and PSH. Among males with undernutrition and normal nutrition, BMI was indirectly associated with PSH through perceived discrimination, whereas in overnourished males, BMI exerted a significant direct correlation with PSH. In contrast to males, perceived discrimination played a significant role in PSH in females with normal nutrition and overnutrition. These gender differences may be attributable to sex-specific body composition patterns and socially constructed body ideals [70]. In many societies, including China, female appearance is more strongly associated with slimness, being overweight often leads to body image anxiety and self-reproach [30]. On the contrary, male body shape is more closely linked to strength and robustness, social aesthetics for males are more inclusive and diverse [71]. Besides, the differences in physical and mental development during puberty may also account for these gender differences. Girls are more sensitive to bodily changes in this period and tend to internalize negative remarks about weight, which may increase the risk of psychological problems [72]. For boys, slight weight gain is generally taken as physical growth, whereas underweight males are more prone to inferiority. Boys tend to release their resentment over unfair treatment through externalizing behaviors such as confrontation and aggression, which cause relatively less psychological trauma compared with internalizing behaviors [73]. In addition, gender differences in family parenting concepts may also partly explain the gender disparities in the association between BMI and PSH. Families generally pay more attention to girls’ body management and appearance control. Frequent comments on weight can plant negative psychological suggestions [74]. When it comes to boys, families focus more on physical fitness and ability development, and rarely make negative judgements about their weight [75]. In conclusion, the gender differences in the association pattern between BMI and PSH may be the result of multiple factors, including individual physiological differences, family education guidance, social and cultural atmosphere, et al. Unlike previous studies that primarily attributed psychological problems associated with overnutrition to weight-related stigma [76], our findings indicate a direct association between BMI and PSH in overnourished males. This suggests that, in addition to social evaluation, biological mechanisms may contribute to psychological vulnerability in this group. One potential mechanism involves a dysregulated hypothalamic–pituitary–adrenal (HPA) axis, a central neuroendocrine system regulating stress responses and emotional regulation [77]. Previous study had indicated that excessive fat accumulation can cause the secretion of proinflammatory factors such as tumor necrosis factor-α (TNF-α) and interleukin-6 (IL-6) [78], which can break through the blood-brain barrier and directly enter the key regions of the brain (such as hypothalamus) [79]. Such hypothalamic inflammation can destroy the negative feedback regulation of the HPA axis by the central nervous system, resulting in HPA axis dysfunction [80]. Animal study also showed that HPA axis dysregulation can directly induce anxiety-like behaviors [81]. Such evidence not only support the rationality of the hypothesis of the current study that obesity related with high risk of PSH, but also indicate that the psychological harm of obesity is the result of the combined effects of social pressure and physiological disorders. However, due to the lack of biological data in this study, its physiological mechanism still requires future research to be verified using multi-omics evidence such as proteomics and metabolomics. Previous research has reported that female body fat percentage, perceived as most attractive, was often slightly below the medically defined healthy range [82]. Meanwhile, dissatisfaction with body shape and concerns of attractiveness to the opposite sex are more prevalent among females than males [83]. These findings suggest that females may internalize unrealistic thinness ideals. Consequently, even those with a normal BMI may experience heightened body dissatisfaction, often driven by distorted body image perceptions [84], which may contribute to PSH. These findings have important implications for targeted intervention strategies. For undernourished and normally nourished males, interventions should prioritize reducing weight-related stigma in school environments and strengthening peer support systems. For overnourished males, integrated approaches combining physiological monitoring with psychological support may address both biological and mental health vulnerabilities. For females, mental health interventions should extend beyond traditional nutritional classification frameworks and emphasize body image education, psychological resilience, and reduced sensitivity to external appearance-based evaluations. Grounded in social exclusion and weight stigma theory, this study takes perceived discrimination as a mediator between BMI and PSH in adolescents. However, mental health problems in adolescents are the result of the combined effects of multiple dimensions such as physiology, psychology, society and environment. Other alternative explanatory pathways between BMI and PSH in adolescents also deserve great attention in the future studies. For instance, chronic inflammation [78], accelerated cellular aging [85] and sleep disorders [6] resulting from extreme BMI values may be all associated with an increased risk of PSH. In addition, this study has several limitations. First, the cross-sectional design precludes causal inference between BMI, perceived discrimination, and PSH. Nonetheless, our findings—specifically the U-shaped relationship between BMI and PSH risk, and the mediating role of perceived discrimination—provide valuable insights into current associations, contribute to hypothesis generation, and inform public health decision-making. Future longitudinal studies are required to clarify temporal and causal relationships. Second, PSH and perceived discrimination were assessed using self-reported measures, which may be subject to reporting and recall biases. Adolescents may have overestimated or underestimated their experiences, especially those with psychological distress, which might have affected the validity of the measures in the current sample. However, self-report instruments are common in large-scale population studies due to their feasibility and efficiency [86], and the measures used in this study demonstrated good reliability and validity under strict quality control procedures. Third, although several covariates were controlled in the current study. Both individual-level factors (such as family economic condition, parental educational level and social status, and early-life origin factors), and group-level factors (such as urban planning and architectural environment, social cultural norms, and political governance and regulations) may all be related to the nutritional status and psychological development of adolescents, thereby interfering with the determination of the relationship between BMI and PSH. Future research incorporating multidimensional data would allow for a more comprehensive understanding of these issues. Moreover, the participants of present study were recruited using cluster sampling at the school level. Although the intraclass correlation coefficient of main variables in the current study indicates that the school-level clustering effect can be ignored, applying multilevel models may be a better way to deal with such nested data in the future, so as to fully capture school-level heterogeneity and demonstrate the influence of school characteristics on the results. Furthermore, missing data from eligible questionnaires were handled using the random forest imputation method only. Taking into account the wide applicability and excellent interpolation performance of random forest interpolation [48], extensive sensitivity analyses were not conducted to assess the robustness of findings. Finally, the sample was drawn from only three provinces in China, which may limit the generalizability of our findings to regions with different cultural or socioeconomic contexts. Studies with more geographically diverse samples may enhance this external validity. Conclusions Grounded in social exclusion and weight stigma theory, we provide a comprehensive examination of the association between nutritional status and mental health in Chinese adolescents. We demonstrate that both undernutrition and overnutrition are associated with adverse mental health outcomes. Perceived discrimination was identified as a key mediating factor linking BMI and PSH in both undernourished and overnourished adolescents. Additionally, BMI in overnourished boys was associated with PSH through both indirect and direct pathways, indicating gender differences in the association pathways between BMI and PSH across different nutritional statuses. Although the cross-sectional design limited causal inference, this large, multi-regional sample underscores the importance of maintaining a healthy body weight for mental health development during adolescence. These findings provide clues to the causes of psychological sub-health problems among adolescents with different nutritional statuses and also point out the direction for future research on developing targeted mental health promotion strategies tailored to adolescents with different nutritional statuses. In the future, the development of a comprehensive intervention program, which not only help adolescents maintaining a healthy weight, but also help to foster a balanced and accurate understanding of body weight, reduce weight-related stigma, and prevent both external discrimination and internalized self-stigmatization, may be of great benefit to the promotion of mental health among Chinese adolescents. Clinically, this study supports the development of a combined nutrition and mental health assessment model for adolescents, enabling targeted interventions for comorbidities and breaking the vicious cycle of physical and mental health issues. In public health, this study can help to identify vulnerable populations for psychological interventions, inform the formulation of health policies and health education programs, and facilitate the targeted prevention of physical and mental health among adolescents. Supplementary Information Supplementary Material 1. Acknowledgements The authors sincerely acknowledge the partial financial support provided by the High-Level Talent Research Funding Program of Jinzhou Medical University. Appreciation is extended to all adolescents, and also staff members involved in data collection and for their valuable cooperation and contributions. Abbreviations psychological sub-health body mass index analysis of variance multivariate analysis of variance odds ratios confidence interval restricted cubic spline structural equation model comparative fit index goodness of fit index incremental fit index normed fit index root mean square error of approximation hypothalamic–pituitary–adrenal intraclass correlation coefficient Authors’ contributions F.G. contributed to conceptualization, supervision, and funding acquisition. H.L. contributed to study design. M.L. contributed to conceptualization, data curation, and drafting of the original manuscript. Y.X. and Y.Y. contributed to manuscript review and editing. X.D., H.Y., X.L., Q.Y., and Y.L. contributed to data collection. All authors read and approved the final manuscript. Funding This work was supported by the National Natural Science Foundation of China [72104094], the Science and Technology Plan Project of Liaoning Province, China [2023-MSLH-041], and the Scientific Research Project from Education Department of Liaoning Province, China [LJ212510160010]. Data availability The datasets generated and/or analyzed during the current study are not publicly available due to the confidentiality of student mental health information but are available from the corresponding author upon reasonable request. Declarations Ethics approval and consent to participate This study was conducted in accordance with the Helsinki Declaration. The study protocol was reviewed and approved by the ethics committee of Jinzhou Medical University [JZMULL2025235]. We obtained the oral informed consent from parents or primary guardian of every participant. Written informed consent to participate in this study was provided by all the participants. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The datasets generated and/or analyzed during the current study are not publicly available due to the confidentiality of student mental health information but are available from the corresponding author upon reasonable request.