Perceived stress, not body image amplification, tracks teen depression risk

A large analysis of more than 200,000 South Korean adolescents has found that perceived stress is the strongest and most consistent correlate of self-reported depressive symptoms, while a widely hypothesized amplification effect—in which body-image concerns would make stress more damaging—did not hold up under prespecified testing. The study, published in BMC Public Health, analyzed data from 201,895 adolescents who took part in the Korea Youth Risk Behavior Survey between 2022 and 2025, splitting the sample so that models were developed on the 2022–2023 responses (99,235 adolescents) and then validated unchanged on the 2024–2025 responses (102,660 adolescents).

The research team set out to test a diathesis–stress hypothesis: that body-image-related vulnerability would be amplified when adolescents experienced greater perceived stress. Both perceived stress and body image are individually associated with adolescent depressive symptoms, but whether the combination produces more-than-additive risk remained unclear. To examine this, the researchers used multivariable logistic regression with an ordinal-linear stress-by-body-image interaction term as their prespecified analysis, and they applied survey-weighted models as sensitivity checks.

The results were clear on the main question. Perceived stress showed a strong association with self-reported depressive symptoms, with an odds ratio of 2.715 (95% confidence interval 2.555–2.885). By contrast, the prespecified stress-by-body-image interaction was not statistically significant, with an odds ratio of 0.996 (95% CI 0.977–1.014, p = .640). When the analysis was repeated with survey weights, the interaction estimate was nearly identical (OR = 0.994, p = .611), indicating that the null finding was not an artifact of the sampling design.

Because any interaction test depends on how a continuous variable is modeled, the researchers also ran post hoc checks on the functional form. A categorical specification of body image improved model fit and produced a significant overall interaction, but the category-specific effects were not uniform, and no individual interaction term remained significant after correction for multiple comparisons in the survey-weighted analysis. A quadratic specification of body image was not supported (p = .168). Taken together, these checks suggest possible category-specific heterogeneity in how body image relates to depressive symptoms, but the authors emphasize that this pattern requires independent confirmation before any substantive conclusions can be drawn.

The study also tested whether a more flexible machine-learning approach could extract signal that the regression models missed. A balanced random forest, used as an exploratory nonlinear complement, was trained on the development data and evaluated on the temporal validation set. It did not outperform the simpler model: logistic regression achieved an area under the receiver operating characteristic curve of 0.736 (95% CI 0.732–0.739), compared with 0.712 (95% CI 0.709–0.716) for the random forest. The authors also caution that the predicted probabilities from the balanced random-forest model overestimated absolute risk and should not be treated as calibrated clinical risk estimates.

The methodological design deserves attention. Classification thresholds were selected in the development data and applied unchanged to the 2024–2025 validation sample, providing a temporal check that the reported associations were not simply products of overfitting to a single survey wave. The scale of the sample—more than 200,000 adolescents across four survey years—lends weight to the null interaction result, although the authors frame their conclusion in terms of what the prespecified test showed rather than what it ruled out.

The authors’ conclusion is measured. The prespecified diathesis–stress amplification hypothesis was not supported in this nationally representative sample. Perceived stress emerged as the most consistent correlate of self-reported depressive symptoms, while the post hoc analyses pointing to possible category-specific heterogeneity in body-image associations remain exploratory. The study is cross-sectional and based on self-reported measures, so it establishes associations rather than causal effects, and the depressive-symptom outcome reflects adolescents’ own reports rather than clinical diagnoses.

For researchers and practitioners, the findings suggest that in this population, perceived stress deserves primary attention as a correlate of adolescent depressive symptoms, and that assumptions about body image acting as a stress amplifier should not be taken for granted without direct evidence. The authors note that the exploratory hints of heterogeneity across body-image categories warrant follow-up in independent datasets before they can inform theory or practice.

Subject of Research: Joint association of perceived stress and body image with depressive symptoms in South Korean adolescents

Article Title: Beyond independent effects: stress and body image in adolescent depressive risk

Article References: Jo, H., Park, S., & Kim, B. (2026). Beyond independent effects: stress and body image in adolescent depressive risk. BMC Public Health. https://doi.org/10.1186/s12889-026-29997-0

Image Credits: AI Generated

DOI: 10.1186/s12889-026-29997-0

Keywords: adolescent mental health, depressive symptoms, perceived stress, body image, diathesis-stress model, Korea Youth Risk Behavior Survey, logistic regression, interaction analysis, temporal validation, random forest, BMC Public Health, South Korea