Dietary Patterns and an Adverse Mental Health Indicator Among School-Attending Brazilian Adolescents: Analyses of the 2015 and 2019 PeNSE Surveys

Background/Objectives: Dietary patterns may be associated with adolescent mental health, but nationwide Brazilian evidence based on empirically derived patterns across survey editions remains limited. This study identified dietary patterns and examined their associations with a survey-specific adverse mental health indicator among school-attending adolescents aged 13–17 years in the 2015 and 2019 National Survey of School Health (PeNSE). Methods: The two independent cross-sectional surveys were analyzed separately, with the 2015 analysis restricted to Sample 2. Two dietary patterns were derived in each edition using principal component analysis. The primary outcome was membership in the highest category defined from the weighted distribution of a non-diagnostic, survey-specific composite score of adverse mental health-related indicators. Survey-weighted modified Poisson regression estimated adjusted prevalence ratios (PRs); continuous, ordinal, between-survey heterogeneity, alternative-outcome, weighted-PCA, and E-value analyses assessed robustness. Results: Both surveys yielded ultra-processed/unhealthy and healthy/traditional patterns. Compared with the highest adherence quartile (Q4), the lowest quartile (Q1) of the ultra-processed/unhealthy pattern was associated with a lower prevalence of the adverse indicator in 2015 (adjusted PR = 0.85, 95% CI: 0.73–0.98) and 2019 (adjusted PR = 0.81, 95% CI: 0.76–0.86). Conversely, Q1 of the healthy/traditional pattern was associated with a higher prevalence in 2015 (adjusted PR = 1.75, 95% CI: 1.45–2.10) and 2019 (adjusted PR = 1.34, 95% CI: 1.26–1.42). No between-survey heterogeneity was detected for the ultra-processed/unhealthy pattern, whereas the healthy/traditional estimates differed in magnitude between surveys; sensitivity analyses were directionally consistent. Conclusions: Greater ultra-processed/unhealthy adherence and lower healthy/traditional adherence were associated with a higher prevalence of the survey-specific adverse mental health indicator. Because the surveys were cross-sectional and used edition-specific measures, these findings do not establish temporality, causality, or temporal change in invariant constructs.