Real-World Fermented Food Consumption and Depression, Anxiety, and Stress Symptoms in a Southeast Asian University Population: A Cross-Sectional Study

2.1. Study Design and Participants

This cross-sectional study was conducted at Universiti Malaysia Sabah (UMS), Sabah, Malaysia. Undergraduate students were recruited between 1 June 2025 and 31 December 2025 through institutional communication channels and social media platforms.

The study was conducted using an anonymous online survey platform. Participants were required to review the participant information sheet, declare any possible exclusion criteria, and provide electronic informed consent before accessing the study questionnaire. The survey platform was configured to retain data only from respondents who consented to participate; therefore, information regarding the number of individuals who received the survey invitation, accessed the survey without consenting, or declined participation was not available. Only responses from consenting participants were recorded and screened for eligibility and completeness prior to analysis.

2.2. Inclusion and Exclusion Criteria

Participants were eligible for inclusion if they were aged between 18 and 30 years and were currently enrolled as undergraduate students at Universiti Malaysia Sabah (UMS) during the study period. Participants were required to be able to understand either English or Malay and to provide electronic informed consent prior to completing the online questionnaire. Only respondents who completed all key components of the survey, including the fermented food frequency questionnaire and the DASS-21, were included in the final analysis.

Students were excluded if they self-reported currently undergoing treatment for a diagnosed psychiatric disorder, such as major depressive disorder or generalized anxiety disorder, as ongoing treatment could influence symptom reporting. Individuals who reported following a medically prescribed diet that restricted or eliminated fermented foods (e.g., low-histamine or low-FODMAP diets) were also excluded to reduce potential dietary misclassification.

2.3. Sample Size

The required sample size for this cross-sectional study was determined based on power analysis for logistic regression with a binary outcome. Assuming a two-sided α of 0.05, 80% statistical power, and an anticipated prevalence of moderate-to-extremely severe anxiety symptoms of approximately 35%, a minimum sample of 210 participants was required to detect an odds ratio of 1.8 or greater for the association between fermented food intake and psychological symptoms.

The final analytic sample comprised 212 participants, meeting the calculated minimum requirement for the primary logistic regression analyses. This sample size was considered adequate for evaluating the primary association between fermented food intake and psychological symptoms while adjusting for key covariates. However, the study was not specifically powered to detect small effects in subgroup, interaction, quartile-based, or non-linear dose-response analyses. These secondary analyses were therefore considered exploratory and are interpreted with appropriate caution [22, 23].

2.4. Measurements

2.5. Data Analysis Methods

All statistical analyses were conducted using R (version 2023.06.1+524). Data were first screened for completeness, implausible values, and internal consistency. Continuous variables were summarised using means and standard deviations (SD) or medians and interquartile ranges (IQR), depending on distribution. Categorical variables were summarised using frequencies and percentages.

Fermented food intake was calculated as total estimated servings/week by converting frequency categories from the food frequency questionnaire to midpoint values and summing across items. Intake was analysed both as a continuous variable (per additional serving per week) and categorised into quartiles based on the sample distribution. The aggregated intake variable was intended to reflect overall habitual fermented food exposure in a real-world dietary context rather than standardised probiotic or strain-specific intake. Given the exploratory epidemiological nature of the study, fermented foods were analysed collectively despite expected heterogeneity in microbial composition and preparation methods.

Depression, anxiety, and stress symptoms were assessed using the DASS-21. Subscale scores were summed and multiplied by two according to standard scoring procedures. The primary outcomes were binary indicators of moderate-to-extremely severe symptoms (moderate+) for each subscale, defined using established cut-offs. The DASS-21 is a validated symptom-screening instrument and does not constitute a clinical diagnosis of depression, anxiety, or stress disorder.

Bivariate analyses were conducted to explore associations between fermented food intake and participant characteristics using χ² tests for categorical variables and t-tests or ANOVA for continuous variables, as appropriate. Correlation analyses (Pearson or Spearman, depending on normality) were used to examine relationships between continuous fermented food intake and DASS-21 subscale scores.

Multivariable logistic regression models were fitted to estimate odds ratios (OR) and 95% confidence intervals (CI) for moderate+ depression, anxiety, and stress. Models were constructed sequentially: (1) unadjusted; (2) adjusted for age and gender; and (3) fully adjusted for age, gender, ethnicity, year of study, living arrangement, chronic illness, and probiotic supplement use.

Dose-response relationships were examined using intake quartiles (lowest quartile as reference) and natural spline models (degrees of freedom = 3) to assess potential non-linearity. Exploratory interaction analyses were conducted to assess potential effect modification by probiotic supplement use and health-motivated fermented food consumption. Given the relatively small number of events in several subgroups, these analyses were considered hypothesis-generating and interpreted cautiously.

To account for multiple testing across the three primary outcomes, false discovery rate (Benjamini-Hochberg) correction was applied. All statistical tests were two-tailed, and a p-value <0.05 was considered statistically significant.

3.1. Participant Characteristics and Fermented-food Intake

A total of 212 participants were included in the complete-case analysis. The mean age was 21.2 years (SD 3.4), with 48% male and 77.4% female. The median fermented food intake was 2.0 servings/week (IQR 0.75-3.50; range 0.00-28.25).

Across intake quartiles (Table 1), median intake increased from 0.25 servings/week in Q1 to 6.12 servings/week in Q4. The proportion of females decreased across quartiles (84.7% in Q1; 67.3% in Q4). Ethnic distribution varied across intake groups, with Dusun participants comprising 22.0% of Q1 and 51.9% of Q4. The proportion of Malay participants decreased across increasing quartiles.

Year of study distribution differed between quartiles, with Year 3 students representing 78.8% of the highest intake group. Probiotic supplement use increased from 10.2% in Q1 to 28.8% in Q4. Chronic illness was reported by 2.4% of the overall sample.

The number of participants meeting criteria for moderate-to-extremely severe symptoms was 29 for depression (13.7%), 75 for anxiety (35.4%), and 22 for stress (10.4%). These event counts were used in the logistic regression analyses examining associations between fermented food intake and psychological symptoms. Symptom prevalence across quartiles did not demonstrate a consistent linear pattern.

The full distribution of DASS-21 symptom severity is presented below (see Table 2):

3.2. Association between Fermented-food Intake and Moderate-to-extremely Severe Symptoms

Table 3 presents logistic regression models examining the association between fermented food intake (per additional serving/week) and moderate-to-extremely severe symptoms of depression, anxiety, and stress.

For depression, the unadjusted model showed an odds ratio (OR) of 1.04 (95% CI 0.95-1.12; p=0.319). After adjustment for age and gender, the OR was 1.03 (95% CI 0.95-1.11; p=0.419). In the fully adjusted model, the OR was 1.04 (95% CI 0.92-1.16; p=0.532).

For anxiety, the unadjusted OR was 1.01 (95% CI 0.95-1.08; p=0.687). The age- and gender-adjusted model yielded an OR of 1.01 (95% CI 0.95-1.08; p=0.702), and the fully adjusted model showed an OR of 1.04 (95% CI 0.95-1.13; p=0.359).

For stress, the unadjusted OR was 0.92 (95% CI 0.76-1.05; p=0.324). After adjustment for age and gender, the OR was 0.92 (95% CI 0.75-1.04; p=0.281). In the fully adjusted model, the OR was 0.87 (95% CI 0.65-1.05; p=0.254).

False discovery rate correction applied across the three fully adjusted primary models did not materially change statistical significance.

3.3. Dose-response and Non-linearity

Compared with the lowest intake quartile (Q1), adjusted odds ratios in higher quartiles were generally imprecise and did not indicate a consistent dose-response pattern (as shown in Appendix B). For example, depression odds were lower in Q2 (aOR 0.29 (0.03, 1.99)) but higher in Q4 (aOR 1.22 (0.25, 6.40)), with wide confidence intervals.

Spline models likewise did not demonstrate a stable monotonic relationship across the intake range, with widening uncertainty at higher intakes (See Appendix B).

3.4. Interaction Analyses

Exploratory interaction analyses from fully adjusted logistic regression models assessed whether the association between fermented food intake (per additional serving/week) and moderate-to-extremely severe symptoms differed by probiotic supplement use or health-motivated fermented food consumption (See Appendix C). No interaction term reached statistical significance after accounting for the limited statistical power available for subgroup analyses. Given the small number of outcome events in several modifier subgroups, particularly for depression and stress outcomes, these findings should be considered hypothesis-generating only and interpreted with caution.

4.1. Null Association between Fermented Food Intake and Psychological Symptoms

In this cross-sectional sample of Malaysian university students, habitual fermented food intake was not independently associated with moderate-to-extremely severe symptoms of depression, anxiety, or stress after adjustment for relevant sociodemographic and health-related factors. No consistent dose-response relationship was observed across intake quartiles or spline analyses, and the observed effect estimates remained small and statistically non-significant across models. These findings suggest that, within the intake range observed in this population, habitual fermented food consumption was not associated with meaningful differences in psychological symptom severity.

The present findings differ from some previous studies that reported beneficial associations between fermented foods or probiotic intake and mental health outcomes. However, much of the existing literature derives from controlled intervention studies using standardised probiotic strains, defined dosages, and more homogeneous exposure conditions [27-29]. In contrast, the present study assessed naturalistic dietary intake within a real-world university population, where fermented food exposure is likely to be intermittent, variable, and influenced by broader lifestyle and cultural dietary practices.

Several explanations may account for the absence of detectable associations in the present study. First, median fermented food intake in this sample was relatively modest, potentially limiting the ability to detect small exposure-related effects. Second, the study population consisted primarily of non-clinical young adults, among whom severe depressive and stress symptoms were relatively infrequent. Dietary influences on psychological symptoms may therefore be more difficult to detect within relatively healthy populations where symptom variability is limited.

Interpretation of potential microbiota-related mechanisms should remain cautious. The present study did not assess gut microbiota composition, microbial viability, inflammatory markers, metabolomic profiles, or broader dietary patterns. Thus, no mechanistic conclusions regarding microbiota-gut-brain interactions can be directly inferred from the current findings. References to possible psychobiotic mechanisms are therefore intended only as theoretical context supported by prior literature rather than mechanisms demonstrated in this study.

4.2. Translational Gap between Psychobiotic Trials and Habitual Dietary Intake

A central issue highlighted by the present findings is the translational gap between controlled psychobiotic research and real-world dietary behaviour. Many randomised controlled trials demonstrating beneficial effects of probiotics or fermented foods on mood have used standardised microbial strains, high Colony-Forming Unit (CFU) concentrations, daily dosing schedules, and relatively short intervention windows [30, 31]. These conditions differ substantially from habitual dietary consumption patterns observed in community settings.

This distinction is critical for the field of nutritional psychiatry. While mechanistic and intervention studies support the plausibility of microbiota-gut-brain interactions, translating these findings into population-level dietary recommendations requires evidence that typical consumption patterns exert measurable effects [32].

These findings therefore contribute to a more nuanced understanding of psychobiotic research, emphasizing that evidence derived from supplement-based or strain-specific trials should not be directly generalized to diverse, real-world fermented food consumption without careful consideration of dosage, microbial characterization, and exposure context.

4.3. Cultural Heterogeneity and Microbial Diversity of Traditional Fermented Foods

An important dimension of the present study is its inclusion of traditional Southeast Asian fermented foods, many of which remain microbiologically undercharacterised. Unlike commercially produced yogurt or kefir, which contain standardised and well-documented probiotic strains, traditional fermented foods such as bosou, tuhau, tempoyak, tapai, and belacan are produced using spontaneous fermentation processes. Their microbial composition may vary substantially depending on ingredients, environmental conditions, household preparation methods, and storage practices.

This heterogeneity introduces both scientific opportunity and analytical complexity. From a mechanistic perspective, traditional ferments may contain diverse microbial communities with potential psychobiotic properties. However, without strain-level identification or quantification of viable organisms, it is difficult to determine whether such foods deliver consistent or biologically meaningful exposure [33, 34]. Aggregating culturally distinct fermented foods into a single “total intake” variable may therefore obscure potential differential effects between food types.

For epidemiological feasibility, the present study operationalised fermented food intake as a cumulative exposure variable representing habitual fermented food consumption patterns rather than attempting food-specific microbial quantification. This approach is consistent with exploratory nutritional psychiatry studies assessing broad dietary exposure patterns, although it may reduce specificity regarding individual fermented food effects [35, 36].

The present findings highlight the need for greater microbial profiling of indigenous fermented foods in Southeast Asia. By examining fermented food intake within a Malaysian university population, this study broadens the geographical and cultural scope of the field.

4.4. Limited Statistical Power for Subgroup and Interaction Analyses

Although exploratory subgroup and interaction analyses were conducted to examine potential effect modification by probiotic supplement use and health-motivated fermented food consumption, these findings should be interpreted cautiously due to limited statistical power. Several subgroup categories contained relatively small numbers of participants and sparse outcome events, particularly for depression and stress outcomes. This resulted in wide confidence intervals and reduced precision of the interaction estimates.

Interpretation of the spline analyses should be cautious. Although spline modelling was used to explore potential non-linear relationships, uncertainty increased substantially at higher intake levels where observations were sparse. As a result, the spline findings do not support strong inference regarding the presence or absence of a non-linear association between fermented food intake and psychological symptoms.

The absence of statistically significant interaction effects therefore should not be interpreted as definitive evidence that no subgroup differences exist. Rather, the current sample may have been insufficiently powered to reliably detect modest interaction effects or context-dependent associations. Sparse-event conditions in logistic regression modelling may also increase instability of parameter estimates and reduce interpretability of subgroup findings.

Additionally, the overall prevalence of moderate-to-extremely severe depressive and stress symptoms was relatively low within this predominantly non-clinical university sample. While the study was adequately powered for the primary regression analyses, it was not specifically designed to detect small subgroup differences or interaction effects across multiple modifiers.

These limitations underscore the importance of cautious interpretation of secondary analyses and highlight the need for larger studies with greater subgroup representation and higher event frequencies. Future research using larger multicentre samples or longitudinal designs may provide greater statistical precision for evaluating whether associations between fermented food intake and psychological symptoms differ according to demographic, clinical, or lifestyle-related factors.

4.5. Implications for Nutritional Psychiatry

The present findings carry several implications for the evolving field of nutritional psychiatry. First, the findings highlight the methodological challenges involved in evaluating fermented foods within nutritional psychiatry research. Variability in microbial composition, food preparation methods, and exposure assessment may contribute to inconsistent findings across studies and populations. They highlight the methodological challenges involved in assessing fermented food exposure within epidemiological studies. Future nutritional psychiatry research may benefit from integrating dietary assessment with microbiome profiling, biomarker analysis, and longitudinal study designs to better clarify potential pathways linking fermented food consumption and psychological wellbeing. Context-sensitive research is necessary before incorporating traditional fermented foods into mental health promotion frameworks.

Second, the results suggest that fermented food intake alone may not function as an independent determinant of psychological symptom severity in non-clinical young adult populations. Mental health outcomes are multifactorial, influenced by complex interactions among genetic, psychosocial, behavioural, and environmental factors. Fermented food consumption, particularly at modest levels, may represent only a small component within a broader lifestyle matrix. Therefore, interventions aimed at improving student mental health may require integrated approaches that address diet alongside sleep, physical activity, social support, and stress management.

Third, the absence of statistically significant associations should not be interpreted as evidence that fermented food intake has no influence on psychological wellbeing. Rather, the present findings indicate that no detectable association was observed within this study population and exposure range. Given the heterogeneity of fermented foods, limitations of dietary exposure assessment, and the possibility of residual confounding, modest or context-dependent effects cannot be excluded.