Participants were recruited from university campuses and surrounding community settings in Lahore, Pakistan, between August and September 2024. The sample consisted predominantly of young adults, with a larger number of women. The percentage of women with depressive symptoms was higher, which was consistent with the epidemiological evidence of the world, with no systematic variables related to smoking or alcohol consumption. Physical activity was measured using a structured self-report measure that categorized participants into low, moderate, and high activity groups based on the frequency of weekly moderate-to-vigorous exercise. A previously invalidated self-reported evaluation of physical activity was utilized, which may have misclassified some participants based on physical activity levels. Physical activity was measured using a non-validated self-report measure, which may have resulted in misclassification. Supplement users were excluded; no additional sensitivity analysis was conducted. Figure 2 presents the participant flow diagram.

The study was conducted in Lahore, the capital city of Punjab province in Pakistan, among university students and community adults.

The sample size was 283 participants; no power calculation was performed in advance, but it is a decent size to detect indicative relationships in logistic regression models. Inclusion criteria were 18–60 years old, healthy, non-medicated, non-supplemented participants, whereas the exclusion criteria were the presence of antidepressant medication or vitamin/mineral supplements, and pregnant women. Those who reported missing data on dietary or depressive symptoms were excluded from the analysis. No formal a priori power calculation was conducted; however, post-hoc power estimation indicated sufficient power (>80%) to detect an odds ratio of 0.70 for vitamin B6 intake.

A total 435 participants were contacted. Among them, 49 failed to fit the inclusion requirements, and 37 refused to take part. The remaining 349 participants answered the questionnaire. After eliminating 66 participants with missing dietary or depressive symptoms data, an analytical sample of 283 participants was used in the analysis (Figure 3).

The study employed a cross-sectional design to investigate the comparative role of the Vitamin B6 diet between two groups, including individuals with and without the symptoms of depression. This study design can evaluate both groups simultaneously to identify possible correlations between nutrient intake and depression status.

A semi-quantitative Food Frequency Questionnaire (FFQ) specifically modified to suit Pakistani dietary habits was used toassess dietary intake. The tool consisted of 145 food items, categorized into the following groups: cereals, legumes, dairy products, meats, vegetables, fruits, beverages, and mixed traditional meals.

Participants were asked to indicate their normal intake during the last 3 months using a scale of seven points, in which the participants indicated their frequency:

The visual portion-size aids (standard cup, tablespoon, bowl size references) were used to provide standard household portion sizes.

Standardized recipes were made for mixed dishes like biryani, daal, and chicken curry based on how people in Lahore usually make them at home. Using entries from the USDA database in Cronometer, each dish was broken down into its individual ingredients and given nutrient values. For instance, we broke down biryani into rice, chicken, oil, and spices by making a few estimates of the proportions (e.g., 60% rice, 20% chicken, 15% oil, and 5% other ingredients). These assumptions could lead to measurement errors that don't differ.

The conversion of frequency responses to daily intake was done based on the following equation:

Daily dietary intake (g/day) = Frequency factor × Portion size (g).

Where frequency factors were put as:

Nutrient intake was computed by multiplying the daily intake (g/day) by the nutrient composition values from Cronometer (Nutritionix database). The complete list of food items (n = 145), frequency response options, and portion-size estimation aids that make up the full FFQ instrument used is included as SDC, File 1. The FFQ has not been officially validated in Pakistan, despite being developed from more than 6 months of training data and adjusted to local dietary patterns. Thus, potential measurement error and misclassification cannot be ruled out.

The severity of depressive symptoms was assessed with the Center for Epidemiologic Studies Depression Scale (CES-D), in which scores >16 indicate depressive symptoms and scores <16 indicate non-depressive individuals.¹⁰

Age, sex, BMI, physical activity (self-reported), and socioeconomic status were collected via questionnaire. The measurement of socioeconomic status (SES) was based on educational attainment (12 years or less, undergraduate, postgraduate degree) and employment status (employed, unemployed/student). The education level has been incorporated as the main SES measure in multivariate models because of its stability and reduced sensitivity to short run fluctuations. Multivariable logistic regression analyses were also adjusted for sex (male/female), age (years), BMI (kg/m²), education level (socioeconomic status proxy), and energy intake (kcal/day).

Data collection from university students and community members was completed in a month. Participants completed a structured survey that included CES-D scores, a 24-h dietary history, and a FFQ. The 24-h dietary recall was used only to estimate portion sizes when the FFQ was completed and was not included in the nutrient intake estimates or in the statistical testing. Cronometer (4.48.x) software was employed to evaluate Vitamin B6 intake, while participants reported on their use of medications and medical treatments to control. Data were analyzed using IBM SPSS Statistics (version 22.0). Descriptive statistics had been used to summarize the characteristics of the participants. The Center of Epidemiologic Studies Depression Scale (CES-D) was used to measure depressive symptoms, and a normal cut-off score (≥16) was used to dichotomize the presence of depressive symptoms. Independent-samples t-tests were used to compare continuous variables, and chi-square tests were used to compare categorical variables to evaluate group differences among participants with and without depressive symptoms.

The most important dietary exposure was vitamin B6 intake. Vitamin B6 was assessed as a continuous variable (mg/day increment) and as a categorical variable (percentage of the recommended dietary allowance: <50%, 50%–74%, 75%–99%, and 100%). The intake of vitamin B6 was modeled as a continuous variable (increase by 1 mg/day) and standardized (increase by 1 standard deviation). To be comparable with previous reporting, models per 0.5 mg/day were kept in additional analyses. All the estimates are presented as odds ratios (ORs) with confidence intervals (CIs). The following variables were included in all multivariable logistic regression models: age (years), sex (male or female), body mass index (kg/m²), energy intake (per 1000 kcal/day), education (categorical), physical activity (low, moderate, or high), vitamin B12 intake (µg/day), and folate intake (per 100 µg/day).

The secondary analysis of other nutrient intakes was conducted as an exploratory study. Vitamin B6 was determined a priori as the key exposure variable. Correlations with other nutrients were considered exploratory, so no correction of multiple comparisons was applied. Therefore, these findings must be taken with caution. Variance Inflation Factors (VIF) was used to determine multicollinearity between covariates. The values of VIF were lower than 2.5, which means that there is no serious multicollinearity. The diagnostic analysis of model fit and influence, based on commonly used goodness-of-fit measures, also indicated that no individual observation significantly influenced model estimates. The statistical significance level was set at P = 0.05.

Alternative energy-adjustment methods were used as sensitivity analyses. First, the residual approach was implemented by regressing vitamin B6 intake on total energy intake and then applying the residuals in logistic regression models. Second, the nutrient density method was used by dividing the vitamin B6 intake by 1000 kcal/day. The primary model,which adjusted for total energy intake as a covariate, was compared with estimates obtained through these methods.

The participants were recruited on university campuses and in the surrounding community in Lahoreduring August and September 2024, after receiving ethical approval from the Office of Research, Innovation and Commercialization (ORIC) at the University of Management and Technology (REF-009-2021).

The design used in this study was a cross-sectional convenience sample of Pakistani university students and community adults in Lahore, Pakistan. Considering the research design, the results indicate a relationship and do not mean causality. The CES-D scale was used to evaluate depression; scores of 16 or higher were assigned to the depressive category, and a score of 16 was considered a depressive sample. The intake of vitamin B6 was measured against the WHO/FAO Recommended Dietary Allowances (1.3–1.7 mg/day) with sex-specific recommendations as necessary.¹¹ The USDA Food Composition Database was the main source of nutrient values, which were then used by Cronometer. About 85%–90% of the entries were matched directly. For local Pakistani foods and mixed dishes, recipe decomposition was used to get close. A sensitivity analysis was performed using several nutritional estimates for key vitamin B6-rich foods, based on regional data sources, to determine whether there was any bias in the non-local ingredient composition data. The results remained in the same direction, suggesting that measurement error related to the database is unlikely to differ.

A cross-sectional design does not allow causal inference. The convenience sample (mostly university students) might restrict the applicability of the results to the rest of the Pakistani adult demographic. Food evaluation was based on self-report and nutrient estimation on non-local databases. Vitamin B6 status (measured, e.g., plasma PLP) was not measured. Residual confounding poses a serious threat to the validity of the findings, given that the study does not account for important behavioral and clinical factors such as smoking habits, alcohol consumption, the quality of sleep, and whether or not participants had chronic disease or were taking medications. While confounding factors such as age, gender, BMI, caloric intake, education level, and the intake of certain nutrients were taken into account for the purposes of the analysis, it is still possible that there were other unmeasured factors that had an impact on the overall results of this study. The results of the sensitivity analysis using E-values indicated that for an unmeasured confounding variable to fully explain the association between vitamin B6 intake and depressive symptoms, it would need to have a moderate correlation with both vitamin B6 intake and depressive symptoms. Physical activity was measured using a non-validated instrument; this may have accounted for the reduced accuracy of adjusting for physical activity when controlling for other variables.

Overall, 148 participants (52.3%) were classified as depressed, while 135 (47.7%) were non-depressed. It was composed of 59.7% females, and they were more likely to suffer from depressive symptoms than males (P = 0.008). Figure 4 illustrates these trends and shows a strong correlation between depressive symptoms and gender. Figure 5 shows that both males and females also reported an inverse relationship between the intake of vitamin B6 and depressive symptoms, although the protective pattern was a bit stronger among female subjects.

Body weight also exhibited a substantial association with depressive symptoms. Figure 6 showed the percentage of participants with depressive symptoms across different BMI categories. Individuals with normal BMI (18.5–24.9) were the least affected (18.4% depression rate). However, depressive symptoms were significantly more common in underweight individuals (66%), overweight individuals (68%), and people who were obese (83%). The BMI of the depressed participants was higher than that of the nondepressed participants (26.3 ± 4.7 vs. 23.7 ± 3.9 kg/m²) (Table 1). This increasing tendency suggests that psychological well-being may be affected by both undernutrition and overnutrition. Figure 7 shows that the reference group was those who took less than 50% RDA. An increase in vitamin B6 was also linked with a lower probability of having depressive symptoms in both groups of body mass index (<25 kg/m² and ≥25 kg/m²), indicating that the relationship was similar in both body mass indexed groups.

While comparing the daily intake of nutrients of the depressive and nondepressive groups, several nutrients were associated with depressive symptoms (Table 2). Vitamin B6, vitamin B12, protein, carbohydrates, phosphorus, and vitamin K levels were lower in depressed individuals than in the nondepressed individuals. On the other hand, mean intakes of folate, potassium, selenium, and vitamin A were higher among participants with depressive symptoms; thus, not all observed associations were attributed to dietary inadequacy. The mean dietary intake of vitamin B6 in the overall sample was 1.59 ± 0.61 mg/day, with a median intake of 1.48 mg/day (interquartile range: 1.12–2.05 mg/day). Figure 8 illustrates that the intake of vitamin B6 was linked to a declining odds of depressive symptoms, meaning that there could be a dose–response relationship with vitamin B6.

A strong correlation (P < 0.001) between depressive symptoms and the % RDA of vitamin B6 is observed in Table 3. Participants taking ≤50% of the RDA had high depression rates, whereas those meeting or above 100% of the RDA had significantly lower depressive symptoms. It supports the concept that adequate Vitamin B6 intake may help alleviate symptoms of depression.

Table 4 reveals that Vitamin B6 (P = 0.025) and Vitamin B12 (P = 0.03) were the only vitamins significantly associated with BMI categories. People with normal BMI had a higher frequency of taking Vitamin B6 and B12 closer to the recommended levels than people in other BMI categories. The vitamin B6 consumption was relatively low among the depressed people. An inverse gradient was observed across increasing categories of vitamin B6 intake, expressed as a percentage of the Recommended Dietary Allowance. It was positively associated with a lower risk of depression in multivariate models (adjusted OR = 0.72, 95% CI: 0.55–0.94, P = 0.015) (Table 5). The BMI was also positively correlated with the intake of vitamin B12 and negatively correlated with depressive symptoms in the female participants.

The relationship between vitamin B6 consumption and depressive symptoms held true in sensitivity analyses using various energy-adjustment techniques (Table 6). The adjusted OR using the residual method was 0.74 (95% CI: 0.57–0.96; P = 0.021). The adjusted OR was 0.70 (95% CI: 0.53–0.92; P = 0.012) using the nutrient–density approach (mg/1000 kcal). These findings indicate the robustness of the primary model.