The Global Diet Quality Score is associated with nutrient adequacy and depression among Vietnamese youths

The Global Diet Quality Score (GDQS) has been recommended as a simple diet quality metric that is reflective of both nutrient adequacy and noncommunicable disease outcomes. It has been validated among women of reproductive age (15–49 years) in diverse settings but not specifically among younger women. This paper examines the relationship between the GDQS and nutrient adequacy, anthropometric outcomes, and depressive symptoms among 1001 Vietnamese young women aged 16–22 years. In energy‐adjusted models, the GDQS was significantly (p < 0.05) and positively correlated with intakes of protein (ρ = 0.23), total fat (ρ = 0.06), nine micronutrients (calcium, iron, zinc, vitamin C, riboflavin, niacin, vitamin B6, folate, and vitamin A) (ρ = 0.12–0.35), and the mean probability of adequacy of micronutrients (ρ = 0.28). Compared to young women with optimal GDQS, those with low and very low GDQS were two to five times more likely to have a mean probability of nutrient adequacy less than 50% and showed two to three times higher odds for depression. No association was observed for GDQS and anthropometric outcomes. In conclusion, the GDQS performed well in capturing nutrient adequacy and depressive symptoms among Vietnamese young women. Further research is warranted to explore the relationship between diet quality and depression in other settings.


INTRODUCTION
Suboptimal diet is a leading cause of adverse health outcomes globally, with an estimated 11 million deaths and 255 million disabilityadjusted life-years attributable to dietary risk factors. 1 Notwithstanding such large negative effects, important gaps remain in the collection of dietary data, particularly in low-and middle-income countries (LMICs) due to bottlenecks such as high cost, time burden, and low capacity. 2 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.© 2023 The Authors.Annals of the New York Academy of Sciences published by Wiley Periodicals LLC on behalf of New York Academy of Sciences.
The Global Diet Quality Score (GDQS) was developed to provide a simple, low-cost, food-based approach to measure diet quality in diverse settings. 3The GDQS metric has been associated with a range of outcomes related to nutritional status and noncommunicable disease (NCD) risk in various settings.For example, the GDQS was positively associated with individual micronutrient intakes and overall nutrient adequacy in 10 sub-Saharan African (SSA) countries, 4 Mexico, 5 India, 6 and China. 7The GDQS was also associated with hematological indicators (including hemoglobin, 8 serum ferritin, 5 serum folate, 5,8 and anemia 4 ) in SSA and with lipid profiles in Mexico 5 and India. 6[5][6]9 For NCDs, higher GDQS was associated with a lower risk of diabetes in US women, 10 hypertension in Ethiopian men, 8 and metabolic syndrome in China. 7Among females, the GDQS has been used mainly on nonpregnant, nonlactating women of reproductive age (WRA, 15-49 years) [11][12][13][14] but no work, to our knowledge, has yet examined the usefulness of the GDQS in examining its association with depression symptoms among young women.Suboptimal diet quality is a well-established risk factor in poor mental health outcomes, such as depression, anxiety, and other psychiatric conditions. 15Conversely, a systematic review of the current evidence suggests that a nutritionally adequate, healthy dietary pattern that is rich in fruits, vegetables, whole grains, low-fat dairy, and lean protein foods is associated with a lower risk of behavioral health disorders. 16,17e biological mechanisms involved relate to the critical role of several nutrients in the neuroendocrine system.For instance, tryptophan, vitamin B 6 , vitamin B 12 , folate, phenylalanine, tyrosine, histidine, choline, glutamic acid, and Omega-3 fatty acids all play a role in the production of neurotransmitters (serotonin, dopamine, and norepinephrine) that are involved in the regulation of mood, appetite, cognition, depression, and anxiety. 18,19Against such evidence, the question arises whether GDQS could also capture, beyond the "usual" NCD risks, behavioral health outcomes among young women in LMICs, an issue that gained saliency with the publication of the Global Burden of Disease 2021, which found that "depression presents the greatest disease burden for women when compared with other diseases," particularly among younger women. 20This question is further relevant in the context of the COVID-19 pandemic which exacerbated many determinants of poor mental health and led to a 27.6% increase in major depressive disorders globally. 20New York Academy of Sciences (NYAS) project aimed at engaging young women in solutions to improve the quality of their diet in Thai Nguyen, Vietnam.This project collected extensive data on a variety of nutrition and health outcomes among Vietnamese young women, providing a unique opportunity to examine the performance of the GDQS in predicting not only nutrient adequacy, micronutrient intakes, BMI, underweight, overweight/obesity, and MUAC but also in assessing the relationship between the GDQS and depressive symptoms.

Study population
The NYAS study was conducted in eight urban and four periurban communities of Thai Nguyen, a city in the Northern province of Vietnam in 2021-2022, during the COVID-19 pandemic.A total of 1001 female participants aged 16-22 years old stratified by main occupational status (201 paid workers, 399 high school students, and 401 college students) were randomly selected.A detailed description of the study design and sampling approach is provided elsewhere. 21

Dietary assessment
Dietary intake was assessed using INDDEX24's multiple-pass, quantitative 24-hour recall methodology 22 via a smartphone app and web-based platform tailored to the Vietnamese language.The recall assessed the consumption of all the foods and beverages a respondent had consumed the previous day from when she woke up in the morning until she went to bed.The quantity of each food or beverage consumed was estimated using previously validated visual portion size estimation aids, such as standard plates, bowls, cups, and other common household utensils. 23Two Vietnamese dietary experts provided weekly data checks to ensure the completeness and accuracy of the food codes associated with the food descriptions.They also reviewed the plausibility of estimated quantities and portion size of the reported foods.To account for within-person variation, repeated 24-hour recalls were conducted in a subset of participants (10%) on a nonconsecutive day, which could be any day of the week.

Global Diet Quality Score
Several diet metrics exist that are valid, robust, and easily operationalized at the population levels, including the Food Group Dietary Diversity Score (FGDS), [24][25][26] the Global Dietary Recommendations (GDR) score, 27 and the GDQS. 3 The FGDS includes 10 food groups and has been shown to be associated with micronutrient intake in WRA in several LIMCs, 24,26 but the FGDS does not provide information on the risk of diet-related NCDs.The GDR is based on 17 food groups and has been validated using data sets from Brazil and the United States against dietary recommendations included in the 2018 WHO healthy diet fact sheet, 28 but has not been validated against micronutrient intake adequacy or NCD-related risk outcomes directly.
The GDQS categorizes diets using both food groups and quantity of intake, and has been validated to reflect both nutrient adequacy and diet-related NCD risk using data sets from 14 countries across low-, middle-, and high-income settings. 3For these reasons, this study used the GDQS, the first metric of diet quality to be validated for global use. 29e GDQS is a food-based metric of diet quality that includes 25 food groups seen as important contributors to nutrient intake and/or NCD risk. 3There are 16 healthy food groups (scored to give more points for higher intake), seven unhealthy food groups (scored to withdraw points for higher levels of intake), and two food groups classified as unhealthy only if consumed in excessive amounts (score 0 for low or excessive intake).Summing points across all 25 food groups provides a summary measure of overall diet quality that ranges from 0 to 49. GDQS ≥23 is associated with a low risk of both nutrient adequacy and NCD risk, GDQS ≥15 and <23 indicates moderate risk, and GDQS <15 indicates high risk.Two GDQS submetrics were also calculated: GDQS+ (consisting only of the food groups the GDQS considers healthy) ranges from 0 to 32; the GDQS− (consisting only of the food groups the GDQS considers unhealthy or unhealthy in exces-sive amounts) ranges from 0 to 17. 3 All GDQS metrics were computed on the first day of recall information.Foods reported as consumed by respondents were classified into food groups according to the definition of the GDQS food groups and rules for classification.

Nutrient intakes
Foods consumed were converted to nutrient intakes using a 2007 Vietnamese food composition table (FCT) 30 updated in 2019 and 2021, including adjustment for nutrient retention of cooked foods 31,32 and for yield factor for raw-cooked food from a food conversion table. 33xed dishes were disaggregated into individual foods based upon recipes derived from the Vietnam General Nutrition Survey. 23All the foods were then reclassified into their corresponding GDQS food groups. 3Foods for which no nutrient information existed were updated based on the Vietnam General Nutrition Survey FCT (2019-2020), Thai FCT, 34 and Asian FCT. 35ual energy and micronutrient intakes were estimated using the intra-person or intra-day variance method. 36The Probability of Adequacy (PA) for 11 micronutrients (vitamin A, vitamin C, thiamin, riboflavin, niacin, vitamin B6, vitamin B12, folate, calcium, zinc, and iron) was calculated by applying Estimated Average Requirements following WHO/FAO recommendations by age and sex, 37 the IZiNCG recommendations for zinc, 38 and Institute of Medicine recommendations for calcium 39 assuming low levels of bioavailability for iron (5%) and zinc (15%).The mean probability of adequacy (MPA) of micronutrient intake was calculated by combining the PAs for the 11 micronutrients.Nutrient inadequacy was operationalized as an MPA <0.50. 24,40

Anthropometric outcomes
Weight, height, and MUAC were measured by trained staff using standard procedures. 41,42All measurements were taken twice with strict adherence to accepted measuring techniques and recording procedures.The average of the two measurements was used in the analysis.BMI was calculated as weight (kg)/ height 2 (m), then categorized as underweight (<18.5),normal (18.5-23.0),or overweight (≥ 23.0) using cutoff values recommended to identify high-risk individuals in Asians. 43Low MUAC was defined as MUAC < 23.5 cm based on previous research conducted in WRA in Thai Nguyen, Vietnam. 44

Depressive symptoms
Depressive symptoms were assessed using Reynolds Adolescent Depression Scale Second Edition (RADS-2). 45The RADS-2 is measured using a brief and easy-to-administer self-report of depressive symptomatology in adolescents and youth.The RADS-2 includes 30 items of four subdomains: Dysphoric Mood, Anhedonia/Negative Affect, Negative Self-Evaluation, and Somatic Complaints.The items use a four-point Likert-type response format that requires the adolescent to indicate whether the symptom-related item has occurred: never, hardly ever, sometimes, or most of the time.The RADS-2 includes seven reverse-scored items that are worded in a positive manner.The depression total score represents the sum of all items on the RADS-2, with a higher score representing a higher risk of depression.
The RADS-2 tool was widely used in various settings as an official measure of depression for adolescents ages 10-20 years in schools, clinics, and hospitals. 46,47

Other variables
Individual and household characteristics were collected in face-to-face interviews by trained enumerators.At the individual level, we collected information on adolescent's age, ethnicity, education, occupation, and body perception.At the household level, we collected information on household size and assets, as well as parent's education and occupation.Household socioeconomic status (SES) index was created from a set of household assets using principal component analysis. 48,49

Statistical analysis
Descriptive analysis was used to report the quantity of intake, the distribution of the GDQS subscore for each food group, and total GDQS, GDQS+, and GDQS−, as well as other health outcomes.The performance of the GDQS, GDQS+, and GDQS− against diet quality were evaluated by computing Spearman correlations between the metrics and energy-adjusted nutrient intakes which were estimated using the residual method. 50Associations between the GDQS and MPA, nutrient inadequacy, anthropometric outcomes, and depression were explored using multivariable logistic regression models adjusted for total energy intake, age, ethnicity, occupations of participants and their parents, household size, SES, and clustering effect at the commune level.Girls' self-image (body perception) was also added in the models of nutrient adequacy and depression.Data were analyzed using STATA v.16.0.

Ethical approval
The study was approved by the Ethical Committee of Thai Nguyen National Hospital in Vietnam.Informed consent was obtained from participants aged ≥18 years and from the parent or guardian of participants aged <17 years as well as the assent of the participant.

RESULTS
Participants' age ranged from 16 to 22 years, with a mean of 19 years (Table 1).A fifth of the sample belongs to ethnic minorities (Tay, Nung, Dao, and San chi).Mean BMI was ∼20 kg/m 2 , with about a third of participants being underweight and 13% overweight/obese.Mean MUAC was 23.8 cm and nearly half of the sample had low MUAC (<23.5 cm).
Table 2 presents the quantity of intake for each of the 25 GDQS food groups.Most of the groups classified as healthy (GDQS+) were consumed at a low level, except liquid oils, cruciferous vegetables, and other vegetables.More than 90% of participants consumed low amounts of deep orange fruits, vegetables, and tubers, followed by 82% of them eating low amounts of dark green leafy vegetables.In contrast, nearly two-thirds of participants (62%) consumed high quantities of cruciferous vegetables.Less than 20% of the study population had medium to high consumption levels of plant protein in the form of legumes, nuts, and seeds.The percentage of the medium and high consumption levels of animal protein in the form of poultry, fish, and eggs were about 10 percentage points higher than that of the plant protein forms.Low-fat dairy and whole grains were not popularly consumed in this population.As for the GDQS category "unhealthy in excessive amounts," more than 70% of girls consumed a high amount of red meat and 15% consumed a very high level of high-fat dairy.As for GDQS's "unhealthy foods," all girls consumed high levels of refined grains and baked goods, and more than a third consumed middle to high levels of sweets and ice cream.Likewise, about 25% of participants had middle to high consumption levels of processed meat.The other groups in the GDQS− were not widely consumed.
Figure 1 shows a large gap between the possible GDQS score (49)   and this sample's average score (17.9).The population mean score of GDQS+ was 25 points lower than the possible score of 32 points.
The gap between possible score and population score of GDQS− was narrower than the gap of total GDQS and GDQS+.Overall, 11% of the sample were at the "low risk" category of the GDQS (≥23 points), nearly two-thirds at the "moderate risk" (15-22 points), and 23% at the "high risk" (<15 points).
Correlations between each score and the energy-adjusted nutrient intakes are shown in Table 3.In general, the total GDQS correlated more with micronutrient intakes than with macronutrients.
In terms of macronutrients, the GDQS positively and significantly correlated with the intake of protein (the Spearman correlation coefficients, ρ = 0.23) and total fat (ρ = 0.06).Carbohydrate intake was negatively correlated with the GDQS+ (ρ = −0.08),but positively correlated with the GDQS− (ρ = 0.07) (results not shown).For micronutrients, the correlation coefficients ranged from 0.12 to 0.35 and showed statistical significance for nine out of 11 listed nutrients (all except thiamin and vitamin B12).The GDQS also significantly correlated with probability of adequacy for most micronutrients and MPA (Table 3).
The GDQS was significantly associated with nutrient inadequacy and depressive symptoms, but not with anthropometric outcomes (Table 4).Nearly, 75% of the participants had inadequate nutrient intake (MPA <0.5) and nearly, 15% of the participants had depressive symptoms (Table 1).The GDQS was significantly associated with lower odds of nutrient inadequacy (OR = 0.87, 95% CI: 0.83, 0.91).Compared to young women with high GDQS scores (≥23), those with low scores (15-22.9)had nearly two times higher odds of nutrient inadequacy (OR = 1.92, 95% CI: 1.20, 3.07), and those with very low scores (<15) had more than five times higher odds of nutrient inadequacy (OR TA B L E 2 Quantity of intake for each food group and the distribution of the GDQS subscore for each food group.= 5.24, 95% CI: 2.72, 10.09).Regarding depressive symptoms, those in moderate or high-risk categories of the GDQS had a higher depressive score and two to three times higher odds of being at risk for depression.

GDQS metric design/scoring system
The GDQS did not show a significant association with underweight, overweight, and MUAC (Table 4).

DISCUSSION
We used dietary data from 24-hour recall to evaluate the performance of the GDQS in predicting nutrient adequacy, anthropometry, and depressive symptoms among young women in urban and peri- urban settings in Vietnam.We found that the GDQS was associated with several measures of micronutrient intake, nutrient adequacy, and risk of depression in this group, but not with anthropometric outcomes.
Overall, the mean GDQS was low in this sample (18 out of 49 possible points), lower than the GDQS among WRA in neighboring China (20.8 points) 7 or other LMICs (23 and 28 points for WRA in India 6 and Mexico, 5 respectively).The GDQS+ submetric was nearly five times less than the possible score (6.9 vs. 32), with a low consumption of most subgroups in the healthy category (deep orange fruits, deep orange vegetables, orange tubers, dark green leafy vegetables, whole grains, and low-fat dairy).These findings echo a global review that found most adolescents to have inadequate fruit and vegetable intake daily, with just over one-third and less than half of adolescent girls, respectively, eating vegetables and fruit on a daily basis. 51Further, the high consumption of refined grains and baked goods and high intake of red meat in our study population, both of which are associated with increased NCD risk, 52 call for tailored nutrition education interventions and the promotion of national guidelines to sensitize adolescents and young women about the importance of balancing the excessive or insufficient intake of specific food groups.Notably, the disaggregation provided by the GDQS classification allows for refining the recommendation beyond the general food group into specific food subgroups.
The GDQS is associated positively with the intake of key macronutrients (protein and fat) and several micronutrients (such as calcium, iron, zinc, vitamin C, riboflavin, niacin, vitamin B6, folate, and vitamin A).The risk of nutrient inadequacy is two times higher among adolescents with low GDQS and five times higher among those with very low GDQS scores.This magnitude of association is similar to findings from previous studies in Mexico, 5 Ethiopia, 8 India, 6 and China. 7It is important to note that the Vietnam FCT table does not have all values for different types of fat, therefore, we cannot distinguish the association between the GDQS and the good or bad fats.We did not observe any association between the GDQS scores and the risks of underweight or overweight/obesity or low MUAC in our young women sample.Previous studies found a much stronger association between higher GDQS and lower overweight and less weight gain. 3,4,9It is worth mentioning that the strongest associations with anthropometric measures have been seen in cohort studies in which an increase in GDQS score over time is associated with change in weight (i.e., lower weight gain). 9The lack of association between GDQS, BMI, and MUAC may be expected given the wide anthropometric fluctuations that characterize this age period and the cross-sectional analytical approach: cohort studies are more appropriate to study this relationship.
Notably, we found a strong association between the GDQS and the risk for depression in this population.Compared to adolescents with optimal GDQS score (≥23), those with low (15-22.9)and very low GDQS scores (<15) were two to three times at higher risk for depression.Although many factors determine the mental health of an individual, the diet has been identified as a key factor because of the role played by various macro-and micronutrients in regulating mood, appetite, depression, and anxiety. 18, 19,53 Mental health problems are a significant burden to adolescents in Vietnam, with overall prevalence estimated to range from 8% to 29% and with girls reported to have higher rates of anxiety and depression than boys. 54,55The burden of mental health disorders, including depression, worsened during the COVID-19 pandemic in Vietnam and many other contexts which require further attention.Our study suggests that dietary modification could be part of a strategy to reduce behavioral health issues among adolescent girls and young women.
Since the initial GDQS validation studies were published in 2021, 3 there has been a significant uptake of using the metric to measure diet quality in several new country contexts, including: Bangladesh, Brazil, Cameroon, the Democratic Republic of the Congo, Ethiopia, India, Lebanon, Nepal, Niger, Nigeria, Philippines, Thailand, the United States, and Zambia.The GDQS app (a technology-based, open 24-hour recall, data collection tool), developed to provide a method for collecting lowcost, time-relevant data on the GDQS, 56 was used to collect primary GDQS data for more than 10 of these studies.This rapid uptake of both the GDQS metric and the GDQS app, along with the results provided here, is indicative of the potential for the GDQS to fill the critical gap of metric availability to report on diets at risk of coexisting burdens of micronutrient inadequacy and NCDs.
TA B L E 3 Correlation between GDQS and energy-adjusted nutrient intakes.since those are increasingly consumed in LMICs. 57Using the GDQS in existing data collection and surveillance systems, therefore, presents a useful addition to existing diet assessment tools in Vietnam and elsewhere, allowing better tailoring and targeting of nutrition education interventions toward adolescents and young women.

Nutrient intakes Correlation between GDQS and nutrient intakes a
Further, at a time when several studies raised the alarm on the specific plight of young women who struggle to construct their body self-image against idealized social media representations, sociocultural expectations, and gender socialization processes, having a tool that links behavioral health to nutrition is a welcome addition.
It was validated in Vietnam by experts at the National Institute of Mental Health where it has been in use since 1995.Depression is categorized into four levels based on the depression total score: None (≤30 points), mild (31-40 points), moderate (41-50 points), and severe depression (51-90 points).For analysis, this study used depressive symptom scores both as a continuous variable and as a dichotomized indicator contrasting all participants scoring ≤30 to all participants scoring >30.

1
Possible GDQS scores and average sample scores.Note: GDQS+ consists only of the food groups the GDQS considers healthy.GDQS− consists only of the food groups the GDQS considers unhealthy or unhealthy in excessive amounts.Abbreviation: GDQS, Global Diet Quality Score.
Sample characteristics and health outcomes.
TA B L E 1Abbreviations: BMI, body mass index; MUAC, mid-upper arm circumference.
Zinc was for refined diets as defined by the International Zinc Consultative Group.Association between the GDQS and nutrient inadequacy, anthropometric, and depression outcomes.a Abbreviations: BMI, body mass index; CI, confidence interval; GDQS, Global Diet Quality Score; MPA, mean probability of adequacy; MUAC, mid-upper arm circumference; OR, odds ratio; SES, socioeconomic status.a Models run for total GDQS and categories of GDQS separately.b Models adjusted for total energy intake, adolescents' body perception, age, ethnicity, occupation of adolescents and their parents, household size, and SES.c Models included underweight and normal weight participants, and adjusted for total energy intake, age, ethnicity, occupation of adolescents and their parents, household size, and SES.d Models included overweight and normal weight participants, and adjusted for total energy intake, age, ethnicity, occupation of adolescents and their parents, household size, and SES. e Models adjusted for total energy intake, age, ethnicity, occupation of adolescents and their parents, household size, and SES.f Models adjusted for adolescents' body perception, age, ethnicity, occupation of adolescents and their parents, household size, and SES.
aValues are Spearman correlation coefficients between the GDQS and the energy-adjusted intake of nutrients.b Iron: assume 10% bioavailability.c the GDQS analysis.One limitation is that the cross-sectional design hampers causality inferences between the GDQS and anthropometry or depressive symptoms.With regard to behavioral health, a poor diet may lead to increase mental health issues; however, adolescents with depressive symptoms may also be more likely to reduce or increase food consumption-or else, to exhibit response bias.It will be impor-lescents but was not associated with anthropometric outcomes.Our study provides a useful addition to the range of conditions associated with the GDQS, a simple and low-cost metric that can be widely applied in LMICs.Unlike other metrics, such as the Minimum Dietary Diversity for Women, the GDQS does consider the deleterious effect of consuming unhealthy foods, a matter of importance TA B L E 4 *p<0.05.**p<0.01.***p<0.001.