The longitudinal association between onion consumption and risk of depressive symptoms: results from the TCLSIH Cohort study and the UK Biobank

Abstract

Background and aims: Onions have shown antidepressant effects but relevant evidence from people was limited. Thus, we aimed to explore the prospective association between onion consumption and risk of depressive symptoms in the general population. Methods: We used data from 2 cohorts: the Tianjin Chronic Low-grade Systemic Inflammation and Health (TCLSIH) cohort included 7739 participants (males, 57.6%) with a mean age of 39.5 years, and the UK Biobank included 169 806 individuals (males, 45.2%) with a mean age of 55.7 years. In the TCLSIH cohort, onion consumption was assessed by a validated food frequency questionnaire from May 2013; depressive symptoms were evaluated by the Chinese version of the Self-Rating Depression Scale (SDS scores ≥ 45) and were assessed up to 2018. In the UK Biobank, onion consumption was measured by 1–5 times validated 24-hour dietary recalls in 2006–2010; depressive symptoms were determined through the linked hospital records and death registries and were assessed up to 2018. Multivariable Cox proportional hazards models were employed to determine the association between onion consumption and risk of depressive symptoms. Results: A total of 1098 and 1924 cases of depressive symptoms were identified during 15 004 person-years and 1 243 832 person-years of follow-up in the TCLSIH cohort and the UK Biobank, respectively. After adjusting for many confounding factors, the fully adjusted HRs (95% CI) of depressive symptoms comparing the higher levels to the lowest level of onion consumption were 0.78 (0.65, 0.94), 0.73 (0.61, 0.87), and 0.77 (0.64, 0.92) in the TCLSIH cohort (p for trend = 0.01); and were 0.79 (0.68, 0.93), 0.81 (0.69, 0.94), and 0.97 (0.85, 1.12) in the UK Biobank (p for trend = 0.07). Similar associations were observed in the sensitivity analyses. Conclusion: Our results indicated that habitual onion consumption was associated with a lower risk of depressive symptoms in two cohorts. However, this association was not statistically significant in the highest level of onion consumption in the UK Biobank.

---

1. Introduction

Depression is an increasingly common disease characterized by a persistent feeling of sadness and/or an inability to experience pleasure.¹ The WHO reported that over 300 million people were estimated to suffer from depression worldwide, equal to 4.4% of the world's population.² This epidemic has been a severe public health issue, resulting in a high economic burden for society.³ Although many patients respond favorably to treatment, inefficiency, residual depressive symptoms, and drug side effects are common. Thus, the search for modifiable factors (e.g., dietary regimen) to decrease the incidence of depression is indispensable.

Onion, a typical example of Allium species, is one of the most widely consumed vegetables among various nations.⁴ It is rich in nutrients, such as dietary fiber, minerals, vitamins, and organosulfur compounds (OSCs).⁵ Particularly, onion is one of the richest sources of dietary flavonoids, largely contributing to the overall intake of flavonoids.^6,7 Flavonoids have been shown to improve neuroinflammation and suppress neuronal apoptosis,^8,9 which play an important role in depression pathophysiology. Previous studies also demonstrate that dietary intake of flavonoid-rich foods (such as tea, soy foods, and blueberry) is associated with a lower risk of depressive symptoms.^10–12 Quercetin and its derivatives are the main flavonoid compounds in onion.¹³ Studies conducted at Wageningen Agricultural University have demonstrated that quercetin from onion shows twice absorption comparable to tea and thrice relative to apple.¹³ In addition, other nutrients in onion, including vitamin B, magnesium, and dietary fiber, have shown antidepressant effects.^14–17 Several animal-based studies prove that oral administration of onion powder could attenuate the depressive-like behaviors of rats.^18,19 Thus, we hypothesized that onion intake might reduce the risk of developing depressive symptoms.

One recent clinical study reported that quercetin-rich onion consumption improved depressive symptoms in healthy elderly people (n = 70).²⁰ However, the study sample was small, and the study population was restricted. Epidemiological evidence related to onion consumption and depressive symptoms is still limited. Thus, we designed this study to explore the longitudinal association between onion consumption and the risk of depressive symptoms in the general adult population. We assessed this association in two large cohorts, namely the Tianjin Chronic Low-grade Systemic Inflammation and Health (TCLSIH) Cohort Study and the UK Biobank, to better understand the underlying role of population characteristics and to improve the external validity of the results.

2. Methods

2.1. Study design and participants

The data used in this study were derived from Tianjin Chronic Low-grade Systemic Inflammation and Health (TCLSIH) Cohort Study, a multipurpose dynamic prospective cohort study centering on the adult populations (≥18 years) living in Tianjin, China. Participants were randomly recruited while taking their annual physical examinations. The follow-up examinations were usually performed once every year. Beginning in May 2013, all volunteers in TCLSIH were invited to complete a health questionnaire survey, including dietary intake, physical activity, lifestyle factors, and sociodemographic characteristics. More details of this cohort have been described previously.²¹ The response rate of participants was >90%.²² This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects were approved by the Institutional Review Board of Tianjin Medical University (number: TMUhMEC 201430). All participants gave written informed consent prior to participate in the study. Between May 2013 and December 2018, we recruited 11 618 participants (20–90 years) who completed the baseline dietary questionnaire and had information on follow-up disease. We excluded participants with implausible energy intake (n = 549), history of CVD (n = 482) or cancer (n = 79), or baseline depressive symptoms (n = 1811). In addition, we excluded individuals who did not complete follow-up health examinations (n = 958; retention rate: 89.0%). Finally, the number of 7739 participants were brought into the analysis (mean ± standard deviation (SD) age: 39.5 ± 10.5 years; men, 4459 (57.6%)).

The UK Biobank is a prospective cohort study of half a million males and females aged 40–69 years recruited from across the UK between 2006 and 2010.²³ The protocol of UK Biobank is available online (https://www.ukbiobank.ac.uk/wp-content/uploads/2011/11/UK-Biobank-Protocol.pdf). Potential participants were identified from National Health Service patient registers. They were invited to participate in baseline assessments at 22 centres across England, Scotland, and Wales, including a touchscreen questionnaire, verbal interview, physical measures, and biosample collection. Permission for access to patient records for recruitment was approved by the National Information Governance Board for Health and Social Care in England and Wales, and the Community Health Index Advisory Group in Scotland. All participants gave written informed consent prior to participate in the study. The present study included 210 469 participants who had completed at least 1 time web-based 24 h dietary assessment during 2009–2012 and had information on follow-up disease events based on the hospital records and the death registry (until January 31, 2018). We excluded participants with undesirable energy intake (n = 5496). In addition, participants who had a history of cancer (n = 17 119) or CVD (n = 8216), or baseline depressive symptoms (n = 9832) were also excluded. Our final analytic sample included 169 806 participants (mean ± standard deviation (SD) age: 55.7 ± 7.9 years; men, 76 691 (45.2%)).

2.2 Assessment of depressive symptoms

In the TCLSIH cohort, depressive symptoms were evaluated by the Chinese version of the Self-Rating Depression Scale (SDS), one of the most widely used for the assessment of depression in Chinese population.²⁴ It consists of 20 questions either positive-symptom or negative-symptom. Each item was scored on a scale of 1 to 4 points, with the summary scores ranging from 20 to 80 points. The greater values indicate increased severity of depressive symptoms. In this study, 45 was used to determine depressive symptoms. In addition, participants who were taking antidepressants were also considered to have depressive symptoms.

In the UK biobank, depressive symptoms were determined through the linked hospital admission and diagnosis records. All diagnoses related to depression were recorded using the International Classification of Diseases (version 10; code ICD-10) coding system. Depression cases were identified by screening the participants in the UK Biobank who has the depression-related ICD-10 codes (F32 and F33).

2.3 Assessment of onion intake

In the TCLSIH cohort, dietary intake was assessed by a structured and validated food frequency questionnaire (FFQ) including 100-item foods (the initial version contained 81-item food until 2015) with specified serving sizes. The FFQ consisted of 7 categories for foods (ranging from almost never eat to two or more times per day) and 8 categories for beverages (ranging from seldom drink to four or more cups per day). Participants were asked to choose the average consumption of every food and beverage item over the past month. An ad hoc computer program was used to calculate the intake of dietary nutrients and energy based on the Chinese Food Composition Tables.²⁵ The reproducibility and validity of FFQ were assessed in a random sample of 150 participants of the TCLSIH study by comparing the data from repeat measures approximately 3 months apart and 4-d weighed dietary records (WDR).²¹ In brief, Spearman correlation coefficients between the FFQ and the dietary records were 0.49 for total energy and 0.69 for onions. Spearman rank correlation coefficients between the two FFQs 3 months apart were 0.68 for total energy and 0.72 for onions. We created a healthy diet score by combining five common elements of healthy dietary patterns, which included vegetables, fruits, fish, unprocessed red meat, and processed meat.²⁶ Total scores range from 0 to 5, with higher values indicating healthier dietary habits. To characterize overall dietary patterns, factor analysis was used to extract dietary patterns and to determine factor loadings for each food item (in g d⁻¹). Three dietary patterns were extracted, including “healthy” dietary pattern, “sweets” dietary pattern, and “animal foods” dietary pattern, the details of which were described previously.²⁷

Onion intake was assessed by the question: “during the past one month, how often on average did you eat onions”. Raw and cooked onions were not measured separately. Participants are needed to choose one of the seven options: almost never eat, <1 time per week, 1 time per week, 2–3 times per week, 4–6 times per week, 1 time per day, and ≥2 times per week. Intake of food items (g per day), including onion, was calculated by multiplying the portion size (g per times) by the frequency of each food item consumed per day (times per day). To correct for potential measurement error, total onion consumption was adjusted for total energy intake according to the nutrient density method and expressed as (g per 1000 kcal per d).²⁸ For further analysis, energy-adjusted onion intake was categorized into four levels: because 14.8% of participants “almost never” ate onion, we defined onion intake which was equal to “0” (g per (1000 kcal d)) as reference (level 1); and we additionally divided the remaining onion intake which was over than “0” (g per 1000 kcal per d) into tertiles (the lowest tertile: level 2, the middle tertile: level 3, and the highest tertile: level 4).

In the UK Biobank, dietary intake was assessed by 5 web-based 24 h dietary recalls (Oxford WebQ), which included up to 206-item foods and 32-item beverages frequently consumed in the UK. Participants were asked to select the number of portions consumed from each food and beverage during the previous 24 hours. The first Oxford WebQ was completed at baseline in the assessment centers, and participants were invited to complete the remaining 4 from 2011 to 2012 at estimated intervals of 6 months through e-mail. Total energy and nutrients were calculated with standard composition food tables in the United Kingdom.²⁹ The validity of the Oxford WebQ was assessed by an interviewer administered 24-hour dietary recall, with Spearman's correlation coefficients for the majority of nutrients calculated from the web-based 24-hour dietary assessment ranging between 0.5 and 0.9 (median of 0.6).³⁰

Onion consumption was assessed by asking: “How many servings of onions (red, white, pickled, shallots, spring) did you have during the previous 24 hours” (This question was only asked to participants who reported consuming vegetables). Similar to the TCLSIH cohort, raw and cooked onions were not assessed separately in the UK Biobank, either. Participants were needed to select one of the six alternatives: none, quarter, half, 1, 2, 3 + (servings per day). The mean frequency of onion consumption (servings per day) was calculated based on the available data. The mean value of onion intake (g per day) was calculated by multiplying the portion size (g per serving) by the mean frequency of onion consumption (servings per day). Also, onion intake was adjusted for total energy intake (g per (1000 kcal d)). In the main analysis, because 66.3% of participants reported no onion consumption, we set “none” onion consumption as the reference group (level 1); and the rest of the participants with onion consumption were categorized into tertiles (the lowest tertile: level 2, the middle tertile: level 3, and the highest tertile: level 4) with the same method used in the TCLSIH Cohort Study.

2.4 Assessment of covariate

In the TCLSIH cohort, baseline height, weight, and waist circumference (WC) were assessed by trained and experienced staff using a standard protocol. Body mass index (BMI) was calculated as the body weight (kg) divided by the square of the body height (m). Data on sex, age, sociodemographic characteristics (marital status, education level, household income, and occupation), lifestyles (drinking status, smoking status, living alone or not, and friends visiting), and family and personal disease histories, were obtained by questionnaire survey. Physical activity (PA) was assessed using the short form of the validated International Physical Activity Questionnaire (IPAQ),³¹ which consisted of walking, moderate activity (e.g. riding, playing table tennis, or household activity), vigorous activity (e.g. running, swimming, playing basketball, and other sports). Total PA levels were assessed by combining separate scores for different activities and expressed as metabolic equivalent hours per week (MET-h per week).

Blood samples were collected from the antecubital vein in siliconized vacuum plastic tubes after an overnight fast. The methods for the measurement of fasting blood glucose (FBG), lipid, triglycerides (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL), high-density lipoprotein cholesterol (HDL) were described in our previous studies.³² Blood pressure (BP) was measured twice from the upper left arm using a device of TM-2655P (A&D Company, Ltd., Tokyo, Japan) and calculated the mean of the two data. The participants would be diagnosed with hypertension if they had systolic BP ≥ 140 mmHg and/or diastolic BP ≥ 90 mmHg, or had a history of hypertension, or had use of anti-hypertension medication. Diabetes was defined as FBG level ≥ 7 mmol L⁻¹ or diagnosed with diabetes by a physician or having self-reported diabetes. The diagnosis of hyperlipidemia was defined as a TC level ≥ 5.17 mmol L⁻¹ or TG level ≥ 1.7 mmol L⁻¹ or LDL-C level ≥ 3.37 mmol L⁻¹ or diagnosed hyperlipidemia or use of anti-hyperlipidemic medication.

In the UK Biobank, information on sex, age, BMI, sociodemographic characteristics (education levels and ethnic background), lifestyle factors (smoking status, drinking status, living alone or not, friends visiting, and diet quality), and family history of disease (hypertension and diabetes) was collected by touchscreen questionnaire survey. Anthropometric indices (such as weight and height) were assessed by trained staff. PA was assessed by IPAQ.³¹ Townsend area deprivation index was derived from the postcode of residence using aggregated data on unemployment, car and homeownership, and household overcrowding.³³ Comorbidities (hypertension, diabetes, cancer, and CVD) and medical history were based on self-report of a doctor diagnosis and verified during the face-to-face interview. A healthy diet score was used to evaluate the overall food combination effect and was calculated by using the following factors: vegetables intake at least four tablespoons each day (median); fruits intake at least three pieces each day (median); fish intake at least twice each week (median); processed meat intake no more than twice each week (median); and processed meat intake no more than twice each week (median). Each one point was given for each favorable diet factor, with the total diet score ranging from 0 to 5.

2.5 Statistical analysis

Continuous variables were described as medians (P25, P75) and categorical values were presented as percentages (%). In the TCLSIH cohort, follow-up time was calculated from the date of completion for the baseline FFQ survey to the date of the first occurrence of depressive symptoms, the endpoint of follow-up (December 2018), or loss to follow-up, whichever came first. In the UK Biobank, follow-up time was assessed from the date of diet questionnaire completion until the time of the first occurrence of depressive symptoms, the time of death, drop out, or the end of follow-up (January 31 2018), whichever was earliest. Multivariable Cox proportional hazards regression models were fitted to assess the association between onion consumption and risk of depressive symptoms, with level 1 as reference. Three models were fitted. Model 1 was a crude model. Model 2 was adjusted for age, sex, and BMI. Model 3 was further adjusted for smoking status, drinking status, education level, occupation (only in the TCLSIH cohort), household income (only in the TCLSIH cohort), total energy intake, Townsend deprivation index (only in the UK Biobank), physical activity, ethnic background (only in the UK Biobank), marital status (only in the TCLSIH cohort), visiting friends, living alone, hyperlipemia (only in the TCLSIH cohort), hypertension, diabetes, family history of disease (including cardiovascular disease, hypertension, hyperlipidemia [only in the TCLSIH cohort], and diabetes), baseline SDS scores (only in the TCLSIH cohort), and healthy diet scores. To examine the linear trend, the categories of onion consumption (level 1: 1, level 2: 2, level 3: 3, and level 4: 4) were used as a continuous variable in the Cox models. I² and P values for heterogeneity test were calculated to assess the heterogeneity in the associations between the two cohorts. Statistical analyses for the heterogeneity tests were performed using Stata version 11.

To examine the robustness of our results, we performed several sensitivity analyses: (1) we included participants who had a history of cancer or CVD; (2) we excluded participants with dietary changes during the follow-up period; (3) because participants in the UK Biobank aged 40–69 years, we repeated the main analysis in the TCLSIH by removing participants who were younger than 40 or older than 69 years old; (4) in the TCLSIH cohort, we adjusted for three dietary patterns instead of a health dietary score.

All statistical analyses were performed by using SAS software, version 9.4 (SAS Institute Inc., Cary, NC, USA). A two-tailed P value less than 0.05 was considered statistically significant.

3. Results

In the TCLSIH cohort, the number of 7739 participants (men, 57.6%) completed the follow-up from May 2013 to December 2018. During a median follow-up of 2.0 years (ranging from 1.0 to 5.0 years) and 15 004 person-years, we documented 1098 incident cases of depressive symptoms (incidence rate per 1000 person-years = 73.2 cases). In the UK Biobank, 169 806 participants (men, 45.2%) were enrolled, with a median follow-up of 7.3 years (ranging from 1.0 to 9.0 years) and 1 243 832 person-years, 1924 participants were identified to have depressive symptoms (incidence rate per 1000 person-years = 1.55 cases).

Table 1 displays the baseline characteristics of participants from the TCLSIH cohort and the UK Biobank. The mean age (P25, P75) (years) was 37.3 (31.5, 45.8) in the TCLSIH cohort, while it was 57.0 (49.0, 62.0) in the UK Biobank. In addition, the mean values of BMI (kg m⁻²), PA (MET h per week), and healthy diet scores were 24.2 (21.7, 26.8), 11.6 (4.43, 24.1), and 2.00 (2.00, 3.00) in the TCLSIH cohort; and were 26.1 (23.7, 29.1), 29.1 (13.8, 55.3), and 3.00 (2.00, 4.00) in the UK Biobank. The baseline prevalence of hypertension and diabetes was 44.6% and 4.73% in the TCLSIH cohort and was 21.6% and 2.88% in the UK Biobank. The baseline characteristics according to four levels of onion consumption were presented in ESI Table 1.†

Table 1 Baseline characteristics of participants in the TCLSIH cohort study (n = 7739) and the UK Biobank (n = 169 806)^a

Characteristics	All	Males	Females
a Abbreviations: TCLSIH, tianjin chronic low-grade systemic inflammation and health; BMI, body mass index; WC, waist circumference; TC, total cholesterol; TG, triglycerides; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; FBG fasting blood glucose; SBP, systolic blood pressure; DBP, diastolic blood pressure; PA, physical activity; MET, metabolic equivalent; SDS, self-rating depression scale; CVD, cardiovascular disease. b Continuous variables are expressed as medians (P25 P75) and categorical variables are expressed as percentages.
The TCLSIH
No. of subjects	7739	4459 (57.6%)	3280 (42.4%)
Age (years)	37.3 (31.5, 45.8)^b	38.4 (32.2, 47.0)	35.8 (31.0, 44.3)
Married (%)	84.2	85.4	82.7
Education level (college or higher, %)	80.8	81.3	80.0
Occupation (%)
Managers	51.8	51.9	51.6
Professionals	18.7	22.1	14.0
Other	29.6	26.0	34.4
Household income (10 000 Yuan, %)	46.4	46.3	46.6
BMI (kg m^−2)	24.2 (21.7, 26.8)	25.5 (23.3, 27.8)	22.2 (20.4, 24.6)
WC (cm)	82.0 (74.0, 90.0)	88.0 (81.0, 94.0)	73.0 (68.0, 80.0)
TC (mmol L^−1)	4.63 (4.09, 5.24)	4.70 (4.15, 5.30)	4.54 (4.03, 5.14)
TG (mmol L^−1)	1.07 (0.75, 1.63)	1.32 (0.92, 1.95)	0.84 (0.63, 1.17)
LDL-C (mmol L^−1)	2.69 (2.20, 3.23)	2.80 (2.31, 3.33)	2.54 (2.10, 3.05)
HDL-C (mmol L^−1)	1.32 (1.10, 1.60)	1.20 (1.02, 1.41)	1.54 (1.30, 1.81)
FBG (mmol L^−1)	5.00 (4.70, 5.30)	5.10 (4.80, 5.40)	4.80 (4.60, 5.10)
SBP (mmHg)	120.0 (110.0, 130.0)	125.0 (115.0, 135.0)	115.0 (105.0, 125.0)
DBP (mmHg)	75.0 (70.0, 85.0)	80.0 (70.0, 85.0)	70.0 (65.0, 80.0)
Total energy intake (kcal day^−1)	2299.6 (1815.8, 2914.9)	2437.0 (1931.1, 3053.8)	2128.4 (1679.7, 2668.9)
PA (MET × hour per week)	11.6 (4.43, 24.1)	13.6 (5.50, 27.8)	10.4 (3.85, 20.7)
Healthy diet scores	2.00 (2.00, 3.00)	2.00 (1.00, 3.00)	2.00 (2.00, 3.00)
Baseline SDS scores	35.0 (30.0, 39.0)	34.0 (30.0, 39.0)	35.0 (30.0, 39.0)
Smoking status (%)
Current smoker	18.0	30.9	0.71
Ex-smoker	5.45	9.10	0.56
Non-smoker	76.5	60.0	98.7
Drinking status (%)
Everyday	3.11	5.02	0.56
Sometime	61.4	76.5	41.2
Ex-drinker	8.94	8.77	9.17
Non-drinker	26.6	9.74	49.1
Living alone (%)	7.22	8.07	6.07
Visiting friends (%)	55.2	50.7	61.3
Individual history of disease (%)
Hypertension	21.5	29.9	10.0
Hyperlipidemia	44.6	52.0	34.2
Diabetes	4.73	6.48	2.35
Family history of disease (%)
CVD	33.3	32.7	34.2
Hypertension	54.5	54.0	55.3
Hyperlipidemia	0.45	0.43	0.49
Diabetes	28.8	27.2	31.0
The UK Biobank
No. of subjects	169 806	76 691 (45.2%)	93 115 (54.8%)
Age (years)	57.0 (49.0, 62.0)	57.0 (50.0, 63.0)	56.0 (49.0, 62.0)
Ethnic background (%)
White	91.1	91.8	90.6
Mixed	3.15	3.38	2.96
Asian	4.20	3.46	4.80
Black	0.49	0.45	0.52
Chinese	0.31	0.24	0.36
Others	0.73	0.64	0.81
Townsend deprivation index	−2.35 (−3.75, −0.02)	−2.39 (−3.78, −0.08)	−2.31 (−3.72, 0.03)
Education level (college or higher, %)	14.8	15.9	13.8
BMI (kg m^−2)	26.1 (23.7, 29.1)	26.8 (24.6, 29.4)	25.4 (23.0, 28.8)
Total energy intake (kcal day^−1)	2039.0 (1720.1, 2404.7)	2206.7 (1866.5, 2575.4)	1918.8 (1633.4, 2239.7)
PA (MET × hour per week)	29.1 (13.8, 55.3)	29.2 (13.7, 56.3)	29.0 (13.9, 54.6)
Healthy diet scores	3.00 (2.00, 4.00)	3.00 (2.00, 4.00)	3.00 (2.00, 4.00)
Smoking status (%)
Current smoker	7.52	9.08	6.23
Ex-smoker	34.4	37.7	31.8
Non-smoker	58.1	53.2	62.0
Drinking status (%)
Current drinker	94.2	95.5	93.2
Ex-drinker	2.66	2.54	2.76
Non-drinker	3.10	1.96	4.05
Living alone	16.6	15.1	17.9
Visiting friends (%)	75.7	71.1	79.5
Individual history of disease (%)
Hypertension	21.6	25.3	18.5
Diabetes	2.88	4.03	1.93
Family history of disease (%)
CVD	55.3	53.2	57.0
Hypertension	49.1	44.4	53.0
Diabetes	20.9	20.1	21.5

---

The association between onion consumption and risk of depressive symptoms was shown in Table 2. In the crude model, the unadjusted hazard ratios (HRs) and 95% CI of depressive symptoms comparing the higher levels to the lowest level of onion consumption were 0.77 (0.65, 0.92), 0.69 (0.57, 0.82), and 0.73 (0.61, 0.87) in the TCLSIH cohort (p for trend = 0.001); and were 0.76 (0.65, 0.89), 0.82 (0.70, 0.95), and 1.05 (0.91, 1.20) in the UK Biobank (p for trend = 0.30). These associations were not subsequently changed after adjustments for age, BMI, and sex in two cohorts. After further adjusting for demographic and socioeconomic characteristics, lifestyle factors, and healthy diet scores, the fully adjusted HRs (95% CI) across depressive symptoms were 1.00 (reference) for level 1, 0.78 (0.65, 0.94) for level 2, 0.73 (0.61, 0.87) for level 3, and 0.77 (0.64, 0.92) for level 4 in the TCLSIH cohort (p for trend = 0.01); and were 1.00 (reference) for level 1, 0.79 (0.68, 0.93) for level 2, 0.81 (0.69, 0.94) for level 3, and 0.97 (0.85, 1.12) for level 4 in the UK Biobank (p for trend = 0.07). In addition, heterogeneity was observed in the associations between two cohorts (P for heterogeneity = 0.04 and I² = 87.7%).

Table 2 The association between onion consumption and risk of depressive symptoms in the TCLSIH cohort study (n = 7739) and the UK Biobank (n = 169 806)^a

	Category of onion consumption (g per 1000 kcal per d)				P for trend ^b
	Level 1	Level 2	Level 3	Level 4
a Abbreviations: TCLSIH, Tianjin Chronic Low-grade Systemic Inflammation and Health; BMI, body mass index; SDS, self-rating depression scale. b Based on categories of onion consumption as a continuous variable. c Model 1 was a crude model. d Model 2 was further adjusted for age, sex, and BMI. e Model 3 was further adjusted for smoking status, drinking status, education level, occupation (only in the TCLSIH cohort), household income (only in the TCLSIH cohort), Townsend deprivation index (only in the UK Biobank), total energy intake, physical activity, ethnic background (only in the UK Biobank), individual history of disease (including hypertension, hyperlipidemia [only in the TCLSIH cohort], and diabetes), friends visiting, living alone, marital status (only in the TCLSIH cohort), family history of disease (including cardiovascular disease, hypertension, hyperlipidemia [only in the TCLSIH cohort], and diabetes), baseline SDS score (only in the TCLSIH cohort), and healthy diet scores. f Hazard ratios (95% confidence intervals) (all such values).
The TCLSIH cohort
Medians	0.00	1.19	2.16	4.91	—
Participants, n	1143	2194	2201	2201	—
Cases, n	203	315	278	302	—
Person-years	2127	4295	4245	4337	—
Model 1^c	1.00 (reference)	0.77 (0.65, 0.92)^f	0.69 (0.57, 0.82)	0.73 (0.61, 0.87)	0.001
Model 2^d	1.00 (reference)	0.77 (0.65, 0.92)	0.69 (0.57, 0.83)	0.73 (0.61, 0.87)	0.001
Model 3^e	1.00 (reference)	0.78 (0.65, 0.94)	0.73 (0.61, 0.87)	0.77 (0.64, 0.92)	0.01
The UK Biobank
Medians	0.00	3.66	8.83	21.4	—
Participants, n	112 649	19 036	19 051	19 070	—
Cases, n	1331	173	185	235	—
Person-years	822 823	141 350	141 131	138 528	—
Model 1^c	1.00 (reference)	0.76 (0.65, 0.89)	0.82 (0.70, 0.95)	1.05 (0.91, 1.20)	0.30
Model 2^d	1.00 (reference)	0.78 (0.67, 0.91)	0.80 (0.68, 0.93)	0.97 (0.85, 1.12)	0.05
Model 3^e	1.00 (reference)	0.79 (0.68, 0.93)	0.81 (0.69, 0.94)	0.97 (0.85, 1.12)	0.07

---

In sensitivity analysis, when we included participants with a history of cancer or CVD, the associations were sustained in two cohorts (ESI Table 2†). The observed associations were not materially changed in two cohorts when excluding individuals who reported dietary changes during the follow-up (ESI Table 3†). Likewise, the association did not substantially change when we restricted the population of the TCLSIH cohort to the ages of 40–69 years old (ESI Table 4†). Adjusting for three dietary patterns instead of a healthy diet score did not materially alter the association, and the corresponding HRs (95% CIs) for onion consumption were 1.00 (reference), 0.78 (0.65, 0.94), 0.71 (0.59, 0.86), and 0.76 (0.63, 0.91) in the TCLSIH cohort (p for trend = 0.008).

4. Discussion

In the present study, we explored the association between onion consumption and risk of depressive symptoms in two large cohorts from China and the UK. Our results showed that habitual onion consumption was inversely associated with a lower risk of depressive symptoms in these two cohorts after adjusting for multiple traditional risk factors. However, this association was not statistically significant in the highest level of onion consumption in the UK Biobank. Several sensitivity analyses confirmed the robustness of our results.

Onions have been demonstrated to have the potential to reduce the risks of various depression-comorbid diseases, such as cancer, CVD, stress, and neurodegenerative disease.¹³ However, available evidence related to onion consumption and depressive symptoms is limited, especially data from epidemiologic studies. Preliminary evidence from one clinical trial suggested the benefits of quercetin-rich onion for improving depressive symptoms.²⁰ However, this study was limited to small sample size, specific population, and short research period. Overall, there is no comparable evidence with respect to whether long-term habitual onion intake affects the risk of depressive symptoms. To the best of our knowledge, this is the first study to evaluate the longitudinal association between onion consumption and risk of depressive symptoms in two large cohorts. Our results partly confirmed previous findings, suggesting an inverse association between habitual onion consumption and risk of depressive symptoms. Nevertheless, these findings need to be replicated in further investigations.

In the UK Biobank, 66.3% of participants reported no onion intake in the dietary assessment, while only 14.8% of participants seldom ate onion in the TCLSIH cohort. In addition, we observed a higher incidence of depressive symptoms in the TCLSIH cohort than in the UK Biobank (cases per 1000 person-years: 73.2 vs. 1.55 for TCLSIH vs. UK Biobank). These differences between the two cohorts might be attributed to the heterogeneity (P for heterogeneity = 0.04 and I² = 87.7%). This heterogeneity could be partly explained by the different races, genetic backgrounds, dietary habits, and cultures in the populations studied,³⁴ different measurements for onion intake, and different methods to define depressive symptoms. For example, onion intake was assessed by 24 h dietary recalls in the UK Biobank and FFQ in the TCLSIH cohort, respectively. Compared to 24 h diet recalls, FFQ can better reflect the long-term dietary intake.^35,36 The diagnosis of depressive symptoms in the UK Biobank was determined through the medical records rather than an active following-up, which might miss milder cases of depressive symptoms or other patients who never went to the hospital. Racial and ethnic differences play an important role in the development and progression of depressive symptoms because psychological impairment may be both expressed and influenced largely by local sociocultural and environmental factors.³⁷ Nevertheless, the results were fairly consistent among the two cohorts, indicating that the association between onion consumption and risk of depressive symptoms was not limited to countries or ethnicities with different cultures and dietary habits.

Several plausible mechanisms may explain the inverse association between habitual onion consumption and risk of depressive symptoms. First, compelling evidence supports a protective role for quercetin on depressive symptoms with direct or indirect mechanisms. For direct mechanisms, quercetin may modulate signaling pathways responsible for suppressing neural apoptosis³⁸ and promoting brain-derived neurotrophic factor (BDNF).³⁹ For indirect mechanisms, quercetin may exert antidepressant effects through reducing neuroinflammation and oxidative stress, inhibiting the activity of monoamine oxidase enzyme A, and suppressing hypothalamic-pituitary-adrenal axis activity.⁴⁰ In addition, onions contain a high content of dietary fiber,^41,42 which are also known as prebiotics.¹⁴ Dietary fiber has been shown to maintain the integrity of intestinal microbiota and normalize the microbiota-gut-brain axis communication,^14,15 the imbalance of which might increase the state of depressive symptoms.⁴³ Finally, the other bioactive compounds in onion, such as vitamin B6, folic acid, and magnesium, may also contribute to the beneficial effects of onion on depressive symptoms.^16,17 Therefore, the beneficial association between onion consumption and depressive symptoms is biologically plausible.

The inverse association between the highest level of onion consumption and risk of depressive symptoms was not statistically significant in the UK Biobank. In addition, in the TCLSIH cohort, we observed a slight increasement in HR of the level 4 participants compared to level 3 in the TCLSIH (HRs: 0.76 vs. 0.71 for level 4 vs. level 3). It seemed that excessive onion consumption might affect this inverse association. Normally, onion and its bioactive compounds are quite safe for humans.⁴⁴ However, several potential risks from onion consumption have raised many concerns, for instance, the residue of pesticides and heavy metals.⁴⁴ Exposure to these contaminants may impair the monoaminergic neurotransmission system, disturb the central nervous system, and cause neurotoxic effects.^45,46 In addition, quercetin, as the prominent bioactive compound in onion, was found to promote the serum levels of substance P.⁴⁷ This substance plays an important role in the pathophysiology of mood disorders^48,49 and is often higher in individuals with depressive symptoms.^50,51 Further randomized clinical trials are required to examine the exact association between an excessive amount of onion intake and mental health.

The strength of the present study was the use of two large-scale prospective cohorts from countries with different amounts and proportions of consumption of onion, as well as different ethnic backgrounds, socioeconomic characteristics, and lifestyles. Additional strengths include the long follow-up period, detailed measurement of diets and lifestyle factors, careful adjustment for a wide range of potential confounders, and robust sensitivity analyses.

This study also had some limitations. First, dietary information was self-reported in two cohorts. Thus, measurement error and recall bias were inevitable. Second, we did not assess raw and cooked onions separately in this study. Although quercetin conjugates in onions are remarkably resistant to degradation during normal processing operations (e.g., refrigerated storage, oven roasting, moderate microwave, and frying, except boiling),^52,53 culinary treatments, especially heating process, can influence the activity and contents of other nutrients (e.g., folic acid)⁵⁴ and may further affect the observed associations. In Tianjin, people often eat raw, cooked, or a mixture of raw and cooked onions,⁵⁵ which is similar in UK.^56–58 However, because data on onion cooking methods were not available in the present study, it is not possible to determine consumption ratios. Future studies are warranted to determine the effects of raw and cooked onions on depressive symptoms. Third, measurement of onion consumption relied on the baseline dietary records; thus, measurement bias was inevitable and dietary changes might occur during the follow-up period. However, the observed associations remained sustained in two cohorts in the sensitivity analysis excluding participants who had dietary changes during the follow-up. Forth, the diagnosis of depressive symptoms in the TCLSIH cohort was determined by SDS scores. This may lead to misclassification bias, though the survey was performed by trained interviewers and the SDS has been widely used in many studies. Fifth, in the UK Biobank, depressive symptoms were diagnosed by hospital inpatient records, which might miss milder cases or severely depressed patients who never went to the hospital unless their lives were threatened. Sixth, although a lot of confounding factors had been adjusted, we could not rule out residual confounding by other unmeasured or unknown factors. Finally, given the nature of observational studies, we cannot infer the causality between onion consumption and depressive symptoms.

5. Conclusion

In conclusion, results from these two cohorts suggested that habitual onion consumption was associated with a lower risk of depressive symptoms in the general population in China and the UK. However, this association was not statistically significant in the highest level of onion consumption in the UK Biobank. Randomized clinical trials are needed to establish causal and mechanistic associations between the roles of onion consumption and depressive symptoms. In addition, it will be interesting to evaluate the effects of different flavonoid-rich foods on depressive symptoms in future studies.

Abbreviations

BMI	Body mass index
BP	Blood pressure
CI	Confidence interval
CVD	Cardiovascular disease
FBG	Fasting blood glucose
FFQ	Food frequency questionnaire
HDL-C	High-density lipoprotein cholesterol
HR	Hazard ratio
ICD-10	International classification of diseases (version 10)
IPAQ	International physical activity questionnaire
LDL-C	Low-density lipoprotein cholesterol
MET	Metabolic equivalent
OSCs	Organosulfur compounds
PA	Physical activity
SD	Standard deviation
SDS	Self-rating scores
TC	Total cholesterol
TCLSIH	Tianjin chronic low-grade systemic inflammation and health
TG	Triglycerides
WC	Waist circumference
WDR	Weighed dietary records.

Data availability

Data from TCLSIH Cohort Study are available from the corresponding authors on reasonable request.

Data from the UK Biobank are available in a public, open access repository. This research has been conducted using the UK Biobank Resource under Application Number 44430. The UK Biobank data are available on application to the UK Biobank (https://www.ukbiobank.ac.uk/).

Author contributions

The authors’ responsibilities were as follows: H. Y. analyzed the data and wrote the paper. H. Y., Y. G., B. Z., G. M., Q. Z., L. L., H. W., S. Z., T. Z., X. W., J. Z., S. S., X. W., M. Z., Q. J., and K. S. conducted research. Y. W., T. H., and K. N. designed the research and had primary responsibility for the final content. All authors had full access to all the data in the study and read and approved the final manuscript.

Ethics approval

This study was approved by the Institution Review Board of Tianjin Medical University (Ethics Approval Number: TMUhMEC 201430) and the National Health Service National Research Ethics Service (ref11/NW/0382, June 17 2011). The written informed consent was obtained from all participants.

Consent to participate

Written informed consent was obtained from each participant prior to enrollment.

Conflicts of interest

None of the authors has any potential conflict of interest.

Acknowledgements

This study was supported by grants from Tianjin Major Public Health Science and Technology Project (No. 21ZXGWSY00090), National Health Commission of China (No. SPSYYC 2020015), China Cohort Consortium (CCC2020003), the National Natural Science Foundation of China (No. 81872611, 81941024, 82103837 and 81903315), Food Science and Technology Foundation of Chinese Institute of Food Science and Technology (No. 2019-12), 2014 and 2016 Chinese Nutrition Society (CNS) Nutrition Research Foundation—DSM Research Fund (No. 2016-046, 2014-071 and 2016-023), China. The authors greatly acknowledge all the people that have made this study. Data from the UK Biobank are available in a public, open access repository. This research has been conducted using the UK Biobank Resource under Application Number 44430.

References

1. R. M. McCarron, B. Shapiro, J. Rawles and J. Luo, Depression, Ann. Intern. Med., 2021, 174, Itc65–itc80.
2. C. Gómez-Donoso, A. Sánchez-Villegas, M. A. Martínez-González, A. Gea, R. D. Mendonça, F. Lahortiga-Ramos and M. Bes-Rastrollo, Ultra-processed food consumption and the incidence of depression in a Mediterranean cohort: the SUN Project, Eur. J. Nutr., 2020, 59, 1093–1103.
3. D. Chisholm, K. Sweeny, P. Sheehan, B. Rasmussen, F. Smit, P. Cuijpers and S. Saxena, Scaling-up treatment of depression and anxiety: a global return on investment analysis, Lancet Psychiatry, 2016, 3, 415–424.
4. T. N. Ly, C. Hazama, M. Shimoyamada, H. Ando, K. Kato and R. Yamauchi, Antioxidative compounds from the outer scales of onion, J. Agric. Food Chem., 2005, 53, 8183–8189.
5. Q. Li, Y. Wang, Y. Mai, H. Li, Z. Wang, J. Xu and X. He, Health Benefits of the Flavonoids from Onion: Constituents and Their Pronounced Antioxidant and Anti-neuroinflammatory Capacities, J. Agric. Food Chem., 2020, 68, 799–807.
6. R. Slimestad, T. Fossen and I. M. Vågen, Onions: a source of unique dietary flavonoids, J. Agric. Food Chem., 2007, 55, 10067–10080.
7. C. Manach, A. Scalbert, C. Morand, C. Rémésy and L. Jiménez, Polyphenols: food sources and bioavailability, Am. J. Clin. Nutr., 2004, 79, 727–747.
8. K. Vafeiadou, D. Vauzour, H. Y. Lee, A. Rodriguez-Mateos, R. J. Williams and J. P. Spencer, The citrus flavanone naringenin inhibits inflammatory signalling in glial cells and protects against neuroinflammatory injury, Arch. Biochem. Biophys., 2009, 484, 100–109.
9. J. P. Spencer, Flavonoids and brain health: multiple effects underpinned by common mechanisms, Genes Nutr., 2009, 4, 243–250.
10. Y. Miyake, K. Tanaka, H. Okubo, S. Sasaki, S. Furukawa and M. Arakawa, Soy isoflavone intake and prevalence of depressive symptoms during pregnancy in Japan: baseline data from the Kyushu Okinawa Maternal and Child Health Study, Eur. J. Nutr., 2018, 57, 441–450.
11. Y. Yao, H. Chen, L. Chen, S. Y. Ju, H. Yang, Y. Zeng, D. Gu and T. P. Ng, Type of tea consumption and depressive symptoms in Chinese older adults, BMC Geriatr., 2021, 21, 331.
12. J. Fisk, S. Khalid, S. A. Reynolds and C. M. Williams, Effect of 4 weeks daily wild blueberry supplementation on symptoms of depression in adolescents, Br. J. Nutr., 2020, 124, 181–188.
13. H. A. Suleria, M. S. Butt, F. M. Anjum, F. Saeed and N. Khalid, Onion: nature protection against physiological threats, Crit. Rev. Food Sci. Nutr., 2015, 55, 50–66.
14. A. Burokas, S. Arboleya, R. D. Moloney, V. L. Peterson, K. Murphy, G. Clarke, C. Stanton, T. G. Dinan and J. F. Cryan, Targeting the Microbiota-Gut-Brain Axis: Prebiotics Have Anxiolytic and Antidepressant-like Effects and Reverse the Impact of Chronic Stress in Mice, Biol. Psychiatry, 2017, 82, 472–487.
15. Z. Liu, L. Li, S. Ma, J. Ye, H. Zhang, Y. Li, A. T. Sair, J. Pan, X. Liu, X. Li, S. Yan and X. Liu, High-Dietary Fiber Intake Alleviates Antenatal Obesity-Induced Postpartum Depression: Roles of Gut Microbiota and Microbial Metabolite Short-chain Fatty Acid Involved, J. Agric. Food Chem., 2020, 68, 13697–13710.
16. K. Mikkelsen, L. Stojanovska and V. Apostolopoulos, The Effects of Vitamin B in Depression, Curr. Med. Chem., 2016, 23, 4317–4337.
17. C. Sun, R. Wang, Z. Li and D. Zhang, Dietary magnesium intake and risk of depression, J. Affective Disord., 2019, 246, 627–632.
18. H. Sakakibara, S. Yoshino, Y. Kawai and J. Terao, Antidepressant-like effect of onion (Allium cepa L.) powder in a rat behavioral model of depression, Biosci., Biotechnol., Biochem., 2008, 72, 94–100.
19. N. Samad and A. Saleem, Administration of Allium cepa L. bulb attenuates stress-produced anxiety and depression and improves memory in male mice, Metab. Brain Dis., 2018, 33, 271–281.
20. J. Nishihira, M. Nishimura, M. Kurimoto, H. Kagami-Katsuyama, H. Hattori, T. Nakagawa, T. Muro and M. Kobori, The effect of 24-week continuous intake of quercetin-rich onion on age-related cognitive decline in healthy elderly people: a randomized, double-blind, placebo-controlled, parallel-group comparative clinical trial, J. Clin. Biochem. Nutr., 2021, 69, 203–215.
21. S. Zhang, S. Kumari, Y. Gu, X. Wu, X. Li, G. Meng, Q. Zhang, L. Liu, H. Wu, Y. Wang, T. Zhang, X. Wang, X. Cao, H. Li, Y. Liu, X. Wang, S. Sun, X. Wang, M. Zhou, Q. Jia, K. Song, Z. Sun and K. Niu, Soy Food Intake Is Inversely Associated with Newly Diagnosed Nonalcoholic Fatty Liver Disease in the TCLSIH Cohort Study, J. Nutr., 2020, 150, 3280–3287.
22. S. Zhang, Y. Gu, S. Bian, M. J. Górska, Q. Zhang, L. Liu, G. Meng, Z. Yao, H. Wu and Y. Wang, Dietary patterns and risk of non-alcoholic fatty liver disease in adults: a prospective cohort study, Clin. Nutr., 2021, 40, 5373–5382.
23. C. Sudlow, J. Gallacher, N. Allen, V. Beral, P. Burton, J. Danesh, P. Downey, P. Elliott, J. Green, M. Landray, B. Liu, P. Matthews, G. Ong, J. Pell, A. Silman, A. Young, T. Sprosen, T. Peakman and R. Collins, UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age, PLoS Med., 2015, 12, e1001779.
24. H. C. Lee, H. Chiu, Y. K. Wing, C. M. Leung, P. K. Kwong and D. Chung, The Zung Self-Rating Depression Scale: Screening for Depression Among the Hong Kong Chinese Elderly, J. Geriatr. Psychiatry Neurol., 1994, 7, 216–220.
25. Y. Yuexin, W. Guangya and P. Xingchang, China Food Composition, Peking University Medical Press, Beijing, China, 2nd edn, 2009.
26. S. Zhang, Y. Gu, S. Bian, Z. Lu, Q. Zhang, L. Liu, G. Meng, Z. Yao, H. Wu, Y. Wang, T. Zhang, X. Wang, S. Sun, X. Wang, M. Zhou, Q. Jia, K. Song, L. Qi and K. Niu, Soft drink consumption and risk of nonalcoholic fatty liver disease: results from the Tianjin Chronic Low-Grade Systemic Inflammation and Health (TCLSIH) cohort study, Am. J. Clin. Nutr., 2021, 113, 1265–1274.
27. S. Zhang, Y. Gu, S. Bian, M. J. Górska, Q. Zhang, L. Liu, G. Meng, Z. Yao, H. Wu, Y. Wang, T. Zhang, X. Wang, S. Sun, X. Wang, M. Zhou, Q. Jia, K. Song, L. Qi and K. Niu, Dietary patterns and risk of non-alcoholic fatty liver disease in adults: A prospective cohort study, Clin. Nutr., 2021, 40, 5373–5382.
28. W. C. Willett, G. R. Howe and L. H. Kushi, Adjustment for total energy intake in epidemiologic studies, Am. J. Clin. Nutr., 1997, 65, 1220S–1228S.
29. B. Holland, A. Welch, I. Unwin, D. Buss, A. Paul and D. Southgate, McCance and Widdowson's the composition of foods, Royal Society of Chemistry, 1991.
30. B. Liu, H. Young, F. L. Crowe, V. S. Benson, E. A. Spencer, T. J. Key, P. N. Appleby and V. Beral, Development and evaluation of the Oxford WebQ, a low-cost, web-based method for assessment of previous 24 h dietary intakes in large-scale prospective studies, Public Health Nutr., 2011, 14, 1998–2005.
31. C. L. Craig, A. L. Marshall, M. Sjostrom, A. E. Bau man, M. L. Booth, B. E. Ainsworth, M. Pratt, U. Ekelund, A. Yngve and J. F. Sallis, International physical activity questionnaire: 12-country reliability and validity, Med. Sci. Sports Exercise, 2003, 35, 1381–1395.
32. T. Ji, X. Li, G. Meng, Y. Gu, Q. Zhang, L. Liu, H. Wu, Z. Yao, S. Zhang and Y. Wang, The association between banana consumption and the depressive symptoms in Chinese general adult population: A cross-sectional study, J. Affective Disord., 2020, 264, 1–6.
33. P. Townsend, P. Phillimore and A. Beattie, Health and deprivation: inequality and the North, Routledge, 1988.
34. P. G. Kittler, K. P. Sucher and M. Nelms, Food and culture, Cengage Learning, 2016.
35. W. C. Willett, Commentary: Dietary diaries versus food frequency questionnaires—a case of undigestible data, Int. J. Epidemiol., 2001, 30, 317–319.
36. W. C. Willett, Nutritional epidemiology, Oxford University Press, 2012.
37. N. Iwata and S. Buka, Race/ethnicity and depressive symptoms: a cross-cultural/ethnic comparison among university students in East Asia, North and South America, Soc. Sci. Med., 2002, 55, 2243–2252.
38. F. Dajas, F. Rivera-Megret, F. Blasina, F. Arredondo, J. A. Abin-Carriquiry, G. Costa, C. Echeverry, L. Lafon, H. Heizen, M. Ferreira and A. Morquio, Neuroprotection by flavonoids, Braz. J. Med. Biol. Res., 2003, 36, 1613–1620.
39. K. Fang, H. R. Li, X. X. Chen, X. R. Gao, L. L. Huang, A. Q. Du, C. Jiang, H. Li and J. F. Ge, Quercetin Alleviates LPS-Induced Depression-Like Behavior in Rats via Regulating BDNF-Related Imbalance of Copine 6 and TREM1/2 in the Hippocampus and PFC, Front. Pharmacol., 2019, 10, 1544.
40. L. Pathak, Y. Agrawal and A. Dhir, Natural polyphenols in the management of major depression, Expert Opin. Invest. Drugs, 2013, 22, 863–880.
41. R. Kahane, E. Vialle-Guérin, I. I. Boukema, D. Tzanoudakis, C. Bellamy, C. Chamaux and C. Kik, Changes in non-structural carbohydrate composition during bulbing in sweet and high-solid onions in field experiments, Environ. Exp. Bot., 2001, 45, 73–83.
42. N. Benkeblia, Free-radical scavenging capacity and antioxidant properties of some selected onions (Allium cepa L.) and garlic (Allium sativum L.) extracts, Braz. Arch. Biol. Technol., 2005, 48, 753–759.
43. T. G. Dinan and J. F. Cryan, The Microbiome-Gut-Brain Axis in Health and Disease, Gastroenterol. Clin. North Am., 2017, 46, 77–89.
44. X. X. Zhao, F. J. Lin, H. Li, H. B. Li, D. T. Wu, F. Geng, W. Ma, Y. Wang, B. H. Miao and R. Y. Gan, Recent Advances in Bioactive Compounds, Health Functions, and Safety Concerns of Onion (Allium cepa L.), Front. Nutr., 2021, 8, 669805.
45. A. M. Cybulska, S. Grochans, M. S. Kamińska, M. Bosiacki, K. Skonieczna-Żydecka and E. Grochans, Are cadmium and lead levels linked to the development of anxiety and depression? - A systematic review of observational studies, Ecotoxicol. Environ. Saf., 2021, 216, 112211.
46. A. Arab and S. Mostafalou, Neurotoxicity of pesticides in the context of CNS chronic diseases, Int. J. Environ. Health Res., 2021, 1–38.
47. M. Nicolau, S. S. Dovichi and G. Cuttle, Pro-inflammatory effect of quercetin by dual blockade of angiotensin converting-enzyme and neutral endopeptidase in vivo, Nutr. Neurosci., 2003, 6, 309–316.
48. K. Ranga and R. Krishnan, Clinical experience with substance P receptor (NK1) antagonists in depression, J. Clin. Psychiatry, 2002, 63(Suppl 11), 25–29.
49. Q. P. Ma and C. Bleasdale, Modulation of brain stem monoamines and gamma-aminobutyric acid by NK1 receptors in rats, NeuroReport, 2002, 13, 1809–1812.
50. L. Li, X. Gao, J. Zhao, X. Ji, H. Wei and Y. Luo, Plasma and cerebrospinal fluid substance P in post-stroke patients with depression, Psychiatry Clin. Neurosci., 2009, 63, 298–304.
51. T. D. Geracioti Jr., L. L. Carpenter, M. J. Owens, D. G. Baker, N. N. Ekhator, P. S. Horn, J. R. Strawn, G. Sanacora, B. Kinkead, L. H. Price and C. B. Nemeroff, Elevated cerebrospinal fluid substance p concentrations in posttraumatic stress disorder and major depression, Am. J. Psychiatry, 2006, 163, 637–643.
52. A. S. Rodrigues, M. R. Perez-Gregorio, M. S. Garcia-Falcon and J. Simal-Gandara, Effect of curing and cooking on flavonols and anthocyanins in traditional varieties of onion bulbs, Food Res. Int., 2009, 42, 1331–1336.
53. K. Ioku, Y. Aoyama, A. Tokuno, J. Terao, N. Nakatani and Y. Takei, Various cooking methods and the flavonoid content in onion, J. Nutr. Sci. Vitaminol., 2001, 47, 78–83.
54. Y. Shulpekova, V. Nechaev, S. Kardasheva, A. Sedova, A. Kurbatova, E. Bueverova, A. Kopylov, K. Malsagova, J. C. Dlamini and V. Ivashkin, The concept of folic acid in health and disease, Molecules, 2021, 26, 3731.
55. J. Zhang, Y. Gu, G. Meng, Q. Zhang, L. Liu, H. Wu, S. Zhang, Y. Wang, T. Zhang, X. Wang, X. Zhang, X. Wang, S. Sun, M. Zhou, Q. Jia, K. Song and K. Niu, Association between dietary onion intake and subclinical hypothyroidism in adults: a population-based study from an iodine-replete area, Endocrine, 2021, 74, 616–624.
56. T. Crowther, H. A. Collin, B. Smith, A. B. Tomsett, D. O'Connor and M. G. Jones, Assessment of the flavour of fresh uncooked onions by taste–panels and analysis of flavour precursors, pyruvate and sugars, J. Sci. Food Agric., 2005, 85, 112–120.
57. B.-E. Van Wyk, Culinary herbs and spices of the world, University of Chicago Press, 2014.
58. L. Mason, Food Culture in Great Britain, Greenwood Publishing Group, 2004.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2fo01640k

‡ Co-first author with equal contribution.

---