Sample size and study design
It was an individually matched case-control observational study. To extract dietary patterns, 110 patients with depression and 220 healthy individuals were recruited. To measure serum folate, vitamin B12, tHcy, Trp, and Trp/Caa ratio a total of 86 individuals (n=43 equally distributed in case and control groups) were randomly selected from the participants.
People with an age range of 18 to 65 years, resided in Tehran, and who have been diagnosed with the disease for a maximum of 3 months were entered in case group.
People with cognitive impairment or other psychotic illnesses, severe depression or lacking the ability to cooperate and answer the questions, hormonal disorders such as Addison’s, Cushing’s disease, hyperthyroidism, hypothyroidism, hyperparathyroidism, cancer, heart disease, diabetes, stroke, fibromyalgia, kidney, or liver failure, multiple sclerosis, Parkinson disease, a history of trauma, cuts, fractures, bleeding, burns, and other similar events in the past three months resulting in unconsciousness and hospitalization, chronic and infectious diseases such as HIV, mononucleosis, tuberculosis, viral hepatitis, and pneumonia in the past two weeks, those who consumed anti-depression drugs, those who had addiction to alcohol or drugs at the time of the study or in the past three months, BMI ≥40 kg/m2, pregnancy and lactation at the time of the study or in the past year, a special diet in the past two months, any type of special diet for more than two months in the past year, B12 injection more than once in the past six months, B-complex injection more than once per month for at least six months in the last year, and any dietary supplement in injection or oral form in the past three months were excluded from the study.
Recruitment and Matching
Our samples were recruited from two psychiatric clinics in Tehran, capital of Iran. The participants' major depressive disorder(s) was diagnosed by psychiatrists using the criteria of the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) (18). These patients had no history of depression in the past year. For the control group, lack of major depression disorder(s) was confirmed based on Beck Depression Inventory questionnaire (BDI-II) (19), standardized in Iran (20). The control group members had no history of depression in the past year.
To invite eligible individuals, the interviewers went to each patient’s residential area and described the aim of the study. Patients were matched with controls based on gender, age, and residential area. Each patient with depression was matched with two members of the control group.
To identify the dietary patterns, the open Epi software was used (21, 22).
Assessment of covariates
A demographic questionnaire was used to collect information about participants' general characteristics and some confounders. To calculate BMI weight and height were obtained from all participants. For quantitative measurement of anxiety, as a confounder, the Iranian standardized Beck Anxiety Inventory was utilized (23). Participants' dietary intakes in the last 12 months were assessed using a valid and reliable semi-quantitative food frequency questionnaire (FFQ) (24). A validated physical activity questionnaire consisting of nine levels of activity was applied (25, 26).
Assessment of biochemical markers
In order to measure the biochemical markers, 5 ml blood samples were collected from participants after 12 hours of fasting from. The samples were transferred into tubes with no anticoagulant. After centrifuging for 20 min at 1,500 g in room temperature, the serum was separated and stored at -70°C. Serum folate and vitamin B12 were measured by gamma method (SimulTRAC-SNB Radioassay Kit; Becton Dickinson, Orangeburg, NY, USA). Total homocysteine (tHcy) concentration was also analyzed using the modified immuno-enzyme method (Axis-Shield, UK). Plasma levels of amino-acids were determined by a modern HPLC device with fluorescent detector. Initial preparation involved the precipitation of serum protein with methanol. Simultaneous derivatization was performed with ortho-phthaldehyde. After derivatization, 10μl of each sample was injected into the HPLC (High-Performance Liquid Chromatography) device. The device model was YoungLin Acm 3000 HPLC (Young Lin Instrument Co. Ltd., Anyang, Korea). Later, HPLC was carried out on a 4.6 × 250 mm column of Inertsil ODS 5 μm, at 37 °C in the wavelength range of 340–420 nm, which took one hour for each sample. Within- and between-assays precisions were 4.8-6.8 and 6.5-8.5 for all amino acids, respectively.
The normality of covariates was tested by Kolmogorov-smirnov. To compare variables in two groups t-test or Mann-Whitney were used. The association between dietary patterns and depression was investigated by logistic regression models after adjusting for the confounders in multiple logistic regression models. In addition, the goodness of fit for these logistic regression models was examined by logical confidence intervals and Hosmer and Lemeshow test. Furthermore, we used multiple logistic regression to test the mediatory variables (27).
The mediatory analysis was designed to determine the mediator variable, among the biochemical markers, in the relationship between dietary patterns and depression. Logistic regression was employed due to the dichotomous nature of the dependent variable. We designed tree models to determine whether a variable was a mediator or not.
After these analyses, we selected the mediator variables in the case that:
- A significant association was confirmed between dietary patterns and depression
- By adding the mediator variables into the first model, the relationship between dietary patterns and depression did not remain significant. Since dietary pattern is related to depression via mediator variables, by adding mediators into the model, the relationship between dietary patterns and depression is transferred into the relationship between mediator variables and depression. Therefore, a third model should be designed to ensure the significant relationship between the mediator variables and depression.
- In the third model, the relationship between the mediator variables and depression should be significant (27).
In the current study, the mediatory analysis was performed after adjusting for the confounding variables. In other words, the confounding variables were present in all three aforementioned models. All statistical analyses were carried out using SPSS (version 20; Chicago, IL).