Design
Multicenter, two-arm, parallel-group clinical trial. One group of patients (arm 1) is assigned to a Mediterranean diet supplemented with extra-virgin olive oil and the second group (control) (arm 2) has no nutritional intervention. The participants are recruited from four centers across Spain: Hospital Dr. Negrín (Las Palmas de Gran Canaria), Clínica Universidad de Navarra (Pamplona), Hospital Universitario (Vitoria), and Clínica Dr. Chiclana (Madrid). The intervention period lasted two years.
Patient recruitment
In a first phase of the selection process, we extract names of potential participants from the clinical records of the hospitals or health centers who are willing to collaborate with the study. Next, the participating psychologists and psychiatrists review the clinical records individually to identify the subjects that comply with the inclusion criteria. The potential participants are contacted by phone or during a clinical visit. When a candidate agrees to participate, a face-to-face interview with the specialist is carried out to exclude participants who do not meet the criteria. Besides the semi-structured clinical interview (Spanish SCID-I) [
15], participants also self-complete the Beck Depression Inventory to assess depressive symptoms [
16]. Information about the participant is completed with questions from the eligibility questionnaire, such as medical conditions or problems related with the adherence to the Mediterranean diet. During the first visit, participants receive a brief explanation of the study, are informed that are going to be given extra-virgin olive oil for the duration of the trial at no cost, and a signed informed consent is obtained. All study participants are asked to provide blood samples (previous appointment with the hospital) and a subgroup of participants is instructed to also provide urine samples (recruitment centers: Vitoria, Pamplona) and stool samples (recruitment center: Vitoria).
Randomization
Study participants are randomly assigned to one of two groups (Mediterranean diet or control) once their data are included in a centralized computer system by the specialists. Various stratification factors are considered for the randomization, sex, age group (< 65 years or ≥ 65 years) and recruitment center. At baseline, psychiatrists and clinical psychologists are blinded to the allocation of the participants, following the CONSORT guidelines for randomized trials to prevent selection biases.
Intervention with the Mediterranean diet and control group
Full-time registered dietitians with experience in the PREDIMED trial are responsible for the dietary intervention in the PREDI-DEP study. The PREDIMED trial is a landmark trial of intervention with a Mediterranean diet in participants at high risk of cardiovascular disease [
17,
18]. The methodology in PREDI-DEP is similar to that in PREDIMED.
Participants allocated to the Mediterranean diet receive intensive training on Mediterranean diet and supplemental foods (Table
2), as well as extra-virgin olive oil (one liter of extra-virgin olive oil rich in polyphenols every two weeks), at no cost. The authors have no conflict of interest with any food company.
Table 2
Intervention with Mediterranean diet
Free food supply | Quarterly | Collected by participants in medical centers |
Book and information regarding Mediterranean Diet | Initial | Collected by participants in medical centers |
Buying foods list, plan of meals, cooking recipes, menus | Quarterly | Collected by participants in medical centers |
Compliance degree, personalized goal achivement | Quarterly | Telephone |
Mediterranean Diet news | Monthly | Web/e-mail |
Clarification of doubts, suggestions | Any moment | Any instrument |
At the beginning of the trial, the dieticians thoroughly explain the reasons to follow a Mediterranean diet to each participant and negotiate changes in his/her diet, working with the subject to determine what he or she considers an attainable goal. The dietician does this every three months by calling the participant over the phone. To assess adherence to the Mediterranean diet and enhance future adherence (Additional file
1), the dietitians use a validated 14-point Mediterranean Diet Assessment Screener (MEDAS) [
19], which was employed in the PREDIMED trial.
Every three months the participants receive written material with information on key Mediterranean foods and seasonal shopping lists, menus and specific recipes for a typical week. This material is discussed in detail with the dietitians.
The website developed for this study (
https://predidep.es/participantes/) is updated monthly with news related to the Mediterranean diet and its health effects. The dieticians take into account any doubt or suggestion made by the participants at any time of the intervention period.
No nutritional intervention is employed with the control group. These participants have an email and access to the website of the project, but with limitations on the contents related with the intervention. To prevent these subjects from withdrawing from a study an incentive at trial termination is offered.
Assessment of exposure
The dieticians have the first contact with the participants over the phone to inform them in which arm they have been allocated and give instructions to the subjects who have been assigned to the Mediterranean diet as where and when to collect the food supply. Moreover, the dieticians complete the information for each participant through: 1) a general questionnaire; 2) a physical activity questionnaire; 3) a semi-quantitative food frequency questionnaire (FFQ); 3) the 14-Item Mediterranean Diet Adherence Screener (MEDAS), and 4) the Short Form-36 Quality of Life Questionnaire (SF-36).
Dietary assessment
Diet is assessed with a semi-quantitative FFQ validated in Spain [
20]. The questionnaire covers 137 foods and is completed at baseline and yearly by the dietitians over the phone. The estimation of nutrient intake will be calculated as frequency multiplied by nutrient composition of the specified portion for each food item, using an ad hoc computer program developed specifically for this purpose. The repeated collection of dietary data allows us to use the PREDI-DEP trial as a unique setting for subsequent cohort studies, analyzed as a prospective observational follow-up study with repeated measurements of diet, thus improving the quality of our dietary assessment.
Changes in adherence to the Mediterranean diet is assessed through the MEDAS questionnaire at baseline and yearly in the control group and every three months in the Mediterranean diet group.
Assessment of the physical activity
Physical activity is assessed using a validated physical activity questionnaire with seventeen activities [
21]. Leisure activities will be computed by assigning an equivalent metabolic score to each activity, multiplied by the time spent in each one and adding up all activities. The baseline questionnaire is completed over the phone. Yearly, participants complete the same questionnaire using the website of the project or by phone with the help of the dietician if they request it.
Assessment of other variables
Information regarding socioeconomic (educational level, employment and marital status), anthropometric (weight, height, and waist and hip circumferences), life-style (tobacco or history of illegal drug use), and medical characteristics of the participants, including medication, family history of mental disorders or the use of psychotherapy or relaxation techniques are obtained from the general questionnaire at baseline. This information is updated on a yearly basis using follow-up questionnaires.
Biological samples
Blood, urine and stool samples are collected at baseline and yearly during medical visits. EDTA plasma tubes, buffy coat, and serum are collected and aliquots are kept at − 80 °C. All the samples are correctly identified and labeled with an alphanumeric code.
Biological compliance markers (plasma proportions of oleic and α-linolenic acid and urinary levels of tyrosol, hydroxytyrosol, resveratrol and ethanol) will be measured randomly in participants from the two arms of the trial at baseline and at the end of the study.
Sample size and data analyses
The original sample size calculation indicated that 250 people per group were required (assuming an attrition of 5%) to provide a statistical power of 80% for detecting a relative risk reduction of 30% in the Mediterranean diet group versus the control group for a two-year follow-up period. The assumed recurrence rate was 50% for the control group and 35% for the group assigned to the Mediterranean diet. The relative risk reduction of 30% was slightly lower than the observed in the SUN cohort study. In this latter study, we found a 40% risk reduction of depression comparing extreme quintiles of adherence to the Mediterranean Diet Score [
23] and the observed among diabetic participants of the PREDIMED trial after three years of intervention with the Mediterranean diet supplemented with nuts [
24].
A researcher blinded to the conditions of the intervention will carry out the analysis of the data. Intention-to-treat analysis will be done. For each participant we will compute person-years of follow-up from the study inclusion date to the date of recurrence of the depression or study termination, whichever comes first.
Log Rank analysis will be used to assess the effect of the intervention on the risk of depression recurrence. If distribution differences of baseline characteristics are detected, the Cox proportional-hazards regression models will be used to check by recruitment center, age group and sex, and adjusting for possible confounding factors (educational level, marital status, prevalence of diseases, body mass index, tobacco use, leisure-time physical activity, alcohol intake, total energy intake, type, dose and use of antidepressants, psychotherapy, presence, number, and time since last of depressive symptoms/episodes, and family history of mental disorders). Hazard ratios (HR) and 95% confidence intervals (CI) will be calculated considering the control group as reference.
We will also carry out a protocol analysis [
25]. Categories of adherence to the Mediterranean diet will be defined using information from the FFQ and MEDAS, and Cox regression models will be fitted. Marginal structural models (inverse probability of exposure-weighted estimators) will be fitted to exposure measures that vary over time [
26]. Finally, to examine quantitative variables such adherence to the Mediterranean diet scorings [
27], other analysis will be explored, e.g., smoothing splines and regression analysis based on fractional polynomials [
28,
29].
Changes in residual depressive symptoms, quality of life or biochemical parameters were evaluated through Generalized Estimating Equation (GEE) models and will be adjusted for possible confounding factors and baseline values of depressive symptoms, quality of life or each biochemical parameter, respectively.
Subgroup and sensitivity analyses will be done under different assumptions. To assess a possible interaction between the intervention and some variables (e.g., age, sex or the presence of non-communicable diseases), product terms will be introduced in the multivariable models. P values for the interaction will be calculated using the log-likelihood ratio test.
Multiple imputation will be employed for handling missing data [
30,
31].
Monitoring of Data
Three independent clinical trial experts in Nutrition, Psychiatry and Statistics made up the data monitoring committee (DMC). They will hold an annual meeting to review the implementation of the protocol, monitor trial progress, and recommend the continuation or termination of the study based on safety, outstanding benefit, or futility criteria.
A web-based system of data access was created (
https://predidep.es/participantes/), from where authorized investigators and the coordinator of the trial can download forms and datasets. For privacy and security, an ID and password are required to access the data and the forms.
Quality control reports are generated for key aspects of the trial, e.g., digit preference and variability. We perform checks on missing and/or inconsistencies of the data. Following data entry, cross-form edit checks are done. Audits are carried out periodically to detect unresolved problems. Standardized edit reports summarizing database problems reassures the quality of the data.