Between December 2013 and December 2016, researchers recruited Chinese adults to participate in the multicenter, multistage, prospective study Objective Diagnostic Indicators and Individualized Drug Intervention of Major Depressive Disorder (OIMDD), project No. 2013CB531305. We conducted a secondary, cross-sectional analysis using the baseline data from OIMDD for participants diagnosed with MDD. Nine hospitals in China participated in the study. All participants provided written consent, and the study protocol was approved by the research ethics board of the institution where it was performed.
Study participants included patients between the ages of 18 and 65 years who had been diagnosed with their first episode of MDD, did not receive any antidepressant treatment in the acute phase of the disease, and had a total score of ≥14 on the 17-item Hamilton Rating Scale for Depression (HAMD-17) . Patients were excluded if they had severe somatic diseases, such as severe heart disease, malignant tumors, or a history of epilepsy. Pregnant or lactating women were also excluded .
Depression Diagnosis and Clinical Assessment
Using criteria from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), psychiatrists made a diagnosis of MDD and then conducted structured clinical interviews with the prospective participants using the Mini-International Neuropsychiatric Interview (MINI), Chinese Version 5.0.0 . After the interviews, HAMD-17 was used to assess depressive symptoms, evaluating functioning in 5 subscales: cognitive impairment, retardation, anxiety or somatization, sleep disturbance, and weight change . Sleep disturbance severity was assessed by evaluating the followingmetrics on the HAMD-17 scale: item 4, difficulty falling asleep; item 5, waking in the middle of the night for any reason except to void; and item 6, waking early in the morning and unable to go back to sleep. Each item could be rated from 0 (no difficulty) to 2 (nightly difficulty), with a total possible score of 6. A sleep subscale score >4 was defined as a serious sleep disorder, and patients in this category were assigned into Group 1; the remainder, were assigned into Group 2. The Hamilton Anxiety Rating Scale (HAMA) was used to assess 2 areas of somatic anxiety: muscular (pains and aches, twitching, stiffness, myoclonic jerks, grinding of teeth, unsteady voice, and increased muscular tone)and sensory (tinnitus, blurring of vision, hot and cold flushes, feelings of weakness, and pricking sensation) [28, 29].
A battery of cognitive tests were performed to evaluate 5 cognitive domains: attention (vigilance), assessed by the Continuous Performance Test (CPT); speed of information processing, assessed by the Animal Verbal Fluency Scale (AVFS), Digit Symbol Coding Test (DSCT), and Color Trial Test (CTT); learning, assessed by the immediate recall of Brief Visual Memory Test-Revised (BVMT-R) and Hopkins Verbal Learning Test-Revised (HVLT-R); memory, assessed by delayed recall of Brief Visual Memory Test-Revised (BVMT-R) and Hopkins Verbal Learning Test-Revised (HVLT-R); executive functioning,assessed by the Stroop Color Word Test (SCWT, which assessed executive inhibition) and color line II (which assessed executive-shifting). We selected the cognitive test battery in our study from the MCCB (MATRICS Consensus Cognitive Battery) of the MATRICS (the Measurement and Treatment Research to Improve Cognition in Schizophrenia) study. The MCCB has been widely used to evaluate the cognitive function of patients with mental disorders. The test scores for each domain were transferred into global deficit scores (GDS) , which were adjusted for age, sex, and education level. A global deficit score≥0.5 was defined as cognitive impairment [30, 31].
In addition to gathering basic demographic data—marital status, living situation (whether the subject lives alone or with others), religious affiliation, work status, the type of work (whether it requires ental labor, physical labor, or both), independence (degree to which the subject depends on someone else for life’s basics), and body mass index (BMI), we obtained information on potential confounding covariables that could alter a person’s risk for cognitive impairment. Some comorbid conditions are known to increase the risk of cognitive impairment, including cardiovascular disease and diabetes. Other suspected risk factors include duration of disorder, severity of depression and/or anxiety, family history of psychiatric disorders, history of alcohol abuse, smoking history, and childhood trauma.
We collected previous disease history at baseline, including hyperlipidemia, hypertension, and diabetes. Diabetes was defined as a fasting blood glucose (FBG) level ≥126 mg/dL (7.0 mmol/L), oral glucose tolerance test ≥200 mg/dL (11.1mmol/L), HbA1c ≥48 mmol/L (6.5%), or a history of diabetes mellitus. Hyperlipidemia was defined as having total cholesterol (TC) ≥5.17 mmol/L, triglycerides (TG) ≥1.7 mmol/L, low-density lipoprotein (LDL) cholesterol ≥3.37 mmol/L, or a history of hyperlipidemia. Social demographic variables were assessed using questionnaires that asked about age, sex, marital status, occupation, alcohol use, and smoking status. Body mass index was calculated as weight in kilograms divided by squared height in meters (kg/m2). BMI was divided into low weight (BMI <18.5), normal weight (BMI ≥18.5 to <24), overweight (BMI >24 to ≤28), and obesity (BMI >28).
Childhood trauma was assessed using the childhood trauma questionnaire (CTQ), which evaluates the subject’s experience of abuse and/or neglect before the age of 18 years. The CTQ includes domains of emotional, physical, and sexual abuse, as well as of emotional and physical neglect. Items are rated from 1 (“never true”) to 5 (“very often true”) according to the frequency with which each event occurred in childhood. We considered the history of trauma as a dichotomic variable (yes/no) if the person rated items as moderate or severe according to the subscale cutoff criteria in at least one type of trauma .
Summary statistics were presented as mean ± standard deviation for variables that conformed to a normal distribution, as medians and quartile for data did not conform to a normal distribution, and as percentages for categorical variables. To make the 2 groups more comparable, we performed propensity score matching.Propensity scores were estimated using a logistic regression model that contained known or suspected covariates that were unbalanced between the 2 groups. To perform the propensity score matching, we included the type of work, marital status, religion, alcohol use, and HAMD-17 weight subscale. Because the independent-variable global deficit scores are functions converted fromstandard scores corrected by age, sex, and education, we did not include these variables in the propensity score matching. Subjects were matched 1:1 without replacement, using a 0.00 caliper width. Effect size was calculated to estimate the balance of the baseline data between the 2 groups. Cohen d was calculated using a t test and φ(phi) or φc (Cramer’s phi) was calculated using the chi-square test. Effect size >0.20 was considered to be an imbalance between the 2 groups.
Statistical analyses were conducted using SPSS 25.0 (SPSS Inc, Chicago, Illinois, USA) and version R 3.3.3(Foundation for Statistical Computing,Vienna, Austria). The R package of MatchIt was used forthe propensity score analysis. Logistic regression analysis was used to estimate the odds ratios for the global deficit scores for each cognitive domain and their corresponding 95% confidence intervals (CIs).