Recruitment of participants
The study was approved by the ethics committee of the University of Münster and the Westphalian Chamber of Physicians in Münster, North-Rhine-Westphalia, Germany. Written informed consent for participation in the study was obtained from all participants. Overall, 2315 participants were recruited for BiDirect-Baseline; 2258 (97.5%) of them matched the final inclusion criteria. Basic demographic characteristics of the three cohorts are summarized in Table
1.
Table 1
Basic demographic characteristics (absolute and relative frequencies) of the BiDirect-Baseline participants stratified by cohort
Sex | | | |
Female | 593 (59.4%) | 49 (14.1%) | 464 (50.9%) |
Male | 406 (40.6%) | 298 (85.9%) | 448 (49.1%) |
Age | | | |
35 to 44 years | 285 (28.5%) | 29 (8.4%) | 194 (21.3%) |
45 to 54 years | 437 (43.7%) | 128 (36.9%) | 304 (33.3%) |
55 to 65 years | 277 (27.7%) | 190 (54.8%) | 414 (45.4%) |
Educational level | | | |
University degree | 238 (23.8%) | 99 (28.5%) | 375 (41.1%) |
University entrance diploma or vocation diploma | 181 (18.1%) | 56 (16.1%) | 157 (17.2%) |
General certificate of secondary education | 278 (27.8%) | 86 (24.8%) | 187 (20.5%) |
Certificate of secondary education or lower | 295 (29.5%) | 104 (30.0%) | 189 (20.7%) |
Net household income1 per month | | | |
≤ 2000 Euro | 345 (34.5%) | 56 (16.1%) | 182 (20.0%) |
> 2000 Euro | 651 (65.2%) | 290 (83.6%) | 726 (79.6%) |
Cohort 1 consisted of 999 patients, who suffered from an episode of depression at the time of recruitment. Recruitment took place at six different psychiatric and psychosomatic hospitals and departments located in and around the city of Münster (radius: 35 km), as well as two resident psychiatrists’ practices located in Münster. The recruitment of outpatients was limited to those who had been hospitalized due to depression at least once during the 12 months period prior to inclusion into the study. Inclusion criteria were (i) age (≥35 and < 66 years) and (ii) current in- or outpatient treatment due to acute depression. Exclusion criteria were (i) compulsory admission, (ii) comorbid dementia, and (iii) comorbid drug abuse (including alcohol). Potential participants were ascertained by trained and certified study psychologists, and eligible patients were invited to participate in BiDirect-Baseline. Appointments were scheduled via telephone or email.
Cohort 2 comprised 347 patients with CVD (excluding cerebrovascular disease) who were recruited by trained and certified study nurses in four different cardiology departments and rehabilitation clinics in and around Münster (radius: 50 km). Inclusion criteria were (i) age (≥35 and < 66 years) and (ii) acute myocardial infarction, or acute coronary syndrome requiring therapy, or treatment of cardiac disease due to myocardial infarction within the last three months. In case of consent, the CVD patients were invited to participate in BiDirect-Baseline two months after recruitment by the study nurse, and appointments were scheduled via telephone or email.
Cohort 3 (reference cohort) included 912 community-dwelling adults (age: ≥ 35 and < 66 years). The participants had been randomly sampled from the population register of the city of Münster and were invited for BiDirect-Baseline via letter; appointments were scheduled by telephone or email.
Baseline response rates for the depression and CVD cohorts could not be estimated. In both cases, recruitment took place on site, and potential participants could be addressed by several different physicians or members of the nursing staff; therefore, the basic populations of those who were asked to participate are not known. The baseline response rate in the reference cohort was 41.5%.
Diagnosis of depression
The thorough assessment of depressive symptoms is a key concern of BiDirect-Baseline and a prerequisite for the testing of the specific BiDirect hypotheses over the course of the study. During BiDirect-Baseline, different measures were combined in order to confirm psychiatric diagnoses and to determine whether a participant suffered from acute depression with melancholic or atypical features. For patients from cohort 1, assessment of depression was conducted during recruitment. For participants from cohorts 2 and 3, (when necessary) diagnosis of depression took place during the BiDirect-Baseline examination at the study center.
For diagnostic purposes, all participants received selected modules (i.e. modules A, A’, B, D, and O) of the M.I.N.I. International Neuropsychiatric Interview (German version 5.0.0) [
19], which assessed whether a participant exhibited acute (first or recurrent) major depression with or without melancholic features, acute dysthymia, acute/former manic/hypomanic episodes, or acute generalized anxiety disorder. Further, for all patients from cohort 1 and for those participants from cohorts 2 and 3 who had received a M.I.N.I. diagnosis of acute major depression, it was clarified whether atypical depression features were currently present by means of six selected items of the Inventory of Depressive Symptomatology (IDS) [
20] (items 8, 12, 14, 27, 29, and 30). In addition, all patients with depression were assessed using the 17 items version of the Hamilton Depression Rating Scale (HAM-D-17) [
21] and the 14 items version of the Hamilton Anxiety Rating Scale (HAM-A-14) [
22]. It was also assessed whether a participant ever had episodes of melancholic or atypical depression throughout lifetime. Eventually, for explorative purposes we formed so-called “probable” subtypes: if a participant showed all but one criterion for melancholic or atypical depression, the diagnoses “probable melancholic depression” or “probable atypical depression” were assigned.
Notably, in some cases melancholic and atypical features of depression were present at the same time (subtype mixed). In other cases, an assignment to one of the categories melancholic depression or atypical depression was not possible (subtype undifferentiated). Table
2 summarizes the frequencies of the different depression subtypes in cohort 1.
Table 2
Absolute and relative frequencies of depression subtypes among N = 920
1
individuals from the depression cohort
Melancholic | 554 (60.2%) |
Atypical | 44 (4.8%) |
Mixed | 95 (10.3%) |
Probably melancholic | 27 (2.9%) |
Probably atypical | 39 (4.2%) |
Probably mixed | 68 (7.4%) |
Melancholic + probably melancholic | 581 (63.2%) |
Atypical + probably atypical | 83 (9.0%) |
Mixed + probably mixed | 163 (17.7%) |
Undifferentiated | 93 (10.1%) |
Constructs and instruments
The BiDirect-Baseline examination included several different modules: (i) computer assisted personal interview (CAPI), (ii) psychiatric interview (where required), (iii) self-administered questionnaires, (iv) assessment of sensory/cognitive functioning, and (v) diagnostic work-up (including MRI of the brain and taking of blood samples). Completing the entire program required an average of three-and-a-half hours.
By means of the
CAPI, which was applied in German language, detailed information regarding socio-demographic and socio-economic background, lifetime medical diagnoses, health care utilization behavior, insurance status, lifestyle and risk behavior (i.e. diet, physical activity, alcohol consumption, smoking), and current medication was obtained, for the most part using standardized sets of closed questions which had previously been used in different studies [
23,
24]. In addition, there were sections devoted to perceived health state. In case of female participants, the number of children, oral contraceptive usage, and certain surgical interventions (i.e. hysterectomy, ovariectomy) were inquired. Notably, for the participants of cohorts 2 and 3, the diagnostic M.I.N.I. interview was included in the CAPI. If indicated by the M.I.N.I. results, an additional appointment was scheduled, where additional non-structured and structured (HAM-D-17, HAM-A-14, selected IDS items) psychiatric interviews were administered by trained and certified study psychologists. The goal of these interviews was to assess the severity of depression and anxiety symptoms, and to evaluate whether or not a participant suffered from melancholic or atypical depression. For patients from cohort 1, the diagnostic and psychiatric interviews had already been conducted during recruitment (cf. section “Diagnosis of depression”).
Several different
self-administered questionnaires or sets of questions, which were combined to form a single booklet, were completed by the participants. The following instruments/sets were used in given order: (i) the Pittsburgh Sleep Quality Index (PSQI) [
25], which assesses sleep quality and disturbances over a one-month time interval; (ii) the Center for Epidemiological Studies Depression Scale (CES-D) [
26], which was designed to measure depressive symptomatology over a one-week time interval; (iii) a headache questionnaire [
27]; (iv) the Pain Sensitivity Questionnaire (PSQ) [
28], a measure of pain perception based on imagined painful daily life situations; (v) a set of six questions regarding perceived memory functioning and recent changes thereof; (vi) the Childhood Trauma Screener (CTS) [
29], which covers general living conditions and traumata during childhood (performed in a subsample of the baseline participants); (vii) a set of seven questions assessing the readiness to change health behavior (i.e. physical activity, diet, weight, alcohol consumption, smoking) according to the transtheoretical model [
30]; and (viii) the EQ-5D [
31], which assesses the perceived health-related quality of life.
In order to test sensory functioning, an orienting assessment of color perception was performed using 14 Ishihara Color Plates. Further, visual acuity was assessed (left vs. right eye). Moreover, odor perception was examined using the Sniffin’ Sticks Screening 12 Test (Burghart Messtechnik GmbH, Wedel, Germany), a forced choice test which allows categorizing participants into anosmic, hyposmic, and normosmic based on 12 different odors. Further, the pressure pain threshold was determined using a pain test algometer (Force Dial FDK/FDN Force Gage, Wagner Instruments, Greenwich, CT, USA) for left and right index fingers separately.
The
cognitive functioning module included the assessment of cognitive processing speed, verbal working and long-term memory, executive functions (verbal fluency, interference disposition, flexibility), emotional processing, and manual dexterity. The following tests were administered in given order: (i) immediate recall of a sequence of 12 (recorded and replayed) spoken German nouns, each with an emotional load (positive, neutral, or negative) (cf. [
32]); (ii) verbal production of as many different animal names as possible within a time interval of one minute [
33]; (iii) Color-Word-Interference Test [
34]; (iv) Trail Making Tests A and B [
35]; (v) delayed recall of the nouns (see above); (vi) rating of the nouns regarding valence and arousal on nine-point rating scales; and (vii) Purdue Pegboard Test [
36] for left and right hands separately. Eventually, the participants self-evaluated their level of attention during the cognitive functioning module on an eleven-point rating scale.
The diagnostic work-up began with the determination of weight (without shoes and heavier clothes), height, and waist circumference. Bioelectrical impedance measurement (Body Impedance Analyzer BIA 2000-S, Data Input GmbH) including the determination of body fat and water, extracellular mass, body cell mass, and basic metabolic rate was applied. The determination of the vascular status consisted of the measurement of a standard 3-channel electrocardiogram (ECG) using extremity leads, the brachial systolic and diastolic blood pressure by oscillometry, the ankle-brachial index by photoplethysmography, the pulse wave velocity, the augmentation index (Vascular Explorer, enverdis GmbH), and the intima-media-thickness of the far wall of the carotid arteries non-invasively by ultrasound (Acuson X300, Siemens). In addition, blood samples were taken (plasma, serum, and DNA), and fundus photography (Digital Retinography System, EyeNovation) was performed in a subsample of the baseline participants. The diagnostic-workup closed with an MRI of the brain, which involved both structural (3D-T1, T2*, FLAIR, and DTI sequences) and functional (resting-state sequence, emotional faces paradigm [
37]) recordings. Table
3 gives an overview regarding the number of completed examinations per cohort.
Table 3
Absolute and relative frequencies of examinations completed
1
by the BiDirect-Baseline participants stratified by cohort
Computer-assisted personal interview | 996 (99.7%) | 346 (99.7%) | 911 (99.9%) |
Sensory functioning | | | |
Pain | 953 (95.4%) | 304 (87.6%) | 795 (87.2%) |
Olfaction | 989 (99.0%) | 341 (98.3%) | 901 (98.8%) |
Visual acuity | 996 (99.7%) | 344 (99.1%) | 907 (99.5%) |
Cognitive functioning | 964 (95.5%) | 332 (95.7%) | 824 (90.4%) |
Anthropometry | 998 (99.9%) | 347 (100.0%) | 890 (97.6%) |
Body impedance | 984 (98.5%) | 331 (95.4%) | 896 (98.2%) |
Vascular status | | | |
Electrocardiogram | 988 (98.9%) | 345 (99.4%) | 905 (99.2%) |
Blood pressure | 987 (98.8%) | 340 (98.0%) | 902 (98.9%) |
Intima media thickness | 987 (98.8%) | 345 (99.4%) | 894 (98.0%) |
Blood sample | | | |
DNA | 929 (93.0%) | 321 (92.5%) | 783 (85.9%) |
Serum and plasma | 895 (89.6%) | 302 (87.0%) | 742 (81.4%) |
MRI | | | |
Anatomical | 735 (73.6%) | 52 (15.0%) | 669 (73.4%) |
Resting state | 719 (72.0%) | 51 (14.7%) | 654 (71.7%) |
Emotional faces | 621 (62.2%) | 43 (12.4%) | 528 (57.9%) |
Quality assurance and data management
The primary goal of all quality assurance measures was to generate high quality data. In principle, we followed the rules for good epidemiological practice as well as the recommendations for the use and storage of blood samples and DNA as provided by the German Society of Epidemiology (cf.
http://dgepi.de/fileadmin/pdf/leitlinien/GEP_LL_english_f.pdf). The storage of biomaterials was split between different refrigerators, all of which were monitored by a central alert system 24 hours per day seven days per week. The BiDirect standard operations manual contains operating procedures for all BiDirect interview and examination components.
Prior to the beginning of BiDirect-Baseline, the members of the study team (i.e. study nurses, study psychologists) underwent an initial three-months training ending with certifications. Moreover, each member of the study staff was trained and certified for a variety of tasks, providing mutual backup in case of illness or changes in personnel. Performance was closely monitored and routinely assessed.
Data management has been carried out in parallel with data collection, based on standardized and partly automated procedures for data processing and plausibility checking. Data backup routines are scheduled on a daily basis.
Biometric concept and statistical analyses
The primary endpoints of BiDirect are incidence and/or progression of both depressive symptoms and (subclinical) arteriosclerosis. Secondary endpoints are onset of depression, myocardial infarction, brain infarction, and death. Current data evaluation is restricted to the baseline cross-sectional analyses, since the first follow-up period is still ongoing. Depending on the outcome scale level (continuous vs. categorical) and time point of assessment (baseline vs. follow-up), the statistical methods comprise multivariable linear or logistic regression analyses, accounting for time-varying predictors and repeated outcome assessment.
The primary endpoints are operationalized by closed questions and non-invasive clinical examinations. The majority of primary endpoints yield continuous variables (e.g. CES-D summary score, systolic/diastolic blood pressure [mmHg], intima media thickness [mm]) with change over time. The computation of arteriosclerosis scores will facilitate the analysis of disease progression in different vascular territories. Power calculations were based on the planned study duration of 12 years and took secondary, frequency-type endpoints (in particular subclinical brain pathology on the MRI as the limiting factor) into account. Loss-to-follow-up between subsequent examinations was expected to be maximally 20% on average across cohorts.