TRAILS TRANS-ID infrastructure
TRAILS TRANS-ID connects two large, ongoing projects, namely TRAILS (TRacking Adolescents’ Individual Lives Survey) and TRANS-ID (Transitions In DepressionFootnote 1). TRANS-ID was designed to prospectively monitor sudden transitions in psychopathology. TRAILS TRANS-ID is one of the studies within the TRANS-ID project, and includes participants from TRAILS. TRAILS is an ongoing, prospective cohort study investigating mental health across pre-adolescence and young adulthood. TRAILS consists of a general population cohort (TRAILS-PC, which started in 2000) and a transdiagnostic clinical cohort (TRAILS-CC, which started in 2004). Participants of TRAILS TRANS-ID were recruited in TRAILS-CC. TRAILS-CC was designed to enrich its general population counterpart by selectively sampling individuals who were at heightened risk for mental ill-health. Inclusion in TRAILS-CC was based on at least one referral before the age of 11 years to the child psychiatric outpatient clinic of the University Medical Center Groningen. Because of this history, TRAILS-CC participants are considered vulnerable for developing (additional) psychopathological complaints later in life (e.g. in young adulthood). This is reflected in heightened mental ill-health in TRAILS-CC compared to the general population cohort of TRAILS . An extensive description of the sampling procedures for TRAILS-CC has been published elsewhere .
TRAILS TRANS-ID consisted of a six-month daily diary study that was designed to capture the ebb and flow of (subthreshold) symptoms of psychopathology in young adults at heightened risk for mental ill-health (see Study population). Immediately before (baseline), directly after (post assessment), and one year after the diary period (follow-up), a clinical interview took place. This interview included a diagnostic interview and two questionnaires on social functioning and life events, respectively. All measures were administered orally and are outlined in detailed below. All assessments and other contact moments with participants (e.g. phone calls) were performed by trained researchers (first two authors and research assistant) in a standardized way. Participants remuneration included up to a maximum of €200,- (for details, see Additional file 1) and a report of their own diary data.
All participants enrolled in TRAILS-CC who participated in at least one of the previous measurement waves of TRAILS-CC and had given consent to be approached for future assessments were eligible to participate in the TRAILS TRANS-ID study (see also TRAILS TRANS-ID infrastructure). At its first measurement wave, TRAILS-CC included 543 children aged between 10 and 12 years old (mean age = 11.1, std. = 0.50, range 10.1–12.4, 34.1% girls). TRAILS TRANS-ID took place when the TRAILS-CC participants were between 21 and 24 years old. Of the 443 TRAILS-CC participants eligible for participation in TRAILS TRAILS-ID, 134 (30.2%) enrolled in the study (see details on recruitment below). The TRAILS-CC cohort study was approved by the Dutch Central Committee on Research Involving Human Subjects (CCMO), in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. Prior to each measurement wave of TRAILS, informed consent was obtained from the adolescents (at earlier waves also from the parents). The TRAILS TRANS-ID study was approved by the standing Ethics Committee, in accordance with the 1964 Declaration of Helsinki. All participants gave written informed consent before enrolment in TRAILS TRANS-ID.
From 18 October 2017 until 12 February 2018, individuals of the TRAILS-CC cohort were invited to participate in TRAILS TRANS-ID. Individuals received an invitation letter by mail, which contained information about the aims and design of TRAILS TRANS-ID. Individuals could indicate whether they were interested in participation through a website, by phone, or by returning the reply card that was enclosed to invitation letter. Participants who did not respond received a reminder by mail (approximately one month after the first invitation).
Upon registering, participants were contacted by phone by a researcher who provided further information about the study. This contact moment was also used to verify whether participants owned a smartphone and whether they anticipated any circumstances that might interfere with the diary period (e.g. long-term stay in remote areas). Participants who did not own a smartphone were provided one for the duration of the diary period. Of the 142 participants who were interested in participating in TRAILS TRANS-ID, 8 participants (5.6%) were not included for various reasons (e.g. because they could not be reached or had changed their minds).
Prior to the clinical interview, participants were once more informed about the study procedures and asked for written informed consent. Thereafter, the clinical interview (see Instruments) was conducted and the diary procedure was explained. Participants completed a practice session in presence of a researcher in order to familiarize themselves with the questions. Participants were instructed to carefully consider their mental state and use the extreme ends only if they could not imagine experiencing a certain symptom more (corresponding to a score of 100) or less (corresponding to a score of 0). To verify whether participants’ interpretation of diary items remained consistent during the diary period – as outlined in our second aim – half of the participants were interviewed about their interpretation of a randomly selected set of ten items after the clinical interview (see Qualitative assessment).
Participants started the diary period immediately after the baseline assessment (Fig. S1, Additional file 1). This period comprised filling in an electronic diary every evening for six consecutive months (i.e. 183 days). Participants could access the diary via a link sent in a text message to their mobile phone. The diary consisted of questions assessing experiences during the past day (for an overview of items, see Table S3, Additional file 1). Participants therefore always received text messages in the evening, at a fixed time according to the participant’s wishes. Although timing of the text messages differed between participants, all participants had 24 h between each measurement point. Participants were asked to complete the questionnaire as soon as possible after receiving the text message. If the questionnaire was not completed within 30 mins, participants received a reminder text message. After this reminder, participants had 2.5 h to fill in the diary. The questionnaires were administered and stored via Roqua (www.roqua.nl).
During the diary period, the researchers contacted the participants at least seven times. The first two phone contacts took place in the first and second week of the study, respectively, to resolve any issues regarding, for instance, the timing of the text message or answer questions about the study procedures. Thereafter, participants were contacted monthly or when they skipped diaries for several consecutive days. These phone contacts served to support participants, to solve any issues that occurred (e.g. technical problems), and to monitor life events that might affect symptom dynamics. The latter were assessed by asking participants whether they had experienced important (positively or negatively appraised) events regarding their health, work or education, social relations, and financial situation. These events were noted and, together with the participant, retrospectively dated. Researchers were also available by telephone and email if participants needed assistance at other times than the pre-specified contact moments.
Post and follow-up assessment
After the diary period, participants were invited for a post assessment, which involved a clinical interview spanning the diary period (i.e. the past six months) and an evaluation of the diary. The latter included rating the burden of completing the diary period on a scale from one (not at all burdensome) to ten (extremely burdensome). Participants were offered a report on their personal data including descriptive information which was explained orally and provided on paper. Those who were included in the qualitative assessment of diary items at baseline, were asked to take part in this assessment again at post (see Qualitative assessment). Finally, one year after finishing the diary period, participants were invited for the follow-up assessment. This assessment involved a clinical interview (mini-SCAN, GVSG, life events questionnaire) that spanned the past year.
The diary questions, listed in Table S3 (Additional file 1), were selected to cover symptoms of all common psychiatric disorders. Items were derived from earlier studies that adopted ambulatory assessments or diary protocols [16,17,18,19] as well as from existing questionnaires such as the Positive and Negative Affect Scale (PANAS) , the Adult Behavior Checklist (ABCL) , and the Symptom Checklist (SCL-90) . Items were considered for inclusion in the diary if they (1) covered multiple diagnostic categories; (2) were sufficiently distinct from other items; (3) were expected to vary from day to day; and (4) were easy to understand. To accomplish the latter, we conducted a pilot study among 12 adolescents with a similar educational attainment level to that of the TRAILS-CC cohort (see Additional file 1 for more information). The final selection consisted of 58 items, which were all rated on a visual analogue scale (VAS) that ranged from 0 to 100. Participants were given the opportunity to add an item to the diary that they personally wished to monitor (e.g. ‘To what extent was I productive at work today?’). At the end of the diary, an optional textbox allowed participants to make notes about their day. Participants were encouraged to report unusual or important events, such as starting a new job or being ill, which might have influenced their responses to the diary that day. Participants’ notes remained confidential and were solely examined during the post assessment for the descriptive report on their personal data.
Qualitative assessment: validity of diary items
To evaluate the validity of diary items – as indicated by the consistency of participants’ interpretation – the first 60 participants who were included completed an interview about their interpretation of a randomly selected set of ten items both at the baseline and post assessment. We interviewed 60 individuals rather than the whole sample because this was considered to yield a reliable indication of interpretation consistency, without needlessly burdening participants. During the qualitative assessment, participants were asked what a certain item, such as feeling stressed, meant to them. Participants were asked about the same set at baseline – when participants practiced with the diary – and during the post assessment – when participants discussed their personal data with the interviewer. At both assessments, participants’ interpretations were recorded and later transcribed. Participants were encouraged to elaborate and to provide examples. In case participants did not understand an item during the first session, the item’s meaning was explained by the researcher. In this case, consistency in interpretation was not evaluated for this item.
Demographic characteristics of the sample, including socio-economic status and current occupation (education, employment), were retrieved from earlier measurement waves of the clinical cohort of TRAILS. Socio-economic status consisted of a standardized score that combined information about household income and the education and occupation of participants’ parents, and was assessed during the first measurement wave when participants were approximately 11 years old.
The clinical interview that was administered at baseline, post assessment, and follow-up consisted of a diagnostic interview complemented by orally administered questionnaires on social functioning and life events (see below). The diagnostic interview consisted of the computer-assisted mini-SCAN which was administered by trained interviewers and assessed the presence of psychiatric disorders . The mini-SCAN is a validated semi-structured interview, which by means of screening questions and skips efficiently assesses whether individuals meet the diagnostic criteria of a broad range of psychiatric disorders. Since current purposes were primarily of scientific, as opposed to traditional diagnostic, nature, we slightly adapted the assessment procedure (see also Table S4, Aditional file 1). First, similar to an earlier study , we rated items on a three-point scale (absent, subthreshold, clinical) rather than a dichotomous scale (absent, present). This allowed for recognizing symptoms that were endorsed by participants, but did not meet the criteria of clinical significance (e.g. with respect to severity, duration, or frequency; see Additional file 1 for details on scoring). For instance, feeling down for multiple consecutive days, but less than two weeks, could classify as a subthreshold symptom of depression. The distinction between absent, subthreshold and clinical symptoms was considered particularly relevant since our participants were at risk for psychopathological symptoms. This at risk status may frequently coincide with mild signs of psychopathology that occupy the grey area between mental health and psychopathology. Adapting the conventional rating procedure of the mini-SCAN allowed for capturing this grey area. A second adaptation concerned the removal of skips that are traditionally imposed by the mini-SCAN. To be able to compare ratings across but also within individuals, we assessed the following sections in each participant: stress, anxiety (including post-traumatic stress disorder (PTSD), generalized anxiety disorder (GAD), panic disorder, social anxiety disorder, specific phobia), obsessive-compulsive disorder (OCD), depression, mania, hallucinations, delusions, substance use, attention deficit hyperactivity disorder (ADHD), autism spectrum disorder (ASD). For the majority of sections, all symptoms were rated regardless of the presence of hallmark symptoms (e.g. for depression these are depressed mood and/or loss of interest). Sections for which skips were preserved were anxiety, hallucinations, and delusions, which all contained symptoms that were conditional upon the hallmark symptom(s). For instance, symptoms of PTSD are conditional upon the experience of a traumatic event and were hence disregarded in the absence of such an event.
The mini-SCAN does not cover oppositional or antisocial behavior, yet such behavior may represent an important precursor or manifestation of psychopathology . Therefore, we complemented the mini-SCAN with the aggressive behavior subscale of the Adult Self Report (ASR, also assessed at earlier measurement waves of TRAILS; 17)), which was orally administered. This subscale consists of 15 items that are rated on a three-point scale (never or seldom, sometimes, frequently or always). Scores on the aggressive behavior subscale were considered indicative of symptoms of oppositional defiant disorder (ODD) and will henceforward be referred to as such.
Social functioning was assessed with the Groningen Social Behaviour Questionnaire (in Dutch: Groningse Vragenlijst Sociaal Gedrag, GVSG), which has good psychometric properties . The GVSG consists of 45 items that assess functioning in several domains, including social relations (i.e. with parents, children, friends, and romantic partners), work, school, household, and leisure-time. Items were rated on a 4-point Likert scale which ranged from ‘never’ to ‘always’. Domain scores that exceeded the upper limit cut-off provided by De Jong and Van der Lubbe (2001, ) were categorized as ‘severe impairments in functioning’ and domain scores that exceeded the lower limit cut-off were categorized as ‘mild impairments in functioning’.
The life events questionnaire comprises an adapted version of the List of Threatening Experiences, which has good psychometric properties [27, 28]. The List of Threatening Experiences consists of 12 items describing major stressful life events, such as experiencing serious illness, losing a job, or ending a steady relationship. Eight items describing pleasant events, such as going on a holiday or graduating, were added to this list, resulting in a total of 20 items. The occurrence of these events (yes/no) was rated.
The data presented in the current paper pertain to the baseline assessment, diary period, and the post assessment. Follow-up interviews were still being conducted. Analyses were conducted in R (version 1.463).
Aim 1: feasibility
Our first aim was to explore whether conducting a six-month daily diary study in at risk young adults is feasible. Feasibility was evaluated by examining descriptives (mean, standard deviation) concerning participants’ drop-out and compliance, and their responses to the diary-evaluation questions in the post assessment interview, which inquired about the burden and reasons for missing diary responses.
Aim 2: three core assumptions
Our second aim was to evaluate three common assumptions of IL assessments of psychopathology. The first assumption is that diary item capture variability in symptom expression. This assumption was evaluated by computing the between- and within-individual variability in symptom reports. Between-individual variability was operationalized as the deviance in mean item scores across individuals. Large values indicate that individuals differ considerably in their scores on the respective item, while small values suggest that individuals score relatively similar. Within-individual variability was examined through calculating the mean deviation in item scores within individuals. Large values indicate that the respective item varies considerably over time within individuals, while small values suggest that scores are relatively stable. As there are no criteria yet for the minimum amount of variability that can be considered informative, we compared different items descriptively.
The second assumption to evaluate was the consistency of participants’ interpretations of diary items over time. This was addressed by comparing, per participant and per item, interpretations at baseline and at post assessment. Item interpretations were coded as consistent when participants used identical or similar words or examples to describe an item at baseline and post assessment, respectively. Inconsistent interpretations, in contrast, were scored in case the interpretations at baseline and post assessment showed marked discrepancies. All interpretations were coded by two raters separately. Interrater agreement was assessed through the B statistic [29, 30] instead of Cohen’s Kappa because ratings of inconsistency were far less frequent than ratings of consistency. The B statistic quantifies the agreement between two raters, accounting for the agreement that would occur due to chance .
To address the third assumption – concerning the correspondence between daily symptom reports and diagnostic measures assessed in a clinical interview – we examined the correlation between mean diary ratings and mean scores on distinct psychopathological domains (assessed by the diagnostic interview at post assessment). Because the data did not fulfill parametric assumptions, Spearman correlation coefficients were computed. We then determined correspondence by evaluating which diary items correlated most strongly with each domain.