Data and design
This study is based on baseline survey data on social and health-related variables from the randomized controlled trial “Supported Employment and preventing Early Disability” (the SEED-trial) [22]. The SEED-trial is an ongoing randomized controlled trial investigating the effect of individual placement and support vs. traditional vocational rehabilitation in individuals at risk of early work disability in Norway. For additional information about the trial, study design and procedures, see Sveinsdottir et al. 2016 [22].
Participants and recruitment
Ninety-six individuals (65 men (68%) and 31 women (32%)) with a mean age of 24 (SD = 3.25), participated in the study. Participants were young adults aged 18–29 in the year of inclusion, were not in employment or undergoing education, were receiving temporary benefits (mainly WWA, or employment scheme benefits), and were considered by their caseworkers at the Norwegian labor and welfare administration to be eligible for traineeships in sheltered businesses. Traineeships in sheltered businesses are only offered to those with impairment and particularly uncertain work capabilities requiring close follow-up. During June 2014 through December 2016, new eligible clients at one central and nine local labor and welfare offices in and around the city of Bergen, Norway, were referred to meetings to receive information about the study. Referrals were also made by a secondary care district psychiatric center in Bergen, with subsequent follow-up at the local labor and welfare office. At the information meetings, eligible participants were screened on two additional inclusion criteria, before being invited to participate in the trial: 1) Sufficient language skills to answer questionnaires in Norwegian, and 2) interest in receiving help to obtain ordinary work. There were no exclusion criteria based on diagnosis, and participants with any type of social and/or health-related problems were invited. A total of 163 participants attended the information meetings, whereof 67 were excluded or declined participation (Fig. 1).
Data collection
Questionnaires were administered to all participants at the information meetings, either electronically or in paper format. Participants received help and assistance to answer the questionnaires upon request.
Electronic responses were collected using iPads with secure survey software (Qualtrics, Provo, UT), and stored in a secure online database. Responses in paper format were stored in a locked filing cabinet. Personal information and contact details were stored separately from the collected data, in a locked and fireproof safe.
Questionnaire and instruments
In the first part of the questionnaire, information on self-reported background and social variables were collected. Dichotomous variables were computed for education level (less than high school vs. other), reading/writing difficulties (yes vs. no), marital status (single vs. other), living arrangements (living with parents vs. other), number of children (none vs. other), immigrant background (immigrants and Norwegian-born to immigrant parents vs. not), previous participation in employment scheme (yes vs. no), previous employment (yes vs. no), and reasons for unemployment (psychological problems vs. not, other health problems vs. not, and other non-health-related reasons vs. not). Participants were also asked to list whether they had received treatment during the last 6 months, and whether they had received consultations by general practitioners, psychiatrists, psychologists, physio−/manual therapists, chiropractors, or other therapists. Dichotomization of continuous background variables was based on visual inspection of the distributions.
Disability level
Disability level was measured using the 12-item self-administered version of the WHO Disability Assessment Schedule 2.0 (WHODAS 2.0), consisting of a sum-score (Cronbach’s α = .87) based on six domains of life: Cognition, mobility, self-care, getting along, life activities, and participation [23]. Each item was scored on a 5-point scale ranging from 0 (none) to 4 (extreme or cannot do). A simple scoring strategy without weighting of individual items was used, ranging from 0 (no disability) to 48 (full disability). In cases of up to five missing items, the mean of the remaining items was calculated and multiplied by 12. In addition to the sum-score, cut-off scores of ≥10 for significant disability were used based on the top 10% of the population in normative data [24].
Health behaviors: Alcohol and drug use
Alcohol use was measured using the 3-item Alcohol Use Disorders Identification Test (AUDIT-C) [25]. Items were scored on 5-point scales ranging from 0 to 4, with higher scores indicating higher frequency and quantity of alcohol consumption. Based on a sum-score, validated cutoff-scores of ≥4 for men and ≥ 3 for women were used to indicate hazardous drinking or active alcohol use disorders [26].
Drug use was measured using the 11-item Drug Use Disorders Identification Test (DUDIT) [27]. Items 1–9 were scored on 5-point scales ranging from 0 to 4, and items 10–11 were scored on 3-point scales as 0, 2, and 4, with higher scores indicating more severe drug use. Based on a sum-score, validated cutoff-scores of ≥6 for men and ≥ 2 for women were used to indicate drug-related problems [28].
Bullying
A new eight-item questionnaire was developed in collaboration with Dan Olweus and Jørn Hetland, researchers within the fields of bullying in school and bullying in working life, respectively. The new measure was specifically developed in order to measure lifetime experiences with bullying victimization and bullying perpetration in different social arenas, for respondents who are currently not in employment, education, or training. The items were preceded by a description of bullying according to Olweus’s definition [29], describing that bullying can be direct and indirect as well as verbal and physical, and that it involves a perceived power imbalance or difficulty defending oneself. It was emphasized that friendly teasing, and fights or arguments between equal parts were not regarded as bullying.
Bullying victimization was measured using five single items, concerning three arenas: School (2 items, bullied by other students or teachers), working life (2 items, bullied by colleagues or leaders), and other social arenas (1 item). The items were scored on a 5-point scale ranging from 0 (never or almost never), 1 (one short period (a few weeks)), 2 (several shorter periods), 3 (one long period (several months)), to 4 (several longer periods of my time in school/working life/other social arenas). Values ≥2 within each arena were coded as bullying in that arena, and an overall dichotomous variable was created for bullying victimization in any arena vs. bullying in no arena. This is in line with the Olweus definition emphasizing repeated incidences over time rather than the length of an incidence [29], and coincides with how bullying has been categorized in other studies [30].
Bullying perpetration was measured with three single items, corresponding to the method and scale used for bullying victimization as described above. The questions concerned whether the participant him/herself had bullied others in three arenas: School (1 item), working life (1 item), and other social arenas (1 item). Values ≥2 within each arena were coded as bullying in that arena, and an overall dichotomous variable was created for bullying perpetration in any arena vs. bullying in no arena.
An additional dichotomous variable was created for those who reported that they were both victims and perpetrators of bullying (bully-victims).
Violence
Violence was measured using a single item concerning whether participants had been the victim of violence inflicted by others (not counting accidents and common children’s fights). If yes, participants were further asked to indicate what types of violent acts they had experienced (being hit, robbery/assault, sexual violence, deprivation of liberty, severe threats, or other), and whether incidents were single or repeated.
Psychological distress
Psychological distress was measured using the 25-item Hopkins Symptom Checklist (HSCL-25), consisting of two subscales: An anxiety dimension (10-items, α = .82) and a depression dimension (15-items, α = .91), in addition to a mean score (α = .93) [31]. Each item was scored on a 4-point scale ranging from 1 (no symptoms) to 4 (severe symptoms). In addition to the mean score, a validated cut-off score of ≥1.75 was used for psychological distress [32, 33].
Fatigue
Fatigue was measured using the 11-item Chalder Fatigue Questionnaire (CFQ) consisting of two subscales: Physical fatigue (7 items, α = .88) and mental fatigue (4 items, α = .67), in addition to a sum-score (α = .86) [34]. Each item was scored on a 4-point scale ranging from 0 (less than usual) to 3 (much worse than usual). In addition to the sum-score, a binary global fatigue score ranging from 0 to 11 was calculated and validated cut-off scores of ≥4 were used for severe fatigue [34, 35].
Sleep problems (insomnia)
Three single items were developed in collaboration with Mari Hysing, researcher within the field of mental health and sleep problems in children and adolescence, to serve as a simple proxy for the diagnostic criteria for insomnia according to the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5) [36]. The first item concerned problems with falling asleep, waking up at night, and/or waking up too early. Respondents were asked to indicate how many nights they experienced each problem during a typical week, on a scale ranging from 0 to 7. If any sleep problems were reported, respondents were asked to proceed to the second and third items, indicating how long the problems had lasted, and how many times a week the problems affected daily life. A dichotomous variable for insomnia was computed based on whether or not one or more sleep problems exceeded three nights a week, had lasted more than 3 months, and affected daily life for more than 3 days a week.
Subjective health complaints
Subjective health complaints were measured using the 29-item Subjective Health Complaints Inventory (SHC), consisting of five subscales: Musculoskeletal pain (8 items, α = .78), pseudoneurology (7 items, α = .73), gastrointestinal problems (7 items, α = .64), allergy (5 items, α = .48), and flu (2 items, α = .56), in addition to a sum-score (α = .82) [37]. Each item was scored on a 4-point scale ranging from 0 (no complaints) to 3 (serious complaints).
Global well-being
Global well-being was measured using a 10-point Cantril Ladder Scale [38], ranging from 1 (the worst life possible) to 10 (the best life possible), asking respondents to indicate on which step of the ladder they feel they stand today, on which step they would say they stood a year ago, and where they believe they will be a year from now.
Social support
Social support was measured using 11-items of the Nondirective and Directive Support Survey [39] as suggested by Øyeflaten et al. (2010), using two subscales: Directive social support (4 items, α = .73) and nondirective social support (7 items, α = .84) [40]. The directive subscale involves instructive support and taking charge of the situation in order to help the recipient, while nondirective support is of a more cooperative nature and involves acceptance of the recipients own thoughts and choices [40]. Each item was scored on a 5-point scale ranging from 1 (not at all typical) to 5 (very typical). The survey also instructs respondents to indicate the specific person to whom they turn for support, and whether this is their doctor, spouse/partner, or “other” including an open response. An additional dichotomous variable was created based on a categorization of whether the support provider was a professional (e.g. doctor, psychologist) vs. personal (e.g. partner, family, friend).
Coping
Coping was measured using the 7-item Theoretically Originated Measure of the Cognitive Activation Theory of Stress (TOMCATS) [41], consisting of three subscales: Coping (1 item), helplessness (3 items, α = .65), and hopelessness (3 items, α = .66). Each item was scored on a 4-point scale ranging from 1 (completely true) to 4 (not true at all). Items were reversed in order for higher scores to reflect higher degrees of coping, helplessness, or hopelessness. Mean scores were calculated for the helplessness and hopelessness subscales.
Illness perceptions
Illness perceptions were measured using the 9-item Brief Illness Perception Questionnaire (BIPQ), where each item measures a different dimension of illness perceptions: Consequences—how much the illness affect your life; timeline—how long you believe the illness will last; personal control—how much control you feel over the illness; treatment control—how much you think treatment can help the illness; identity—how much you experience symptoms from the illness; concern—how concerned you are about the illness; coherence—how well you understand the illness; emotional response—how much the illness affects you emotionally; and a causal attribution item [42]. Items 1–8 were scored on 11-point scales ranging from 0 to 10, with higher scores reflecting an increasingly threatening view of the illness. Item 9 was open-ended and concerned causal attribution: “Please list in rank-order the three most important factors that you believe caused your illness”. Participants who did not perceive themselves as having any illness, were told to skip this questionnaire.
Data analyses
Descriptive statistics were calculated for the total sample, and by gender. Gender differences were analyzed by chi-square tests for the dichotomous variables and independent t-tests for the continuous variables.
Responses to the open-ended item in the BIPQ regarding causal attribution were categorized using thematic analysis, as described by Joffe & Yardley [43]. Themes were identified and data was categorized into coding categories using a descriptive and inductive approach. A coding manual including category definitions was prepared (Additional file 1), and categorization was performed independently by two authors to determine inter-rater reliability. In cases of inconsistency, categorization was discussed until consensus was reached.
In order to maintain the anonymity of respondents, values with fewer than five respondents are not reported.
Statistical analyses were performed using IBM SPSS Statistics, Versions 24.0 and 25.0. The significance level was set to α = .05.