Design
We conducted a three-armed parallel pilot RCT in the Municipality of Aarhus in Denmark (340,421 inhabitants) and intended to include 24 participants in each trial arm, in total 72 participants. We assessed that this sample size was justified to meet the objectives of our pilot trial [24]. The inclusion criteria were wide as reflecting the real-life setting. Participants had all contacted Aarhus Municipal Health Care Center due to stress-related problems. Furthermore, participants had to be 18 years or older and able to understand, speak, and read Danish. Exclusion criteria were: 1) acute treatment-demanding clinical depression or a diagnosis of psychosis or schizophrenia, 2) abuse of alcohol, drugs, medicine, and 3) pregnancy.
Process of the study
The municipal health care center referred all individuals who contacted Aarhus Municipal Health Care Center with stress-related problems between January and April 2018 to the researchers in the project at Aarhus University. A member of the project team contacted each individual by telephone and gave information about the project according to a written guideline. Using a web-based booking system, individuals who met the inclusion and exclusion criteria had an appointment for baseline measurements. They also received detailed written information about the project via e-mail. At the appointment for baseline measurements, a member of the project team also verbally informed the individuals about the project to make sure that individuals fully understood the implications of participation. All attendants at the appointment for baseline measurements agreed to participate and provided written consent. A project team member collected baseline measurements and randomisation was performed. Data collection and randomisation were conducted using the REDCap electronic data capture tools hosted by Aarhus University. REDCap (Research Electronic Data Capture) is a secure, web-based application designed to support data capture for research studies [25]. An independent data manager had programmed the randomisation algorithm in REDCap. The randomisation was stratified for self-reported history of mental disorder. Previous research indicates that self-reported history of mental disorder is associated with a smaller effect of MBSR [17].
Two different staff members performed the baseline measurements using written guidelines, initiated the randomisation in REDCap and informed participants of the results of the randomisation process. The same staff members collected follow-up measurements. Hence, the data collectors were not blinded for group assignment.
We conducted baseline measurements and randomisation in two rounds defined by time-frames to achieve similar intervention and measurement courses of events. In the first round, baseline measurements were conducted between 12 March 2018 and 23 March 2018, interventions between 10 April 2018 and 29 May 2018, and 12-week measurements were conducted between 4 June 2018 and 15 June 2018. In the second round, baseline measurements were conducted between 3 April 2018 and 16 April 2018, interventions between 25 April 2018 and 20 June 2018. Twelve-week measurements were conducted between 18 June 2018 and 3 July 2018. All measurement were made at baseline and at 12 weeks.
Interventions
The PaT Plot [26] in Fig. 1 shows research activities in all three randomised groups.
MBSR
The curriculum- and evidence-based MBSR course can only be delivered with fidelity by a trained MBSR teacher. It consists of 2.5-h weekly group sessions during 8 weeks with one seven-hour silent retreat day and 45–60 min of daily homework for 6 days a week [13]. The orientation session was performed after randomisation. An MBSR teacher from the Danish Center for Mindfulness taught MBSR in the current pilot trial. She was not a member of the research group. She received supervision and had co-taught two MBSR circles with a MBSR teacher from the research group in order to secure MBSR fidelity. At Danish Center for Mindfulness, the current practice is to allow up to 30 participants in each MBSR group in accordance with recommendations [13]. In order to reflect the usual group size of MBSR, which is twice as high as the group size in LSR, participants in the MBSR course participated in groups along with approximately 12 self-paying participants at the Danish Center for Mindfulness at Aarhus University.
LSR
The LSR elements are not curriculum-based but developed by psychologists employed in the municipality to provide interventions to people experiencing stress-related problems, and the content of the interventions may vary. At the time of this study, LSR was based on the principles of the psychological therapeutic method ACT [23]. It was an eight-week group-based course with 12 participants and weekly 2.5-h sessions. Between the sessions, participants were recommended to spend 45 min daily on homework. Prior to group enrollment, all participants had an individual consultation with the psychologists teaching the LSR. The effect of LSR and inter-therapist variation have not previously been evaluated.
MBSR and LSR have many similarities, such as structure, timeframe, amount of homework and the group format. Both interventions also include mindfulness training as a central component and encourage participants to create a different relationship with difficult experiences. The main differences between the interventions relate to the content of the interventions and the duration of the meditations. Furthermore, in the LSR intervention, emphasis is placed on a more psychoeducational, or cognitive approach, whereas MBSR is mainly practice-based as the aim is to move towards an embodied experience rather than conceptual knowing of mindfulness [27]. We did not monitor intervention fidelity in this study.
The waiting list group
We offered the participants allocated to the waiting list group to participate in either LSR or MBSR in August of 2018.
To reflect real-life, there were no restrictions for any of the participants in the three trial arms regarding participation in other stress-reduction interventions during the trial period.
Outcomes
Primary outcomes of this pilot RCT were: 1) recruitment rate, 2) retention rates regarding intervention participation, 3) retention rates regarding follow-up at 12 weeks, and 4) proportions of allocated participants who had participated in other stress reduction interventions during the trial period. Secondary outcomes were: 1) completion and score rates of study questionnaires, and 2) proposed effect estimates of outcome measures considered to be used in the real RCT.
Self-reported measures
The perceived stress scale (PSS)
Cohen’s 10-item perceived stress scale is a self-report measure of subjective stress [8]. It consists of ten questions indicating how often respondents have found their life unpredictable, uncontrollable, and overloaded in the past month. All items are scored on a five-point Likert scale (total sum scores: 0–40), and higher scores indicate higher levels of stress. The instrument has demonstrated good validity and reliability [28,29,30]. Scores of PSS have been associated with mortality in a dose-response relationship [7]. Cronbach’s α was 0.84 in the present study sample.
The Hopkins symptom check List-5 (SCL-5)
SCL-5 is a five-item self-report measure of symptoms of anxiety and depression [31]. All items are scored on a four-point scale, ranging from 1 (not bothered at all) to 4 (extremely bothered). The score is calculated as the average of the five items with higher scores indicating greater symptoms of anxiety and depression. The SCL-5 originates from the 25-item Symptom Check List (SCL) and correlates at r = 0.92 with the SCL. The alpha reliability for the SCL-5 has been found to be 0.85 [32] A SCL-5 score > 2 has been found to predict mental illness as assessed independently by psychiatrists [32]. Cronbach’s α was 0.85 in the present study sample.
The WHO-5-wellbeing scale (WHO-5)
WHO-5 is a five-item self-report measure of wellbeing. It consists of five questions indicating the extent to which respondents have been feeling well during the last 2 weeks. Each question is scored on a five-point scale indicating how often respondents have experienced specific feelings. The points are added and multiplied with four, calculating the total score ranging from 0 to 100; higher scores indicating higher level of wellbeing. The WHO-5 wellbeing scale is considered to be a valid measure of the overall wellbeing of respondents [33]. Cronbach’s α was 0.78 in the present study sample.
The brief resilience scale (BRS)
The BRS is a six-item self-report measure of resilience [34]. The BRS consists of six questions. All items range from 1 to 5 (total sum range 6–30). A summary score is created that averages across the six items (range = 1–5), with higher scores indicating a greater perceived recovery from stress [34]. The following cut-off points have been suggested: Scores from 1.00–2.99: low resilience; 3.00–4.30: normal resilience; 4.31–5.00: high resilience [34]. Cronbach’s α was 0.82 in the present study sample.
The Amsterdam resting state questionnaire (ARSQ)
The ARSQ is a self-report questionnaire to sample thoughts and feelings during rest, i.e., an awake state characterised by the absence of goal-directed cognitive activity. The scale consists of 21 statements scored on a Likert scale from 1 (completely disagree) to 5 (completely agree) [35]. The ARSQ identifies seven dimensions of resting state cognition: Discontinuity of Mind, Theory of Mind, Self, Planning, Sleepiness, Comfort, and Somatic Awareness. During the resting state, the mind typically wanders in a way that represents habitual ways of thinking. MBSR targets those habitual, normally persistent patterns of thoughts and feelings, hence the programme has the potential of inducing change. In the present study sample, Cronbach’s α was: Discontinuity of Mind (0.86), Theory of Mind (0.59), Self (0.46), Planning (0.77), Sleepiness (0.79), Comfort (0.85), and Somatic Awareness (0.71).
The self-compassion scale (SCS) - short form
The SCS is a 12-item self-report measure of self-compassion [36]. It originates from the Self-Compassion Scale (SCS) and has shown good internal consistency and a close to perfect correlation with the long SCS [36]. Self-compassion is proposed to be a contributing mediator of the effect of mindfulness [27] and increased self-compassion has been found to mediate the beneficial effect of MBCT on post-treatment symptoms of depression [37]. Cronbach’s α was 0.87 in the present study sample.
The five facet mindfulness questionnaire (FFMQ-15)
FFMQ-15 is a 15-item self-report measure of the dispositional tendency to be mindful in daily life [38]. It is developed from the original FFMQ-39 and has been found to be reliable and valid [38]. It consists of five facets of mindfulness including observing, describing, acting with awareness, non-judgment and non-reactivity. Cronbach’s α was 0.86 in the present study sample.
The experiences questionnaire (EQ) - decentering sub scale
The EQ - decentering sub scale is a validated 11-item self-report measure of decentering [39]. Decentering refers to the ability to observe thoughts and feelings as temporary and automatic events in the mind, rather than facts or true descriptions of reality. The items of the decentering factor assess three facets: the ability to distinguish one’s self from one’s thoughts, the ability not to automatically react to one’s negative experiences and the capacity for self-compassion. All items are scored on a five-point Likert scale ranging from 1 to 5 and the total sum score ranges from 11 to 55. Cronbach’s α was 0.81 in the present study sample.
Clinical measures
Blood pressure
Brachial blood pressure was measured after a 10-min rest with an automated blood pressure monitor. Three measurements were made and the average of each pressure index constitutes the values of systolic and diastolic blood pressure in mmHg.
Weight
Weight was measured twice without shoes in light indoor clothing to the nearest 0.1 kg. The measurements were completed twice, and the average constitutes the value of weight in kg.
Waist circumference
Waist circumference was measured with the participant in the standing position at the mid-point between the lower costal margin and the level of the anterior superior iliac crest to the nearest millimeter. The measurements were completed twice, and the average constitutes the value of waist circumference in cm.
The rationale for collecting the clinical outcome data was the allostatic load theory.
The allostatic load theory is a theoretical framework for physiological pathways that may explain relations between mental and physical well-being [40,41,42]. Furthermore, an allostatic load index has been developed, consisting of both primary mediators of the stress response in conjunction with clinically relevant biomarkers representing secondary outcomes. The index has been found to be a strong predictor of disease development and mortality [43, 44]. Blood pressure, weight and waist circumference are parts of this index.
Data analysis
We estimated proportions and proposed effect estimates with 95% confidence intervals (CIs). We estimated the participation rates in the two intervention arms using two cut-off points: participation ≥5 sessions and participation ≥7 sessions. Furthermore, we estimated medians with 1st and 3rd quartiles of the total number of sessions. We performed a loss to 12-week follow-up analysis for age, gender, educational level, employment, sick-leave, history of mental disorder, participation in ≥5 sessions of allocated intervention, and pre-scores of PSS, SCL-5, WHO-5 and BRS by t-test and chi2-test. We also performed a loss to course participation analysis (participation in < 5 sessions of allocated intervention) for age, gender, educational level, employment, sick leave, history of mental disorder, and pre-scores of PSS, SCL-5, WHO-5 and BRS by t-test and chi2-test.
We estimated the mean changes in the mental health and clinical outcomes from baseline to 12-week follow-up in the three groups. We compared the differences in the mean change in the mental health outcomes between the groups using linear regression models adjusting for age, sex, educational level, history of mental disorder and baseline scores of mental health (PSS, SCL-5, WHO-5 and BRS). When we compared the differences in the mean change in the clinical outcomes between the groups, we adjusted for age, sex, educational level, history of mental disorder, baseline blood pressure, body mass index (BMI), weight and waist circumference using linear regression models. We adjusted for the above potential confounders because they were not evenly distributed at baseline across the three groups due to the small study sample. We present proposed effect estimates with 95% CIs without p values as suggested by the CONSORT statement, as pilot trials are not powered for testing hypotheses about effectiveness [22]. Bonferroni correction should be considered in a real effectiveness RCT. We did not make analyses to take into account missing data. All analyses were performed in STATA 14.