Informed by guidance on the development of complex healthcare interventions [23] and feasibility studies [24], we conducted coMforT in two overlapping phases (Figure 1).
<INSERT FIGURE 1 HERE>
We published a detailed protocol elsewhere [25] and registered the feasibility trial [26]. An advisory group of women with lived experience of DVA advised the study team. A steering committee of independent experts oversaw the conduct of the study and advised on progression to a full-size RCT.
Qualitative methods
We conducted qualitative semi-structured interviews in both phases. Interviews were audio-recorded, transcribed verbatim, anonymised, and uploaded into NVivo 12 software. We used inductive and deductive coding, applying constant comparison techniques [27, 28], to develop themes about trauma-specific adaptations to the curriculum (intervention refinement phase), and about the acceptability of the intervention and trial procedures (feasibility trial). We applied a framework method [29]. Our initial analytical coding frame was based on the framework for modifications and adaptations of evidence-based interventions, and the framework of mindfulness-based programmes fidelity [30]. We organised codes under two analytical themes of trauma-specific adaptations to the curriculum content and course context. We applied the initial coding frame across all qualitative datasets and refined it as the analysis progressed.
Phase one. Intervention refinement
The aim of Phase one was to develop a TS-MBCT intervention that is acceptable to the DVA population and feasible to deliver in an RCT. The intervention refinement work involved evidence synthesis from an updated 2014 literature review, qualitative interviews with women and professionals, and a 2-stage consensus exercise with experts in trauma and mindfulness (Figure1, table 1).
Table 1. Phase one. Intervention refinement. Study assessments
Procedures
|
Timepoint
|
Measurement tool
|
Socio-demographic data -women
|
At the start of the qualitative interviews
|
Bespoke questionnaire (age, ethnicity, education, employment, relationship status, time since end of abusive relationship, previous talking therapy)
|
Face-to-face/phone semi-structured qualitative interviews with women
|
Before the first mindfulness group
|
Topic guide exploring experiences of recovery after DVA, and any psychological support received (Supplementary file 1)
|
Socio-demographic data -professionals
|
At the start of the qualitative interviews
|
Bespoke questionnaire (age, gender, job title, years of delivering psychological therapy to people affected by violence and trauma)
|
Face-to-face/phone/Skype semi-structured qualitative interviews with professionals
|
Before the first mindfulness group
|
Topic guide exploring experiences of providing psychological interventions to people affected by violence and trauma (Supplementary file 1)
|
Online survey with experts in trauma and mindfulness
|
Before the first mindfulness group
|
Bespoke questionnaire: years of practice, 15 statements about trauma-specific modifications to a standard mindfulness-based cognitive therapy course grouped under three categories: participants, curriculum, and teacher. 5-point Likert scale (strongly disagree to strongly agree), free text comments (Supplementary file 2)
|
Remote consultations with experts in trauma and mindfulness
|
• Before feasibility trial
• Six months post-randomisation
|
• Agenda covering seven statements from the survey that produced contention
• Agenda covering areas of uncertainty identified through the process evaluation of the trial mindfulness groups
|
Data collection and analysis
Literature review
We searched MEDLINE for studies of any design that evaluated mindfulness-based programmes for people affected by childhood abuse or DVA. Results and discussion sections of included papers were treated as primary qualitative data with deductive coding of text which reported trauma-specific adaptations to standard mindfulness interventions.
Qualitative semi-structured interviews with women and professionals
Interviews explored participants’ views on proposed TS-MBCT-1. Recruitment via email to patient and professional networks drew a purposive sample of (i) women aged 18+ years who self-identified as having experienced DVA and mental health problems, and (ii) professionals working with people affected by trauma. Researchers conducted interviews at university offices or online. Topic guides included a vignette about the TS-MBCT-1 prototype and questions about intervention feasibility, acceptability, barriers, and facilitators to women, group logistics, and trial procedures (Supplementary file 1).
2-stage consensus exercise with experts in trauma and mindfulness
Our literature review and qualitative interviews identified conflicting perspectives on proposed adaptations and raised new questions about the inclusion/exclusion of participants, the content and format of the TS-MBCT-1 prototype, and the teachers’ training and qualifications. The aim of the consensus exercise was to develop pragmatic solutions for further intervention refinement and trial procedures. We sent an expression-of-interest email with an online survey to networks of mindfulness teachers inviting those who had taught mindfulness-based programmes to people affected by trauma, including DVA. The anonymous survey included 15 statements relating to the remaining uncertainties (Supplementary file 2). We imported the responses into Microsoft Excel and analysed quantitative data with descriptive statistics and narrative text. We manually coded free text comments on the framework themes of trauma-specific adaptations to curriculum content and context. We then invited survey respondents to two virtual consultations with the study researchers and the mindfulness teacher. In the first consultation, attendees formulated pragmatic resolutions of conflicting statements. In the second, carried out post-trial, the respondents agreed further refinements to the intervention. Discussion about remaining areas of uncertainty and agreed solutions were summarised narratively.
Evidence synthesis
We integrated the findings from the literature review, qualitative interviews, and consensus exercise into an evidence matrix. We categorised evidence from the three sources as no evidence, limited evidence (single study), agreement (different studies/contributors reported converging perspectives) and disagreement (different studies/contributors reported diverging or conflicting perspectives). We further scrutinised gaps in the evidence, limited evidence, and disagreements until consensus was reached in the form of agreed terms and conditions.
Phase two. Feasibility trial
We conducted an individually randomised parallel group feasibility trial with pre-specified progression criteria and a traffic light system, and an embedded mixed-method process evaluation and health economics evaluation. Women were screened, assessed at baseline, randomised 2:1 into intervention (TS-MBCT group) and control (self-referral to psychological therapy in the National Health Service (NHS)), and followed-up 6 months post-randomisation. We ran two consecutive TS-MBCT groups, with findings from the interim process evaluation informing refinements of the TS-MBCT curriculum and trial procedures between the two groups (Figure 1).
Data collection
We used convenience sampling to recruit trial participants during 2-month periods (plus 1 buffer month) prior to pre-scheduled TS-MBCT group dates.
Initial screening
The recruitment sites were four charitable-sector specialist DVA agencies in Southwest England, UK. The study researchers trained agency support workers on the recruitment protocol, and regularly visited the sites. To ensure that potential participants received specialist DVA support, agency workers were asked to carry out an initial screening of all female clients near to case closure, or on a waiting list for any of the agency services (Table 2).
Table 2. Phase two. Feasibility trial with embedded process evaluation and economic evaluation. Study assessments
Procedures
|
Initial screening by recruitment site worker
|
Final screening by study researcher
|
Baseline assessment by study researcher
|
6-month follow up by study researcher
|
Measurement tool
|
Eligibility assessment
|
Age
|
x
|
|
x
|
|
Bespoke socio-demographic questionnaire: age, ethnicity, education, employment, marital status, children, housing, and financial situation
|
Speaking and understanding English
|
x
|
|
|
|
Recruitment site worker’s judgement
|
Diagnosed psychosis, bipolar disorder, personality disorder
|
x
|
|
|
|
Recruitment site case record
|
Current psychological therapy
|
x
|
|
|
|
Recruitment site case record
|
Readiness to start mindfulness group or alternative talking therapy on the NHS
|
x
|
|
|
|
Woman’s own judgement
|
Post-traumatic stress disorder (PTSD)
|
|
x
|
|
|
The Primary Care PTSD Screen for DSM-5 (PC-PTSD-5) [31]
|
Current alcohol dependence
|
|
x
|
|
|
The Alcohol Use Disorders Identification Test Consumption (AUDIT-C) [33]
|
Current drug dependence
|
|
x
|
|
|
The Drug use Disorders Identification Test (DUDIT) [34]
|
Suicide history
|
|
x
|
|
|
Bespoke questions:
• I made plans to end my life in the last 2 weeks
• I made attempts to end my life in the last 12 months
|
Trial assessments
|
Live events
|
|
|
x
|
|
The Life Events Checklist for DSM-5 (LEC-5) Standard [50]
|
Clinically important symptoms of PTSD
|
|
|
x
|
x
|
The PTSD Checklist for DSM-V (PCL-5) [50]
|
Clinically important symptoms of complex PTSD
|
|
|
x
|
x
|
The International Trauma Questionnaire (ITQ) [51]
|
Depression
|
|
|
x
|
x
|
The Patient Health Questionnaire-9 (PHQ-9) [60]
|
Anxiety
|
|
|
x
|
x
|
Generalized Anxiety Disorder-7 (GAD-7) [61]
|
Adverse childhood experiences
|
|
|
x
|
|
Brief screening version of the Childhood Trauma Questionnaire [62]
|
Domestic violence and abuse
|
|
|
x
|
x
|
Composite Abuse Scale Revised-Short Version [52]
|
Dissociative symptoms
|
|
|
x
|
x
|
The Severity of Dissociative Symptoms—Adult (Brief Dissociative Experiences Scale [DES-B]—Modified) [63]
|
Self-compassion
|
|
|
x
|
x
|
Self-Compassion Scale – Short Form (SCS-SF) [64]
|
Embedded health economic evaluation
|
|
Health related quality of life, women
|
|
|
x
|
x
|
EQ-5D-5L [35]
|
Health-related quality of life, index child
|
|
|
x
|
x
|
KIDSCREEN-10 Index. Health Questionnaire for Children and Young People. Parent Version [36]
|
Resource use
|
|
|
|
x
|
Bespoke
|
Intervention cost
|
|
|
|
|
Study documentation
|
Embedded process evaluation
|
|
Qualitative semi-structured interviews with trial participants
|
|
|
|
x
|
Topic guide about experiences of the trial and interventions (Supplementary file 3)
|
Feedback form (intervention arm)
|
|
|
|
x
|
Bespoke
|
<INSERT TABLE 2 HERE>
Initial eligibility criteria were: female, aged 18+ years old, able to speak and understand English, no diagnosis of psychosis or bipolar or personality disorder, no current psychological therapy, feeling ready to attend a mindfulness group or an alternative NHS psychological therapy. The agency workers emailed the research team with contact and safety information of clients who met the initial eligibility criteria, were interested in the study, and agreed to the sharing of their details.
Final screening
The study researcher made four attempts to contact each potential participant and arrange face-to-face screening at a safe and convenient location. Women completed four questionnaires. At the final screening stage, we included women who self-reported clinically important symptoms of PTSD on the Primary Care PTSD Screen for DSM-5 (PC-PTSD-5) [31]. They did not have to have a formal diagnosis of PTSD established by a clinician [32]. We excluded women who had current alcohol dependency on the Alcohol Use Disorders Identification Test (AUDIT) [33], drug dependency on the Drug Use Disorders Identification Test (DUDIT) [34] or who had made suicide plans or attempts in the last 12 months.
Baseline assessment
Women completed baseline questionnaires (Table 2). If the woman had children, the researcher used a bespoke Access database to randomly select an index child for inclusion in the study.
Randomisation
The recruiting researcher phoned/emailed the Bristol Trials Centre (BRTC) remote randomisation service. The BRTC Database Manager used blocked randomisation to form the allocation list for the two trial arms. A computer random number generator was used to select random permuted blocks with a block size of 18 and a 2:1 ratio. The researcher notified women of their allocation, and conveyed details of those randomised into the intervention arm to the mindfulness teacher. Due to the nature of the psychological interventions in both arms, it was not possible to blind participants, intervention providers, or researchers to the group allocation.
Follow-up 6-month post-randomisation
A study researcher maintained contact with participants via monthly check-in texts, emails, and phone calls. Participants received a £20 shopping voucher and reimbursement of travel and childcare expenses for attending each assessment and intervention session. At 6 months post-randomisation, the researcher arranged a face-to-face assessment, at the woman’s choice of location. Women completed follow-up questionnaires. Due to COVID-19 lockdown, one participant received and returned the questionnaire via post.
Process evaluation
We collected process data after the first TS-MBCT group in the intervention arm, and at the 6-month follow-up in both arms. Women in the intervention arm completed a Home Practice Feedback Form after each session and an Evaluation Form at the end of the course. During the monthly check-in contacts, a study researcher obtained women’s permission to be contacted for a qualitative interview, and conveyed these details to two qualitative researchers who arranged face-to-face interviews in university offices or the woman’s home. The feedback forms and interviews (Supplementary file 3) explored women’s experiences of the interventions and trial procedures.
Health economics evaluation
We recorded information about the resources used in the delivery of the TS-MBCT intervention. At 6-month follow-up, women completed a bespoke resource use questionnaire capturing theirs, and an index child’s, use of health and social care services over the study period. Quality of life was measured with the EQ-5D-5L (women) [35] and KIDSCREEN (an index child) [36] questionnaires during the baseline and 6-month follow-up assessments.
Outcomes and measures
Feasibility outcomes
We calculated the rates of recruitment (primary feasibility objective), intervention uptake, retention, and follow-up, and qualitatively assessed the acceptability of the intervention and trial design. We extracted data from study logs, mindfulness teacher records, TS-MBCT Feedback and Evaluation Forms, and administrative NHS data.
Candidate clinical outcomes in a full trial
We tested standardised validated questionnaires for measuring clinically important symptoms of PTSD and CPTSD (candidate primary outcomes), dissociative symptoms, depression, anxiety, DVA re-experience, and self-compassion (candidate secondary outcomes).
Adverse events
During the monthly checks, 6-month follow-up, and process evaluation interviews, study researchers asked women about any adverse events. The TS-MBCT teacher was available at the beginning and end of sessions, and between sessions, if women felt that their symptoms were worsening and wanted to talk. Adverse events were processed according to the trial protocol [25].
Analysis
We entered questionnaire responses into REDCap software, and prepared and analysed quantitative data in STATA16. To give an acceptable precision for estimating the proportion of eligible women consenting to participate, we aimed to screen 120 women and to recruit 36 (recruitment ratio: 30% of those eligible). Assuming a 33% drop out rate before the start of the intervention, we inflated the sample size to 54 [25]. Following recommendations on developing progression criteria for pilot studies [37], we prespecified progression criteria and a traffic light system for decision-making about progression to a full trial. All green targets met: progress to the full trial. Green and/or amber targets met: Study Steering Group discuss problems with study management group, and progress if strategies are available to improve the intervention and trial protocol. Red targets met: consult Steering Committee and funder if progression is justified [25]. We reported feasibility outcomes by mapping on the progressions criteria and traffic lights. For criteria that were based on quantitative and qualitative data, we integrated findings through joint display of statistics-by-themes [38]. Since this was a feasibility study, we did not perform significance testing for effect sizes in candidate clinical outcomes for a full trial, instead calculating descriptive statistics with 95% CI. We reported differences in means and 95% CIs between the two arms at 6 months.