Protocol for a mixed methods study to examine social workers' implementation of national guidelines with children in care who self-harm

Background/rationale

Child self-harm, involving intentional self-poisoning or injury, irrespective of motivation (NICE, 2022a), in young people under the age of 18, is an increasing problem globally (Cairns et al., 2019; Griffin et al., 2018; McManus et al., 2019; Morgan et al., 2017). Care-experienced children, defined as children who have spent more than 24 hour in the care of the local authority (NSPCC, 2024) are at greater risk of self-harm (Stanley et al., 2005), with studies indicating a two and four-fold higher risk compared with the general population (Evans et al., 2017; Hjern et al., 2004; Katz et al., 2011; Pilowsky & Wu, 2006; Stanley et al., 2005; Vinnerljung et al., 2006). Despite clear evidence of elevated self-harm and suicide risk among care-experienced children, the assessment and management of self-harm during periods of statutory care remains markedly under-researched (Johnson et al., 2017). Review evidence identifies a persistent gap in the literature, with no epidemiological studies reporting the prevalence of self-harm specifically while children are in care (Evans et al., 2017).

The National Institute for Health and Care Excellence (NICE) guidelines on self-harm: assessment, management, and preventing recurrence (NICE, 2022a) aim to improve the quality and care of individuals who self-harm in all services and settings, including children in care. The guideline provides evidence-based recommendations for practitioners on 15 areas (displayed in Table 1), including information-giving, psychosocial assessment, safety planning, and requirements for training, supervision, and consistent implementation.

Although the guidelines are advisory and are not intended to supersede other national or local processes and procedures involved in safeguarding and decision-making, they should be fully considered when making professional judgments and in consultation with individuals, their families, and carers. The current NICE guideline on self-harm (NG225): assessment, management and preventing recurrence updates and replaces two earlier guidelines: CG16 (2004), which addressed short-term management, and CG133 (2011), which focused on longer-term management of self-harm. Most of the recommendations in NG225 reinforce best practices and do not require additional resources to implement if CG16 and CG133 have been followed (NICE, 2022b).

This study addresses several gaps in the literature. A recent rapid systematic review, conducted by the authors, revealed an absence of research on social workers’ implementation of the NICE guidelines on self-harm (NG225, CG16 and CG133) and their implementation with children (Lever et al., 2025b). Whilst NICE guidelines purport to be evidence-based, much of the evidence derives from clinical studies and may not be generalisable to other areas (Scullard et al., 2011), such as residential children’s homes.

Adherence to NICE guidelines on self-harm is an under-researched area, and to date, no studies on awareness/implementation rates of NG225 have been published (Lever et al., 2025b). Implementation research further demonstrates a persistent practice gap, generally, whereby guideline recommendations are applied inconsistently or not used at all, across professional contexts (Nilsen, 2015; Eccles & Mittman, 2006). Together, these gaps emphasise the need for empirical investigation into how NG225 is used within children’s social care.

Theoretical framework

Implementation science focuses on how, why, and under what conditions guidance is adopted, integrated into everyday practice and sustained over time. The COM-B model proposes that for implementation to occur, practitioners need to have the capability (knowledge and skills), opportunity (time, organisational support, and enabling environments) and motivation (beliefs about importance and priorities) to carry out the recommendations (Michie et al., 2011). Existing evidence suggests that awareness and implementation of NICE guidance on self-harm among professionals is often low (Lever et al., 2025b) and influenced by factors related to capability, opportunity and motivation (Leather, 2022). Examining these behavioural determinants in the context of social workers’ implementation of the NG225 with children in care is therefore critical for identifying specific modifiable barriers and facilitators, determining whether recommendations are operationalised, and informing the development of targeted implementation interventions.

Aims and objectives

This mixed-methods study will provide insights into the management and prevention of self-harm in children in care. Exploring the barriers and facilitators to implementing NG225 with children in care by social workers offers a unique contribution to the knowledge base, whilst addressing key limitations arising from the lack of diverse populations and sectors in the literature.

The study has four objectives:

Research Questions are:

It is hypothesised that social workers and students of social work who have recently supported children in care who have self-harmed will demonstrate low levels of awareness and implementation of NG225. This expectation is informed by findings from a rapid systematic review showing that health professionals’ awareness and use of NICEself-harm is generally low (Lever et al., 2025b). This pattern is consistent with wider evidence highlighting limited professional awareness of policies and guidelines (Keyworth et al., 2018) and inconsistent implementation (Nilsen, 2015; Eccles & Mittman, 2006).

Study design

Here, we outline a protocol to explore social workers’ knowledge of the NICE guidelines on self-harm and the barriers and facilitators to implementing these guidelines with children in care, using a sequential explanatory mixed-methods study. First, a large-scale survey will be conducted. Objectives 1–3 will be addressed using quantitative survey data, and objective 4 will be partially addressed through qualitative data generated from a free-text item designed to capture preliminary information on facilitators and barriers to implementation. In phase 2 of the study, semi-structured interviews will explore these issues in depth to generate an understanding of the underlying mechanisms and to contextualise the quantitative findings of determinants of implementation, with integration at the interpretation stage. Analysis of the Phase 1 findings will inform Phase 2 through purposive sampling, ensuring the full spectrum of guideline awareness and implementation experiences salient COM-B determinants will also influence refinement of the interview topic guide, enabling in-depth exploration and explanation of quantitative patterns. Preliminary Phase 1 analyses will be conducted prior to finalising the Phase 2 interview guide. This ensures that Phase 2 builds directly on the insights from Phase 1, enabling a richer understanding of implementation challenges and supporting the identification of targetable behaviour change techniques that are relevant across diverse practice contexts.

This protocol is reported in accordance with the ObsQual reporting checklist (included in the supplementary material) and the study will be reported in line with STROBE, COREQ, and GRAMMS guidelines as appropriate.

Trauma-informed approach

A trauma-informed approach informs the design and ethical delivery of the research. A trauma-informed approach takes account of the impact trauma has had on people’s lives (Dowding, 2021). Social workers are at elevated risk of both primary and vicarious trauma. The helping professions generally attract people with significant life adversities (Newcomb et al., 2015), and “the negative effects of caring about and caring for others” (Perlman & Maclan, 1995, p.31) are well documented in the social work profession (Dykes, 2016; Esaki & Larkin, 2013; Peart, 2023; Sellers & Hunter, 2005; Thomas, 2016). The key principles of trauma-informed practice: safety, trust and transparency, choice, collaboration, and empowerment (Fallot & Harris, 2009) inform the research design, implementation, data analysis, and dissemination of the research findings.

Theoretical Domains Framework (TDF)

The TDF (Cane et al., 2012; Michie et al., 2005) integrates 33 theories of behaviour, incorporating 84 theoretical constructs of behaviour change and culminating in 14 distinct theoretical domains (1. knowledge, 2. skills, 3. social/professional role and identity, 4. beliefs about capabilities, 5. optimism, 6. beliefs about consequences, 7. reinforcement, 8. intentions, 9. goals, 10. memory, attention, and decision processes, 11. environmental context and resources, 12. social influences, 13. emotion, and 14. behavioural regulation). Its theoretical basis promotes the elicitation of beliefs in data collection (Dyson et al., 2011) and the identification of behaviour determinants that are instrumental in behaviour change interventions (Michie et al., 2005). The TDF is recommended as a framework for interviews to identify factors influencing behaviour (Michie et al., 2014). In this study the TDF guides data collection (interview schedule) and qualitative analysis (coding framework).

Capability, Opportunity, and Motivation (COM-B) model of behaviour

The COM-B model consists of three interacting factors – capability, opportunity, and motivation – which must be present to generate behaviour (Michie et al., 2011). The three key domains are further divided into six subdomains: physical and psychological capability, physical and social opportunity, and reflective and automatic motivation. The COM-B model is purposefully designed to be broad and flexible whilst being rooted in the scientific theory of behaviour implementation, ensuring its methodological rigour. The COM-B model of behaviour is widely used in public health and policy to identify which factors need to be targeted for a behaviour change intervention to be effective (West & Michie, 2020). Although the COM-B model is a distinct framework from TDF (Tomas et al., 2023), it is complementary (Fahim et al., 2020; Ojo et al., 2019), enabling the TDF to be condensed into its three key domains: capability, opportunity, and motivation. The TDF domains of knowledge, skills, memory, attention, decision processes, and behavioural regulation map onto capability. Environmental context, resources, and social influences map onto opportunity, while intentions, goals, beliefs about capabilities, optimism, beliefs about consequences, reinforcement, social/professional role, and identity, and emotion map onto motivation. In this study, the COM-B model informs survey measures and interview schedule (adapted from Keyworth et al., 2020) and analysis, examining associations between COM-B constructs and NG225 implementation.

Phase 1 – Survey study

The survey will be conducted online between February 2025 – February 2026. Participants will be purposefully recruited during the same period using a multi-modal recruitment strategy, comprising letter drops and emails to every local authority children’s service, child and adolescent mental health service (CAMHS), and social work degree or apprenticeship training provider in England, alongside targeted social media advertisements, and the British Association of Social Workers (BASW) e-bulletin (distributed to approximately 22,000 social workers and social work students). This approach is designedto yield a large sample size, mitigate the margin of error, and produce meaningful results. As online surveys tend to generate low response rates, typically below 20% (Fricker et al., 2005), we offer an incentive to participate for the chance to win a £50 gift voucher through a prize draw. This is included as a proportionate strategy to support modest improvements in engagement (Abdelazeem, et al., 2023; Singer & Ye, 2013).

Data collection

The survey (provided in the supplementary material) consists of 19 questions (three screening questions (1–3), two assessing awareness and implementation levels (4–5), and six assessing opportunity, motivation, and capability to implement the NICE guidelines on self-harm, adapted from Keyworth et al., 2020 (6–11) based on the COM-B model (Michie et al., 2011); a free response question asking participants if anything has helped or hindered NICE guideline implementation with children in care who self-harm (12); seven questions addressing demographic information (13–19): area of social work specialism, number of years of experience, regional area of work, age, ethnicity, gender and disability status).

Items 16–19 ask about personal information of protected characteristics, constructed using Diversity and Inclusion Survey (DAISY) guidance (EDIS & Wellcome, 2022) and Sex and Gender Equity in Research (SAGER) guidelines (Heidari et al., 2016). These questions contain an optional response of “prefer not to respond.” 18 of 19 items are forced-choice questions, requiring participants to insert a response, meaning items cannot be skipped. Item 12, the free response text item, is the exception.

Inclusion criteria are occupation as a children’s social worker or student of social work, practising in England, and having worked, within the last 12 months, with at least one child in care who engaged in self-harm (i.e., intentional self-poisoning or self-injury) during this time. The study is restricted to social workers and students of social work who have recently supported a child in care who has self-harmed because the COM-B constructs (capability, opportunity, and motivation) can only be meaningfully assessed among practitioners who have had a recent opportunity to implement the NG225 in practice. The 12-month timeframe was selected to minimise recall bias and to ensure that reported awareness and implementation of NICE guidelines reflect current practice. Exclusion criteria are social workers practising in the devolved nations and outside of the UK and children’s social workers who have not worked with a child in care who has self-harmed within the last 12 months. Practitioners from the devolved nations and outside of the UK are excluded because statutory definitions of children in care and safeguarding frameworks differ outside of England. The restriction supports contextual comparability and the interpretability of findings.

The primary outcome is the extent to which practitioners implement the NG225 with children in care. Implementation is operationalised as a percentage, whereby respondents indicate on a scale of 0–100 the estimated proportion of children in care who had self-harmed with whom they had implemented the NG225. This approach enables the study to capture partial or inconsistent implementation, which is theoretically important for understanding behaviour change mechanisms.

The six COM-B subdomain (physical capability, psychological capability, physical opportunity, social opportunity, automatic motivation, and reflective motivation) determinants of implementation are secondary outcomes, measured on a 0–10 scale, with higher scores reflecting higher presence of the determinant.

A potential exposure variable is practitioners’ awareness of the NG225, defined as respondents’ self-reported prior knowledge. However, given that professionals often enact guideline-consistent behaviours without explicit awareness of the originating guideline (Keyworth et al., 2018), it is plausible that children’s social workers follow local organisational protocols derived from NICE guidance (NG225) without being explicitly aware of the guideline itself. It is therefore imperative to examine both awareness and implementation, treating them as related but distinct dimensions. The survey is structured to first assess awareness of the NICE guidelines on self-harm, using a 5-point Likert scale (1 = never heard of, 2 = heard of, 3 = know a little, 4 = know a fair amount, 5 = know well) and subsequently assess self-reported implementation. Importantly, respondents can view the NICE guideline before answering the implementation question. This design allows participants who may not have prior explicit awareness of the guidance to still report on whether their practice aligns with guideline recommendations, thereby capturing implementation via organisational protocols, also.

Main predictor variables include respondents’ awareness of the guideline, COM-B determinant scores, and the number of children in care who had self-harmed that respondents have supported.

Potential confounders or effect modifiers in the associations between awareness/COM-B determinants and implementation are: years of professional experience (for qualified social workers), region of work (e.g., North-East, London), area of social work specialism (e.g., safeguarding, child and adolescent mental health), and the total number of children in care who had self-harmed that the respondent has supported in the past 12 months.

Several procedures will be implemented to minimise potential sources of bias in the design, administration, and analysis. The survey is structured to provide initial eligibility screening and uses skip logic to prevent ineligible respondents from contributing data and reducing unnecessary burden. Progress indicators are utilised to mitigate attrition, as shorter perceived task duration is associated with higher completion rates (Burruss & Johnson, 2021; Couper et al., 2001).

Survey items have been reviewed by a trauma-informed expert and public contributors with lived experience of self-harm to ensure language sensitivity and enhance acceptability. The survey was piloted with students of social work to test acceptability, clarity and burden. This informed minor refinements to wording and layout to reduce misinterpretation and cognitive load. Data validity is also aided by enabling word clarification and term definitions via pop-up information when participants click them (Burruss & Johnson, 2021).

We have tried to reduce the burden on respondents by using a checkbox survey to prevent the high attrition rates associated with long, time-consuming surveys (Burruss & Johnson, 2021). The concise design (5–10 minutes to complete, inclusive of reading the participant information sheet and indicating consent via checkbox items) minimises satisficing and inaccurate recording. Fixed-response formats and forced-response settings are used to minimise the risk of missing data and prevent out-of-range or blank responses, which are common sources of measurement error in online surveys (Couper et al., 2001; Wright, 2005).

Multiple responses from the same participant will be identified using the deduplication procedure for online surveys, as described by Konstan et al. (2005). This involves screening all completed surveys for duplicate internet service provider (ISP) entries and eliminating any surveys that contain identical ISP and demographic data (region, social work specialism, etc.).

Exceptionally rapid completion times will be flagged as potential indicators of bot activity or invalid responses and reviewed during data cleaning (Burruss & Johnson, 2021). Although the platform allows participants to resume incomplete surveys on the same device, this does not prevent submissions from multiple devices; therefore, multiple data-quality checks are used in combination.

It is challenging to calculate a sample size due to the lack of available data on the number of social workers who work with children in care and how many of those children self-harm. Therefore, the assumption has been made that 10% of the 33,119 child and family social workers (FTE) practising in England (Community Care, 2024) will, in the last 12 months, have worked with a child in care who has self-harmed. This estimate is used solely to guide the desired sample size to obtain a sufficiently large and diverse sample. It does not imply or require recruitment of the entire eligible population. Using a 95% confidence level and 5% margin of error, the estimated population (3312) yields an ideal sample size of 345. As a non-probability sample is used, findings will be interpreted as indicative rather than population-representative estimates of awareness and implementation.

Data analysis

As the eligible population (social workers with recent experience of working with children in care who self-harm) cannot be directly enumerated from national datasets, demographic (e.g., age, gender) and professional characteristics (e.g., years of experience) will therefore be compared descriptively with the wider national workforce data (e.g., Social Work England) as the closest available contextual benchmark. These comparisons will be interpreted cautiously, as the eligible subpopulation may differ systematically from the broader workforce.

A pre-specified analysis plan will be followed to maintain transparency and specificity in analysis. Descriptive statistics will summarise professional characteristics, COM-B domain scores, and the rates of NG225 awareness and implementation. Multivariate analyses will control for a priori-identified confounders: professional experience, region of work, and work specialism. Regression models or suitable alternatives, if assumptions are violated, will be employed to explore associations between awareness and implementation and COM-B constructs and implementation.

Exploratory subgroup analyses will be conducted where sample size allows, aiming to identify differences between groups in opportunities, capabilities, and motivations to implement the guidelines. These will be assessed using analysis of variance (ANOVA) or the Kruskal-Wallis test (if ANOVA assumptions are not met).

All data, including from partial survey completion will be included in the analysis. Information on partial completion and response-time patterns will be used for data-quality checks to identify whether attrition clusters around specific items. The analysis will be descriptive and does not form part of the substantive statistical analysis.

Phase 2: Semi-structured interviews

Phase 2 addresses research question 4, exploring the barriers and facilitators to social workers’ implementation of NG225 with children in care, using semi-structured interviews.

Social work participants for the semi-structured interviews will be recruited during Phase 1. Participants will be asked whether they would like to take part in follow-up semi-structured interviews. Those who express interest and meet the eligibility criteria will be contacted by email in September–October 2026 and provided with a participant information sheet (see supplementary material). Participants will be given at least seven days to consider the information before providing informed written consent to take part in an audio-recorded interview via Microsoft Teams. Exclusion criteria are social workers who did not participate in phase 1 of the research, participants who did not provide consent to be contacted about participating in semi-structured interviews in research phase 2, and participants who did not consent to audio recording. If an insufficient number of phase 1 respondents consent to be interviewed, we will recruit additional participants through the same channels used in phase 1, applying the same inclusion criteria.

Whilst the flexibility of online interviews enables participants to be interviewed in their home environment, this risks greater distraction and third-party influence/compromised confidentiality (Rahman et al., 2021; Saarijarvi & Bratt, 2021). We mitigate this challenge by following the recommendations for online interviewing described by Saarijarvi and Bratt (2021) by sending participant information a week in advance and resending it the day before the scheduled interview. This will contain the interview questions (provided in advance, in accordance with a trauma-informed and neuro-affirmative approach), a digital link to the interview, and interview preparation advice (to enable video-camera on a computer/smartphone; ensure stable wi-fi connection; ensure there is a private, quiet, and calm environment for the duration of the interview). All participants will be encouraged to wear headsets to reduce background noise and enhance audio quality, whilst also promoting privacy and confidentiality. At the start of the interview, the researcher will check with participants that they are in a safe and quiet environment with minimum distraction, that they have received and read the information sent to them in advance and offer to answer any questions before proceeding. Consent will be reconfirmed verbally at the start of the interview to ensure ongoing willingness to participate. The sample size is determined according toFrancis et al.’s framework for theoretically rooted interviewing. This framework sets a priori minimum sample size of 10 and ceases data collection at the point when three interviews do not identify new themes (Francis et al., 2010). Data collection and analysis will occur concurrently. Analysis will begin immediately after the first interview: each transcript will be anonymised, coded, and added to the Framework matrix, enabling real-time identification of emerging themes. This iterative charting process facilitates ongoing comparison and allows the researcher to monitor when no new concepts are emerging. Sampling will remain iterative and ongoing analysis will inform recruitment to ensure that emerging gaps in perspectives (e.g., specific practice settings, variation in behavioural determinants) are addressed until sufficient conceptual depth and diversity are achieved, in line with the stopping criterion specified.

The use of online semi-structured interviews has increased in recent years (Lobe et al., 2022), and lessons have been taken from some of the challenges encountered. As unfamiliar technology can present a challenge for both participants and researchers (Lobe & Morgan, 2021), we shall conduct interviews using the platform Microsoft Teams, which is widely used by Local Authorities, the NHS, and the third sector. Thus, participants are expected to be familiar with this platform. However, we take the precaution of offering a 5–10-minute pre-session with all participants to address technical issues and ethical considerations, as well as the provision of a software guide (please refer to the supplementary material).

Data collection and analysis

The provisional interview schedule comprises ten open-ended questions to encourage social work participants to provide detailed accounts of barriers and facilitators, affecting their knowledge of, and implementation of NG225 with children in care. Questions were developed using semi-structured interview guides from previous studies with health professionals, constructed using the COM-B model to elicit key determinants of professionals’ implementation of NICE self-harm guidance (Leather et al., 2022, 2023). Consistent with recommendations by Atkins et al. (2017), questions are framed using language that is meaningful and accessible to practitioners, and the schedule is structured to support conversational flow, following principles outlined by Rubin and Rubin (2005). Prompts related to TDF domains were developed, as recommended by Michie et al. (2014), to expand exploration of specific influences on behaviour. Questions and prompts were reviewed by a trauma-informed expert and public contributors with lived experience of self-harm to ensure language sensitivity and minimise the risk of re-traumatisation. We incorporate an open final interview question, asking, “Is there anything that we have not covered that you feel is important/relevant?” to mitigate the risk of issues that are important to the participant being neglected (McGowan et al., 2020). The schedule will be finalised following analysis of Phase 1 survey data to ensure that emerging patterns in capability, opportunity, and motivation are reflected. It will then be piloted with social work students to assess clarity, acceptability, and burden prior to roll-out.

Participants are explicitly informed in advance and at the beginning of the semi-structured interviews that there are no right or wrong answers. The confidentiality statement will be read out at the beginning of the semi-structured interview and repeated at the end. Participants will be reminded of their right to withdraw at any time without needing to provide a reason. The researcher has undergone training in conducting challenging qualitative interviews. Their professional background includes extensive work in child mental health, safeguarding, and trauma-informed practice, which involves routinely identifying, monitoring, and responding to signs of emotional distress. Whilst the researcher will be vigilant for signs of participant distress and will offer breaks, emotional support, and the option to cease the interview, it is acknowledged that it can be more difficult to detect and respond appropriately to signs of participant distress in the online forum (Lobe et al., 2022). However, cues can be drawn from intonation and facial expression (Saarijarvi & Bratt, 2021), and participants have more agency to withdraw than they would in physical face-to-face interviews, where there can be greater implicit demand to continue, by terminating the interview with the click of a button (Lobe et al., 2022; Thunberg & Arnell, 2021).

The duration of the semi-structured interview is 45–60 minutes. Field notes will be made during and following the semi-structured interviews, as well as during data analysis. These are for debrief and reflexivity purposes and will not be included in the data analysis.

Qualitative data analysis: Framework analysis

Data from the semi-structured interviews will be transcribed verbatim and analysed using framework analysis (Ritchie & Spencer, 1994) in the computer-assisted qualitative analysis software, NVivo v.14. During transcription, all identifying information – including names, workplaces, colleagues, local authorities, specific cases, locations, and any other contextual details that could enable identificationwill be removed or replaced to ensure full anonymisation.

Framework analysis identifies data similarities and differences and relationships between data, enabling both descriptive and explanatory conclusions to be drawn (Klingberg et al., 2023). The coding framework for analysis is developed using the TDF, as advised by Atkins et al. (2017). Framework analysis will facilitate the interpretation of diverse views and experiences and enable systematic identification of areas of similarities and differences within and between children’s social workers, encountering children in care who self-harm. Rich insights into the influence of personal and organisational factors on the management of self-harm in children in care and its impact will be drawn. Framework analysis is well suited to semi-structured interviews since its matrix output provides a structure for analysis by participants and code. In this way, the views of each participant are connected within the matrix, and the context of individual views, patterns, and nuances is apparent (Klingberg et al., 2023).

The semi-structured interview data will be analysed according to the five stages of framework analysis adapted from Gale et al. (2013), to systematically identify facilitators and barriers to NICE guideline implementation.

Stage 1: Data Immersion Verbatim transcriptions of the semi-structured interviews will be observed repeatedly for data immersion.

Stage 2: Familiarisation Initial thoughts and impressions will be recorded on the transcripts.

Stage 3: Coding & Indexing Predefined codes, based on the TDF, will be applied to describe interpretations of the data.

Stage 4: Charting Data into the Framework Matrix Data will be charted by summarising it by category into the matrix. Illustrative quotations will be tagged in NVivo v.14 onto the chart.

Stage 5: Interpreting Researchers will journal ideas, insights, and early interpretations and discuss these regularly. Analytic memos will be used to stimulate discussion, as described by Charmaz (2006). A collaborative approach to the analysis will be endorsed. Two researchers will analyse the transcripts independently. Any disagreement will be resolved by discussion with the wider research team. Codes and themes will be documented in key domains. In this final stage, commonalities and differences will be identified, connections between categories mapped, and descriptive and/or explanatory interpretations will be generated. If the data is rich enough, predictions can be made about the extent to which the guidelines are likely to be implemented in distinct conditions.

Gale et al. (2013) caution that the final phase of framework analysis, interpreting the data, is time-consuming, often taking longer than expected. Accordingly, additional sessions are set aside should they be needed for sufficient data interpretation. Researchers will be flexible and adaptive to enable the generation of rich and nuanced insights. Critical reflection is a core feature throughout the research process in recognition that how qualitative research is conducted and analysed is inevitably influenced by researcher characteristics. The researcher will keep a journal to record reflections, insights, ideas, impressions, and early interpretations about the research and analysis, as Gale et al. (2013) recommended. Academic supervision and examination of journals/field notes documenting the researcher’s perceptions of the interview process and their emotional responses will promote objectivity and help to minimise the risk of researcher bias, influencing data interpretation (Cowles, 1988) as well as supporting researcher resilience and wellbeing (Dowding, 2021; Silverio et al., 2022).

We are mindful that qualitative explorations of behavioural determinants, influences, and experiences should be flexible and open to identifying all emergent ideas and themes (Layder, 1993; McGowan et al., 2020). Although we use a predominantly deductive approach, whereby data is coded into TDF domains, inductive content analysis will be applied to emergent sub-themes within each TDF domain as per guidance from Atkins et al. (2017). We will apply the coding, as described in Arden et al. (2019), whereby two researchers independently code under the TDF domain that best matches the data, and instances where data relates strongly to multiple TDF domains will be coded in all that apply. As the evidence indicates that some TDF domains measure an amalgamation of factors rather than distinct behaviour (Huijg et al., 2014), double coding should enhance the identification of related themes and domains. Data under each domain will then be subjected to inductive content analysis to create sub-themes, identified according to the procedure outlined by Atkins et al. (2017) when: mentioned by several participants; participants report conflicting beliefs and experiences; or there is strong importance expressed that may affect the target behaviour. Any discrepancies in coding will be discussed and resolved using the wider research team.

Underlying data

No data are associated with this article.

Extended data

1. Figshare: A mixed methods study to examine social worker’s implementation of national guidelines with children in care who self-harm. https://doi.org/10.48420/30069739 (Lever et al., 2025a)

This project contains the following extended data:

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).

2. Figshare. https://doi.org/10.48420/29605634 (Lever, 2025a)

This project contains the following extended data:

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).

Reporting guidelines

1. Figshare. https://doi.org/10.48420/30192934 (Lever, 2025b)

This project contains the following extended data:

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).