‘It felt like there was always someone there for us’: Supporting children affected by domestic violence and abuse who are identified by general practice

One in five children in the UK are affected by domestic violence and abuse. However, primary care clinicians (GPs and nurses) struggle to effectively identify and support children and young people living in homes where it is present. The IRIS + (Enhanced Identification and Referral to Improve Safety) training and advocacy support intervention aimed to improve how clinicians respond to children and young people affected by domestic violence and abuse. IRIS + training was delivered as part of a feasibility study to four general practices in an urban area in England (UK). Our mixed method design included interviews and questionnaires about the IRIS + intervention with general practice patients, including children and young people as well as with clinicians and advocacy service provid-ers.


| INTRODUC TI ON
In the United Kingdom (UK), domestic violence and abuse (DVA) is defined as any incident or pattern of incidents of controlling coercive or threatening behaviour, violence or abuse between people aged 16 or over who are or have been intimate partners or family members regardless of gender or sexuality (House of Commons Library, 2018).
It is estimated that one in five children are affected by DVA (Radford, Corral, et al., 2011), and this can have a significant negative impact on health and wellbeing across the life course (Holt et al., 2008;McTavish et al., 2016). Furthermore, DVA often co-occurs with child maltreatment due to the increased risk of physical harm and emotional abuse (Hamby et al., 2010).
General practice is well placed to identify and respond to children affected by DVA. This is because the GP is often the first place families will go to for help when experiencing DVA . The long-term relationships and continuity of care provided by primary health care clinicians (GPs, nurses and allied health professionals) potentially provide a supportive backdrop to disclosures of DVA (Woodman et al., 2013). Clinicians also have a duty to consider referral to children's social care where there are child protection concerns, which is known to be a factor in some situations where there is DVA (Department for Education, 2018;GMC, 2012).
Previous research has found uncertainty amongst clinicians in general practice about how to safely ask, record and support children affected by DVA (Szilassy et al., 2016). In particular, clinicians are reluctant to talk directly to children about DVA (Larkins et al., 2015).
There is also a lack of evidence to support the effectiveness of interventions to improve healthcare responses to children experiencing DVA (Howarth et al., 2016). To date, training and support provision have been patchy and a key message from clinicians involved in previous training interventions in this field, such as IRIS (Feder et al., 2011), HERMES (Williamson et al., 2015) and RESPONDS (Szilassy et al., 2016;Lewis et al., 2017), was that they would prefer DVA training to be integrated with one simple referral route.
In response, IRIS+ (Enhanced Identification and Referral to Improve Safety) builds on and extends the successful IRIS programme in the context of a feasibility study. The IRIS trial showed that a brief training intervention together with a referral pathway to a DVA advocacy service and ongoing support to practices significantly increased identification and referral of female patients in general practice to specialist support (Feder et al., 2011). IRIS+ expanded the IRIS model into an integrated programme which responded to the needs of women, men and children affected by DVA. The study reported here assessed the feasibility and acceptability of the IRIS+ model for children. More detailed integrated findings of the feasibility and acceptability of IRIS+ for all patients affected by domestic abuse is reported elsewhere (Szilassy et al., 2021).

| The IRIS+ feasibility study
The IRIS+ model provides DVA training for primary healthcare clinicians and a simple referral pathway to a local specialist DVA service (referred to as the IRIS+ hub). The three patient groups supported by IRIS+ are.

• female victims-survivors (v-s) and perpetrators
• male v-s and perpetrators • Children (18 and under) affected by DVA IRIS+ training was delivered to four general practices in England between May-August 2017, to a total of 30 clinicians. Training was delivered in two 2-hr sessions by a clinical lead and an advocate educator (AE) based in the specialist DVA service. As well as delivering training, the AE received referrals from clinicians, and provided expert advocacy (Rivas et al., 2019) to referred female and male adults as well as children affected by DVA. The core activities of the IRIS+ AE were providing legal, housing, and financial advice; facilitating access to and use of resources such as refuges, emergency housing and emotional support; giving safety planning advice; and providing ongoing support. In the second session, content about child safeguarding was delivered by a local children's social worker.
IRIS+ training was designed to support clinicians to appropriately identify, ask and respond to men and women who may experiencing What is known about this topic • Domestic violence and abuse can have significant and long-lasting impacts on children • Clinicians are well placed to identify and respond to children affected by domestic violence and abuse but there is a lack of training to support them

What this paper adds
• The implementation of IRIS+ (a specialist training and support service) led to a substantial number of children being referred for specialist support by clinicians • Children and young people benefitted from direct and indirect support (via their mother) from the IRIS+ specialist service • There were a significant number of children who could have been supported through the service but were not. These missed opportunities reflect entrenched barriers to children receiving support in their own right around domestic violence and abuse or perpetrating DVA and to children who may be experiencing DVA.
With the patient's consent, clinicians were encouraged to refer patients onto the IRIS+ hub. The IRIS+ hub provided women and menwhether v-s or perpetrators-with support from an AE. Children and young people were supported by a children's worker based within the IRIS+ hub. The referral pathway and hub were operational between May 2017-March 2018.
To test the overall feasibility and acceptability of the training and support intervention, a mixed-method research study was conducted (Szilassy et al., 2021). This paper reports on the findings of the feasibility study relating to children.

| ME THODS
Data sources for this study were as follows:

| Agency data
Data were collected from the IRIS+ hub about referrals received from the four general practices during the pilot period (May 2017-March 2018).

| Medical records data
Electronic medical records (EMR) were searched in the four pilot practices for codes specific to DVA. The medical records of patients with a DVA code were reviewed to establish how DVA was identified (e.g., direct disclosure to clinician, information from another agency), whether DVA was discussed at subsequent consultations, and any offer of support such as IRIS+.

| Interviews with IRIS+ patients
Interviews were completed at two time points with IRIS+ adult patients who consented to be part of the study: immediately following referral to the IRIS+ hub and 3 to 6 months after baseline. In total, 10 mothers were recruited to the study. No fathers (or male v-s) were referred to IRIS+ and therefore none were interviewed. Seven mothers completed both the baseline and follow up interview, with three completing on one occasion. In interviews, mothers were asked about their experiences of being referred to the IRIS+ hub and of receiving support. Participants were also asked about their children and what impact support had upon them.
Three children (aged 9-16) were interviewed, all of whom had received support from the children's worker within the IRIS+ hub.
Children were recruited via their non-abusive parent (the mother in all cases) who, in turn, were introduced to the study by the AE.

| Interviews and questionnaires with clinicians participating in IRIS+
All thirty clinicians who did the training were invited to be interviewed. In total nine interviews were conducted with clinicians immediately after the IRIS+ training with eight completing follow-up interviews 6 months post-training. The interviews focused on clinicians' experiences of the training, their views on the service and what enablers and barriers they experienced in asking about DVA and referring patients to the IRIS+ hub.
An online questionnaire, the PIM+ (adapted from the HERMES study, Williamson et al., 2015) was also administered to clinicians prior to the IRIS+ training and nine months after the final training session. In total, 25 clinicians completed the survey, with 18 completing at both time points. The questionnaire measured participants' knowledge, confidence and preparedness to ask, respond, identify, refer, record and support adult and child patients experiencing or perpetrating DVA. Interviews were also conducted with the two professional IRIS+ trainers as well as the IRIS+ AE and children's worker (four in total).

| Stakeholder interviews
Consultation was undertaken with key stakeholders outside the IRIS+ intervention to understand the key challenges raised during the feasibility study: namely, the identification and referral of men and children, and use of reports from external agencies in GP consultations (see Pitt et al., 2020). Stakeholder participants (n = 9) were identified through professional networks and included GPs, clinical leads for DVA and safeguarding leads in clinical practice. These interviews were conducted between April and July 2018.

| Data collection
All interviews (with IRIS+ clients, children and professionals) were semistructured, informed by a topic guide and conducted by JR and ES. The topic guides were developed based on a review of relevant literature, and researchers observations of training sessions. After initial interviews with participants, the topic guides were modified to reflect emerging issues. Interviews with IRIS+ clients and children were done face-to-face in a safe environment, chosen by the participant. Interviews with professionals were conducted either face-to-face or over the phone.

| Data analysis
All interview data were transcribed and loaded into qualitative data analysis software (NVivo v.10). The qualitative interview data was then analysed using thematic analysis (Bryman, 2016). Two researchers (JR and ES) independently read and re-read the transcripts identifying recurrent and salient themes. These initial themes were then shared between researchers and a coding framework was developed. The interviews were then coded thematically using the framework created by JR and ES.
All questionnaire data were entered into REDCAP, an online data capture system for clinical research. Raw data were then transferred into SPSS (v.21) and basic descriptive statistics were produced. Data from medical records were extracted by a member of the practice team in the presence of a researcher who did not have access to personal information. This procedure ensured that linkage to identifiable patient data was not possible. This anonymised and redacted data were transferred into Excel for analysis.

| Ethics
The study was guided by panels of professional and service user experts as well as project advisory and data monitoring experts.
The study was given a favourable ethical opinion by South West-Frenchay Research Ethics Committee (REC 17/SW/0098) and the Health Research Authority (HRA).

| FINDING S
Three main themes emerged from the qualitative and quantitative data: 1. The identification and referral of children affected by DVA by clinicians 2. The barriers and enablers to the identification and referral of children 3. Type and impact of specialist support on children and their parents.

| Identification and referral of children
EMR data indicated that, across the four pilot practices, 49 children were identified as being exposed to DVA with the majority coming from one practice [table 1 here] (practice D). Only two of these children were identified following GP consultation. The remaining 47 children were identified as a result of information shared for safeguarding purposes by an external organisation with the GP which was then coded on the EMR. Police reports-following a callout to a DVA incident where a child was present-were the most common source of external information which was then coded onto the EMR.
Despite there being information in children's EMR about DVA, consultation records showed little to no evidence that clinicians spoke directly to children or raised the issue of DVA with their parent(s). This was the case when children were presenting with medical conditions potentially associated with DVA such as behavioural and mental health problems.
What is important to highlight is that the majority of the 49 children identified in medical records were not referred to IRIS+. For example, EMR search of Practice D found 46 children affected by DVA but only four referrals were made to IRIS+ by this practice (see Table 2). The mismatch between EMR and referral data and the information available on children's EMR suggest there were potentially missed opportunities to identify and support children affected by DVA.
Thirty-five children were referred to the IRIS+ hub across the ten months of the study; again, the majority came from one practice, Practice B (see Table 2). All these referrals arose from a consultation with children's mothers-and this may explain why the information did not appear on the child's EMR. Of the 35 children referred to IRIS+ hub, 13 children did not receive any support. In some cases, this was because the mother did not want ongoing support and in some cases the child already had support. Twenty-two children received support either indirectly through the parent (n = 16) or directly from the children's worker (n = 6). The nature and impact of this support are discussed below under theme 3.

| Barriers and enablers to identifications and referrals
There were two groups of children who were eligible to receive support from IRIS+ but did not: the first were children referred to the IRIS+ hub who did not receive support (n = 13) and the second were children identified in medical records as experiencing DVA but who were not referred to the IRIS+ hub or engaged with by clinicians.
Through interviews with patients and clinicians, our analysis indicated there were enablers and barriers to the identification, referral and support of children.

| Enablers for parents and clinicians
All the children who were referred to the IRIS+ hub and were identified after consultations clinicians had with their mothers. As such, it is important to consider what enabled mothers to seek support from their GP clinicians. Two main factors were identified: the first was their relationship with and skills of the clinician.
Mothers interviewed (n = 10) identified a series of clinician skills which enabled disclosure of DVA including being trustworthy, being easy to talk to, giving the patient time to talk and understanding the impact of DVA: It didn't matter that I had a 10-minute appointment, she was prepared to sit there and talk to me and listen and that sort of stuff. She was just sympathetic, wanted to hear and just really wanted to help you. M1 Mothers also identified that clinicians making the referral for them to the specialist support service was key to accepting support: I said, "I can't do it myself. I realise now that I need help but I can't force myself to do it, I'm too scared." She said, "Do you want me to do it?" I said, "Yes, please." And she did. The day after I got a call from The second enabler to seeking support was children themselves.
Mothers were aware of the physical and emotional impact that DVA was having on their children and this was often their motivation to feeling more 'comfortable' (GP7) in talking to children and another spoke about how talking to children had been 'more at the forefront of my mind' (GP3). However, as noted above, evidence of direct talk with children was not evident in the records.

| Barriers for parents and clinicians
Two main barriers prevented mothers from seeking or accepting support for their children: fear of professional intervention and fear of the perpetrator. Similarly, clinicians noted that some of their patients had a mistrust of professionals: …there is a large amount of suspicion in our area, certainly for social services…Almost the first thing that people say when there's any suggestion of -well, even depression -people will say, "What matters to me most is not having someone thinking about taking my children away". GP2 Another-very real-fear was that the perpetrator (in many cases the child's father) would find out about any support the child was receiving. If the mother was still living with the perpetrator or the children still had contact with the perpetrator, then it was considered too risky for the child to be directly supported in case they found out. As one mother said (whose children were being indirectly supported via her): I need to make sure that everything is in place, so that their father will not know if possible.

| Engagement with, and impact of support for, children
Twenty-two children received IRIS+ support tailored to the child and family, either indirectly or directly. The AE and children's worker explained that in all cases support was offered to the children. Six children aged between 7 and 16 years received direct support.
This included one-to-one support sessions and a 10-week group programme on healthy relationships. Sixteen children (aged 1-16) received support indirectly. This took the form either of emergency refuge accommodation with their mother (four children) or the AE's/ children's worker working with their mother. In most cases indirect support was provided because it was assessed by the AE/children's worker or mother not to be appropriate or safe to provide direct support. This normally because the child was too young (e.g., under 5) or because the child was still living with the perpetrator, and therefore, direct support involving the child could have increased the risk of harm.
Interviews with AE, children's worker and mothers indicated that indirect support for children (

| D ISCUSS I ON
In this study we evaluated the feasibility and acceptability of the IRIS+ training and referral pathway for children. Thirty-five children were identified and referred on for specialist support by clinicians.
Previous studies have demonstrated that specialist training increased the identification and referral of women experiencing DVA (Feder et al., 2011). The IRIS+ pilot has demonstrated that children can also be directly and indirectly supported through this model, even in the context of increased pressure on general practice (Baird et al., 2016).
Analysis of interviews showed that a small number (n = 6) of children benefitted from tailored, 1:1 support from the children's worker.
The type of support offered depended on the child's situation, what they wanted and what they would benefit from. A larger group of children (n = 16) were indirectly supported through their mother. In many cases, this indirect support was the only feasible way for children to be supported due to their age or the fact they were still living with perpetrator. Importantly, our study's findings do suggest that children benefitted from indirect support, for example, as a result of improved parent-child communication and parent reports of the child's wellbeing. A similar effect has been observed elsewhere: a trial of mothers and children affected by DVA found that children's behaviour problems significantly improved during and up to 24 months after their mother (only) had received intervention and support (McFarlane et al., 2005).
Indirect support may also be beneficial because it acts as an enabler for children to receive direct support later on. For example, a recent evidence synthesis by Howarth et al. (2016) found that mothers are more ready for their children to receive support when the immediate danger of leaving the perpetrator is over and they have received support in their own right. As such, while we anticipated more children to supported directly by IRIS+, the indirect support provided was still important and beneficial for children: both as a means of improving their wellbeing (through the mother) and potentially supporting the mother to feel ready and safe for their child to receive support in the future.
Our feasibility study revealed that there were a group of children who were eligible for IRIS+ support but were not identified or referred into the IRIS+ hub by clinicians. The study identified practical barriers to clinicians identifying and supporting children affected by DVA. A key issue raised by clinicians was the lack of clarity around how they should respond to information from other agencies (see Pitt et al., 2020 for further discussion). Other barriers identified reflect wider issues in general practice, such as the well documented lack of time for consultations (Baird et al. 2016) which is likely to impede the capacity of clinicians to talk through issues of DVA with patients and children.
There were also structural barriers specifically relating to DVA and children which are worth considering in more depth. As found in previous studies (e.g., Stanley et al., 2012), mothers are often scared of disclosing DVA because it may lead to their children being removed from their care. This was an identified barrier in the present study and is likely to have meant that some mothers were unwilling to disclose or accept a referral to the IRIS+ hub. Going forward, it is important that clinicians can reflect on and recognize this concern in consultations with women who are experiencing DVA.
Another barrier we identified was clinicians' unwillingness to talk directly to children. While clinicians reported feeling more confident and skilled in talking to children about DVA, there was little to no evidence that they did so. Furthermore, the data from medical records indicated that clinicians were not asking parents or children about DVA even in situations where they were aware that DVA had occurred due to information from other agencies.
There may be good reasons why the clinician chose not to talk to children-for example, the child's age, concerns about safety, and who the child was with during the consultation. But, it would be expected that-given the specialist training and the number of DVA identifications in EMR's-some clinical encounters would have involved talking directly to children. It seems likely that, as Larkins et al., (2015) found, clinicians may have been reluctant to talk to children because of concerns about children's competence and their own clinical skills. This unease in talking to children about sensitive topics is not unique to clinicians, and is evident across health, education and social care professionals (CRAE, 2010;Radford, Aitken, et al., 2011). Concerns about children's vulnerability, their presumed inability to express themselves and the potential for distress (Callaghan et al., 2017) have all been cited as reasons why children are not (or should not be) spoken to. Critically, these concerns are based on paternalistic ideas of children as passive victims, without agency (Katz, 2015) or make an (incorrect) assumption that children all experience DVA in the same way. However, research shows that children do want to talk to professionals about DVA (Stanley et al., 2012), moreover it is their right to do so (UNCRC, 1989) and it is important to understand children's own perspectives on their lives (Arai et al., 2019). A consultation with an appropriately trained clinician in a safe environment away from home is a good opportunity to do that.
Addressing the reluctance of clinicians to engage with children is important. It potentially explains why clinicians did not speak to children directly about DVA despite specialist training. It also explains why EMR identifications did not necessarily translate into referrals to the IRIS+ service. In turn, this impacts on the number of children who were able to receive support, indirectly or directly. Further development work is being undertaken to strengthen the child specific elements of IRIS+ however it is likely that these issues may continue to represent a significant barrier to supporting children through general practice.

| LI M ITATI O N S
Our feasibility study only covered four general practices in one city.
We interviewed a modest number of clinicians and patients about the IRIS+ training and support intervention. As such, the findings cannot necessarily be transferred to other general practice settings in other geographical locations. The interviews and questionnaires with patients and clinicians were planned to be conducted at two time points but there was attrition and we lost one quarter at follow up. It may be that clinicians and patients who viewed IRIS+ favourably were more likely to engage at both time points, and this introduces the possibility of selection bias.
Finally, the study only interviewed three children which limits our understanding of how children experienced IRIS+. It is of note that the research team encountered a number of barriers trying to recruit children into the study-many of which reflect the issues highlighted above. It remains critical that children and young people are given the chance to engage in research which is relevant to their lives and wellbeing.

| CON CLUS ION
The IRIS+ intervention resulted in clinicians identifying and referring children exposed to DVA to a specialist service. Several enablers to identification and referral were identified, including clinicians' increased confidence and skills, the presence of a dedicated support service (IRIS+ hub) and parents feeling listened to and respected in consultations. The study's findings also indicate that children benefitted from being identified and referred into the IRIS+ hub. While small in number, children who received support directly valued it.
Indirect support for the child also appeared to be beneficial to both the child and mother.
However, barriers to identification and referral were identified.
Some may be amenable to change, for example through further development of the IRIS+ training intervention relating to information shared by external agencies. However, other barriers reflect systemic issues in general practice, such as the lack of time in consultations. Other barriers reflect entrenched issues relating to DVA, and to children's agency, for example mothers' fear that children will be removed from their care and clinicians' reticence to talk directly to children. These issues require systematic and significant changes in how we-society, professionals, academics-conceptualise and approach children's experiences of DVA.

ACK N OWLED G EM ENTS
Authors would like to thank all the professionals, experts and service users who contributed to the development and delivery of the intervention and to the development of data collection tools and protocols. We would also like to thank all our project partners who facilitated our data collection and all the professionals and patients who gave their time to take part in the study. We are grateful to the members of the REPROVIDE team for their support and to the Care.

CO N FLI C T O F I NTE R E S T
The authors declares no conflict of interest.

AUTH O R CO NTR I B UTI O N S
JR and ES conducted the surveys, interviews and observations and collected agency data. KP, JR and ES collected data from EMR. JR and ES analysed the data. JR drafted the manuscript. All authors were involved in the interpretation of the data, revised the drafts and contributed to the final version of the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.