Case series

Overview

Phase 2a of the study consisted of a small case series. Using a preliminary version of the treatment manual, resulting from the work of the Manual Development Working Group, we trained a group of practitioners in the modified Video-feedback Intervention to promote Positive Parenting and Sensitive Discipline, Foster Care (VIPP-FC) to work with a target of six children in foster care and their carers. We proposed to purposively recruit a sample reflecting variations in age and presentation of RAD symptoms, to road-test the modified VIPP-FC treatment manual and examine its clinical suitability for these children in real-world clinical practice. During this phase of the study, we recruited from a single local authority and identified a small number of cases using screening measures of RAD with the help of the children’s services. The case series also allowed us to undertake preliminary tests of the feasibility and acceptability of the screening process and the proposed research assessments. The case series was originally planned to run in two local authorities. However, shortly before we were due to begin work, the partner trust in one site lost their contract to deliver CAMHS locally, which meant that we were unable to proceed with that site for the case series.

Methods

Ethics considerations

Two key ethics issues required careful consideration in this study: consent and the RAD label/diagnosis. The determination of who must provide consent for a child in foster care to take part in a research study and how this should be done is recognised in the literature as a complex problem.35 An important aim of the current study was to investigate the range of views on how this should be achieved, from local authorities in different regions and from parents and carers. However, the starting point for the purposes of this study was to follow a procedure used in another NIHR pilot trial being conducted by our group for older children in foster care [aged 6–16 years; NHS Research Ethics Committee reference number 15/EE/0032; award ID: PB-PG-0614-34079 (https://fundingawards.nihr.ac.uk/award/PB-PG-0614-34079)]. The consent protocol for that study was developed from a review of the literature and consultation with a local authority director of social care and their legal department. The general principle informing the protocol was that written, informed consent to take part in the study must be received from all parents and from the local authority when holding parental responsibility, and that every effort should be made to secure the consent of parents when they do not have sole legal parental responsibility.

On the basis of these considerations, to take part in the intervention phase of the study (i.e. after initial screening), parents of children on full care orders were sent information sheets and an opt-out letter (with a 14-day period for opting out) in all but exceptional circumstances. Specifically, children on a full care order could enter the study without parental consent if the local authority [which shares parental responsibility with the birth parent(s)] deemed it not in the child’s best interests to seek it, or when every attempt had been made to inform parents of the study and to secure consent, but this had failed. For all other children, full written consent was obtained from the child’s parent(s). The child’s social worker initially made contact with the parent(s) and provided parents with the information sheets. It was at the discretion of the social worker whether to seek verbal consent for parents’ contact details to be passed on to the research team so that a meeting could be arranged to sign the full written consent or to seek the full signed consent from the parents themselves. In either case, the research team’s contact details were always provided to parents in case they wished to ask more in-depth questions. Active refusal of consent by a parent was always respected, regardless of the care order. Furthermore, a parent’s decision to withdraw their child once the child had started in the study was always respected, regardless of the care order.

Full informed consent was obtained from all participating foster carers and from the local authority. As the children were all aged ≤ 6 years, they did not provide written consent, but the researchers were trained to obtain verbal assent in an age-appropriate fashion for children old enough to do so (age ≥ 4 years).

Throughout our discussions with colleagues in the local authorities, a recurrent theme was professionals’ concerns regarding the diagnostic term ‘RAD’. In particular, professionals expressed conerns that if this diagnosis became a part of a child’s social work record, this may have direct implications for the child’s subsequent care, including their likelihood of being adopted. For a more in-depth discussion of this issue, see Chapter 4: stakeholder interviews drew attention to the potential difficulties associated with using diagnostic terms and a diagnostic criterion as part of the eligibility criteria for entry to the study, as well as the complications associated with information-sharing regarding diagnoses.

Because we included children presenting with varying degrees of RAD-type behaviour and as our assessments were based on research instruments rather than full clinical assessments, we did not use the term ‘RAD’ in our communications with participating families (e.g. in participant information sheets or verbally) to refer to the set of behaviours the study was focused on. Instead, we described the specific behaviours that are considered under the rubric of RAD, which tend to carry fewer unhelpful connotations (e.g. difficulties in seeking comfort from available caregivers). The same avoidance of diagnoses applied to written correspondence with foster carers and social workers about each child.

Target population

The target population was children in foster care aged between 11 months and 6 years, and their foster or kinship carer(s).

Eligibility criteria

Inclusion criteria

Families recruited to the study were eligible if the following criteria were met:

• parental figure –
  • foster carer(s) who is (are) primary carer(s) for the child
  • aged ≥ 18 years
  • proficient in English
• child –
  • living with foster carer(s) in a placement planned to last at least 4 months
  • has lived with foster carer for at least 4 weeks
  • presence of RAD symptom(s) as defined by the DSM-5
  • aged between 11 months and 6 years.

Exclusion criteria

• Parental figure: already engaged in a similar parenting intervention.
• Child: severe developmental disability.

As a pragmatic study, we deliberately minimised exclusion criteria; this was to enable the later feasibility trial to reach a participant population that is broad and reasonably representative so that we could address the clinical suitability of VIPP-FC for a wide range of children in foster care. We did, however, collect detailed information about factors that might be relevant to the suitability of the VIPP-FC intervention for children with RAD, such as co-occurring conditions, placement variables and care order.

It is a pragmatic requirement of this form of intervention that the foster carer has some proficiency in English language because of the reliance of this programme on dialogue between the carer and practitioner as part of the video-feedback procedure. Thus far, the Video-feedback Intervention to promote Positive Parenting and Sensitive Discipline (VIPP-SD) has not been tested using interpreters, and we considered that this project was not the appropriate context in which to do such testing. Furthermore, on the basis of consultation with colleagues in the local authorities, we determined that comparatively few foster carers would not have adequate command of English language to participate in the programme. The research assistant in contact with the foster carer made a judgement regarding the suitability of their English-language proficiency, making sure that they met the minimal requirements necessary to take part in the study.

Although the programme is suitable for children with a range of levels of cognitive ability, and would be appropriate for children with physical and mild to moderate disabilities, it has not been used before among children with very severe global delay; such global delay would also substantially complicate, if not preclude, a reliable RAD assessment. Therefore, children with global developmental delay would need to be excluded from the study. This exclusion criterion was assessed by asking carers if the child had any medical diagnoses relating to severe developmental delays.

Owing to the risk of participant overburden and the contamination/confounding of treatment effects, it would be problematic for families to be engaged in similar parenting interventions simultaneously; hence, this was also an exclusion criterion. We did not consider it to be problematic that families were involved in other forms of social, psychological or medical treatment; therefore, no such exclusion criteria were applied. Furthermore, we explicitly assessed the presence or absence of comorbid psychiatric conditions in the child because these may be important factors influencing both clinical management and outcome, and are also likely to be quite prevalent. Exclusion on that basis would therefore be inappropriate. However, when other diagnoses were identified that warranted intervention (e.g. attention deficit hyperactivity disorder), we notified the child’s social worker and foster carer and recommended a referral.

Extensive consideration was given to the range of care arrangements that should be eligible for this pilot project. Children in foster care can be subject to a range of different care orders and placement arrangements (e.g. interim care order, full care order, section 20 voluntary placement36). Children placed in family and friends care may be placed under a SGO, a residence order or a care order and, rarely, as an informal placement. These issues are primarily relevant to the feasibility of intervention delivery, rather than child mental health need (as children in all of these circumstances are likely to present with raised rates of attachment problems and RAD). In consultation with colleagues in the local authorities, we decided that this is an important feasibility question and exclusion on the basis of placement setting or type of order would unhelpfully limit the learning that the feasibility study could produce. Regarding family and friends care in particular, there was a clear view from our consultation, which was highly consistent with recommendations made by NICE, that family and friends care, which is increasingly common, should be within the scope of this study.13 Children placed under a section 2036 (a voluntary placement, whereby the parents retain full parental responsibility) or on an interim care order may present the most difficulties from a practical point of view because the duration of the placement will be uncertain. Nevertheless, it is not uncommon for such placements to last at least 6 months (i.e. long enough for an intervention to be put in place) and duration cannot be predicted in advance. Therefore, our view, which was supported by our discussions with local authorities, was that an inclusive strategy should be taken.

One critical point to emerge from these discussions was that there may be great value in considering adaptations to the intervention that allow for the programme to ‘follow the child’ when there is a transition in the placement arrangements (e.g. to adoption, to a more permanent foster care placement or back to the child’s parents). A lack of continuity in care is a problem highlighted by many in the field and by children in care themselves.13,27 We planned to discuss this process carefully with the study advisory groups and we proposed to test, in a small number of cases as they arose, the workability of the plan they recommended. In such cases, we anticipated that the work would not begin again, but would focus on transferring what had been learnt during treatment to the subsequent placement and supporting a smooth transition.

Recruitment

Screening and identification stage

The screening process involved close collaboration with the partner local authority, which sent out information sheets, initial consent forms and screening measures to the foster carers in the study locality. In addition, as noted previously, the local authority sent out a letter to all parents explaining the study and the opting-out process, including the contact details of the research team. They communicated to their staff members within the children’s social care team to raise awareness of the study and draw attention to the support required from children’s social workers. We originally planned to invite all foster carers or special guardians registered with the local authority caring for children aged ≤ 6 years. In practice, the local authority applied its own exclusion criteria at this point (e.g. excluding highly complex cases undergoing assessments for court or cases with restricted files), as we note in Results and Conclusion.

We also contacted and sent leaflets to several foster care agencies inviting them to take part in the study. When direct self-referrals or clinician referrals were received by the research team, foster carers/clinicians were asked for the child’s social worker’s contact details in order for the local authority to initiate the screening process, including contacting the birth parents. Regardless of the route of recruitment to screening, a minimum 2-week period was given between sending out the letter to parents and initiating the child’s participation in the study.

Once initial consent forms and screening measures had been received by the research team, research assistants contacted participants, thanking them for their response and reiterating the research process. Invitations and follow-up telephone calls were made to all foster carers to take part in a face-to-face interview assessment to confirm eligibility, unless it was clear from the information provided at that point that they would not be eligible (e.g. the child had already left placement or the placement was very far from the borough). Owing to the comparatively small number of carers who could be invited to take part, we opted to invite all foster carers who returned the questionnaires to take part in the DAI so that we gained experience of the acceptability of the interview and a reasonable level of initial data on the performance of the screening questionnaires. As the screening/identification stage involved no direct contact with the child and involved minimal burden or risk, we obtained active consent from the foster carer only. However, a letter was sent to all birth parents with an explanation of the study and the process of opting out (except in circumstances in which the local authority deemed it inappropriate to inform the parents). If, after the 2-week period for opting out, the local authority or the research team had not been contacted by the parent to opt out, the screening process was initiated. Foster carers received an information sheet explaining the study fully and a contact telephone number of a member of the research team in case they had any questions. The information sheet explained that completing the screening measures did not entail any obligation to participate in the subsequent stages of the study (i.e. the baseline research assessments, treatment and follow-up assessments).

Post-screening stage

If eligibility was confirmed, foster carers were invited to the next stage of the study, involving baseline assessments, receiving the VIPP-FC and completing follow-up (i.e. post-treatment) assessments. The baseline assessment took place as soon as possible after all consents had been obtained (the target was within 4 weeks). At this point, we initiated the more comprehensive consent process (i.e. consent to baseline/follow-up assessments and treatment). Consent from birth parents (if applicable) and consent from the social worker or team manager were obtained before the baseline study visit. At the baseline study visit, we obtained the foster/kinship carer’s consent, as well as the child’s assent, if they were aged > 4 years.

The baseline assessment took place in our partner CAMHS, in the local authorities’ premises, or at UCL, depending on the preference of the foster carer. Carers were paid £30 at both the baseline and follow-up (i.e. post-treatment) outcome assessments to thank them for their time, and were reimbursed any travel expenses. Children were given a sticker after each visit.

Sample

For this (non-randomised) case series, we aimed to recruit six cases for piloting the clinical intervention, and testing screening and assessment procedures. As noted previously, we proposed to include a mixture of cases obtaining high RAD scores on the screening measures and cases falling below this threshold to explore whether or not the intervention required adjustment for these groups of children, and to maintain the pace of manual development in the event that prevalence was lower than expected.

Screening measures

There is no pre-existing established tool to screen for RAD. This is a significant gap in health technology in this field, as early detection may be critical for improving outcomes for children in care. In this study, we used two different instruments that each have their individual limitations, but may be useful in combination for screening for symptoms of RAD. The Attachment Screening Assessment (ASA) (Danya Glaser, Saul Hillman, Yael Shmueli-Goetz and Vivian Prior, the Anna Freud National Centre for Children and Families and UCL, 2013, personal communication) is a new instrument designed specifically to capture behaviours reflective of the DSM-5 symptoms of RAD and DSED, as well as standard Ainsworth-type attachment classifications.37 In pilot work (Ching Yiu Ng and Saul Hillman, the Anna Freud National Centre for Children and Families and UCL, 2015, personal communication), the RAD items in particular have been found to discriminate well between children in care and those from the community, as well as children with greater overall psychopathology on the Strengths and Difficulties Questionnaire (SDQ). However, the ASA has thus far not been validated against formal diagnostic assessments, which limits the extent to which it can be used alone as a screening measure. The Development and Well-Being Assessment (DAWBA) includes a recently revised set of items for DSM-5 RAD.38 We proposed to use these two instruments so that we could test the optimal configuration of measures for case identification.

Primary outcome: attachment measure

The DAI is a semistructured interview to evaluate the presence of signs of disordered attachment.39 For each of 12 sections, the interviewer asks multiple questions and follows up with probes sufficient to yield a rating reflecting the degree of evidence of disturbed or disordered attachment: 0 = none/never, 1 = somewhat/sometimes and 2 = considerable/frequently. The first five sections of the interview address signs of emotionally withdrawn/inhibited attachment disturbance (i.e. RAD), yielding total scores ranging from 0 to 10 points. The inhibited attachment disturbance items are 1 – does not differentiate among adults, 2 – does not actively seek comfort when hurt/upset or does not seek comfort preferentially, 3 – does not respond to comfort when hurt/frightened, 4 – does not respond reciprocally with familiar caregivers and 5 – does not regulate emotions well. The next three sections address signs of indiscriminate behaviour and the last four sections address other signs of disturbances in attachment behaviour (known as ‘secure base distortions’). There is sound evidence for the validity of this measurement system. First, it distinguishes between institutionalised and non-institutionalised children, reliably identifies signs of attachment disturbance in maltreated children and validly distinguishes RAD from DSED.5,39–41 Second, it has shown very good inter-rater reliability and high levels of internal consistency.39,40,42 Third, criterion validity has been demonstrated through the convergence of attachment disorder diagnosis by different measurements. Construct validity has also been demonstrated by the association of DAI ratings with quality of care.43

Secondary outcome measures

Attachment insecurity and disorganisation

Strange Situation Procedure

Participants were observed with their foster carer while in an unfamiliar room for the Strange Situation Procedure. The Strange Situation is a standardised procedure lasting a maximum of 21 minutes. It is divided into a series of episodes involving brief separations and reunions with the parent/caregiver; exposure to an unfamiliar, but friendly, female adult; and a period of time alone. These episodes are video-recorded for subsequent rating by trained coders, resulting in a classification of the infant–caregiver dyad into one of the four attachment categories: secure, insecure-avoidant, insecure-resistant or disorganised. For children aged > 24 months, an adaptation of the administration of the Strange Situation (mainly concerning the timing of each episode) and the respective coding system for the pre-school years was implemented (Jude Cassidy, Pennsylvania State University, 1992, personal communication; Robert Marvin, University of Virginia, 1992, personal communication; and the MacArthur Working Group on Attachment, University of Virginia, 1992, personal communication). Even though the pre-school rating system was originally created for children up to the age of 54 months, we did not anticipate problems in extending it to the 6-year-old participants. Indeed, the pre-school system was created based on the assumption that the ‘strategies’ associated with each one of the three organised attachment categories remain the same throughout development.

Co-occurring disorders/difficulties

The DAWBA was used to assess the presence of psychiatric symptomatology co-occurring with RAD through the foster carer interviews. The DAWBA is a well-validated diagnostic interview schedule that has been recently extended to children as young as aged 2 years and has been updated for the DSM-5. The DAWBA has been used extensively in clinical trials and epidemiological studies. The system has high levels of inter-rater reliability and provides automated algorithms for scoring and assigning diagnoses. It should be noted that, owing to some limitations in the validity of the DAWBA for assessing RAD, we relied on the DAI as the main instrument to identify RAD symptoms.

The Child Behavior Checklist (CBCL) is a very widely used carer-reported questionnaire for assessing children’s emotional and behavioural problems, and yields well-validated continuous measures of internalising problems, externalising problems, total problems and specific ‘syndrome’ scales for narrow-band symptom areas (such as attention problems, aggression, social problems).44 It is valid for children from the age of 1.5 years and has age-appropriate forms for children aged 1.5–5 years and ≥ 5 years.

Broader impact measures

To assess quality of caregiving, we rated the observed carer sensitivity and other related parameters of parenting (e.g. co-operation, positive regard for the child). We used standardised observational procedures for assessing carer sensitivity during free play and routine caregiving interactions. Specifically, we video-recorded an interaction between carer and child for 10 minutes (excluding instructions), during which a set of structured instructions given to the carer were followed: (1) to teach the child how to play with a challenging toy for their age, (2) to then play with the child using no toys, (3) free play with the toys and (4) to help the child tidy up the toys. These video-recordings were subsequently scored blind to treatment condition and status (pre vs. post) by trained and reliable coders. The scoring system used was the National Institute of Child Health and Human Development (NICHD) sensitivity scales, which have been validated in numerous studies.45

We also administered two parenting-related questionnaires. The Parenting Stress Index-Short Form was used to assess carer well-being, carer strain and the carer–child relationship.46 The Brief Parental Self-Efficacy Scale is a measure of beliefs and confidence about parenting skills.47

The Child and Adolescent Service Use Schedule (CA-SUS) assesses resource use for health economic evaluation.48 Previous versions of the CA-SUS measuring service use by pre-school children with autism and young looked-after children formed the basis of a modified version suitable for the current population, with modifications being determined, initially, by expertise in the research group and, subsequently, by testing for acceptability and comprehensiveness. Specifically, this initial draft version was tested and modified in two ways. First, after a sufficient number of cases, the interviewers and the CA-SUS authors discussed necessary amendments and clarifications that were incorporated into the previous (draft) version. Second, the measure’s comprehensiveness in capturing all relevant services was assessed. Testing was important to identify items that were redundant or important services that had been omitted.

The proposed battery of outcome assessments was similar to those we have used in other trials and cohort studies,25,49,50 and our previous experience was that it would not be excessively burdensome. However, we planned consultation with foster carers (through the foster carers advisory group), as well as with those taking part in the study, regarding the level of acceptability and burden. Establishing the optimal balance between participant time/burden and the robustness of the outcome assessments was an important aim of this project as preparation for a full-scale trial.

The research team attempted to follow-up all families involved in the study, including children who moved to a different placement during the time of the research. Any additional relevant consent that was required as a result of placement changes was obtained (e.g. consent from the new carer).

One of the aims of the case series was to establish initial feasibility of the intervention protocol and the acceptability of the programme and outcome assessments to foster carers. Accordingly, research assistants closely monitored the recruitment and assessment processes and requested continuous feedback from carers to make any necessary adjustments to the protocol for the feasibility RCT (phase 3). In addition, treatment progress and process were closely monitored, and feedback was presented and discussed by the Manual Development Working Group (from phase 1) to allow consideration of further appropriate manual modifications. Practitioner feedback forms were used to provide additional qualitative feedback about the experience of using the VIPP-FC in practice. The results of the case series also informed revision of the research protocol in preparation for phase 3.

Intervention

Intervention description

The VIPP-SD is a brief home-based attachment and parenting intervention. It is highly acceptable to families, has a clear and empirically supported model of change and has strong evidence of efficacy with families of young children in many settings.21 We proposed to have staff who had received training in the intervention visit families at home for six 90-minute sessions over 4 months. Sessions 1–5 take place biweekly and focus on improving carer sensitivity to the child. Session 6 is a booster session; it can take place approximately 1 month after the fifth session and focuses on consolidating the information provided in sessions 1–5 and teaching carers sensitive discipline skills.

Sessions involve both the carer and child, and start by video-recording carer–child interactions, which provides the basis for themed discussion in the next session when the intervener feeds back the clips, pausing the video at certain moments to highlight and provide messages according to manualised themes. Themes include recognising the child’s attachment signals and expressions, providing prompt and adequate responses to them, promoting empathy for the child, praising positive behaviour and appropriate ignoring of negative behaviour. Carers are also given exercises and tips. Session content is consistent across families, although its presentation and the video feedback are tailored to the specific needs of each family.

The VIPP-SD has recently been modified specifically for the Dutch foster care context by the developers of the programme in the Netherlands. This modified programme (VIPP-FC) pays particular attention to the need to help carers recognise signals that are specific to foster children, which may be quite challenging and difficult to understand, so that they are better equipped to respond sensitively and to support the child’s secure attachment to them as their carer. The VIPP-FC is thus designed to specifically address the attachment difficulties often shown by children in foster care. In particular, these children tend to demonstrate attachment signals that are very subtle, highly distorted or absent, for example not crying when hurt because they are not used to being comforted, or becoming angry or fearful when comforted. By helping foster carers become aware of these subtle, distorted or absent signals and training them to respond appropriately, the child can learn that foster carers can be relied on in times of need and that it is safe to show their distress and seek comfort.

Many children in foster care have extremely negative experience of physical interactions with carers; hence, the VIPP-SD modified for foster care also specifically addresses how to gently and sensitively support a child’s needs for physical contact and comfort.28

The intervention manual and delivery were further modified for this study to meet the systems and circumstances of the UK foster care context, as described in Chapter 2. We will refer to the modified intervention as VIPP-FC.

Training of interveners

We trained a group of individuals from an NHS Specialist Looked-after Children’s CAMHS team in the delivery of the VIPP-FC modified intervention. The interveners of the case series were already VIPP-SD trained, experienced practitioners. Each intervener was trained by an accredited VIPP-SD trainer and was required to undertake supervised practice before becoming a study therapist. Accredited supervision face to face, by telephone and by e-mail was provided to all practitioners.

Treatment fidelity

We ensured fidelity through supervision of three visits per case. All interveners completed logbooks for each visit, outlining how closely they followed the manual, if they completed all the filming tasks and if they included key messages in their feedback to the carer; logbooks were required for each different visit. They were also required to audio record at least one of their feedback sessions per case.

Results

Screening response rates

We present the overall response and retention rates in the form of a Consolidated Standards of Reporting Trials (CONSORT) flow diagram in Figure 1. As is evident from the diagram, we achieved a relatively good response rate of 60%, or 12 out of 20 carers.

FIGURE 1 The CONSORT flow diagram for the case series. (Figure not reproduced here; see the original article.)

Performance of the screening questionnaires

We found a moderate prevalence of RAD symptoms in this sample. Specifically, five out of the nine completing the DAI scored positive for at least one RAD symptom/sign, and four scored positive for more than one. It is unclear, of course, if the prevalence in this small sample is representative of the population of children in foster and kinship care or placed with guardians in the UK, and, with such a small sample size and low sample rate, the estimate has a very high degree of uncertainty attached to it.

Treated cases

Of the nine children whose carers completed the DAI, only three were eligible to proceed to treatment (see the reasons for exclusion in Figure 1). These three children underwent all baseline assessments with their carer. After the assessments were completed, one of them left the study because the child’s birth parents (who had received opt-out letters) contacted the social worker to refuse the child’s participation.

Hence, two children went on to receive the VIPP-FC, each provided VIPP-FC by a different CAMHS practitioner. They both completed all six sessions and all of the follow-up (i.e. post-treatment) research assessments.

Process observations: recruitment and consent

It is important to note that the 60% return rate was achieved only after substantial effort, which involved the study team calling social workers multiple times and asking them to encourage carers to return questionnaires, which, in turn, often required social workers speaking to carers several times. Another key learning point from the case series was that only a very small proportion (15%) of those potentially eligible (i.e. the 20 included in the mail-out) were eligible and willing to consent to the second stage of the study, during which they would receive the intervention.

Although return of the questionnaires was challenging, once we were in direct contact with the foster carers, they were generally very willing to hear more about the study and to take part in the subsequent stages. In spite of that, six out of the nine who completed the post-screening interview were, for various reasons, ineligible to take part in the baseline/follow-up assessments and receive the intervention. The feedback of those who took part in the more comprehensive assessments (i.e. at baseline and follow-up) was that these were acceptable and not overly tiresome.

Before and during the recruitment period, a member of the research team attended social worker and supervising social worker meetings to disseminate the study and encourage their involvement. Social workers appeared interested in the research and, in some instances, were able to identify cases that were potentially appropriate for taking part. On the other hand, some negative reactions were also experienced, such as disappointment regarding the limited age range of the target population and some confusion regarding the nature of the clincal presentation we were focusing on. Attempts to join foster carer groups were not successful.

The research team was able to explore and record the reasons for refusal from the birth parents who withdrew consent for their child’s participation, because the father consented to a member of the research team contacting him. The reasons for refusal provided by the father can be summarised as (1) frustration and anger at the local authority for having removed the child from his care; (2) an unwillingness to ‘help’ the local authority and what was seen as one of its partners (the research team); and (3) an unwillingness to allow a service to be offered to the child in the context of his relationship with his current (foster) carers, which was seen as contributing to the quality of the placement. The local authority and the TSC, in response to this, raised the more general question of whether or not there were circumstances in which parents may be considered to be not acting in the best interests of the child, for example when there has been limited contact and there are fraught relationships with the local authority. The TSC recommended that this issue be raised with the Research Ethics Committee. On discussion with the chairperson of the Research Ethics Committee, we agreed a revision of the initial consent procedure for cases on a full care order. Specifically, for the subsequent rounds of recruitment, local authorities were instructed to consider carefully if any birth parents with a child on a full care order might not be in a position to act in the child’s best interests. In such circumstances, the local authority could be permitted to give consent unilaterally.

Process observations: joint working with the local authority

This project involved working closely with the children’s services within a local authority; indeed, the initial step of making contact with foster carers and parents could be completed by the local authority only. Therefore, the study team needed to establish a strong partnership with the local authority, and considerable effort was focused on working and continuously improving the collaboration with the local authority to support the smooth running of the study.

Although the case series involved a single local authority only, a number of key practical implementation issues arose that were critical to consider in preparing the subsequent pilot RCT in other sites. Here we summarise the main findings regarding implementation with respect to the screening process and joint working with the local authority:

• The importance of social worker involvement and agreement – social workers play a central role in implementing a workable screening system in this context. At least two key issues arose in this respect. First, information in local authority digital records was not always up to date, and social workers sometimes held the most up-to-date information about children and their situations, including details and nuances about children’s placements and plans that might not be possible to gather from their electronic files. Because children’s circumstances can change quite often and quickly, close liaison with a child’s social worker was critical. This was not always easy, as social workers were generally very busy and were not always able to return telephone calls or reply to e-mails. This link to social workers for rapid communication of important case information or updates regarding location or suitability of cases is a critical potential vulnerability and a considerable amount of study team time/resourcing was required to make this work. Under-resourcing this aspect of project infrastructure could jeopardise a trial. Second, as representatives of the child’s interests, individual social workers’ agreement or ‘approval’ for a child’s participation in the study was generally expected (by the social workers themselves and, often, by the foster carers), even if someone with the relevant authority had granted permission already. This reinforces the importance of close working relationships with the children’s social workers, as well as the wider team within the local authority.
• Importance of the role of a ‘link’ member of staff – it was clear during this work that it is essential to have a contact person at the local authority who links the study team to the rest of the children’s services team. In the first study site, an internal member of staff at the local authority (initially, a senior social worker and, at a later stage, an administrative support worker) fulfilled this role of connecting the study team and the children’s services. Owing to the changing and somewhat unpredictable environment of these settings, it may not always be realistic to maintain a continuous dialogue with a large number of members of staff working directly with children. Therefore, this ‘link person’ is crucial to make communication effective and not diffuse responsibility for taking the project forward within the local authority. It is likely to be helpful if this is the same person who is designated to undertake the administrative tasks required for recruitment of participants to the study, namely obtaining a list of potential participants, mailing out letters and screening packs, and contacting carers.
• Limited resourcing – it was clear from the first phase of work that local authorities’ resources are limited and that social workers and support staff tend to be under considerable pressure and have little time available to dedicate to a research project, despite their willingness in principle. We made extensive efforts to make sure that the study was ‘on the radar’ of colleagues in the local authority, but the sheer volume of work they face often slowed study procedures significantly (obtaining up-to-date lists of potential recruits, gaining local authority consents, sending out letters, speaking to foster carers on behalf of the study team, etc.). We explored the possibility of bringing a member of the research team into the local authority to support these administrative activities and to lessen the impact on core staff (e.g. from within our team or a clinical support officer from the Clinical Research Network). However, we faced some anxiety from the local authority about allowing an external person into the authority (with an appropriate honorary contract/appointment in place) to support its participation in the study. The reluctance to pursue this possibility appeared to be partly apprehension about governance and training issues, but was also related to an unwillingness to invest the time required to set up such a contract, which was perceived to be an arduous bureaucratic process.
• Lengthy delays in obtaining data from the local authority – gathering information and, particularly, obtaining a list of potential participants (even with relatively simple criteria, such as children aged ≤ 6 years in certain types of placement) took longer than anticipated, usually at least several weeks after all the agreements had been put in place and the go-ahead had been given by a relevant senior manager to generate a list. Therefore, recruitment needs to be planned well in advance and with this time frame in mind. Because children’s circumstances can change quite rapidly, this can also mean that the list becomes out of date by the time it is acted on and may need updating.
• Researcher presence at social workers meetings – we expected that it would be important for the research team to attend social workers’ team meetings to disseminate the study and engage social workers and supervising social workers. It was clear that it would not be enough to go to a single meeting as the study could quickly become forgotten and we would not be able to meet all staff in a single meeting. However, we found that getting invited to these meetings was a challenge in itself, despite the local authority’s willingness to invite the team in principle. Our impression was that this largely reflected the very busy working environment and the stretched resources within local authorities.
• Lack of availability of social workers – social workers (particularly children’s social workers) tended to be extremely busy and were often away from their desks on home visits. They were, in the majority of cases, difficult to reach by e-mail or telephone. It was clear that the study would work best if it relied on them as little as possible and used what little time they had available for the most critical functions. For example, it did not appear to be feasible to ask individual social workers to mail out screening packs or to ask foster carers to return the packs via the social worker. Furthermore, although relying on individual social workers to follow up on questionnaire completion from carers is often necessary, it is important to also have alternative strategies in place to supplement that work (e.g. having an administrator in the local authority to make direct telephone calls to carers).
• We tended to find that supervising social workers were easier to get a hold of and could be very helpful in following up foster carers about screening questionnaires. They were also interested in promoting the study as an aspect of improved foster carer training.
• Social workers appreciated receiving feedback from the research assessments, which, in some cases, could be incorporated into relevant clinical reports about the child.
• Finally, we had no success recruiting from foster care agencies, partly because we would still be dependent on the local authorities to obtain consent to screen specific cases and partly because they were difficult to engage. Therefore, we were not able to recruit any cases through these agencies, despite contacting > 40 of them.

Process observations: working with foster carers

• Foster carers tended to be very busy, often caring for more than one child, and were responsible for completing a great deal of paperwork for the local authority as part of their formal role responsibilities. This was a challenge for the completion of the screening questionnaires and some foster carers fed back to us that if there is no one specifically highlighting the study, then questionnaires may just be ‘thrown on the pile’. The feedback reinforced the importance of promoting the study at multiple levels and in multiple forums, and, in particular, the importance of finding ways to encourage social workers (supervising social workers and children’s social workers) to promote the study directly to foster carers.
• Foster carers tended not to take the initiative to contact the research team if they had questions. In several instances, foster carers had queries about the questionnaires or about whether or not they were allowed to take part [i.e. whether or not the local authority had approved it (even though the cover letter indicated that it had)], but did not use the contact details provided to ask them. Follow-up telephone calls from the local authority on behalf of the research team were very important in these cases.
• Virtually no foster carers completed the research questionnaires without some chasing from the children’s social worker or the supervising social worker (or both), often on several occasions.
• Social workers’ encouragement of foster carers to complete the questionnaires was vital, but, given that it was very challenging to get social workers to do that, we proposed a protocol amendment to the Research Ethics Committee to allow us to contact foster carers directly if they gave the local authority/social worker verbal consent for us to do so. This was very helpful as it gave the research team the opportunity to follow up on the questionnaires, clarify any doubts and show that they were available to help, and to do so without depending on a social worker to make several calls on their behalf. We suspect that making direct personal contact in this way was also reassuring to foster carers and helped build a direct relationship with the study team.
• Some foster carers did not complete the questionnaires because they thought that the child would not be eligible for several reasons. This would tend to mean that non-completion could be non-random and selectively over-represented by low-scoring cases. This may be less of a problem for simply identifying cases, but is problematic for estimating the proportions of cases meeting RAD criteria. We tried to mitigate this in preparation for the RCT by emphasising in the information sheet that we were keen to hear from all foster carers, including those caring for children who did not have any emotional difficulties, but it was unclear how much impact this information had.
• We concluded in discussions with our foster carer team member that an incentive to return questionnaires, in the form of a voucher offered to the carer, might help with return rates; although we could not implement this in time for the case series, we were able to introduce this change to the feasibility RCT.
• Research assessments – the baseline and follow-up assessments took longer to complete than initially expected. To manage this, we implemented a number of strategies that generally worked well. Specifically, to make the study visit of a manageable duration, we offered flexible options for completing any assessments that did not require a formal assessment/laboratory context to complete (i.e. all but the Strange Situation Procedure and interaction observation). Thus, we offered families the opportunity to complete paper questionnaires before the visit and to complete the DAWBA online at home, as well as the opportunity to pay a second visit to a carer’s home to complete these, if needed. We explored the possibility of doing the DAWBA or the DAI over the telephone, but concluded that this was not feasible, as reading aloud the questions and options for the former took too long, and being unable to see and establish rapport with the interviewee made the latter more challenging.
• Foster carers appreciated receiving feedback about the assessments and found it helpful. This also proved to be very valuable for social workers, both for their own use and because they appreciated the support it offered foster carers.
• Although based on a very small number of cases, we found that, once initial contact was established and the team met foster carers in person, foster carers tended to remain engaged with the study and complete all of the assessments and all VIPP-FC sessions.

Process observations: working with parents

• Parents often did not live at the same address; thus, the decision of whom to send the information letter to (i.e. one or both parents) was made by the local authority.
• Even though the research team was available to seek consent from birth parents directly and in person, social workers preferred to be the ones requesting consent from parents in person, usually when they were meeting them for other reasons. However, this meant that obtaining written consent from parents could take a long time, as we had to wait for the social workers to obtain this.
• Although parents have a right to refuse consent for their child to take part in the study, we encountered an example in which the local authority considered that the parents were refusing for reasons that the authority deemed were not in the child’s best interest. Clearly, in some cases, relationships between parents and the local authority can become strained and parents may have limited involvement in their child’s life. After discussion with the local authority and the TSC, we met with the chairperson of the NHS Research Ethics Committee, who advised us that the protocol provided scope, when a child is subject to a full care order, for local authorities to exercise discretion about whether or not the parent is in a position to give consent in the best interests of the child. Thus, on the recommendation of the TSC, we moved to asking local authorities to verify, case by case, which parents should be notified about the study before seeking their consent (both at the screening stage and at the baseline stage).

Process observations: working with children

• In this young cohort (i.e. aged ≤ 6 years) placements are often short term and there is some unpredictability in terms of the duration of placements. A key question we needed to consider was when a child should be considered eligible for the study, which involves a minimum 4-month intervention, when the likely duration of placement is difficult to predict. Through discussions with social workers and direct experience through the case series, we concluded that the best approach is to contact as many foster carers as possible at the screening stage (i.e. not limit screening based on expected placement length initially) and to have discussions on a case-by-case basis regarding potential progression to the intervention stage. Maintaining ongoing communication with those cases for whom there was uncertainty about placement length was important (e.g. if a final hearing was scheduled during the early phases of engagement or screening). Allowing flexibility and individualised decision-making would be beneficial for a future study, given that such cases are unlikely to be rare.
• Research assessments: we found, albeit in a small set of cases, that the children were happy to come to the study visits and do the assessments. It was also clear that it was important to keep the visits relatively short, as children could become impatient or tired after a while, particularly when they were not involved directly (i.e. when the carer was being interviewed).
• When the carer needs to bring siblings along to the study visit, it is crucial to have at least a second member of the research team to help.

Conclusion

The reason for including a testing phase, in the form of a series of case studies, was to road-test the mechanisms of a study that we predicted to be challenging. Specifically, this phase aimed to pilot the recruitment and screening procedures, as well as the modified VIPP-FC, and to examine its clinical suitability for children with RAD symptoms in real-world clinical practice. The implementation of the case series was indeed challenging, but highlighted many crucial practical issues about what works and what does not, allowing us to introduce important modifications to the research protocol for the feasibility RCT.

The most important learning points from this first stage of the study were as follows:

• Staff in local authority children’s services are extremely busy and find it difficult to effectively support research; therefore, when possible, researchers need to minimise the burden on social work staff, and local authorities more generally, to allow them to participate and focus exclusively on aspects of the study that are most important or ‘mission critical’ and cannot be done by others.
• Foster carers are also very busy and ‘desensitised’ to requests for completing paperwork, thus they need encouragement and support to complete research forms and questionnaires.
• The adaptations of the VIPP-FC manual for this study were mainly concerned with specificities of the population (i.e. looked-after children in the UK), and not, for the most part, in the fundamental intervention approach, including whether or not it is suitable for children with RAD in particular. The VIPP-FC was generally considered to be suitable and appropriate, and the intervention was considered helpful and acceptable by the foster carers who received it. A substantial number of adaptations to the VIPP-FC programme concerned the practicalities of the implementation and surrounding context.
• The percentage of conversion from numbers approached to numbers recruited to the intervention was low (15%).

We faced unexpected delays and challenges to recruitment from using a single mail-out in a single recruitment site. A key issue was that the initially lengthy list of potential participants was substantially reduced after the social work teams removed inappropriate cases or cases expected to be very short placements. As a result, we could not easily recruit further cases from this site to continue to the case series. The data we obtained suggested comparatively good return rates for the screening measures from foster carers (60%), and we encountered no difficulties with the data quality or in the administration of the baseline and follow-up research instruments. In addition, the Manual Development Working Group took the view that the manual was working smoothly and was unlikely to require further changes. On that basis, and with the agreement of the TSC, it was concluded that we should expand the number of recruitment sites and proceed to the main pilot trial.

To conclude, the case series highlighted some significant challenges to recruitment and organisational complexities, but, at the same time, it gave a promising indication for the application of this intervention to improve mental health outcomes among looked-after children in the UK.