Defining the intervention
DDP practice and service delivery was found to vary greatly throughout the UK. Therefore three core models of DDP were constructed to broadly represent current service delivery in the UK: DDP Full-Basic, DDP Home-Based and DDP Long-Term. These three models were considered ‘full’ versions of DDP where DDP was used as a therapeutic approach with child and carers together and often with separate work done with carers alone . A distinct, ‘lighter’ approach was found to be practiced when DDP principles of Playfulness, Acceptance, Curiosity and Empathy (PACE)  were used to structure conversations, training and work with carers . Differences exist in service delivery between the UK and original model developed by Dan Hughes  and piloted by Arthur Becker-Weidman  in the USA. In Becker-Weidman et al.’s  pilot study, 23 two-hour sessions over 11 months were the average service delivery requirement of DDP. Carer-only sessions (without the child) took place before and after family sessions, while the child waited in a separate room. Such an approach would not be seen as acceptable the UK (unpublished observations: Quilter, MT; Follan, M; Blower, A; and Minnis, H). The transcripts revealed that in the UK, carer-only sessions tend to take place over the phone, email or on a separate occasion.
This model is termed ‘Full-Basic’ as it represents the full DDP service delivery model with the fewest total sessions and shortest duration per session. This model has an average duration of four months, in which each case will receive 15 family sessions lasting one hour and including both carers and child. Twelve additional sessions take place with carers only lasting 30 minutes each. This model is based on a practice whereby 30% of family sessions take place with two therapists. The model reflected this increase in therapist time as an increased number of family sessions (i.e. 20 sessions, rather than 15) to account for additional NHS provider time. The facilities in this model are most compatible with the original DDP model ,. The therapeutic space has video recording/playback for review after sessions and sofas and toys to make the setting comfortable for the entire family. In this ‘Full-Basic’ model of service delivery, an average of 12 cases per year would be seen per therapist.
This model of DDP delivery varied considerably from the USA model in that the service is delivered in the family’s home. A DDP therapist will travel to each family home, seeing each case for approximately 14 months, consisting of 28 family sessions and a further 28 sessions with carers only. In this model therapists arrive at the carers’ home 45 minutes before the child comes home from school to set up video equipment and meet with carers. The family session would then take place and be recorded for another 45 minutes. In this ‘Home-Based’ model of service delivery, an average of 6 cases per year would be seen per therapist. It is important to note that in this model, some resources are transferred between the NHS and families. There is less of a travel burden required for families travelling to NHS facilities. NHS travel expenses increase; however, there will be lower costs to the NHS as some of the overheads for facilities are transferred to the families, possibly freeing up therapeutic space for other NHS interventions.
This model represents the most resource intense delivery in terms of number and duration of sessions. Each case lasts for one year, with an average of 19 family sessions lasting one hour and 45 minutes each. In addition, 24 sessions lasting one hour and 30 minutes take place with carers only. Due to the intensity of delivery of this model, a DDP therapist would only see approximately four cases per year. With regards to therapeutic space, no video recording, sofas or toys are provided in addition to what is already available in a standard therapy clinical space. This aspect again diverges from the original model developed and delivered in the USA .
Resource use and unit costs
Costs were attributed to three main areas: NHS, families and social services. The resource use in the NHS consisted of professional time which was used for: consultations with carers only, consultations with families, preparation and review of the case and peer supervision. DDP needs to be provided by a trained clinical psychologist (NHS staff salary Band 8) . DDP is also resource intensive in terms of carers time, which may involve time off work for one or both carers; a cost to families. Additionally, in the Home-Based model, use of the carers’ home was a resource burden that fell on the family. In the Full-Basic and Long-Term models, the time and expense for carers travelling to the clinic is also considered. Carers are able to claim reimbursements from social services for travel expenses if they choose. Table 1 details the resource use and costs for each of the three DDP models. The unit cost for a clinical psychologist was £135 per hour . This is a societal cost including wages, salary on-costs (employer incurred national insurance and pensions contributions), overheads, capital overheads, costs to the provider for office, telephone, education and training, supplies and services, and utilities such as water, gas and electricity . For the DDP Home-Based model, the non-societal unit cost for a clinical psychologist was used (£60 per hour ) as overheads and use of a separate clinical space were not costs associated with use of an NHS facility.
Unit costs for materials were taken from market prices and estimates for total years of use were applied and discounted to present value. The discounted price of materials was then divided by the total number of cases seen per year by each DDP model.
As DDP is intended for foster or adopted children, it was assumed at least one caregiver would look after the child full-time. Full-time foster carers in the UK receive the minimum wage and, therefore, foster carers’ time spent in sessions was reimbursed at that wage . The second carer was assumed to work full time, and therefore the average wage in the UK was used to reflect the opportunity cost of their time away from work . Use of the home for sessions in the Home-Based model was accounted for by a £5 ‘token’ cost to the family to reflect the opportunity cost of using their facilities. This token cost was considered appropriate because even though the facility cost was transferred from the NHS to the family in this model, the family would not incur heavy NHS overheads, but was assumed to still keep ‘the lights on’ regardless of a DDP session taking place or not.
For travel expenses, transport was assumed to be by car and a £0.40/mile reimbursement rate was applied as per Department for Transport policy guidance . The DDP Home-Based model assumed a mean travel cost per case of £200 per month based on clinical opinion. Travel expenses in the DDP Full-Basic and DDP Long-Term models assumed an average mileage that covered the service area which would also be reimbursed by social services.
Comparator interventions that were mentioned by experts included: Individual Psychotherapy (IP), Consultation with Carers (CwC), Cognitive Behavioural Therapy, Play Therapy, Multi-Dimension Treatment Foster Care and many others. There was however, little agreement between experts regarding which therapies should be offered to which children  and none of these therapies were available in all areas. In the absence of DDP, children with MAPP in the UK could therefore be treated with a wide variety of interventions. This resulted in the decision that, if a future trial were undertaken, Child and Adolescent Mental Health Services (CAMHS) and services-as-usual (which include CwC and IP interventions) would be the most likely comparator as there is no ‘gold standard’ treatment used in every area. As CwC and IP are the most common standardised interventions in current CAMHS practice, these were used in this study as exemplars for the evaluation of control interventions in any future trial.
Consultation with carers
The CwC model consists of 10 consultation sessions lasting one and a half hours where a clinical professional such as a clinical nurse specialist or clinical psychologist (NHS staff salary Band 8)  advises carers on issues they may be having with their children and give parenting advice focused towards children with mental health problems. Consultations are complete in 10 months and each specialist can see approximately 24 cases per year. Approximately one out of 10 of these sessions take place at the carers’ home. Ideally both carers participate in sessions, but in reality both carers are present about half of the time and this is reflected in the model.
IP is a resource intensive intervention. Therapy involves 60 sessions over two years with each therapist seeing approximately eight cases per year. Each session lasts 50 minutes and 30 minutes are needed for preparation and review of the session. The health professional delivering psychotherapy would be an experienced clinical psychologist (NHS staff salary Band 8) . Toys are the only material item provided and were estimated to cost £25 per year per family.
A carer does not take part in IP, but their time is still needed to drop the child off and wait to pick them up. Therefore, their time driving to and from (one hour assumed) plus the length of the session is a cost to the family in terms of time away from work or other activities. Table 2 details resource use and costs for the two comparator interventions. Similar assumptions made in the DDP models such as travel time, mode of travel and discounting if the intervention lasts over one year, were applied to comparator models except for the unit cost for toys which was given. Table 3 summarises total costs to NHS, families and social services of the five different models.
Total cost per case for all models ranked from lowest to highest are: DDP Full-Basic (£6,700), CwC (£7,100), DDP Home-Based (£7,200), IP (£11,400) and DDP Long-Term (£14,500).
The least expensive treatment per case was DDP Full-Basic, with the fewest sessions and the shortest session duration. CwC was the second least costly treatment model, however, it should be noted that the time required for preparation and review of each case takes three times as long as providing the actual sessions. IP costs considerably more than CwC and the Full-Basic and Home-Based DDP models. The main cost driver is the large number of sessions, on average 60 per case. The most expensive treatment was DDP Long-Term due to its increased number and longer duration of sessions. This model was also the most costly to families as it required the most time from carers, representing a large opportunity cost of their time.
Appropriate principal outcome measure
It was evident from the interview transcripts that few DDP practices consistently measured outcomes of their cases; and some did not consider evaluation of their cases or the ‘impact’ of DDP at all. When prompted, clinical opinion on appropriate outcome measures for DDP and its comparators varied greatly amongst the DDP therapists interviewed. The most common outcome measures suggested in interviews were the Strengths and Difficulties Questionnaire (SDQ)  followed by the Kim Golding Carer Questionnaire. The later does not have any population-based validity; however it was developed specifically for DDP to address MAPP and behaviours that DDP tries to remedy. In the literature, the Child Behaviour Check List (CBCL) and Rutter Questionnaire are commonly cited. The CBCL was used in the Becker-Weidman  pilot study, however the SDQ was developed to cover the same domains as the CBCL and Rutter, and is now by far the most commonly used internationally. It was also directly validated against the Rutter Questionnaire .
The SDQ is well validated and available in 50 different languages. It is completed by carers or teachers for children aged 4–17 or by self-completion of children aged 11–17 years. It has 25 items divided into five symptom scales making up positive and negative attributes which are: emotional symptoms, conduct problems, hyperactivity, peer relationships and prosocial behaviour . The symptom scales are then summed (not including prosocial scale) to make up a Total Difficulties score. The SDQ is a measure of mental health and can be used to measure change as a result of treatment, however, it does not measure reactive attachment disorder. The Relationship Problems Questionnaire addresses this problem and is well validated. This questionnaire however has been used previously in a small-scale audit of DDP and was found not to be responsive to change (personal communication, May 2010, Julie Hudson).
Given the information from the interview transcripts and outcomes used in similar studies to date, it seems that the SDQ is likely to be an appropriate outcome measure for evaluating DDP in an RCT. The SDQ could be appropriate for determining short-term effectiveness in terms of mental health, however in terms of an economic analysis; a longer-term health gain is required. The short-term cost-effectiveness of DDP could be calculated by determining the incremental cost per improvement on the SDQ scale (improvement in mental health); however, a longer-term focus is required to show wider health and economic gains. Quality-adjusted life years (QALY) are the reference case outcome preferred by NICE for demonstrating the cost-effectiveness in health technology assessment and also for public health interventions ,. QALYs consider the life expectancy impacts of health care interventions and adjust any life year gains by the quality of that life. To calculate QALYs a preference based utility measure is needed; however, currently there are few validated preference based utility measures for children. The Child Health Utility 9D (CHU9D) measure of health related quality of life may be a suitable option. It is a generic paediatric preference based outcome measure for children ages 7–17 years and can be used for QALY calculation . Therefore it would be suitable for the majority of children in the age range for DDP treatment. CHU9D is a self-completion questionnaire, and could be easily incorporated into any future trial of DDP as a secondary outcome measure.