Abstract
Objective:
Shared decision making (SDM) between service users and providers is increasingly being suggested as a key component of good healthcare. The aim of this research was to explore whether child- and parent-reported experience of SDM was associated with child- and parent-reported improvement in psychosocial difficulties and clinician-reported functioning at the end of treatment in child and adolescent mental health services (CAMHS).
Method:
The sample comprised N = 177 children (62% female; 31% aged 6–12 and 69% aged 13–18) with a variety of mental health problems from 17 services where routinely collected data consisted of presenting problems at outset, child- and parent-reported change in symptoms between Time 1 and Time 2 (Strengths and Difficulties Questionnaire; SDQ), clinician-reported change in functioning between Time 1 and Time 2 (Children’s Global Assessment Scale; CGAS), and experience of SDM at Time 2 (as measured by responses to the Experience of Service Questionnaire; ESQ).
Results:
Analysis revealed that both child- and parent-reported experience of SDM were associated with higher levels of child- and parent-reported improvement in psychosocial difficulties. However, child-reported experience of SDM was only associated with higher levels of child-reported improvement when their parents also reported higher levels of SDM.
Conclusion:
In CAMHS, involving both children and parents in decision making may contribute to enhanced treatment outcomes.
Keywords
There is an increasing demand from policy makers and others to include more shared decision making (SDM) in healthcare (Department of Health, 2012; The Health Foundation, 2013). In particular, the Chief Medical Officer (2013) recently called for more SDM in the provision of services for children and young people. There is increasing evidence that ensuring collaborative practice and SDM in interventions for those with long-term physical or mental health conditions may contribute to improved self-management and patient activation along with better treatment outcomes (Bodenheimer, Lorig, Holman, & Grumbach, 2002; Storm & Edwards, 2012). While the mechanism of change of SDM is not fully clear, evidence across a range of healthcare settings suggests that implementing SDM tools may improve autonomy and self-confidence, which are in turn associated with better treatment adherence and improved clinical outcomes (Burton, Blundell, Jones, Fraser, & Elwyn, 2010; Center for Mental Health Services Substance Abuse and Mental Health Services Administration, 2010; Desroches, Lapointe, Deschenes, Gagnon, & Legare, 2011; Edwards & Elwyn, 2006; Joosten, De Jong, de Wert-van, Sensky, & van der Staak, 2011; Simmons, Rice, Hetrick, Bailey, & Parker, 2012; Swanson, Bastani, Rubenstein, Meredith, & Ford, 2007; Thompson et al., 2007; Wilson et al., 2010). SDM tools include the provision of information about treatment and options available and supporting patients to discuss their preferences and concerns, for example, with option grids and brief decision aids (see Department of Health, 2014).
SDM has been defined as “a process in which clinicians and patients work together to select tests, treatments, management or support packages, based on clinical evidence and the patient’s informed preferences” (Coulter & Collins, 2011; p. vii) involving a collaboration between both the patient and the clinician in understanding the goals of treatment, information about options and preferences, and agreeing next steps and how to move forward (Charles, Gafni, & Whelan, 1999; Coulter, Edwards, Elwyn, & Thomson, 2011; The Health Foundation, 2012, p. iv; Makoul & Clayman, 2006).
Defining what SDM might look like in the context of child and adolescent mental health services (CAMHS) raises particular complexities. As we have noted elsewhere, Attempting to implement SDM in CAMHS raises particular challenges in terms of complex conversations with vulnerable and often highly stressed or disturbed young children, involving on-going and complex relationships over time (rather than in a single decision point) and balancing multiple perspectives (e.g. child and parent). (Wolpert et al., 2012; p. 4)
Conceptualizing SDM in this context is not the same as in a context of one off decisions with only one patient: “SDM in CAMHS has been conceived of as involving five key aspects: (1) Agreeing key problems and goals, (2) Understanding options available, (3) Agreeing which option will be tried, (4) Reviewing progress, (5) Making changes as necessary” (Wolpert et al., 2012; p. 4). There is as yet not one validated measure for all these aspects of SDM but the experience of SDM may be captured by elements of patient experience questionnaires that focus on the experience of being consulted and listened to throughout the therapeutic process (Wolpert et al., 2012).
There is an increasing commitment to embed SDM in routine practice in CAMHS (Kent & Read, 1998; Richter, Halliday, Grømer, & Dybdahl, 2009; Soffe, Read, & Frude, 2004). Young people have the capacity to be involved in treatment decisions and doing so may have benefits to self-esteem and transition to adulthood (Alderson, Sutcliffe, & Curtis, 2006; Audit Commission, 1999; Costello, 2003; Green, McGinnity, Meltzer, Ford, & Goodman, 2005; Huffine, 2005; Scott, Knapp, Henderson, & Maughan, 2001). Young people want to be involved in decisions about their treatment (Alderson et al., 2006; Costello, 2003; Coyne, 2006; Dogra, 2005; Kelsey, Abelson-Mitchell, & Skirton, 2007) and report feeling under-empowered when they have limited access to information and involvement in care (Gyamfi, Keens-Douglas, & Medin, 2007). Children in acute care settings, for instance, report feeling alienated when clinicians speak only to their parents and that a lack of communication is a barrier to SDM (Kelsey et al., 2007; O’Brien, Crickard, Rapp, Holmes, & McDonald, 2011). Studies in child mental health settings have found that interventions that include SDM reduce (child-reported) decisional conflict and improve treatment outcome, in terms of (child-reported) depression, quality of life, and satisfaction (Asarnow et al., 2009; Richardson, McCauley, & Katon, 2009; Westermann, Verheij, Winkens, Verhulst, & Van Oort, 2013).
There are a number of particular challenges to the use of SDM in CAMHS. Concerns by professionals include the young person’s capacity to be involved in decision making (Fonagy, Steele, Moran, & Higgitt, 1991; Ruhe, Wangmo, Badarau, Elger, & Niggli, 2015), how best to involve parents who may have different experiences to the young person or their own psychopathology (Yeh & Weisz, 2001), and how to balance any safeguarding concerns with the involvement of young people and parents in decision making (Abrines-Jaume et al., 2014). Evidence suggests that young people, their parents, and therapists may have different perceptions of presenting problems and reasons for attending therapy (Bloemsma et al., 2012; Hawley & Weisz, 2005; Jepsen, Gray, & Taffe, 2012). Consequently, there may be tensions over who the patient is—the young person or their parents—making user involvement all the more problematic.
There is some evidence to suggest that for SDM to be associated with better treatment outcome in CAMHS, children may need their parents’ support to self-manage their mental health difficulties, and therefore, SDM should be considered from multiple perspectives (Alegría et al., 2008; Hibbard, Stockar, Mahoney, & Tusler, 2004; Storm & Edwards, 2012). Depending on their age and independence from adult caregivers, it may be that children can only be active patients, through SDM, when their parents are equally active. Evidence from physical and mental child health settings suggests that the use of SDM approaches is associated with higher levels of child–parent agreement, child-reported satisfaction with the decision-making process, and decision-making quality (Feenstra et al., 2014; Stacey et al., 2014). In a sample of children with special healthcare needs, higher levels of parent-reported SDM over time were associated with lower levels of (parent-reported) behavioral impairment (Fiks et al., 2012). In a sample of parents of children with mental health problems, higher levels of (parent-reported) SDM were associated with higher levels of satisfaction with care and information received and lower levels of symptomatology (Butler, 2014; Fiks, Localio, Alessandrini, Asch, & Guevara, 2010; Golnik, Maccabee-Ryaboy, Scal, Wey, & Gaillard, 2012).
The above evidence suggests that SDM may be associated with better treatment outcomes for children and families accessing mental health services. Still, to the best of our knowledge, there is no published evidence from data collected under routine conditions from child mental health services on the relationship between experience of SDM and treatment outcome and whether this is consistent between child- and parent-reported SDM. Previous studies exploring SDM in child services have captured the perspective of parents and clinicians, but this is the first study to explore the perspective of children themselves (Brinkman, 2011; Butler, Elkins, Kowalkowski, & Raphael, 2015; Fiks et al., 2012; Westermann et al., 2013). Examinations of data collected under routine conditions are needed to understand current practice and, therefore, how interventions can best promote SDM in child mental healthcare.
Aim of this research
The aim of this research was to explore whether child and parent experience of SDM was associated with treatment outcome in data collected under routine conditions from child mental health services. We expected that higher levels of both child- and parent-reported experience of SDM with clinicians during therapy would be associated with lower levels of child- and parent-reported psychosocial difficulties over time and higher levels of clinician-reported functioning over time.
Methods
Participants and procedure
The Child Outcomes Research Consortium (CORC, 2014b) is a learning collaboration of child mental health clinicians, managers, and funders, the members of which are professionals providing services for children and young people with mental health and well-being difficulties. Members routinely collect data, which is used to inform clinical practice, for service improvement, and for annual collation and analysis by the CORC central team. Member services include voluntary and statutory services providing therapeutic interventions (e.g. cognitive behavior therapy, systemic family practice).
Children from this dataset were included if their data were reported in or later than 2007, their case was closed, they had complete demographic characteristics and variables of interest (see the “Measures” section), and data were present for their service for more than one child. This resulted in a final sample of N = 177 (62% female; 31% aged 6–12 and 69% aged 13–18) children from 17 CAMHS in the United Kingdom (UK) with data from 3 to 46 children per service. Data were provided by a range of services: 16 statutory and 1 voluntary, inclusive of targeted, specialist, and highly specialist community and clinic-based teams, including therapeutic communities. In terms of ethnicity, 78% of children were White, 6% Asian, 5% Mixed, and 4% Black, with 5% not stating their ethnicity and 2% recording another ethnicity. Compared to the overall dataset, there was a larger proportion of females (CORC = 50% female, χ2(1) = 8.10, p < .01) and children aged 13–18 (CORC = 57%, χ2(1) = 10.28, p < .001) in the present study, but no significant differences in ethnicity: χ2(5) = 3.32, p > .05.
Overall, 10% of children had special educational needs recorded (obtained by the clinician at assessment or referral), which is less than the prevalence of 20% found in a previous study (Meltzer, Gatward, Goodman, & Ford, 2000). The presenting problems recorded were emotional disorder (77%), self-harm (12%), conduct disorder (9%), eating disorders (7%), hyperactivity (5%), autism (3%), learning difficulties (3%), habit disorder (2%), substance abuse (2%), developmental difficulties (1%), psychosis (1%), and other problems (14%) (multiple responses permitted). The treatment types recorded were cognitive behavior therapy (33%), child psychotherapy (2%), family therapy (12%), creative therapy (2%), drug treatment (6%), parent training (4%), other parenting interventions (8%), neuropsychology (1%), counseling (10%), and other therapies (18%) (multiple responses permitted).
According to the CORC protocol, questionnaires are completed by children, parents, and/or clinicians at assessment (Time 1 or T1) and again 4 to 6 months later (Time 2 or T2) or, if sooner, case closure (CORC, 2014a). Overall, 60% of children had treatment duration recorded: 31% were in the service for less than 6 months, 23% for 7–12 months, and 6% for 13–24 months. The measures were taken from a secondary analysis of routinely collected data so ethical review was not relevant (NHS, 2015).
Measures
Child- and parent-reported experience of SDM (T2 only)
To measure child- and parent-reported experience of SDM, the Experience of Service Questionnaire (ESQ; Attride-Stirling, 2002) was used. The ESQ is a patient-reported experience measure that captures perceptions of satisfaction with care received and environment (Brown, Ford, Deighton, & Wolpert, 2014). Four items were selected based on literature reviews of definitions of SDM (Makoul & Clayman, 2006) and existing measures (Elwyn et al., 2013; Right Care, 2012; Scholl et al., 2011). The research team reached consensus that these items assess the key components of SDM: (a) provision of information about the presenting problem or possible treatment options (i.e. “I have been given enough explanation about the help available here”) and (b) discussing and considering patients’ preferences regarding the presenting problem or treatment options (i.e. “I felt that the people who saw me [my child] listened to me,” “It was easy to talk to the people who saw me [have seen my child],” “My views and worries were taken seriously” (parent-reported version in square brackets)). There is an array of measures of SDM for adults, but there are none specifically developed for children (Right Care, 2012; Scholl et al., 2011). Children and parents responded on a three-point scale from certainly true (1) to not true (3), with a fourth option (don’t know) not used in scoring. The ESQ is widely used in CAMHS in the UK, and it has demonstrated inter-rater reliability and construct validity (Brown et al., 2014). As shown in the note of Table 1, internal consistency was .71 for child-reported experience of SDM (.69 for children aged 6–12 and .71 for children aged 13–18) and .62 for parent-reported experience of SDM (.60 for parents of children aged 6–12 and .63 for children aged 13–18), which are acceptable for research purposes (Nunally, 1967).
Multilevel regression models with SDM predicting child-, parent-, and clinician-reported treatment outcome.
Note. N = 177. Child SDM M (SD) = 1.23 (0.35); Cronbach’s alpha = .71. Parent SDM M (SD) = 1.11(0.24); Cronbach’s alpha = .62. Child-reported psychosocial difficulties T1 M (SD) = 17.22 (6.55), T2 M (SD) = 13.81 (6.72). Parent-reported psychosocial difficulties T1 M (SD) = 17.67 (6.79), T2 M (SD) = 12.69 (6.79). Clinician-reported functioning T1 M (SD) = 55.01 (10.46), T2 M (SD) = 70.59 (13.19). Higher scores of SDM indicate lower levels of SDM. Higher scores of psychosocial difficulties indicate higher levels of psychosocial difficulties. Higher levels of clinician-reported functioning indicate higher levels of functioning. Models 1 and 2 included demographic characteristics. SDM = shared decision making.
p < .05, **p < .01, ***p < .001.
Child- and parent-reported psychosocial difficulties (T1 and T2)
To measure child- and parent-reported psychosocial difficulties, the 20-item Strengths and Difficulties Questionnaire (SDQ; Goodman, 1997; Goodman, Meltzer, & Bailey, 1998) was used. The SDQ measures psychosocial difficulties and is comprised of four subscales assessing difficulties (i.e. conduct problems, emotional problems, peer problems, and hyperactivity). The SDQ is a widely used measure of psychosocial difficulties; in particular, the internal consistency of the SDQ has been reported as .71 (child-report) and .82 (parent-report) (Goodman, 1997; Goodman et al., 1998).
Clinician-reported functioning (T1 and T2)
To measure clinician-reported functioning, the unidimensional Children’s Global Assessment Scale (CGAS; Shaffer et al., 1983) was used. The CGAS is a measure of global functioning rated on a scale from lowest functioning (1) to excellent functioning (100). The CGAS is a widely used measure of functioning and has demonstrated reliability and validity in previous studies (Rey, Starling, Wever, Dossetor, & Plapp, 1995; Shaffer et al., 1983).
Analytic strategy
To examine the relationship between child- and parent-reported experience of SDM and child-reported psychosocial difficulties, parent-reported psychosocial difficulties, and clinician-reported functioning, multilevel modeling (MLM) was performed using SPSS 21 (IBM Corp, 2012). MLM was used given the hierarchical structure of the data, with time nested within children nested within services. For each dependent variable (child-reported psychosocial difficulties, parent-reported psychosocial difficulties, and clinician-reported functioning), two random intercept models were tested.
In Model 1, the associations between demographics and treatment outcome were examined, and one level-1 predictor was entered: time (coded 0 for Time 1 (T1) and 1 for Time 2 (T2)). The patient-level demographic characteristics were entered as level-2 predictors: gender (coded 1 for female); age (coded 1 for 6–121); White, Asian, Mixed, Black, and Other (each dummy coded 1, with not stated as the reference category); special educational needs (coded 1 for present); and emotional disorder, self-harm, conduct disorder, eating disorder, hyperactivity, autism, learning difficulties, habit disorder, substance abuse, developmental difficulties, psychosis, and other problems (each coded 1 for present). The intraclass correlation coefficient (ICC) was computed to examine the variance explained at the service-level (“service” refers to one commissioned provider or organization, likely to encompass many teams that provide a range of intervention and treatment types) and at the patient-level.
In Model 2, the association between experience of SDM and treatment outcome was examined, and six SDM variables were included as predictors, outlined below. To examine the relationship between experience of SDM and child-reported psychosocial difficulties, parent-reported psychosocial difficulties, and clinician-reported functioning, grand mean centered child-reported SDM, grand mean centered parent-reported SDM, and child-reported SDM × parent-reported SDM were included. To examine the relationship between experience of SDM and change in the three criterion variables from T1 to T2, time × child-reported SDM, time × parent-reported SDM, and time × child-reported SDM × parent-reported SDM were included. The likelihood ratio test was then used to compare the fit of Model 2 to Model 1. Descriptive statistics for all variables are reported in the note of Table 1.
Results
Results for child-reported psychosocial difficulties are discussed first, and then parent-reported psychosocial difficulties, followed by clinician-reported functioning. Results are shown in Table 1.
Child-reported psychosocial difficulties
In Model 1, 2% of the variance in child-reported psychosocial difficulties was explained at the service-level and 44% at the patient-level. Child-reported psychosocial difficulties significantly reduced over time. Adding experience of SDM in Model 2 significantly improved the model fit; likelihood ratio test: χ2(6) = 16.79, p < .05. Both time × child-reported SDM and time × parent-reported SDM significantly predicted lower levels of psychosocial difficulties over time. Moreover, a three-way interaction effect was significant: time × child-reported SDM × parent-reported SDM. To facilitate interpretation, the interaction was plotted in Figure 1, with T2 child-reported psychosocial difficulties predicted at high (+2 standard deviations (SD)), medium (0 SD), and low (−2 SD) levels of child-reported SDM (x-axis) and at high (+2 SD), medium (0 SD), and low (−2 SD) levels of parent-reported SDM (lines) (Fürst & Ghisletta, 2009).
Three-way interaction effect of time × child-reported SDM × parent-reported SDM predicting child-reported psychosocial difficulties at Time 2.
On the one hand, at higher levels of child-reported SDM, higher levels of parent-reported SDM were associated with lower levels of child-reported difficulties at T2, whereas lower levels of parent-reported SDM were associated with higher levels of child-reported difficulties at T2. On the other hand, at lower levels of child-reported SDM, higher levels of parent-reported SDM were associated with higher levels of child-reported difficulties at T2, whereas lower levels of parent-reported SDM were associated with little change in child-reported difficulties at T2.
Parent-reported psychosocial difficulties
In Model 1, 1% of the variance in parent-reported psychosocial difficulties was explained at the service-level and 51% at the patient-level. Parent-reported psychosocial difficulties significantly reduced over time. Adding experience of SDM in Model 2 significantly improved the model fit; likelihood ratio test: χ2(6) = 42.13, p < .05. Both time × child-reported SDM and time × parent-reported SDM were significant. Children with higher levels of child-reported experience of SDM or parent-reported experience of SDM had lower levels of parent-reported psychosocial difficulties over time than children with lower levels of child- or parent-reported experience of SDM.
Clinician-reported functioning
In Model 1, 8% of the variance in clinician-reported functioning was explained at the service-level and 29% at the patient-level. Clinician-reported functioning significantly improved over time. Adding experience of SDM in Model 2 did not significantly improve the model fit; likelihood ratio test: χ2(6) = 10.77, p > .05.
Discussion
The aim of this research was to explore whether child- and parent-reported experience of SDM were associated with child-, parent-, and clinician-reported treatment outcome in data collected under routine conditions from child mental health services. Child- and parent-reported experience of SDM were not significantly associated with improvement in clinician-reported functioning over time. However, both child- and parent-reported experience of SDM were signficiantly associated with higher levels of improvement in treatment outcome over time, as measured by child- and parent-reported psychosocial difficulties. Regarding child-reported psychosocial difficulties, however, child-reported experience of SDM was only associated with lower levels of child-reported psychosocial difficulties at higher levels of parent-reported experience of SDM—at lower levels of parent-reported experience of SDM, it was associated with higher levels of psychosocial difficulties.
These findings are in line with previous studies with samples of children and parents in physical and mental health settings. In particular, the finding that child-reported experience of SDM was associated with higher levels of child- and parent-reported improvement is in line with findings that interventions that include service users in shared decisions may help reduce depression and improve quality of life and satisfaction in child mental health settings (Asarnow et al., 2009; Richardson et al., 2009; Westermann et al., 2013). The finding that parent-reported experience of SDM was associated with higher levels of child- and parent-reported improvement is in line with findings that parent-reported SDM was associated with lower levels of psychosocial difficulties in samples of parents of children with mental health problems (Butler, 2014; Fiks et al., 2010; Golnik et al., 2012).
Given that child-reported experience of SDM was consistently associated with higher levels of improvement as rated by children and parents, these findings suggest that involving children in decision making may be important. Alternatively, this may be an artifact of children who had higher levels of improvement feeling as though they perceived levels of SDM to be better in retrospect. Communicating times when children may feel less involved in decisions may help children feel informed about, even if not involved in, shared decisions. The 10 steps of SDM may help to structure this dialogue (Simmons et al., 2012). This may involve discussing with a child their expectations and goals of treatment (i.e. what do they want to achieve from attending the service?), how to ask any questions they may have, their experience of sessions and whether issues they wanted to discuss were covered, and possible options for moving forward.
Parent-reported experience of SDM was associated with higher levels of improvement in parent-reported psychosocial difficulties, and only when parents reported higher levels of experience of SDM were higher levels of child-reported experience of SDM associated with higher levels of improvement in child-reported difficulties. One possible explanation for this interaction effect may be self-management and patient activation, in that SDM may support improvements in these areas (Alegría et al., 2008; Hibbard et al., 2004; Storm & Edwards, 2012). For children’s SDM to be associated with better treatment outcome, they may need their parents’ support both to access services (Arai, Stapley, & Roberts, 2014) and to self-manage their mental health difficulties given the systemic context of therapy with children. Perhaps children can only be active patients, through SDM, when their parents are equally active.
For parents to be active “patients” they may need to be supported to make shared decisions. Identifying, discussing, and raising problems with SDM with parents, and encouraging parents to do the same, may help promote SDM and treatment progress. It may be important to also highlight times when SDM may be less likely. There are a number of tools and decision aids for use with adult patients (The Health Foundation, 2013; NHS, 2012) although fewer for parents of children accessing services (Children’s Hospital of Eastern Ontario, n.d.; Crickard, O’Brien, Rapp, & Holmes, 2010; Mobilizing Minds, 2011).
Limitations should be considered when interpreting the findings of the present research. First, this study used data collected under routine conditions opposed to those collected under controlled conditions, which presents the associated limitations of confounding variables and selection bias (Gilbody, 2002). However, demographic characteristics were controlled for in all analyses (see the section “Analytic strategy”). Moreover, without a randomized prospective trial, we cannot infer causality, and children and parents may have rated higher levels of experience of SDM when they perceived improvements in psychosocial difficulties. Still, the findings of the present research are the first to triangulate perspectives from the child, parent, and clinician regarding experience of SDM and treatment outcome in current practice.
Second, the measure of SDM comprised items taken from a measure of service experience (Attride-Stirling, 2002) (see the “Measures” section). Future research should explore the convergent validity of the measure of experience of SDM used in this study with a measure developed specifically for children in mental health services. The four-item SURE measure (Légaré et al., 2010) and three-item CollaboRATE measure (Elwyn et al., 2013) developed with adults are promising options given their simple language and short items, which may mean they can be readily adapted for children. Finally, given the inclusion criteria of this study, there was a relatively small sample included (see the section “Participants and procedure”), meaning the findings may not generalize beyond the present sample. Moreover, demographic characteristics on parents and clinicians are not collected in the CORC dataset, and future research should explore the association between these variables with experience of SDM and treatment outcome.
Notwithstanding the above limitations, this study presents the first ecologically valid, longitudinal findings on the association between child-and parent-reported experience of SDM and treatment outcome in child mental health services. Reports of treatment outcome were triangulated across children, parents, and clinicians. Both child- and parent-reported experience of SDM were associated with higher levels of improvement in treatment outcome over time, as measured by child-and parent-reported psychosocial difficulties. However, for child-reported psychosocial difficulties, child-reported experience of SDM was only associated with lower levels of difficulties over time when their parents also reported higher levels of SDM: At lower levels of parent-reported experience of SDM, children reported higher levels of difficulties.
In CAMHS, therapists face the challenging task of involving both children and their parents in understanding the goals of treatment, sharing information about options and preferences, and mutually agreeing how to move forward (Coulter et al., 2011) within a complex therapeutic context. Findings of this study may suggest that both children and parents should be involved in SDM for families to have enhanced treatment outcomes.
Footnotes
Acknowledgements
The views expressed are not necessarily those of the Department of Health. The authors would like to thank members of CPRU: Terence Stephenson, Catherine Law, Amanda Edwards, Ruth Gilbert, Steve Morris, Helen Roberts, Cathy Street, and Russell Viner. The authors would also like to thank all members of CORC, its committee at the time of writing—(including M.W.): Ashley Wyatt, Duncan Law, Tamsin Ford, Evette Girgis, Julie Elliott, Ann York, Mick Atkinson, Alan Ovenden, Kate Martin, and Neil Humphrey—and the CORC central team at the time of writing (including J.J.): Isobel Fleming, Elisa Napoleone, Victoria Zamperoni, Lily Levy, Charlotte Payne, Kallum Rogers, Kate Dalzell, Craig Hamilton, Sally Wilson, Mark Garbett, Deborah Sheppard, Alison Ford, Amy Ellis-Thompson, and Jeni Page.
Ethics approval
In line with the Governance Arrangements for Research Ethics Committees (GAfREC), ethical review was not applicable as the data were taken from a secondary analysis of routinely collected information where patients or service users were not identifiable to the research team.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Policy Research Unit in the Health of Children, Young People, and Families (CPRU) is funded by the Department of Health Policy Research Programme. This is an independent report commissioned and funded by the Department of Health.