Exploring harm in psychotherapy: Perspectives of clinicians working with children and young people

The potential for harm to occur from therapy has been recognised for decades (Barlow, 2010). Yet despite this recognition, the field has not progressed sufficiently (Parry, Crawford, & Duggan, 2016). One reason is the lack of consensus around what is defined as harm and how it links with other frequently used terms, such as adverse events, unwanted effects, negative effects, side effects and clinical deterioration (Klatte, Strauss, Flückiger, & Rosendahl, 2018; Linden, 2013; Parry et al., 2016; Vaughan, Goldstein, Alikakos, Cohen, & Serby, 2014). Despite this, how harm can be caused has been previously explored in adult patients (Dimidjian & Hollon, 2010; Duggan, Parry, McMurran, Davidson, & Dennis, 2014; Linden, 2013). The causes are likely to be multifactorial and to happen because of a combination of patient, clinician and therapeutic factors (Hardy et al., 2019; Jonsson, Johanson, Nilsson, & Lindblad, 2016; Mohr, 1995). For example, therapist factors interact with patient Received: 29 April 2020 | Revised: 14 August 2020 | Accepted: 14 August 2020 DOI: 10.1002/capr.12347


| Literature around harm from psychological therapies in adult populations
Literature around harm has often focused on the prevalence of adverse or unwanted events (Moritz et al., 2015(Moritz et al., , 2018Schermuly-Haupt, Linden, & Rush, 2018). Recent estimates in patients with depression found that 39% reported at least one side effect (Moritz et al., 2018), whilst for patients with OCD, 93% of patients reported one side effect (Moritz et al., 2015). In another study focusing on unwanted and side effects from CBT, negative effects on well-being/increased distress as a result of treatment was reported in 27% of patients, a worsening in symptoms in 9% of patients, and strains in family relationships in 6% of participants (Schermuly-Haupt et al., 2018). Moreover, 21% of patients reported these negative outcomes as severe or very severe, whilst a further 5% reported these as persistent.
Clinicians' perspectives of negative effects have also been explored. In a study with 74 licensed practitioners, different types of negative effects were identified, and later classified as 'short term negative effects', 'no treatment effect', 'deterioration', 'dependency', and 'impact on other life domains' (Bystedt, Rozental, Andersson, Boettcher, & Carlbring, 2014). Causal factors for harm were also identified. These included 'incompetence and inadequately applied methods', 'failed ethical judgement and professional conduct', and 'discontinuing treatment'. However, this study is limited to the results of clinicians who practiced CBT, and how this translates to other talking treatments remains unclear. Bowie, McLeod, and McLeod (2016) carried out interviews with therapists, who as clients experienced unhelpful therapy. Four superordinate themes were gathered: 'sourcing a therapist', 'the therapist working to their own agenda', 'the pivotal moment-deciding not to continue' and 'the impact of receiving unhelpful therapy'.
The authors concluded that unhelpful therapy may be related to the breakdown of the therapeutic relationship and that there was a difficulty in letting the therapist know about their unhelpful experience.
Nevertheless, the study only included participants that had completed their own psychotherapy training and had private therapy, which would differ in terms of the power dynamic and stakeholders involved when it comes to therapy with children and young people.

| Literature around harm from psychological therapies in children and young people
There is a paucity of studies when it comes to harm from talking therapies with children and young people. One review explored how therapies could be classified as harmful by using the concept of PHT combined with four specific adverse childhood experiences: physical hurt of humiliation, physical abuse, not feeling loved or important, and neglect (Mercer, 2017). Using this framework, four therapies were identified as harmful: Conditioning/Operant Punishment, Holding Therapy/Attachment Therapy and Diagnosis, Festhaltetherapie (Holding Time, Prolonged Parent-Child Embrace) and Conversion therapy (Mercer, 2017). The author posited whether the identified therapies should be made less readily available or even prohibited. Whilst useful, this study does not take into account how harm or side effects may occur for children and young people in more mainstream therapies.
To date, only one qualitative study on this topic has been conducted (Jonsson et al., 2016). This study explored Swedish practitioners' experiences of possible adverse effects of psychological therapy with children and adolescents. Four themes were classified from the interviews: 'Vagueness of the concept', 'Psychotherapistclient interaction', 'Consequences of the child' and 'Family effects'.
The authors concluded that the concept of harm appeared unfamiliar and vague to many practitioners, and as a result, these may go unnoticed (Jonsson et al., 2016). Whilst helpful in beginning to aid understanding of how therapy might be harmful, this study was skewed towards therapists who mainly practiced within a CBT framework. Thus, how practitioners of other therapeutic domains view harm is unexplored.
Given the paucity of research in the field and the unique considerations which children and young people face when coming to therapy, further research is needed, both with other types of clinicians and within a UK context. As such, the aim of this study is to explore UK clinicians' perspectives on harm when working with children and young people. Questions will focus both on the types of harm that can occur as well as how it occurs (e.g. mechanisms).

| Participants
To be eligible to participate in the study, participants had to meet the following criteria: a. Be currently practicing a therapeutic approach in the UK b. Work with children and/or adolescents (up to 25 years old) c. Have a good grasp of English d. Be willing to take part in an interview, either face-to-face or over the phone.
Overall, 11 clinicians were interviewed. This included four males and seven females. Nine clinicians outlined that they worked within an integrative framework, one within a CBT framework, and one within a systemic framework. Clinicians had been practising for a 17.5 years on average (SD = 9.42).

| Procedure
A semi-structured interview schedule was constructed to explore clinicians' notions of harm. Unlike previous research (Jonsson et al., 2016), a definition of harm was not provided at the start of the interview in order to minimise bias and understand how clinicians' thought of harm and how it could occur (see Appendix A for the developed interview schedule).
Participants were recruited by convenience sampling via social media and through The Anna Freud National Centre for Children and Families networks/newsletters. Clinicians who expressed an interest in taking part were provided with an information sheet and a time was arranged to follow up to see if they had any questions and wanted to participate. Overall, one clinician did not proceed with the study and did not provide a reason why. Those that agreed to take part signed a consent form which was either given to the researcher in person (for face-to-face interviews) or via post/email (if the interview was conducted on the phone). All interviews were undertaken by the primary author (BCB) as part of their MSc dissertation and audio recorded.
After taking place, interviews were transcribed verbatim.
Analysis was undertaken in a series of steps, which included familiarising oneself with the data, the generation of codes, searching for themes, the reviewing of themes, defining and naming themes, and producing a report. Coding was first undertaken by the primary author (BCB) and was then shown to the senior author (DH) for critical appraisal and refinement. Following this, codes on similar topics were then collated, refined and grouped into a list of themes by both authors.

| Ethical considerations
This project was reviewed and approved by University College London Research Ethics Committee (Project ID Number: 14957/001).

| Types of harm that can occur from talking therapies
All participants identified at least one type of harm that could occur from psychotherapy.

| Clinical deterioration
When asked about harm in psychotherapy, most participants highlighted clinical deterioration: … [If] partway through therapy or by the end of the therapeutic process, a child is functioning less well, or feels less well than they were when they started.

(Participant 11)
Whilst getting worse in some circumstances was acknowledged as inevitable, it was felt that the worsening, or lack of progress with symptoms, could create a situation where children, young people or parents may not seek help in the future as therapy was seen as ineffective: So many young people I've worked with, have had, 10 therapists? They've tried every modality under the sun, and will say 'quite frankly, it's bollocks', or, 'I never wanna go there again'.

| Retraumatisation
As well as clinical deterioration, participants also mentioned retraumatisation as a possible harmful effect of psychotherapy. This was prominent in clinicians who had background in trauma, abuse and neglect: … You can run the risk of pushing something that actually the child or young person isn't ready to think about. And that can probably or could be potentially retraumatising.

(Participant 8)
To mitigate against this, some participants outlined how support networks around the young person were critical. If support networks were not able to withstand or hold the treatment effects of trauma work, this could lead to the possibility of harm occurring outside of the clinical context: Well, for trauma work, I would be looking at how containing the support network is around them. So, I wouldn't start working with a young person on a trauma if they didn't have a strong, stable, emotionally supportive environment that is going to be able to manage any emotional material that came up in between sessions.

| Theme I: Factors related to therapeutic administration
Many clinicians mentioned factors related to therapeutic administration which could result in harm. These focused on the assessment process, timing and choosing the right intervention.

Subtheme I: Assessment processes
Participants mentioned that a thorough assessment was important in helping mitigate against harm. During the first session(s) where assessment was undertaken, clinicians felt it was important to get to know the patients and their support networks, as this would create the set-up for subsequent meetings: The first few sessions is finding out about, with their parents, finding out about their past and their story and feeling my way through how the family talks about it or doesn't talk about it, how the child responds to TA B L E 1 Themes and subthemes in relation to clinicians' views of harm Ultimately, this may not be the right 'fit' and could result in the patient not getting better or dropping out.
Similar views were also posited more broadly in relation to types of treatment, with some clinicians querying whether talking therapies would be the best option when other types of support were available: … There are lots of routes in to go to mental health and talking therapy is just often a tiny tiny part of it…I don't think we have a good idea of when it is [the best treatment option] and when it isn't.
(Participant 9) Involving other professionals was seen as useful here, as well as discussing options with the patient.

| Theme II: Relationship factors
Around half of participants mentioned that harm could occur as a result of power dynamics, as the young person was in a weaker position of authority than both the clinician and their parent/guardian.

Subtheme I: Child-Caregiver relationship
Whilst clinicians highlighted that parents and guardians had an important role to play, some questioned whether harm could be caused as a result of children and young people being forced to attend therapy. As a result of this, they would not be assenting to treatment:

| Theme III: Therapist factors
Subtheme: Therapists' ability to reflect Participants mentioned that the inability of the therapist to reflect on if or why treatment was not working had the potential to result in harm: I think one of the dangers is, well, is I know that some therapists and some psychologists don't tolerate easily or acknowledge when therapy isn't working.
Reflecting on when an approach was unsuccessful meant changes could be made, rather than harming the patient via continuing with in- Having open and honest conversations in this instance was thought to help mitigate this.

| Theme IV: Contextual factors
Three subthemes were identified as mechanisms for harm under contextual factors: insufficient support, lack of time, and the individualised focus of 1-1 therapy.

Subtheme II: Lack of time
Clinicians thought there was a tension between wanting to allow the patient space to explore difficulties, whilst also knowing there were time limits imposed on how many times the patient could be seen: There's little flexibility in the number of sessions you provide for instance, which is often the case of the NHS or even here, you can run the risk of pushing something that actually the child or young person isn't ready to think about.

(Participant 8)
As a result, this tension could result in the need to progress things faster than what would be optimal, or leaving the child open and exposed.
Participants also expressed that harm could be caused from discharging patients if they failed to attend a set number of sessions.

| D ISCUSS I ON
The aim of this study was to explore clinicians' perspectives on harm in psychotherapy when working with children and young people.
Specifically, it set to explore what types of harm clinicians could identify, as well as mechanisms around how harm is caused. Each of these will be explored below.

| Types of harm
Clinicians identified two types of harm that could occur: clinical deterioration and retraumatisation. Clinical deterioration as a type of harm has previously been identified within the literature, both in quantitative studies and qualitative studies (Jonsson et al., 2016;Lilienfeld, 2007;Mohr, 1995;Schermuly-Haupt et al., 2018). Indeed, it has been suggested that this may be one of the most frequent side effects from talking therapies (Schermuly-Haupt et al., 2018).
Deterioration in clinical samples involving children and young people has been shown to be around 7.7% for anxiety and 3.7% for depression (Edbrooke-Childs, Wolpert, Zamperoni, Napoleone, & Bear, 2018). Not all clinicians identified this as a type of harm which fits with the notion that additional factors need to be taken into account when making these judgements Linden, 2013). One reason for this is some patients do get worse before they get better (Kramer & Stiles, 2015), and clinicians may view this as part of the therapeutic process.
Retraumatisation was the other type of harm identified. Unlike clinical deterioration, which could occur with any patient group, this is more nuanced and linked to the patient having to relive an identified trauma. None of the models reviewed in this paper address retraumatisation as an adverse event Lilienfeld, 2007;Linden, 2013). Thus, this may be considered a novel type of harm. Whilst there is overlap with similar concepts already identified, such as the emergence of new symptoms (Linden, 2013), this is different as it suggests a deliberate reliving of a pre-existing, buried emotion, which causes distress.
When compared to domains of the UE-ATR checklist, clinicians are not able to readily identify some of the categories identified. This appears to be particularly around events that occur outside of clinical settings. However, differences in terms of use, as well as not providing clinicians with a definition of harm at the beginning, such as in other research (Jonsson et al., 2016), may have contributed to these differences. Regardless, exploring clinicians' experience and notions of harm is relevant to enrich previous findings and identify different types of harm that could then be classified according to different criteria.

| Mechanisms of harm
In terms of how harm can be caused, four overarching categories were identified as follows: administrative factors, relationship factors, therapist factors and contextual factors.
Clinicians felt that in order to mitigate against harm, undertaking a thorough assessment was essential. This allowed the clinician to understand the presenting difficulties and formulate the best treatment plan. Additionally, selecting the 'right' treatment, which was congruent with the patients' values and preferences, was also con- The therapist being in a dominant position over the patient has been previously highlighted as leading to harm. Findings here partially fit with previous research exploring clinicians' perceptions of negative events (Jonsson et al., 2016), as both highlight how a clinician, when overbearing, could lead to a compliant young person who continues treatment without expressing concerns or worries.
Additionally, whilst the finding that parents too can dominate a conversation is not novel (Hayes, Edbrooke-Childs, Town, Wolpert, & Midgley, 2018), this is the first empirical study highlighting its role as a mechanism of harm. To counteract power dynamics, clinicians may wish to engage in shared decision-making and make sure they have time alone with the young person (Hayes, Fleming, & Wolpert, 2015).
The reflective capacity of the clinician was also deemed to be an important factor. One aspect of this focused on the clinician reflecting on their relationship with the patient, with harm caused when the therapist was not being mindful. The therapeutic relationship is an important aspect in determining outcomes (Shirk, Karver, & Brown, 2011). In terms of harm, clinicians have highlighted this poor therapeutic alliance as a contributing risk factor (Jonsson et al., 2016). The other aspect around reflective capacity focused on the clinician not recognising when the treatment was ineffective, which could cause harm if the treatment was continued. The ability of clinicians to predict treatment outcome and deterioration is poor and to overcome this, it has been suggested that regular monitoring and real-time change are needed (Lambert, 2011).
Wider contextual factors focused on insufficient support, a lack of time, and the individualised focus of therapy. Insufficient support focused on not getting good quality supervision, as well as clinicians having high workloads. Inadequate supervision and its effect on harm have previously been documented and are potentially more harmful when the therapist is inexperienced or has a complex case . However, even when supervision is available, it may not be used appropriately for issues around harm due to time constraints .
Linked with insufficient support was also a lack of time.

| Limitations of the study
Whilst this study contributes to an under-developed field of research, limitations exist. In particular, there was a skew towards clinicians who worked in trauma, as well as an integrative framework, meaning some types of mechanisms may have been missed, or overrepresented in these findings. Another further limitation is the sample size, which might have meant not reaching saturation of themes.
A further aspect to consider is that in terms of the types of harm involved or the mechanisms by which it may happen, some types of harm may fall outside the awareness of clinicians, due to relational or other factors such as not adhering to therapeutic principles. Indeed, some clinicians spoke about theoretical types of harm that could occur, rather than harm that had occurred in their practice or harm they knew patients who had come to them had previously experienced.

| Further lines of research
Further lines of research may include exploring harm with clinicians working in specific settings. This is relevant given the finding of 'retraumatisation' as a type of harm highlighted by those working in that field. Additionally, exploration with children, young people and parents around harm is also needed, particularly as there is a lack of congruence between them around the types of mental health difficulties and treatment options (Yeh & Weisz, 2001). This may allow for the identification of harmful experiences which remain out of the awareness of particular stakeholder groups, especially since patients' and therapists' understanding of negative effects are divergent (Curran et al., 2019).

| Implications
• Clinicians should be aware of the possibility of different types of harm happening and how this could impact on their own practice. This is especially true given that clinicians sometimes struggle to identify when things go wrong (Lambert, 2011).
• The importance of assessment has been highlighted, as well as the need for clinicians to have good quality, regular supervision.
• Including patients in care and treatment decisions can help mitigate against harm, as the option will be aligned with the values and help create an open and honest relationship.
• Health institutions should consider utilising data to understand when patients are deteriorating and the possible reasons for this. • Are there any risks for the clinician as well?

O RCI D
• Do you think children or adolescents can be more vulnerable to harm in psychotherapy than adults? If so, why? If not, why?
• What aspects are necessary to consider when making a decision about something being potentially harmful?