What helps distressed Australian adolescents impacted by cancer? Mechanisms of improvement of the PEER program

Abstract PEER is a four‐day residential program for adolescents impacted by their own or a relative's cancer, with both psychosocial (acceptance and commitment therapy, self‐compassion) and recreational components. This study aimed to determine whether previously observed improvements in quality of life amongst highly distressed participants were mediated by improvements in processes targeted by psychotherapeutic elements of the program (psychological flexibility, mindfulness, self‐compassion, peer support, distress). Adolescents attending PEER completed surveys assessing the quality of life and proposed mediator variables at pre‐program, post‐program and two‐month follow‐up. Adolescents experiencing high/very high levels of baseline distress (n = 52; 5 patients/survivors, 31 siblings/offspring, 13 bereaved siblings/offspring) were previously identified as experiencing clinically significant improvements in psychosocial well‐being; here, mediation analyses explored whether these improvements were associated with improvements in process variables. Findings evidenced improvements in quality of life amongst distressed PEER participants, mediated by increases in psychological flexibility and self‐compassion, and reductions in distress. Peer support and mindfulness were not significant mediators. Together, this suggests that the psychosocial benefits of PEER observed for highly distressed adolescents are linked to the specific therapeutic approaches used in the program, rather than being non‐specific effects of peer connection or recreation. Findings from this evaluation provide further evidence for the efficacy and mechanisms of the effect of PEER for supporting distressed adolescents impacted by cancer. The study also demonstrates the viability and utility of the therapeutic approaches (acceptance and commitment therapy, self‐compassion) used, showing that they have psychosocial benefits for this population.


| INTRODUC TI ON
For adolescents, a personal or familial cancer diagnosis can cause significant psychosocial disruption at an already dynamic developmental life stage. Just over 40% of young people diagnosed with cancer experience clinical levels of distress (Patterson, D'Agostino, et al., 2021); anxiety and post-traumatic stress symptoms, feelings of difference from peers, and difficulties integrating their cancer experience with their personal identity are also common (Baird et al., 2019;Kim et al., 2016). Those whose parent or sibling has, or has died from, cancer can also experience elevated levels of distress and unmet needs (Long et al., 2018;Patterson, McDonald, Ciarrochi, et al., 2017;Varathakeyan et al., 2017;Walczak et al., 2017).
Common to the experiences of adolescent patients/survivors, offspring and siblings is the need for help adjusting to their cancer situation, and for support from other young people who are similarly affected (Kaluarachchi et al., 2020;McDonald et al., 2014;Walczak et al., 2017).
Whilst some previous evaluation research has evidenced the potential for such programs to improve psychosocial well-being (e.g. Antonetti et al., 2020;Packman et al., 2005;Patterson, McDonald, Kelly-Dalgety, Luo, & Allison, 2021), few studies have examined the mechanisms underpinning their therapeutic benefits. This paper presents a mediation analysis to build on a previous process evaluation of the acceptance and commitment therapy (ACT)-based PEER program (Patterson, McDonald, Kelly-Dalgety, Luo, & Allison, 2021), to determine whether improvements in quality of life observed in highly distressed participants are associated with changes in ACT process variables.

| Acceptance and commitment therapy
Like other third-wave cognitive-behavioural therapeutic approaches, the focus of ACT lies in how individuals relate to their thoughts and emotions, rather than their specific contents (Marshall & Brockman, 2016). The approach aims to improve psychological flexibility, which describes an individual's ability to allow themselves to experience their emotions, rather than attempting to challenge, change or avoid unwanted thoughts and feelings (Hayes et al., 2006;Hulbert-Williams et al., 2015). This is developed through six key processes (the hexaflex model): acceptance of thoughts and emotions, self as context (distinguishing mental states from the self), cognitive defusion (weakening the influence of mental states on behaviour), mindfulness, defining values and committing to values-consistent actions (Hayes et al., 2006;Hulbert-Williams et al., 2015). These processes have been linked to improved psychosocial well-being in adolescents (Halliburton & Cooper, 2015), including those with chronic health conditions (Ernst & Mellon, 2016) such as cancer , and offspring impacted by parental cancer (Patterson, McDonald, White, et al., 2017). Indeed, ACT has been identified as particularly promising for canceraffected populations (Clarke et al., 2020;González-Fernández & Fernández-Rodríguez, 2019;Ing et al., 2019), where difficult emotions and cognitions may be reasonable responses to ongoing health risks and challenges (e.g. fear of cancer recurrence) and where learning to live with these internal processes may better facilitate adjustment to the cancer situation (Hulbert-Williams et al., 2015).
ACT-based programs have been demonstrated to influence the underlying therapeutic processes, such as psychological flexibility and experiential avoidance (avoiding difficult thoughts, feelings and internal experiences-a key process of psychological inflexibility) (Feros et al., 2013;Halliburton & Cooper, 2015;Patterson, McDonald, White, et al., 2017), and thereby elicit improvements in psychosocial well-being (Feros et al., 2013;Patterson, McDonald, White, et al., 2017). Of particular relevance are studies demonstrating the process of change for individuals impacted by cancer participating in ACT-based interventions: for example, a nine-week ACT intervention was found to produce lasting improvements in What is known about this topic?
• Acceptance and Commitment Therapy (ACT) and selfcompassion show promise in facilitating adolescents' adjustment to their own or their parent/sibling's cancer.
• PEER is an ACT-and self-compassion-based program developed for adolescents impacted by cancer.
• Highly distressed participants have been shown to experience greater improvements in quality of life after participating in PEER.

What this paper adds
• Highly distressed PEER participants experienced improved quality of life from baseline to two-month follow-up, mediated by changes in therapeutic processes (psychological flexibility, self-compassion).
• The benefits of the PEER program are linked to the specific therapeutic approaches used, rather than peer connection or recreation.
• ACT and self-compassion have demonstrated psychosocial benefits for distressed adolescents impacted by cancer.
quality of life and emotional well-being amongst cancer patients, mediated by increased psychological flexibility (Feros et al., 2013), whilst another ACT program demonstrated decreases in depression associated with reduced experiential avoidance amongst young people impacted by parental cancer (Patterson, McDonald, Ciarrochi, et al., 2017;. Likewise, González-Fernández and Fernández-Rodríguez's (2019) review of 19 studies examining the effects of ACT in the cancer context indicated psychosocial benefits of this therapeutic approach, although they note the limited ability to identify the specific mechanisms underlying these improvements as a limitation of this literature. This is particularly the case where ACT processes may not be the only variables impacted by an intervention (e.g. group-based interventions may also improve well-being through peer support), or where different therapeutic approaches are used in combination.

| Mindfulness
Whilst mindfulness is conceptualised as a core component of ACT (Hayes et al., 2006;Hulbert-Williams et al., 2015), it can also be used as an independent therapeutic approach. Its focus on deliberate, non-judgemental presence at the moment (Kabat-Zinn, 2003) may be useful in helping young people impacted by cancer to adapt to their situation and cope with their thoughts and emotions . Preliminary evidence suggests that mindfulnessbased interventions are feasible, acceptable and improve psychological well-being amongst AYA patients/survivors (Nissim et al., 2020;Van der Gucht et al., 2017). Therefore, in evaluating the impacts of ACT-based interventions, it may be informative to compare the contribution of mindfulness to those of other mechanisms unique to ACT (e.g. cognitive flexibility).

| Self-compassion
Combining ACT with other complementary approaches may also enhance therapeutic benefits for participants (Marshall & Brockman, 2016;Patterson, McDonald, Kelly-Dalgety, Luo, & Allison, 2021). Self-compassion may facilitate adjustment to cancer by encouraging identification and non-judgemental response to personal struggles (mindfulness), finding commonalities in challenges (common humanity) and being understanding to oneself during difficult times (self-kindness) (Neff, 2003a). Often used in combination with mindfulness, interventions informed by selfcompassion have been shown to be feasible and effectively improve psychosocial well-being for healthy adolescents (Bluth et al., 2016;Bluth & Eisenlohr-Moul, 2017;Rodgers et al., 2018) and young adults diagnosed with cancer (Campo et al., 2017;Lathren et al., 2018)-although the approaches appear to have distinct impacts on well-being (Marshall & Brockman, 2016), and may, therefore, offer greater therapeutic benefits when used in combination.

| PEER: A place for enablement, empowerment and relationships
PEER is a four-day therapeutic and recreational program for adolescents (12-17 years) impacted by their own or a relative's cancer, which draws from ACT and self-compassion in its therapeutic approach. The program involves seven psychosocial sessions and two-month follow-up, with moderation analyses indicating that these psychosocial benefits were greatest for participants with higher distress and lower psychological flexibility, mindfulness and self-kindness at baseline (Patterson, McDonald, Kelly-Dalgety, Luo, & Allison, 2021). That is, the program appears to be most effective for those in greatest need of coping strategies to help adapt to their cancer experience. However, the extent to which these improvements are linked to the specific therapeutic processes targeted by the program (as opposed to the general benefits of peer support and connection) remains unclear.

| Present study
This paper reports on the processes of change underlying improvements in quality of life (QOL) observed amongst PEER attendees who presented with high or very high levels of distress at the beginning of the program, as this group was previously noted to derive the greatest benefit from program participation (Patterson, McDonald, Kelly-Dalgety, Luo, & Allison, 2021). To better understand the mechanisms underlying these improvements, we use mediation analyses to evaluate whether changes in QOL are attributable to changes in the therapeutic processes targeted by PEER (psychological flexibility, self-compassion, mindfulness), over and above changes in peer support associated with purely recreational programs. Since this was a highly distressed sample, we also examined whether improvements in quality of life are secondary to improvements in distress, or whether they occur independently of any change in distress (which may be influenced by external factors such as the progress of cancer). We did not treat distress as a primary outcome in this paper, since the selection of a highly distressed sample could artificially increase any obtained improvements in this outcome due to regression to the mean (Barnett et al., 2004).

| Participants
As previously reported, the 2017 evaluation of PEER involved 148 adolescent participants who attended the program and completed the evaluation at two or more time points (Patterson, McDonald, Kelly-Dalgety, Luo, & Allison, 2021). This paper reports on 52 program participants who completed measures at all three time points and had high to very high levels of distress (Australian Bureau of Statistics, 2012) 1 before beginning the program. These participants were aged between 12 and 17 years, predominantly female (79%), and most had a sibling/parent with cancer (63%). Further demographic details are presented in Table 1. Compared to participants experiencing low to medium distress before PEER, those included in these analyses were more likely to be female (p = 0.039), the person with cancer was more likely to still be in active treatment (p = 0.033), and on average that person had been diagnosed more recently (p = 0.018). There was no difference in participant age, cancer experience category (patient/survivor, sibling/offspring or bereaved), ethnic background, employment status or length of engagement with Canteen 2 between the high/very high participants included in this analysis and those with low/medium distress.

| Measures
2.3.1 | Primary outcome: Quality of life (QOL) QOL was measured using a single item asking "How happy are you with your life as a whole?". Participants responded using an 11-point Likert scale (very sad/unsatisfied to very happy/satisfied), with scores ranging from 0 to 10). This item is derived from the Personal Wellbeing Index-School Children (Cummins & Lau, 2005) with which it is strongly correlated, demonstrating convergent validity (Tomyn et al., 2013); more generally, single-item measures have been successfully used to assess QOL, including in the cancer context (Bush et al., 2010;de Boer et al., 2004;Wasson, 2019), with 11-point scales appearing to have the highest reliability and validity (Kroh, 2006; Organisation for Economic Co-operation and Development, 2013).   (Hafekost et al., 2016) and service evaluations (Bassilios et al., 2017), and has shown adequate reliability and good predictive validity for depression in Australian adolescents.

| Peer support
The 11

| Psychological flexibility
The 8

| Self-kindness
The five-item self-kindness subscale of the Self-Compassion Scale (Neff, 2003c) was used to measure this construct. Participants were asked to rate how often they engage in different behaviours (e.g. "I'm kind to myself when I'm experiencing suffering") on a 5-point Likert scale (almost never to almost always). Total scores ranged from 5 to 25.
The full scale has demonstrated reliability, validity and theoretical coherence, including in adolescents (Cunha et al., 2016;Neff, 2016), and the self-kindness scale specifically has adequate reliability, and good convergent and divergent validity (Cunha et al., 2016).

| Mindfulness
The 10-item version of the Child and Adolescent Mindfulness Measure (CAMM; Greco et al., 2011) was used to assess mindfulness.
Participants rated how true each item (e.g. "I push away thoughts I don't like") was for them, using a 5-point Likert scale (never true to always true). Total scores range from 0 to 40, with higher scores indicating lower mindfulness. The CAMM has demonstrated reliability and confirmed single-factor structure with adolescents (Kuby et al., 2015).

| Other measures
Participants and facilitators completed measures of program fidelity (training quality, adherence and engagement) and satisfaction, as previously reported (Patterson, McDonald, Kelly-Dalgety, Luo, & Allison, 2021), as well as the Brief COPE measure of coping strategies (Carver, 1997) and self-judgement subscale of the selfcompassion measure (Neff, 2003a). These measures were not used in these analyses.

| Data analysis
Data analysis involved two separate procedures. First, we examined the change in distressed participants' quality of life over time using a mixed linear model, to confirm that the PEER program was associated with significant improvements in this outcome. Then, we examined possible mechanisms underlying this improvement using mediation analyses. All analyses were conducted in IBM's Statistical Package for the Social Sciences (SPSS). As missing data was minimal (maximum 3% for each item), values were not imputed for incomplete items.

| Mediation analysis
The mediation analyses were conducted using the MEMORE (MEdiation and MOderation for REpeated measures) macro for SPSS, which allows testing of mediation effects in repeated measures designs with two occasions of testing (Montoya & Hayes, 2017). We selected the pre (T0) and 2-month follow-up (T2) time points for this analysis. Changes in the outcome (QOL) and mediators (i.e., process variables: distress, peer support, psychological flexibility, mindfulness and self-kindness) between T0 and T2 are assumed to be attributable to program participation (Duarte & Pinto-Gouveia, 2017;Montoya & Hayes, 2017). Mediation effects were tested separately, given the limited sample size and the correlation between process variables (Preacher & Hayes, 2008).

| Analysis of changes in QOL for distressed participants
Results of the mixed linear models exploring changes in QOL for distressed participants are presented in Tables 2 and 3. As can be seen in  Table 3, there was a significant linear effect of time on QOL in the final model, suggesting that for distressed participants, QOL increases following program participation and continues to improve across the two-month follow-up.
No other effects reached significance in the model. Figure 1 presents the path diagram of the mediation analysed, for each of the five tested mediators; psychological flexibility and self-kindness-are linked to improvements in QOL following participation. Reduction in distress also mediated this change, whilst mindfulness and peer support were not statistically significant mediators. This builds on findings from the previous process evaluation of PEER which identified that the program resulted in clinically significant improvements in QOL and distress for some participants, with those who reported high and very high levels of baseline distress appearing to benefit more from the program (as did those with lower levels of process variables: psychological flexibility, mindfulness and self-kindness) (Patterson, McDonald, Kelly-Dalgety, Luo, & Allison, 2021). Together, these findings suggest that participation in the PEER program has psychosocial benefits for young people in greatest need of support in coping with their cancer experience (regardless of what this experience is), and that for this group, these benefits are attributable to the specific therapeutic teachings of the program (ACT and self-compassion) rather than non-specific effects of peer connection or recreation.

| Mediation analysis
The evidenced mechanisms of the effect of the PEER program provide further evidence to support the use of ACT and self-compassion as therapeutic approaches when working with adolescents impacted by their own or a family member's cancer, particularly those experiencing higher levels of distress.

| CONCLUSIONS
The findings of this study not only confirm the benefits and underlying therapeutic processes of the PEER program; they also demonstrate the viability and impact of ACT and self-compassion for adolescents who are experiencing high levels of distress as a result of their own or a family member's cancer. Improvements in QOL amongst distressed adolescents were specifically mediated by increases in psychological flexibility and self-kindness, suggesting that therapeutic interventions can improve these processes in this population, with resulting psychosocial benefits. The finding that mindfulness and peer support did not mediate improvements in quality of life suggests that incorporating ACT and self-compassion as part of therapeutic intervention may be more effective than programs based solely on peer support and/or mindfulness. Whilst further research comparing different interventions would be needed to confirm this, healthcare providers and community organisations may want to prioritise programs and services that incorporate ACT and self-compassion when considering the most appropriate referrals to make, or support services to offer, distressed young people impacted by cancer.

FU N D I N G I N FO R M ATI O N
Canteen is a registered charity supported by government funding and donations from individuals, corporations, trusts and foundations. The evaluation of the PEER program received no specific funding.

ACK N OWLED G EM ENTS
The authors would like to acknowledge Dr Richard Tindle for his contribution to the statistical analysis, the Canteen staff who Librarians.

CO N FLI C T O F I NTE R E S T
All authors on this paper are affiliated with Canteen, which owns the intellectual property rights to the PEER program; there are no other conflicting interests associated with this project.

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared due to privacy and ethical restrictions.

E N D N OTE S
1 The moderation analysis in the previous process evaluation indicated that participants with a score of 23.9 or higher on the Kessler-10 measure of psychological distress experienced significant improvements in QOL after program participation. Here a cut-off of 22 was used; this is more clinically meaningful, as it corresponds to the cut-off for high or very high levels of distress (Australian Bureau of Statistics, 2012).
2 Canteen is an Australian not-for-profit organisation which supports AYAs impacted by their own or a family member's cancer (including informational, psychosocial, recreational and peer support options). The organisation developed and runs the PEER program.
3 Demographic information collected in the T0 survey was supplemented by additional data from Canteen client records, with participant assent. 4 The Kessler 10 measure of psychological distress was not included in the post-program survey as it assesses distress over a 30-day period, and is, therefore, inappropriate for detecting changes in distress over a four-day program.