Abstract
People thrive in compassionate environments. For autistic people, their environments are often non-conducive to care and support. We propose that cultivating autistic people’s self-compassion may be an antidote to the mental health difficulties arising from environmental causes. The current study aimed to pilot a self-guided online self-compassion program for autistic adults to assess its feasibility, acceptability, and outcomes related to completing this program. The program was co-produced with autistic individuals. Thirty-nine autistic adults aged 20–77 years (Mage = 45.28, SDage = 11.92) completed the online program. They also provided feedback about their self-compassion practices every week. They completed pre- and post-intervention online surveys measuring self-compassion, symptoms of anxiety and depression, positive and negative affect, emotion regulation difficulties, and psychological well-being. As hypothesized, there were significant group improvements in all outcome measures. Preliminary results suggest that the program is feasible and accepted by participants. Approximately half of the autistic participants reported experiences of backdraft, which are adverse reactions associated with self-compassion practices. Based on our findings, we provided clinical recommendations on how self-compassion can be taught to autistic adults.
Lay abstract
Self-compassion is when we are aware of our feelings and thoughts, are friendly toward ourselves, and realize everyone feels pain and makes mistakes. Self-compassion is associated with having better mental health and well-being in autistic and non-autistic people. But we do not know if autistic people’s self-compassion can be improved through psychoeducation and self-compassion practices. We co-produced an online self-guided self-compassion program based on evidence-based self-compassion practices for autistic adults called the Self-compassion Program for Autistic Adults. This program included live-experiences videos of autistic adults reflecting on their self-compassion and self-critical experiences. This study piloted the program with 39 autistic adults. We wanted to see if these autistic adults’ self-compassion, emotion regulation, mental health, and psychological well-being improved after completing this program. We found that the autistic participants’ self-compassion, emotion regulation, mental health, and psychological well-being improved significantly after completing the program over 5 weeks. We also found that just over half of the participants reported experiencing negative reactions associated with self-compassion practices. We suggested some clinical implications, including a recommendation for emotion regulation interventions to incorporate self-compassion to help promote access to the affiliative system. In addition, autistic adults who are psychologically vulnerable may need to work with mental health professionals while developing self-compassion to help manage the possible negative reactions associated with some self-compassion practices.
Keywords
People thrive in compassionate environments. Compassion encompasses cognitive, emotional, and behavioral aspects, such as recognizing suffering, understanding its universality, feelings of sympathy, empathy, or concern for the sufferer, tolerating the distress associated with this suffering, and being motivated to act or acting to reduce the suffering (Strauss et al., 2016). Compassion can be directed from self to others, others to self, and self to self (Gilbert et al., 2017). Research findings indicate that supportive environments where people receive compassion from themselves, and others lead to better life outcomes across contexts, including work, tertiary education, and parenting (Chan et al., 2022; Conti, 2015; Dutton et al., 2006; Jefferson et al., 2020; Mayer & Oosthuizen, 2020; Shuck et al., 2019). Sadly, many autistic children and adults have experienced environments, such as schools, work places, and communities, that are not compassionate.
Self-compassion and mental health in autistic and non-autistic people
The rates of mental illnesses, including anxiety, depression, suicidality, and suicidal ideation, are high in autistic individuals (Hedley & Uljarević, 2018; Lai et al., 2019). Mandy (2022) offered several suggestions to address the current “mental health crisis,” including acknowledging the environment’s role in causing mental health issues in autistic people. This is consistent with Pellicano and colleagues’ (2022) capabilities approach to understanding and supporting autistic adults. At the core of this approach is the understanding that a person’s capability is not an intrinsic ability but rather an opportunity to be or do something constrained by both the individual and broader contexts in which the individual lives. For autistic people to thrive, their environments must be conducive to care and support. Autistic people are often stigmatized and victimized (Grossman, 2015; Morrison et al., 2019; Sasson et al., 2017; Swaim & Morgan, 2001; Trundle et al., 2023). Although a variety of interventions and approaches to reduce the societal stigma of autism are being implemented, including “autism-friendly” spaces, positive media representation, and educational and psychosocial training for the public and professionals (Turnock et al., 2022), these implementations tend to be confined within small communities. Broader societal changes are unlikely to occur promptly. Therefore, one action that may act as a buffer against the negative effects of stigma and discrimination and related mental health consequences may be to help autistic people improve their self-acceptance and develop positive views about themselves. We propose that cultivating self-compassion is one way to do this.
Self-compassion is a kind and gentle way of relating to oneself in daily life, especially during challenging moments (e.g. when we fail at something or make mistakes). Conceptually, it is the opposite of self-criticism (Warren et al., 2006). Neff (2003) defined self-compassion as having three components: mindfulness, self-kindness, and common humanity. Mindfulness entails acknowledging one’s painful experiences in a balanced way that neither ignores nor fixates on disliked aspects of oneself or life. Mindful awareness of personal suffering is necessary to extend compassion to oneself. However, it is also crucial to stay grounded and prevent getting carried away by the narrative driving the suffering. Self-kindness is characterized by showing care and understanding to oneself instead of being overly critical. When one recognizes a negative aspect of their personality, they approach themselves with a kind and supportive tone instead of harsh and critical ones. This means that even when identifying unproductive behaviors that need change, the self is still offered warmth and unconditional acceptance. In times of stress, a kind response involves offering comfort and solace before attempting to control or solve the problem. Common humanity involves the acknowledgment that all humans are imperfect and face challenges. Self-compassion promotes a broader perspective toward oneself by connecting one’s flaws to the shared human condition. People often feel that failures and disappointments are unique, leading to isolation and the false belief that others lead better or more “normal” lives; common humanity counteracts this tendency.
Self-compassion is related to the construct of mindfulness, defined as “paying attention in a particular way: on purpose, in the present moment, and non-judgmentally” (Kabat-Zinn, 2003, p. 145). Mindfulness is the first step of self-compassion since we need to be aware of our distress and suffering in an accepting way. However, unlike mindfulness, self-compassion extends it by adding the components of responding to moments of suffering and perceived failure with loving kindness toward the self and acknowledging that all share these feelings of suffering. Due to the focused nature of self-compassion in moments of distress and suffering, studies have found self-compassion to be a more potent predictor of well-being than mindfulness in clinical contexts, particularly among individuals seeking assistance for anxiety (Van Dam et al., 2011).
Having compassion for oneself is associated with good mental health and a range of other positive outcomes in clinical and non-clinical populations, including greater levels of happiness, well-being, resilience, better sleep, and lower levels of stress, anxiety, depression, and self-harm (Brown et al., 2021; Cleare et al., 2019; MacBeth & Gumley, 2012; McArthur et al., 2017; Shattell & Johnson, 2018; Zessin et al., 2015). It has been previously suggested that self-compassion may play an important role in the mental health of autistic people (Cai & Brown, 2021); indeed, recent empirical research using self-reported surveys found that autistic adults tended to have lower self-compassion than non-autistic adults, and higher self-compassion levels are associated with lower symptoms of anxiety and depression in autistic adults (Cai, Gibbs, et al., 2023; Galvin & Richards, 2023). Cai, Gibbs, and colleagues (2023) conducted interviews with autistic participants and found that overall, participants understood the value of self-compassion and acknowledged that although it was challenging to be self-compassionate, the ability to do so can build over time through practice. For autistic women, being diagnosed in adulthood allowed them to be kinder and less critical toward themselves, enabling them to engage in self-care (Wilson et al., 2023).
Researchers and clinicians have suggested that the mechanism by which self-compassion improves mental health is through better emotion regulation. According to Berking and Whitley (2014), responding to stressful events with self-criticism triggers negative emotions and activates the amygdala, which increases the body’s stress response. This leads to a vicious cycle of even more self-criticism and difficulty regulating emotions. By contrast, responding with self-compassion does not activate the neural networks associated with threat, resulting in less negative emotions and arousal. This makes it easier to regulate emotions with adaptive strategies like cognitive reappraisal. This theory also aligns with Gilbert’s (2014) philosophy underlying Compassion-Focused Therapy, which argues that compassion is crucial for emotion regulation and supporting the affiliative system. He suggests that negative emotions, such as shame, anger, and anxiety, often stem from self-criticism and self-judgment, which can activate the threat system in the brain and create a vicious cycle of negative emotions. Compassion, on the contrary, can activate the soothing system, which can help regulate negative emotions and create positive feelings of warmth and safety. Gilbert also believes that compassion can support the affiliative system, which is responsible for social bonding and connecting with others. By cultivating compassion toward oneself (and others), individuals can improve their relationships and increase their sense of belonging and connection.
There is emerging evidence for this hypothesis of an interrelationship between self-compassion, emotion regulation, and mental health in non-autistic samples (Diedrich et al., 2017; Finlay-Jones et al., 2015). A systematic review by Inwood and Ferrari (2018) found that emotion regulation mediated the relationship between self-compassion and various mental health outcomes, including stress, depression, and posttraumatic stress disorder. More recently, this pattern of emotion regulation mediating the relationship of self-compassion with anxiety and depression has also been found in autistic adults (Cai, Love, et al., 2023). Considering the known challenges with emotion regulation among autistic individuals, the exploration of the potential emotion regulation benefits of enhancing self-compassion becomes a relevant avenue of inquiry.
Proposed consequences of self-compassion practice for autistic adults
Accumulating evidence suggests that self-compassion is a skill that can be enhanced through practice in non-autistic clinical and general populations. A recent meta-analysis of randomized controlled trials (RCT) found that self-compassion programs significantly improved 11 psychosocial outcomes, including self-compassion, anxiety, and depression (Ferrari et al., 2019), with programs ranging from single-session workshops to multiweek sessions with trained therapists. Emerging evidence supports the acceptability and effectiveness of online self-compassion programs for various non-autistic populations (e.g. Finlay-Jones et al., 2017). No research has investigated the feasibility or outcomes of self-compassion programs for autistic individuals. One study has examined changes in self-compassion levels in autistic adults after completing a mindfulness intervention. Lunsky and colleagues (2022) examined the feasibility of a 6-week online program based on the mindfulness-based stress reduction (MBSR) curriculum. They measured numerous outcomes through self-reported measures, including self-compassion, and found that self-compassion levels significantly increased at post-intervention and 12-week follow-up. This MBSR curriculum included various meditations, breathing techniques, a body scan, and mindful movement. One of the meditations in the MBSR curriculum is a loving-kindness meditation, which is very similar to the loving-kindness meditation used in self-compassion programs such as the Mindful Self-Compassion (MSC) program. Although the online program used by Lunsky and colleagues (2022) only used one self-compassion practice, their self-compassion outcome findings are promising. Interventions focusing specifically on self-compassion will likely benefit autistic people even more in improving self-compassion levels and well-being.
Online self-compassion programs seem to be suitable for autistic adults, with preliminary evidence supporting virtual mindfulness programs (Lunsky et al., 2022; Redquest et al., 2022). Gaigg and colleagues (2020) conducted a waitlist-controlled trial of online self-guided cognitive behavioral therapy (CBT) and mindfulness-based therapy (MBT). They found that autistic adults in active conditions significantly decreased anxiety, which was maintained over 3 months. They concluded that online self-help CBT and MBT tools might provide a cost-effective way of delivering mental health support. We anticipate that online self-compassion programs may be another effective way of delivering mental health support to autistic adults.
Although self-compassion is associated with many long-term health benefits, some people may initially have adverse reactions to self-compassion practices. This distress, called backdraft, occurs when self-compassion practices activate old memories and associated cognitions and emotions. Backdraft can present itself as (a) negative and critical thoughts, (b) unpleasant emotions such as shame, grief, or anxiety, (c) body sensations like aches and pains, and (d) behaviors such as withdrawal or aggression (Neff & Germer, 2022). Neff and Germer (2022), founders of the MSC program (Neff & Germer, 2013), state that backdraft is “an intrinsic part of the transformation process of self-compassion.” Therefore, experiencing backdraft may be an inevitable part of healing from past pains and trauma. Nevertheless, due to the possibility of having adverse reactions in psychologically vulnerable people, Cai and Brown (2021) have previously suggested that the development of self-compassion in autistic adults should be approached cautiously and informed by empirical evidence. Mistretta and Davis (2022) pointed out in their meta-analysis of self-compassion interventions for adults with chronic illness that although research has mentioned potential “backdraft” effects of self-compassion practices, none of the interventions actively monitored adverse effects. We were also unable to find any research on backdraft in the context of self-compassion interventions (Cai, in press). Hence, we conclude that there is currently no empirical research on the experiences of backdraft, even though there is ample evidence for the benefits of self-compassion programs (Ferrari et al., 2019).
The current study
This study examined an online self-compassion training program to improve the emotion regulation and mental health of autistic adults called the Aspect Self-compassion Program for Autistic Adults (ASPAA). This self-guided program contains psychoeducation, self-compassion practices, and self-reflective exercises based on evidence-based self-compassion-based programs, including the MSC program (Neff & Germer, 2013) and Compassion-Focused Therapy (Gilbert, 2014). Participants were asked to complete a module at the start of each week (for week 1, a brief introductory module was completed in addition to Module 1) and encouraged to try the self-compassion practice(s) from each module daily for the rest of the week. An overview of the program modules, exercises, and practices is shown in Table 1.
ASPAA overview.
The ASPAA content was co-produced by non-autistic and autistic researchers (R.C. and C.E.) and incorporated videos of autistic adults reflecting on their self-compassion and self-critical experiences. At the start of program design, R.C. (MSC trained) consulted an experienced MSC teacher trainer who had worked with autistic clients to identify the self-compassion exercises and practices that may be appropriate for autistic adults. Specific suggestions were provided—for example, the MSC teacher trainer thought it necessary to offer participants suggestions for other-touch for the soothing touch practice if self-touch is uncomfortable (e.g. touching a cat or dog). The MSC teacher trainer also thought affectionate breathing would not be suitable, as breathing practices may be uncomfortable or distressing for those with high anxiety. R.C. then identified an initial list of self-compassion exercises and practices for C.E (autistic researcher) to try. This initial list was modified based on the feedback from C.E. The iterative co-production process also involved an advisory group of three autistic adults. After the first draft of the module content was developed, including the psychoeducation content about self-compassion, the initial exercises and practices were emailed to the three advisors for them to try. One-on-one interviews were conducted with our autistic advisory group members to gather detailed feedback. Then, the content of the six modules was updated based on the advisors’ feedback, reproduced in PowerPoint, and migrated to Articulate 360. The advisors then reviewed all modules, provided individual comments for module pages in Review 360, and discussed their feedback during a 2-h focus group. The co-producers of the program decided to give participants 1 week to complete each new module and try the daily practices. Based on this feedback from the autistic advisors, the ASPAA content and structure were refined before starting the current research study.
The current study aimed to pilot a new online self-guided self-compassion program for autistic adults co-produced by autistic and non-autistic researchers. The lived experience findings of the participants who completed ASPAA are presented in a separate qualitative study (Edwards et al., 2024). The following research questions guided the current study:
How feasible and acceptable was ASPAA, according to autistic adults participating in the program?
Would completion of ASPAA improve the self-compassion levels of autistic adults from pre- to post-program completion? It was hypothesized that there would be significant improvements in self-compassion levels.
Given that self-compassion programs have been shown to improve various psychosocial outcomes in non-autistic samples (Ferrari et al., 2019), would completing ASPAA improve the emotion regulation, mental health, and well-being outcomes of autistic adults? Specifically, increased positive affect, decreased negative affect, reduced symptoms of anxiety and depression and difficulties in emotion regulation, and improved psychological well-being. It was hypothesized that there would be significant increases in positive affect and psychological well-being, and significant decreases in negative affect, symptoms of anxiety and depression, and emotion regulation difficulties.
As there is currently no research on backdraft, we were curious about how many autistic participants reported experiencing backdraft as a result of practicing self-compassion and what types of backdraft did they experience?
Methods
Procedure
Ethics approval was obtained from the University of Sydney Human Research Ethics Committee (Application ID: 2022/598), and all participants gave informed consent before participating. We embraced a participatory approach throughout the entire project to co-design and implement the research study. Participants were recruited through the social media and mailing lists of Autism Spectrum Australia (Aspect), autism support networks, and word of mouth. Recruitment began in September 2022 and ended in December 2022. Autistic adults interested in completing the ASPAA were asked to complete a suitability survey to assess their eligibility.
All suitable adults were invited to complete a pre-program survey, which included consent to participate in the study. After completing this survey, autistic participants received a weekly schedule detailing specific dates for completing the modules of the ASPAA. The introductory and first modules were to be completed at the start of week 1, with the remaining modules spread over the subsequent 4 weeks. Given that the program is self-guided, the time taken to complete each module varied widely among participants, based on individual pace and the time dedicated to exercises/practices. Consequently, no specific duration for module completion was prescribed. Participants did not receive workbooks; instead, they were encouraged to take personal notes while reflecting on the exercises and practices. Weekly support was provided in the form of brief one-on-one catchups with either R.C. or C.E., conducted via email, phone, or Zoom. These sessions were scheduled with each participant after they completed the pre-program survey and aimed to gather feedback on each module and document any backdraft experiences. Upon completing the 5-week program, participants were invited to complete a post-program survey within 1 week.
Participants
To be eligible for the study, participants needed to (a) have a diagnosis of autism (confirmed through diagnostic reports or letters from healthcare professionals) or self-identify as autistic, (b) for those who self-identified as autistic, meet the cut-off for autism spectrum disorder through the Autism Spectrum Quotient–Short (AQ-Short; Hoekstra et al., 2011), (c) have no intellectual disability, (d) be aged 18 years or above, (e) scored low to moderate on the Self-Compassion Scale (SCS; Neff, 2003; range: 1 to 3.5), (f) live in Australia, and (g) have access to the required technology (good internet connection and access to computer or laptop with audio outlet). To minimize harm to psychologically vulnerable participants who may experience severe backdraft while practicing self-compassion exercises, exclusion criteria included the presence of a current or lifetime psychotic disorder, current suicide ideation or suicidal behaviors, and recent initiating or stopping psychiatric meditation or psychotherapy. Additional safety processes were put in place to minimize harm: V.G. is a clinically trained psychologist and was on hand to provide additional support to participants when needed, including having a conversation with the participants and making external referrals to mental health professionals. In addition, a list of organizations that can provide mental health support was also provided to all participants.
To determine the program’s suitability to potential participants, people interested in the study were first directed to complete a survey containing questions relating to the eligibility criteria. A total of 104 people completed the suitability survey: 60 people were suitable for the study and provided their contact information; 7 people did not respond to follow-up contact from the researchers, which resulted in 53 autistic adults enrolling in the program; 13 people did not start the program due to life circumstances (n = 7; car accident, moved house, work commitment, death in family, holiday), technical issues (n = 2), and not responding to researcher contact (n = 3); and 1 participant withdrew without providing any reason. This resulted in 40 participants starting the program. One participant completed most of the program but did not finish it due to work commitments.
The final sample consisted of 39 autistic adults aged 20–77 (Mage = 45.28, SDage = 11.92), including 28 (72%) who identified as a woman, 5 (13%) as a man, and 6 (6%) as non-binary or agender (see Table 2 for demographic information). Most participants had a formal autism diagnosis (n = 30), and the remainder self-identified as autistic. We included self-identified people to account for the often-significant delays in gaining a clinical diagnosis (Lai & Baron-Cohen, 2015) or active choice of not getting a formal diagnosis due to poor affordability or accessibility of assessments. There was no significant difference in the AQ-Short scores between participants who reported a formal diagnosis (M = 90.37; SD = 9.78; range: 67–110) and those who did not (M = 95.78; SD = 8.49; range: 83–111), t(37) = −1.50, p = 0.511.
Participant demographic information (N = 39).
ADHD: attention-deficit/hyperactivity disorder; OCD: obsessive compulsive disorder.
Confirmation of formal autism diagnosis was sighted by R.C.
Measures
The study used three online surveys hosted in the university’s Qualtrics system: suitability, pre-program, and post-program surveys. The suitability survey consisted of questions that covered the study’s eligibility criteria, including measures of autistic traits (AQ-Short) and self-compassion levels (SCS). The pre-program and post-program surveys contained the same questionnaire measures of positive and negative affect, emotion regulation abilities, symptoms of anxiety and depression, and psychological well-being. All of these measures have been used in autistic samples for research. The pre-program survey had additional questions capturing participant demographics. The post-program survey contained the additional measure of SCS and questions measuring the program’s acceptability.
Acceptability and feasibility of the program
Several items designed specifically for this study were used to measure the program’s acceptability in the post-program survey. Participants were asked to rate the overall program (Poor to Excellent on a 5-point Likert-type scale), whether they would recommend the program to other autistic adults (Yes or No), whether self-compassion would help the participant (Yes or No), whether they will use the self-compassion skills learned during the program in life moving forward (Yes or No), and whether the training program would be helpful for other autistic people (Yes or No). The feasibility of the program was determined through the ease of recruitment and attrition levels.
Autistic traits
Autistic traits were assessed using the 28-item AQ-Short (Hoekstra et al., 2011), a condensed version of the original 50-item questionnaire developed by Baron-Cohen et al. (2001). Participants rated each item on a 4-point Likert-type scale, ranging from Definitely agree (score of 1) to Definitely disagree (score of 4). Examples of items include statements like “I find social situations easy” and “I am fascinated by numbers.” The correlation between the AQ-Short and the full 50-item AQ is very strong, with correlation coefficients (r) ranging from 0.93 to 0.95. The sensitivity (0.97) and specificity (0.82) estimates for autism using an AQ-Short cut-off score of 65 are also comparable to the original questionnaire.
Self-compassion
Participants’ self-compassion levels were assessed using SCS (Neff, 2003). The scale consists of 26 items that measure 6 sub-domains of self-compassion: self-kindness, self-judgment (reverse-scored), common humanity, isolation (reverse-scored), mindfulness, and over-identification (reverse-scored). Each item is rated on a 5-point Likert-type scale, ranging from Almost never (score of 1) to Almost always (score of 5). An example item is, “I try to be loving towards myself when I’m feeling emotional pain.” Sub-domain and total scores are obtained by averaging the item responses. Total scores range from 1 to 5. Previous research has demonstrated the SCS’ good internal reliability (Cronbach’s α = 0.92) and test–retest reliability (Cronbach’s α = 0.93) over 3 weeks, indicating its consistency over time. More recently, a study examining the factor structure of the SCS using a diverse sample found an excellent fit for a model consisting of one general factor of self-compassion and six specific subscale factors (Neff et al., 2019). The SCS has been used previously with autistic adults (e.g. Cai, Gibbs, et al., 2023). In addition, it has been employed in studies involving conditions frequently co-occurring with autism, such as attention deficit hyperactivity disorder (Beaton et al., 2022), obsessive-compulsive disorder (Wetterneck et al., 2013), and eating disorders (Pullmer et al., 2019). Furthermore, evidence supports the measurement invariance of the scale across different populations (Tóth-Király & Neff, 2021), indicating its suitability for use in diverse samples.
Emotion regulation
Participants’ ability to regulate emotions was assessed using the Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004), consisting of 36 items. This scale comprises six subscales: non-acceptance of emotional responses, difficulty engaging in goal-directed behavior, impulse control difficulties, lack of emotional awareness, limited access to emotion regulation strategies, and lack of emotional clarity. Participants rated each item on a 5-point Likert-type scale, ranging from Almost never (score of 1) to Almost always (score of 5). A sample item is, “I care about what I am feeling.” Some items are reverse-scored. The subscale and total scores are obtained by summing the item scores, with total scores ranging from 36 to 180. The DERS exhibits high internal consistency (α = 0.93), good test–retest reliability (α = 0.88), and adequate construct and predictive validity.
Positive and negative affect
Positive and negative affect was measured using the Positive and Negative Affect Schedule (PANAS; Watson et al., 1988), which consists of 20 self-report items—10 items assessing positive affect and another 10 items assessing negative affect. An example positive affect item is, “Interest,” and an example negative affect item is, “Upset.” Participants were asked to indicate the extent to which they have felt the emotions over the past week by rating each item on a 5-point Likert-type scale, ranging from Very slightly or Not at all (score of 1) to Extremely (score of 5). Higher scores on the negative affect subscale indicate greater levels of negative affect. Similarly, higher scores on the positive affect subscale indicate higher levels of positive affect. Both subscales demonstrate high internal consistency (Cronbach’s α = 0.86–0.90) and good test–retest reliability (ranging from 0.39 to 0.71).
Anxiety symptoms
The Diagnostic and Statistical Manual of Mental Disorders-5 Generalized Anxiety Disorder Dimensional Scale (DSM-5 GAD-D; Knappe et al., 2013) was utilized to assess anxiety symptoms. This self-reported scale is norm-referenced and comprises 10 items. Participants rate each item on a 5-point Likert-type scale, ranging from Never (score of 0) to All of the time (score of 4). An example item is, “During the past month, I have avoided, or did not approach or enter, situations that made me anxious.” Total scores, ranging from 0 to 40, are obtained by summing the item scores.
Depression symptoms
To assess depression symptoms, the Patient Health Questionnaire-9 (PHQ-9) developed by Kroenke et al. (2001) was employed. The PHQ-9 is a 9-item norm-referenced scale specifically designed to screen for depression. Participants rate each item on a 4-point Likert-type scale, ranging from Not at all (score of 0) to Nearly every day (score of 3). The questions focus on experiences over the past 2 weeks and include items such as “Little interest or pleasure in doing things.” Total scores, ranging from 0 to 30, are obtained by summing the scores of all items.
Psychological well-being
The Warwick–Edinburgh Mental Well-being Scale (WEMWBS; Tennant et al., 2007) is a self-report questionnaire comprising 14 items that assess both hedonic and eudemonic aspects of well-being. Subscales include positive affect, positive psychological functioning, and positive relationships with others. Participants rated each statement on a 5-point Likert-type scale, ranging from None of the time (score of 1) to All of the time (score of 5). An example item is, “I’ve been feeling relaxed.” To calculate a total score, the item responses were summed, resulting in a score range of 14–70, with higher scores indicating a higher level of well-being. The WEMWBS has demonstrated excellent internal consistency (α = 0.91) and good test–retest reliability over 1 week (Tennant et al., 2007).
Weekly catchups
Weekly catch-up notes (email, phone, Zoom) were kept for each participant.
Data analysis
First, we assessed the program’s feasibility by reviewing the ease of recruitment and attrition levels. We also examined the program’s acceptability by examining the descriptive responses on the acceptability items described previously.
Next, to address the first hypothesis (there would be significant improvements in self-compassion levels after completing the program), we conducted a paired-samples t-test on participant scores on the SCS. To address the second hypothesis (there would be significant increases in positive affect and psychological well-being and significant decreases in negative affect, symptoms of anxiety and depression, and emotion regulation difficulties after completing the program), further paired-sample t-tests were conducted on participants’ scores on the PANAS, WEMWBS, DSM-5 GAD-D, PHQ-9, and DERS. Missing items (four data fields of two participants, two for DERS and two DSM-5 GAD-D) were addressed using linear interpolation method. Before conducting the paired-samples t-tests, tests of normality using Kolmogorov–Smirnov statistics indicated that all differences in outcome scores were normally distributed (p > 0.05). We used SPSS Statistics version 21 for Mac for all statistical analyses.
Finally, to determine the number of participants who experienced backdraft, R.C. reviewed the weekly catch-up notes and identified the number of participants who provided feedback referencing backdraft (self-critical thoughts, unpleasant emotions associated with doing the practices, or painful bodily sensations).
Community involvement
The research team comprised a researcher with lived experiences of autism (C.E.), a researcher who is a parent of an autistic child (V.G.), and a sibling of an autistic adult (A.L.), consistent with Australian-developed best-practice participatory and inclusive practice guidelines for autism researchers (den Houting et al., 2022). The non-autistic and autistic researchers (R.C. and C.E.) co-managed the entire project from design to write-up, with input from the rest of the research team during the design, analyses, and write-up stages.
Results
Feasibility and acceptability
The program’s feasibility was demonstrated through the ease of recruitment; 53 eligible participants enrolled within 3 months. Of these 53 participants, the attrition rate was 26%. Most participants who discontinued did so for personal reasons unrelated to dissatisfaction with the program (i.e. car accident, moved house, work commitment, death in family, holiday, and technical issues).
Most participants rated the overall program as either good (n = 22; 56%) or excellent (n = 13; 34%). The remaining participants rated the program average (n = 4; 10%). No participant rated the program as poor or bad. In addition, all participants who responded to the questions reported that they would recommend the program to other autistic adults, thought self-compassion would help the participant, that they would use the self-compassion skills learned during the program in life moving forward, and considered the program as potentially helpful for other autistic people (see Figure 1 for detailed findings on those responses—no participant has responded “No” to any question).
Acceptance of ASPAA.
Five participants took 6 rather than 5 weeks to complete the program for personal reasons (one was on holiday, one did not have access to a computer for a week, and three other unknown personal reasons), where they had to delay a module by 1 week.
Changes in self-compassion levels, emotion regulation, mental health, and well-being outcomes
Descriptive statistics, including reliabilities, were calculated for all measures and presented in Table 3.
Descriptive statistics of outcome variables.
AQ-Short: Autism Spectrum Quotient-Short Form; SCS: Self-Compassion Scale; PANAS: POS Positive and Negative Affect Schedule: positive affect, PANAS: NEG Positive and Negative Affect Schedule: negative affect; DERS Difficulties in Emotion Regulation Scale; DSM-5 GAD-D: Diagnostic and Statistical Manual of Mental Disorders-5 Generalized Anxiety Disorder Dimensional Scale; PHQ-9: Patient Health Questionnaire-9; WEMWBS: Warwick–Edinburgh Mental Well-being Scale.
Seven paired-sample t-tests were conducted to evaluate the impact of the intervention on all outcomes. There were statistical increases in self-compassion (Mdiff = −0.85, SDdiff = 0.58), t(38) = −9.14, p < 0.001, η2 = 0.69, positive affect (Mdiff = −3.95; SDdiff = 1.29), t(38) = −3.07, p = 0.004, η2 = 0.20, and psychological well-being (Mdiff = −5.38, SDdiff = 6.92), t(38) = −4.86, p < 0.001, η2 = 0.38, all with large effect sizes. In addition, there were statistical decreases in negative affect (Mdiff = 4.13 SDdiff = 6.02), t(38) = 4.28, p < 0.001, η2 = 0.33, anxiety symptoms (Mdiff = 3.22, SDdiff = 6.49), t(38) = 3.10, p = 0.004, η2 = 0.20, depressive symptoms (Mdiff = 2.97, SDdiff = 5.40), t(38) = 3.44, p = 0.001, η2 = 0.24, and emotion regulation difficulties (Mdiff = 14.88, SDdiff = 17.64), t(38) = 5.27, p < 0.001, η2 = 0.42, again, all with large effects. Effect sizes of improvements were strongest for self-compassion, emotion regulation, and psychological well-being.
Experiences of backdraft
Just over half of our participants reported experiences of backdraft (n = 21; 54%). The two practices most associated with backdraft were the loving-kindness meditation and feeling the soles of your feet. Ten participants reported experiencing backdraft when they first tried loving-kindness meditations, which included unpleasant emotions and self-criticism: “A bit hard to get used to and felt a little uncomfortable. I think it bought up some past negative feelings or anxiety about how I feel about myself,” “Inauthentic. Painful. My mind just kept going around and around in turmoil, thinking of all the things I need to hear, but never do,” and “It was really hard. I had really strong backdraft. My inner monologue started straight away—That’s not true. No, you do this—It was challenging.” Of note, for most of these 10 participants, their experiences either improved after the initial attempt or they anticipated the backdraft would reduce over time: “I feel more comfortable with doing it now and will try to do this every day or when I feel down or hard on myself,” “Still emotional, but there is less backdraft. Sitting with it feels somewhat peaceful now,” “Only done it a couple of times, might feel a little bit better but still needs time,” and “I think it’s worth persevering.”
Six participants reported experiencing backdraft when first trying the practice of feeling the soles of your feet, and the backdraft mainly was associated with the discomfort of feeling their feet: “Honestly, I found it quite uncomfortable . . . focusing on the weight of my body on my feet and paying close attention to how my body felt as I moved was slightly upsetting,” “I was first curious and got really angry focusing on my feet. Just wanted to stop. Don’t like feet being touched, it’s a physiological thing,” and “I have chronic pain in various spots in my body (including my feet) that I try to not focus on most of the time, so bringing awareness to a part of my body/physical sensation isn’t helpful for me.” Unlike the loving-kindness meditation, most people reported having unpleasant sensations doing the feeling the soles of your feet practice even after the initial try: “Much the same. It seems pointless and causes stress that I am not doing it right,” and “I kept at it throughout the week but my feeling on physical grounding exercises hasn’t changed. It mostly ranges from neutral to unenjoyable for me.”
Discussion
The current study aimed to pilot an autism-specific online self-compassion program, assessing its feasibility, acceptability, and the outcomes for autistic adults who completed it. The program appears feasible due to the ease of recruitment and moderately low attrition rate. A total of 90% of participants rated the program as good or excellent, and the remainder rated the program as average. All participants who responded to our question stated they would recommend the program to other autistic adults. Therefore, the program was accepted by all participants.
As expected, there were significant improvements in self-compassion levels, positive affect, psychological well-being and reductions in negative affect, symptoms of anxiety and depression, and difficulties in emotion regulation after completing the program. Our findings align with previous meta-analyses that showed various positive outcomes associated with completing self-compassion programs for non-autistic samples, including self-compassion, anxiety, and depression (Ferrari et al., 2019; Finlay-Jones et al., 2017; Mistretta & Davis, 2022). Our study extends this work, demonstrating for the first time that completing a self-guided self-compassion program over 5 weeks can improve autistic people’s emotion regulation abilities. Therefore, our findings support the theory that self-criticism activates the body’s stress response and threat system, while self-compassion helps to soothe our mind and body to support the affiliative system and improves emotion regulation (Berking & Whitley, 2014; Gilbert, 2014). Based on these emerging findings, we recommend that interventions targeting emotion regulation for autistic individuals should also incorporate content relating to self-compassion, helping to promote access to the affiliative system.
We found that just over half of the autistic adults who completed the self-compassion program experienced backdraft. Mistretta and Davis (2022) and Cai (in press) recommended that qualitative studies could be one way to provide details of backdraft. In this study, we have reviewed the feedback of participants to identify the proportion of participants who reported experiencing backdraft when talking about their self-compassion practices and the types of backdraft reported. We noted that although loving-kindness meditations brought out backdraft for a quarter of the participants after the first attempt, almost all participants stated that they anticipated their backdraft would reduce or felt that it diminished over time with practice. By contrast, the feeling the soles of your feet practice caused physical discomfort (one form of backdraft) for six participants at their initial attempts, and these feelings did not abate over time. It is possible that these experiences of sensory discomfort are not backdraft but personal preferences; however, due to the definition ambiguity of backdraft (Cai, in press), it is unclear how to categorize these sensory experiences. For the purpose of this study, we have used Neff and Germer’s (2022) broad definition of backdraft, which includes physical discomfort and distress caused by self-compassion practices.
These backdraft findings have clinical implications, specifically, whether sensory experiences should be categorized as backdraft. Self-compassion practices such as loving-kindness meditations should be introduced to autistic adults, even if they bring out discomfort initially. By contrast, the soles of your feet practice may not be suitable for some autistic adults due to the sensory nature of the practice. Programs or clinicians introducing practices associated with noticing bodily sensations could provide alternative practices. Another approach may be adapting self-compassion practices to meet the sensory needs of some autistic people. For example, participants in Lunsky et al.’s (2022) mindfulness intervention project suggested adaptations that can be made to mindfulness practices, such as using mindful movement rather than keeping still. Further work is needed to identify the possible adaptations of self-compassion techniques suitable for autistic people with sensory differences.
Our backdraft findings have an additional clinical implication on how self-compassion practices may be taught to autistic adults. Given that backdraft is an intrinsic part of developing self-compassion (Neff & Germer, 2022), many autistic people would inevitably have adverse reactions to the initial practices of self-compassion due to their history of trauma and painful experiences, specifically victimization, stigma, and interpersonal violence (Botha et al., 2022; Gibbs & Pellicano, 2023; Mazumder & Thompson-Hodgetts, 2019; Trundle et al., 2023). Our study shows that autistic adults who are not psychologically vulnerable (e.g. no recent experiences of self-harm ideation or behaviors, no severe mental illnesses such as schizophrenia or complex posttraumatic stress disorder) can complete self-guided self-compassion programs, even when painful emotions, memories, or physical sensations were brought up without any significant adverse effects. However, for autistic adults who are psychologically vulnerable, further research is needed to determine the best approach to support these individuals. For instance, it may be best for these adults to work with a compassionate mental health professional to help them manage the adverse reactions associated with some self-compassion practices.
While the findings from this pilot study are promising, several limitations exist. The primary limitation is the relatively small sample and lack of randomization or a control condition. Recruitment of a self-selecting sample with no co-occurring intellectual disability reduces our ability to generalize the results for all autistic adults, especially those with an intellectual disability or who have limited verbal communication. Outcome measures relied on self-reports, and we have not included follow-up measures. Finally, more autistic women volunteered to participate in this study than autistic adults of other gender identity. Galvin and Richards (2023) found that autistic women had lower self-compassion than autistic men. Similarly, Yarnell and colleagues (2015) found men tend to have slightly higher levels of self-compassion than women via a meta-analysis. They observed that most of the variance in male-female levels of self-compassion is shared and suggested not to overemphasize this gender difference. The unequal gender distribution of our sample may influence our findings; however, further research on gender influence on intervention effects are needed. Future research should conduct an RCT of ASPAA to demonstrate efficacy, determine the mechanisms of change, and measure if improvements are maintained over time. Another way to identify the mechanisms of change is to test the outcomes of individual self-reflective exercises and self-compassion practices. Examining individual practices can also help us better understand the adverse effects of specific practices. Finally, it is likely that unsupportive environments contribute to the lack of self-compassion often observed in autistic adults. It would be helpful to examine the impact of past experiences of stigma and bullying on autistic people’s self-compassion levels.
ASPAA is the first self-compassion program developed to help improve the mental health and psychological well-being of autistic adults. Theoretically, self-compassion improves mental health through better emotion regulation (Berking & Whitley, 2014; Gilbert, 2014). Given the known emotion regulation difficulties experienced by many autistic children, youth, and adults (Cai et al., 2018; Cibralic et al., 2019), this connection between self-compassion and emotion regulation makes self-compassion an even more relevant target to focus on from an intervention perspective. ASPAA’s innovation as an intervention stem from the co-production process utilized to ensure the program is tailored for autistic adults and the use of evidence-based self-compassion practices. Since completing this research, we have refined the ASPAA modules based on the participants’ feedback, and ASPAA is now freely available online for all autistic adults. The refinements include adding more lived-experience videos of participants reflecting on the self-compassion practices, providing alternatives to self-compassion practices that may cause sensory discomfort, and simplifying instructions and design of module pages. Through co-production, we ensured that ASPAA is relevant, relatable, and accessible to the target population. ASPAA can be used by autistic adults in a self-guided manner on their own or with the support of trusted individuals or mental health professionals.
Footnotes
Acknowledgements
The authors are extremely grateful to all who participated in this study and provided their valuable feedback about the self-compassion program. They are also thankful to the autistic advisory group (Aiyah Membery, Anna Gould, and Chris Day) for their valuable insight and reflections about the program.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Autism Spectrum Australia (Aspect) provided the funding for this research study.
