Healthy Relationships on the Autism Spectrum (HEARTS): A feasibility test of an online class co-designed and co-taught with autistic people

Abstract

This study tested the feasibility and preliminary efficacy of a six-session online class on healthy relationships for autistic individuals ages 18–44 years old (N=55). The content of the Healthy Relationships on the Autism Spectrum (HEARTS) class was informed by formative research with 25 autistic individuals, and developed collaboratively by two non-autistic professionals and seven autistic self-advocates. Fifty-five autistic people participated in HEARTS and completed pre- and post- surveys. The study found that it was feasible to deliver HEARTS online. Pairing an autistic and non-autistic person to co-teach was well-received. Participants decreased hostile automatic thoughts (p<.05), involvement in dating abuse in intimate relationships (p<.05), fight-or-flight response (p<.05), and rejection sensitivity (p<.001). Participants experienced improved flourishing (p<.001), coping with rejection and jealousy (p<.001), motivation to engage with others for socializing (p<.05), self-compassion (p<.05), and positive thinking (p<.05). Scores on a measure of interpersonal competence did not change, and loneliness did not decrease. The majority of participants reported high satisfaction with the class. HEARTS is a promising healthy relationships promotion class that should be evaluated through a randomized controlled trial.

Lay abstract

The Healthy Relationships on the Autism Spectrum (HEARTS) class is unique because autistic people helped to develop it and co-taught it. It is an online, six-session class. The class was piloted in 2020–21 with 55 autistic people who were ages 18 to 44 years old. This feasibility study found that most people who took the class liked it. Surveys filled out by the students before and after the class showed that they became less sensitive to rejection, used more positive thinking skills, and were more interested in being social. However, the class may not have made them feel less lonely. The team that invented the class is using the feedback to improve it. The class holds promise for improving the quality of friendships and dating relationships for autistic adults and should be tested further.

Autistic people, like all people, need and deserve education about how to have healthy relationships. In general population studies, one-third of people experience contact sexual violence, physical violence, or stalking by an intimate partner during their lifetimes (Garcia-Moreno et al., 2006; Smith et al., 2018), and one-third of secondary school children experience bullying (Zych et al., 2015). A growing body of literature suggests that autistic individuals are bullied more frequently than non-autistic peers, with prevalence rates as high as 46–96% (Morton, 2021). In a population-based study of women, those higher in autistic traits were 1.4 times as likely as those with few autistic traits to have experienced sexual abuse in childhood (40% vs. 27%), and 1.7 times as likely to have experienced physical or emotional abuse (24% vs. 14%) (Roberts et al., 2015). Not only are autistic people more likely to have these traumatic interpersonal experiences, but many report challenges making and maintaining friendships, and have lower quality friendships than non-autistic peers (Bauminger & Shulman, 2003; Finke et al., 2019)—perhaps as a result of traumatic experiences. A recent study of 126 autistic and 125 non-autistic youth ages 18–24 years old found that autistic people were 3.4 times more likely to be dissatisfied with their current number of friendships (60% vs. 18%)(Finke et al., 2019). A deficit model perspective would interpret these relationship challenges as the result of deficiencies in autistic people in need of remediation, while a neurodiversity perspective would interpret them as resulting from autistic neurological differences being met with stigma, ableism, elitism, and the widespread acceptance of exclusionary behavior in society.

Programs that teach healthy relationship skills have been developed for the general population. These include, for example, school-based dating and sexual violence curricula such as Fourth R, Safe Dates, Coaching Boys Into Men, and Shifting Boundaries (De La Rue et al., 2014), as well as programs for married and engaged couples, and prospective parents (Hawkins & Fackrell, 2010), which have been tested through large-scale and methodologically rigorous randomized controlled trials and found to be effective. Alternative approaches to promoting healthy relationship skills with non-autistic populations include support groups for people who have experienced abuse in intimate relationships (Allen et al., 2021; Schechter et al., 1985; Tutty et al., 1993), intervention groups for people who have perpetrated partner violence (Snead et al., 2018), and social and emotional learning programs for children (Corcoran et al., 2018).

Healthy relationships skills are not the same as social skills. Social skills describe the ability to interact with others in a given social context in socially acceptable ways that may only be pleasing to other people, but not necessarily to the individual performing these skills (Combs & Slaby, 1977; Green & Forehand, 1980). Although definitions of social skills vary, most are limited to discrete behaviors such as initiating and turn-taking in conversations, interpreting others’ verbal and non-verbal actions, and adjusting behavior according to social cues (Crooke et al., 2008). Healthy relationship skills, in contrast, tend to be less task-oriented, amorphous, and generally improve relationship quality for both people. They include, for example, effective interpersonal communication, conflict resolution, obtaining consent, establishing boundaries, rapport building, and expressing intimacy (Wood et al., 2012).

While numerous social skills training curricula have been developed and tested for effects on autistic individuals’ social cognition, social communication, social anxiety and other outcomes (Gates et al., 2017; Pallathra et al., 2019), to our knowledge there are only three evidence-based programs that promote healthy relationship skills for autistic individuals. As identified by a recent systematic review, these are: Supporting Teens with Autism on Relationships (STAR), Tackling Teenage Training (TTT), and Peers Engaged in Effective Relationships-Decision-Making (PEER-DM) (Holmes et al., 2021). Notably all three programs were designed for adolescent participants, and to our knowledge all were designed by non-autistic research and intervention experts and are delivered to groups of autistic participants by trained non-autistic interventionists. The lack of inclusion of autistic people in the development or delivery such programs may increase the likelihood content encourages autistic masking, and the resulting interventions may be less likely to address extrinsic factors that contribute to negative experiences in relationships (Leadbitter et al., 2021). As such, there is still a need for additional evidence-based healthy relationships interventions that serve autistic people, and a particular gap in the field related to healthy relationships interventions that are designed by autistic people, for autistic people, and delivered in partnership with autistic people.

In 2020 our research team collected formative data from 25 autistic individuals about what kinds of skills they would like to learn from a healthy relationships class, and how such a class would ideally be delivered in terms of setting, class composition, and teacher characteristics (Rothman & Graham Holmes, 2021). Findings resulting from that formative work were that an ideal healthy relationships class would take place online to accommodate a range of participation styles (e.g., verbal, typing/chat only), would include people of all genders, would be taught or co-taught by an autistic person, and would explicitly use a neurodiversity model (as opposed to a deficit model) as the basis for the content. Findings also suggested that the class should take a “sex positive” approach, meaning one that affirms participants sexuality, and promotes inclusion of alternative intimate relationship structures (e.g., polyamory), and sexualities (e.g., asexual, kinky or interested in bondage, discipline, dominance, submission, sadism, masochism (BDSM)). It was clear from the formative research that the teachers would need to be personally experienced in online dating. In addition, autistic individuals expressed wanting to learn how to overcome anxiety rooted in bad prior friendship and dating experiences, including when it was safe to take emotional risks and stop automatic hostile thoughts about others, how to cultivate reciprocity in relationships, and how to identify, communicate and respect sexual and emotional boundaries. Building upon these findings, in 2021, our team—which includes seven autistic self-advocates—collaboratively developed a six-session healthy relationships class to be delivered online by a team of one autistic and one non-autistic teacher. The class, called Healthy Relationships on the Autism Spectrum (HEARTS), was delivered five times during 2021 using an autistic/non-autistic team-teaching approach.

Current Study

The purpose of the current feasibility study was to evaluate the acceptability and preliminary efficacy of the (HEARTS) class.

METHODS

This feasibility study used a non-experimental one-group pre- and post-test design. All procedures were reviewed and approved by the Institutional Review Board (IRB) at the first author’s institution.

Sample and Data Collection

All participants registered for the HEARTS program through the US-based national autism service and advocacy organization Asperger/Autism Network (AANE; www.aane.org). AANE charged participants $10–60 for the six-session class, using a sliding scale. AANE advertised the availability of the class to their existing network via email. The first author also advertised the class using Twitter. Approximately 20 individuals registered for each iteration of the class for a total of 100 individuals across five iterations of the class taught in 2020–21. Once individuals were registered for the HEARTS program, they were invited to participate in one-on-one intake interviews with a class facilitator. During the interview, they were offered the opportunity to consider participating in the research study evaluating HEARTS. They were assured that they could still participate in HEARTS even if they declined participation in the research. Of the 100 individuals who participated in HEARTS classes, 84% (n=84) opted to be screened for research eligibility.

All interested individuals were eligible for class participation. However, to be eligible for the research study, individuals had to be 18–44 years old, able to communicate in English, state that they had one or more relationships (with a friend, intimate partner, or other peer) that they wanted to improve, affirm that they had access to a computer with Zoom and a private space from which to participate in class, state that they were autistic (either self-diagnosed, or with a diagnosis of autism spectrum disorder, Asperger’s, or PDD-NOS), and be responsive to interviewer questions. This final criterion was important because unlike a purely didactic or asynchronous online course, the success of a psychoeducational group depends in part upon the class members’ willingness to share about themselves, listen to one another, and collectively create a dynamic learning environment. Therefore, for research participation purposes, the investigators assessed group participation skills using a standard technique; each participant was asked to talk a bit about how they met their friend or partner, what they enjoy doing with that person, and one thing that they want to improve about that relationship. The content of the responses was not evaluated, but the quality of the social response in terms of social communication (e.g., showing responsiveness to the questions), social response (e.g., sustaining a comfortable interaction), quality of rapport (e.g., capacity for two-way social interchange without extreme or marked discomfort), and anxiety were rated on a 4-point scale. Individuals who scored ≥2 on social communication, social response, quality of rapport, or anxiety were considered ineligible for research (though they were still welcome to participate in the HEARTS class, intermixed with research participants), while those who scored ≤1 were eligible for the research study. The rating tool is available upon request from the first author. The reason for limiting research eligibility to 18–44 years old was to strike a balance between two competing needs. On the one hand, some clinicians believe that it is advantageous to limit interventions for participants in narrow age groups because they are more likely to be developmentally similar, and interventions should be tailored for developmental stage (National Research Council and Institute of Medicine Committee on the Prevention of Mental Disorders and Substance Abuse Among Children, 2009). On the other hand, chronological age is a less meaningful predictor of developmental stage for autistic people than non-autistic people, and individuals of all ages expressed interest in HEARTS. Therefore, the research team selected to limit the age range to young and middle-aged adults (18–44 years old), based on the idea that this group would be roughly in the same developmental period and would likely be facing similar challenges related to making friends and navigating intimate partner relationships.

Of the 84 people screened for eligibility, 90% (n=76) were eligible for research; four did not qualify based on social communication, two were >44 years old, and two were repeat students who had already participated in the study. Of the 76 eligible individuals, 95% (n=72) consented to participate in research. Those who consented completed the pre-test survey online, which took approximately 40 minutes, prior to the first HEARTS class session. Participants also completed the post-test survey four weeks following the final HEARTS class session. Participants received a $200 gift card for completing all surveys.

HEARTS

HEARTS was designed as a six-session psychoeducational class for autistic adults to be delivered online using Zoom. HEARTS was collaboratively designed by the authors of this paper, an autistic Advisory Board, and the autistic co-teachers of the class (i.e., seven autistic self-advocates and two non-autistic researchers). The content reflected formative research drawing on interviews with 25 autistic people about what they would want to learn in a healthy relationships class (Rothman & Graham Holmes, 2021). HEARTS was manualized before implementation. The majority of the content is available in a Powerpoint slidedeck, one for each week. The slides were made available to the students approximately two days prior to class so that they could familiarize themselves with the material if they wished.

Each HEARTS class was taught by one non-autistic and one autistic team teacher. The feasibility trial involved two non-autistic teachers and three autistic teachers in total. Qualifications for teachers, whether autistic or non-autistic, included: (a) prior experience dating online; (b) comfort discussing polyamory, kink, same-sex sexuality, pornography, and asexuality; (c) a commitment to anti-racist, anti-ableist, and gender and sexual minority supporting practice; (d) firsthand experience making friends, ending friendships, falling in love, and ending a serious intimate partnership; (e) prior experience facilitating group discussions online; and (f) capacity to attend HEARTS-related teaching meetings weekly in preparation for class sessions. Two of the three autistic teachers had prior experience facilitating online groups for AANE and were recommended by AANE, and the third autistic teacher was an acquaintance of the first author who had extensive prior experience facilitating online discussions about autism and relationships.

Each week all teachers met for one hour for a supervision session to discuss class progress, logistics, and review material to be taught in the week ahead. This weekly supervision session was designed to improve fidelity across classes. The autistic teachers were invited to give feedback on HEARTS Powerpoints during supervision. They made slight change suggestions, such as word choices (e.g., replace “nice” with “friendly,”) additions (e.g., add “masking” to bullet point list), or slide formatting (e.g., change color to grayscale).

Each week the session followed the same routine. First there was a “check-in,” during which the participants reported on how their week was and their thoughts on the homework practice assignment. Next the co-teachers presented didactic information on the session theme and guided the participants in interactive activities, such as card sorting using a Padlet.com board. At the end of the session students were asked to consider engaging in a homework practice assignment. Two or more optional homework assignments were offered each session. The weekly session themes are presented in Table 2. They include: (1) Defining healthy relationships; (2) Launching new relationships; (3) Neurohealth for relationship health; (4) Meeting new people; (5) Boundaries; and (6) Ending relationships. Each class session was 90 minutes long. Autistic individuals participated from a private space in their own home, without parents, spouses, or others present. The class was delivered via Zoom and students were explicitly told that it was acceptable to participate with cameras on or off, to participate verbally or exclusively using the typing-chat function, and that they could use assistive devices for communication as desired. Fidelity to the original curriculum plan was assessed using a three-point rating scale for planned session activities (i.e., 1=skipped this activity, 2=partially completed, 3=completed). Mean average fidelity ratings for activities ranged from 2.6 to 3.0, and the mean average fidelity score across all activities for all five classes was 2.92 (Supplementary Table 1).

Table 2.

HEARTS topics

Session	Theme	Topics
1	Defining healthy relationships	-Healthy vs. unhealthy relationships
		-Abusive behaviors
		-Warning signs of unhealthy relationships
2	Launching new relationships	-Being a trustworthy friend or partner
		-Developing curiosity in other people
		-Active listening
		-Avoiding relationship pitfalls
3	Neurohealth for relationships	-Fight, flight or freeze responses
		-Coping with relationship anxiety
		-Why eating, sleeping, exercising and relaxation are important for relationship health
4	Meeting new people	-How to meet people
		-Reconnecting with old friends
		-Perspective-taking and empathy
		-Handling jealousy
		-Disclosing autism to new friends
		-Progressing relationships
5	Boundaries	-Understanding boundaries
		-Setting boundaries
		-Respecting boundaries
		-Renegotiating boundaries
		-Apologizing and accepting apologies
6	Ending relationships	-Domestic violence and sexual assault hotlines
		-Signs a relationship should end
		-Reasons it can be hard to end relationships
		-How to end relationships

Measures

All data were collected through two self-administered online surveys, a pre-test (implemented before the intervention) and a post-test (implemented four weeks after HEARTS concluded). The surveys included the following measures:

Flourishing was assessed through the 8-item Flourishing Scale (Diener et al., 2010). Sample items include “I lead a purposeful and meaningful life,” and “My social relationships are supportive and rewarding.” Responses were on a Likert scale from 1 to 7, where 1 represented strongly disagree and 7 represented strongly agree. A high score represents a person with many psychological resources and strengths (Diener et al., 2010). The Cronbach’s alpha in this sample was 0.86.

Hostile automatic thoughts were assessed through a 25-item version of the Hostile Automatic Thoughts Scale (HAT) (Snyder et al., 1997). This instrument instructs respondents to consider how often they had various thoughts in the past week about other people. Sample items include “I hate this person so much I could kill them,” and “I’ll show this person!” The original scale was modified for this study by eliminating four items and expanding response options from two to five categories. For the present study, response options were on a 5-point Likert scale where 1 represented “not at all” and 5 represented “all the time.” A high score represents a person with frequent hostile automatic thoughts. The Cronbach’s alpha in this sample was 0.95.

Dating abuse in the past three months was assessed through selected items from the Measures of Adolescent Relationship Harassment and Abuse (MARSHA). Respondents answered 7 dating abuse victimization questions and 7 dating abuse perpetration questions. Sample items include “I slapped, pushed, shoved or shook them,” and “They hit, punched, kicked or choked me.” Response options were “yes,” “no” and “prefer not to answer.” The endorsement of one or more victimization or perpetration items was classified as presence of dating abuse in the participant’s relationship. In addition, a total scale score was calculated by summing across items. Higher scores represented more dating abuse. The number of items endorsed by participant ranged from 0–5, with a median of 1.

Coping with rejection and jealousy was assessed through a 16-item original instrument. Sample items include “I am so afraid of rejection that I don’t try to meet new people,” and “I’ve been bullied in the past, so I get anxious about making friends.” Response options were on a 5-point Likert scale where 1 represented “totally disagree” and 5 represented “totally agree.” All responses were summed for a scale score. Higher scores indicate better coping with rejection and jealousy. The Cronbach’s alpha in this sample was 0.56.

Interpersonal competence in peer relationships was assessed through the 40-item Interpersonal Competence Questionnaire (Buhrmester et al., 1988). The questionnaire assesses respondents’ confidence in five domains, including initiating relationships, self-disclosure in relationships, asserting displeasure with others’ actions, providing emotional support, and managing interpersonal conflict. A sample item is: “Letting down your protective ‘outer shell’ and trusting a close companion.” Minor wording modifications were made. The phrase “I’m poor at this” was changed to “I’m bad at this,” and items referring to peers as s/he or his/her were changed to gender neutral “they.” Response options were on a 5-point Likert scale ranging from 1 (“I’m bad at this”) to 5 (“I’m extremely good at this”). Subscale scores were used. Cronbach’s alpha in this sample was 0.95.

Motivation to engage with others for socializing in the past month was assessed through an original 16-item scale. Respondents were asked how motivated they felt in response to 16 items on a scale from 1 to 5, where 1 represented “not at all motivated” and 5 represented “very motivated.” Sample items include “send a text message or email to a friend,” and “talk to someone that I don’t already know.” A higher score indicates more motivation to engage with others. Cronbach’s alpha in this sample was 0.92.

Positive thinking was assessed through the 22-item Positive Thinking (PT) scale (Diener et al., 2009). A sample item is “I believe in the good qualities of other people.” Two modifications were made to the original version. The word “salient” was changed to “memorable” and “shortcomings” was changed to “faults.” Response options were “yes” and “no.” A scale score was derived by assigning a value of 1 to each response of “yes” and summing across all items. Higher scores represent more positive thinking. The Cronbach’s alpha in this sample was 0.42.

Tendency to have a fight-or-flight type response to a stressful incident was assessed via a modified version of the Predominant Response Questionnaire (Jones et al., n.d.). Sample items from this questionnaire include “I acted without thinking rationally (lashing out), “I misread events because I expected the worst,” and “I felt frozen or stuck [in terms of decisions].” Response options were “yes” or “no.” A scale score was created by assigning a value of 1 to each response of “yes” and summing across items. Higher score indicates greater likelihood of having a fight-or-flight response. Cronbach’s alpha in this sample was 0.88.

Rejection sensitivity was assessed through a 10-item rejection sensitivity scale inspired by an existing instrument, but rewritten to reflect contemporary modes of socializing (Downey & Feldman, 1996). This scale asks respondents to consider 10 different scenarios in which people might find themselves and imagine whether they would be worried that they had done something wrong, and whether they would expect the other person in the scenario to want to stop socializing with them. Worry is scored on a 6-point Likert scale, and expectation that the other person will stop talking is assessed on a 6-point Likert scale. For example, respondents are asked to imagine: “I ask a friend to get together and hang out but they say that they are too busy,” and then asked how worried they are that it’s their fault (where 1 represents “not at all worried” and 6 represents “very concerned”), and asked how likely they would be to expect that the friend doesn’t like them anymore (where 1 represents “very unlikely” and 6 represents “very likely”). Higher scores represent more rejection sensitivity. Cronbach’s alpha in this sample was 0.96. The mean score on the scale was used for analytic purposes.

Self-compassion was assessed using the short form State Self-Compassion Scale, which is a 6-item scale. Response options were modified to be yes, sort of, and no. A sample item is “I feel intolerant and impatient towards myself.” Higher scores represent more self-compassion. The Cronbach’s alpha in this sample was 0.50. The total scale score was used for analytic purposes.

Loneliness was assessed using the 20-item UCLA Loneliness Scale. The scale asks respondents to rate how often a series of statements is descriptive of them, such as “how often do you feel alone?” Response options are on a 4-point Likert scale where 1 represents “never” and 4 represents “often.” Higher scores represent more loneliness. The Cronbach’s alpha in this sample was 0.34.

Knowledge about intimate partner violence was assessed using 6 original questions. Example questions are: “I know the warning signs of abuse in dating relationships” and “If a dating partner ever made me afraid for my safety, I would know how to get help.” Response options are on a 5-point Likert scale where 1 represents “totally disagree” and 5 represents “totally agree.” Higher scores represent more knowledge about dating abuse. Responses for each question were tabulated individually.

Data analysis plan

Quantitative survey data were cleaned and coded using SAS v9.4 (SAS Institute, Cary, NC 2016). Descriptive statistics for pre- and post-test scores are presented as mean and standard deviation for continuous variables and as number and percent of participants for categorical variables. Within subject differences were assessed using paired t-tests and McNemar’s tests, as appropriate. Tests were not adjusted for the number of tests conducted. Responses to two open-ended questions on the post-test survey were analyzed using a content-based approach; all responses to a question were read for a sense of the whole, themes were identified by the first and second authors, and illustrative quotes were selected for presentation.

Community involvement

HEARTS was collaboratively created by all authors of this paper, which includes seven autistic self-advocates (five of whom are members of the project Advisory Board, and two of whom were co-facilitators of the class). In addition, the class was designed based on formative data collected from 25 autistic individuals about their priorities and preferences for a healthy relationships class (Rothman & Graham Holmes, 2021).

RESULTS

Sample characteristics

Of the 72 individuals who consented to participate in research, three did not complete the baseline survey, eight missed three or more intervention sessions and were therefore removed from the analytic dataset, and six did not complete post-test, leaving a final analytic dataset of N=55. Participants were 31% male, 55% female, and 11% non-binary gender. Eighty percent identified as White, 6% identified as Black, 6% identified as Asian, 7% identified as multiracial, and 1% identified as another race or did not report race (Table 1). Eight percent identified as Hispanic/Latinx. Twenty-five percent (n=14) were self-diagnosed autistic, and 75% had a written report from a doctor or other healthcare professional indicating that they had a diagnosis of autism, Asperger’s or PDD-NOS. Raw SRS-2 scores ranged from 40 to 146, and T-scores ranged from 50 to 87. Eighty-four percent of the sample had SRS-2 scores in the mild, moderate or severe range, and 16% scored in the normal range.

Table 1.

Participant demographics (N=55)

	% (n)
Gender
Cis-Male	31% (17)
Cis-Female	55% (30)
Non-binary	11% (6)
Other	4% (2)
Age (Mean, SD)	28 (7)
Range: 20–43 years
Race
Asian	6% (3)
Black or African-American	6% (3)
White	80% (44)
Multiracial	7% (4)
Ethnicity
Hispanic	8% (4)
Non-Hispanic	93% (49)
Sexual orientation
Heterosexual or straight	53% (30)
Bisexual, gay, lesbian, or pansexual	38% (21)
Asexual	4% (2)
Other	6% (3)
Region or Country
Australia	1% (1)
U. S. Northeast	53% (29)
U. S. South	20% (11)
U. S. West	15% (8)
U. S. Midwest	11% (6)
Ever diagnosed by a professional with:
Autism, Asperger’s or PDD-NOS	75% (41)
Intellectual disability	9% (5)
Attention deficit disorder (ADD) or ADHD	46% (25)
Depression	76% (41)
Anxiety disorder (including OCD)	82% (45)
Substance use or alcohol use disorder	7% (4)
Schizophrenia	6% (3)
Bipolar disorder	6% (3)
Eating disorder	13% (7)
Oppositional/defiant disorder or conduct disorder	7% (4)
Post-traumatic stress disorder	26% (14)
Self-diagnosed autism	25% (14)
Currently has a spouse, dating or intimate partner	38% (21)
Currently reside with parents	47% (26)
Ever received free or reduced-price meals at school	24% (13)
Employed full-time	35% (19)
Employed part-time	25% (14)
SRS-2 score (Mean, SD)	68 (9)
SRS-2: Severe or moderate	66% (36)
SRS-2: Mild	18% (10)
SRS-2: Normal	16% (9)

Class participation

The majority of participants (67%) attended all six class sessions; 53 participants (93%) either missed one class or attended them all.

Outcomes of interest

Participants decreased hostile automatic thoughts (score 41.42 vs. 38.05, p<.05), dating abuse (score 1.08 vs. 0.71, p<.05), fight-or-flight response (score 11 vs. 9, p<.05), and rejection sensitivity (mean score 12.84 vs. 9.51, p<.001) (Table 3). Participants experienced improved flourishing (score 37.93 vs. 41.73, p<.001), coping with rejection and jealousy (score 48.69 vs. 54.45, p<.001), motivation to engage with others for socializing (score 44.35 vs. 55.16, p<.05), self compassion (score 5.73 to 7.33, p<0.05), and positive thinking (score 7.44 vs. 8.05, p<.05) (Table 3). Interpersonal competence did not improve, and loneliness did not decrease significantly. In addition, the percentage of participants who agreed that that they knew how to call the national domestic violence hotline increased from 42% to 69% (p<.001), the percentage who knew how to call a sexual assault crisis organization increased from 40% to 66% (p<.001), and the percentage that totally agreed with the statement “I know the warning signs of abuse in dating relationships” increased from 73% to 91% (p<.01)(Table 3). The percentage who agreed with the statement “I am not sure how to keep my relationships healthy” decreased from 47% to 22% (p<.001)(Table 3).

Table 3.

Pre- and post-test scores on outcomes of interest (N=55)

	Pre-test Mean (SD)	Post-test Mean (SD)	Paired t-value, p-value	Cohen’s d
Flourishing	37.93 (7.99)	41.73 (6.81)	3.79, 0.00	0.51
Hostile automatic thoughts	41.42 (14.01)	38.05 (12.55)	−2.20, 0.03	−0.25
Dating abuse (scale score)	1.08 (1.18)	0.71 (1.04)	−2.26, 0.04	−0.34
Coping with rejection and jealousy	48.69 (10.54)	54.45 (9.62)	5.14, 0.00	0.57
Interpersonal competence
Domain 1: Initiating relationship	21.00 (2.80)	21.09 (2.30)	−0.23, 0.82	0.04
Domain 2: Self Disclosure	24.11 (2.48)	23.84 (3.42)	−0.57, 0.57	−0.09
Domain 3: Asserting Displeasure with Others	23.60 (2.86)	24.26 (2.88)	1.81, 0.08	0.23
Domain 4: Providing Emotional Support	26.42 (3.40)	26.35 (3.59)	−0.16, 0.87	−0.02
Domain 5: Managing Interpersonal Conflicts	22.85 (3.63)	23.69 (2.47)	1.50, 0.14	0.27
Motivation to engage	44.35 (12.98)	55.16 (10.87)	2.75, 0.02	0.91
Positive thinking	7.44 (2.57)	8.05 (2.65)	2.40, 0.02	0.24
Fight-or-flight response	10.64 (4.78)	9.45 (5.53)	−2.35, 0.02	−0.23
Rejection sensitivity (mean score)	12.84 (8.43)	9.51 (8.52)	−4.35, 0.00	−0.39
Self compassion (scale score)	5.73 (3.03)	7.33 (3.03)	4.10, 0.00	0.53
Loneliness	12.33 (5.13)	11.69 (5.14)	−1.39, 0.17	−0.12
	Pre-test % (n)	Post-test % (n)	Odds Ratio (95% CI), p-value
Dating abuse (present in relationship)	22% (12)	16% (9)	0.6 (0.19–1.90), 0.18
If a dating partner ever made me afraid for my safety, I would know how to get help	64% (35)	82% (45)	2.57 (1.07–6.19), 0.03
Knowledge about domestic violence hotline	42% (23)	69% (38)	3.11 (1.42–6.81), 0.00
Knowledge about sexual assault organization	40% (22)	66% (36)	2.84 (1.31–6.17), 0.00
I plan to ask people that I trust for help as I work on my relationships	72% (39)	82% (44)	1.69 (0.68–4.20), 0.17
I know the warning signs of abuse in dating relationships	73% (40)	91% (50)	3.75 (1.26–11.20), 0.01
I am not sure how to keep my relationships healthy	47% (26)	22% (12)	0.31 (0.14–0.71), 0.00

Note. Cohen’s d effect size 0.2 = small, 0.5 = medium, and 0.8 = large.

Participant satisfaction

On a scale from 1 to 10, with 1 representing the worst possible score and 10 representing the best, participants scored the class as a 7 overall. On a scale from 1 to 10 where 1 represented “very sorry” and 10 represented “very glad,” in response to the question “how glad are you that you decided to take this workshop, the average score was 8. Approximately 29% reported that the workshop improved their ability to meet new people, 35% reported that they made new friends or met a new dating partner in the 6 weeks prior to the post-test survey, 71% reported that they workshop improved their ability to have healthy relationships, and 65% reported that the workshop taught them how to improve their relationships.

The post-test included two open-ended questions that asked participants to reflect on what worked well about the class and what could be improved. Three themes related to what participants appreciated about the class were identified. These were: (a) enjoying interacting with other autistic students; (b) appreciating that a co-teacher was autistic; and (c) valuing particular activities, handouts or other content. For example, 16 participants made comments about enjoying other autistic students, such as: “So much interaction with and validation from fellow autistic people, great topics, astute leaders, made me think,” “I made new friends. It made me more open to others,” “I found it helpful/motivating to listen to other people talk about their relationship experiences when I was avoiding socializing entirely at the start of the course. I started dating again and I think the course motivated me to do so,” and “I appreciated being in course with people on the spectrum who have been married and people who have children. I had a lot of hopelessness previously because I didn’t know if these things were possible for me.” In terms of appreciating the autistic co-teacher, 13 participants wrote comments including: “I liked that a person on the spectrum was assisting in teaching the class, it’s nice to have someone on the spectrum talk about their own experiences and how they handle certain situations,” “I was grateful that there was an autistic moderator,” and “I liked hearing the perspectives of other autistics and of a non-autistic professional.” Comments about particular activities that were especially valuable to participants centered on an active listening exercise, the “positivity sandwich” activity, and the presentation of information about neurohealth (i.e., fight-or-flight responses).

Participants also contributed feedback about what could be improved. Themes noted in these comments were: (a) there were too few sessions and too much material to cover (10 participants commented on this); (b) some topics felt less relevant to some students (6 participants), and they felt bored during those sessions; and (c) frustrations with the way that some participants used the chat feature on Zoom (4 participants). In addition, there was one participant who felt that the class needed to be “completely overhauled” because it reminded them of a social skills class. That participant commented: “Basically everything about this class is your typical social skills class nonsense with a nice shiny neurodiversity spin on it.” A second person had similar feelings, and commented: “Although the class was supposedly being taught through a neurodiversity lens, I saw little evidence of that.” Although there were only two commenters expressing this sentiment, we include them here in the interest of full transparency. Other participants commented on wanting more class time: “90 minutes felt like a long time to be on a call with 15–20 people. I would have preferred to meet for 9 weeks instead of 90 minute sessions,” “I think either the sessions should be longer or there should be more of them: I feel like we ended up skipping or glossing over certain ideas for the sake of time,” and “I wish we were able to get through all of the content each week.”

DISCUSSION

This study found that it was feasible to deliver the HEARTS healthy relationships online class to autistic participants. Further, pairing an autistic and non-autistic person to team teach HEARTS was well-received by participants. Overall, the class was highly rated by participants, with the majority expressing satisfaction.

Comparison of pre- to post-test survey responses indicated that participants may have benefitted from participating in HEARTS. Specifically, participants learned warning signs of relationship abuse, and about resources for those experiencing domestic or sexual violence. In addition, participation in HEARTS was associated with decreased rejection sensitivity, hostile attribution bias, fight-or-flight responses, and dating abuse involvement, and associated with improved flourishing, coping with rejection and jealousy, motivation to engage with others for socializing, self-compassion, and positive thinking. Given that rejection sensitivity is associated with depression, aggression, and relationship dissatisfaction, lessening rejection sensitivity may lead to individuals behaving in a less defensive manner with others, and improved quantity and quality of social relationships (Kawamoto et al., 2015). Furthermore, negative repetitive cognition (i.e., ruminating) about relationships is common among autistic people and can erode the quality of social relationships (Keenan et al., 2018). That HEARTS participation was associated with an improvement in positive thinking, flourishing, self-compassion, and coping with rejection may signal that it could interrupt negative ruminating helpfully.

There was no change on participants’ scores on a measure of interpersonal competence, and on loneliness. It is possible that six sessions were too few to make an impact on participants’ interpersonal competence, or their loneliness; taking action and practicing new ways of interacting with others, and establishing new relationships (or deepening existing relationships) may take longer or require more intensive forms of personalized coaching. We note that more than one-third of participants reported making a new friend or meeting a new dating partner in the six weeks prior to the post-test survey, so a substantial subset did meet new people, but those relationships may not have deepened enough to address loneliness. If the follow-up period were longer, we might have detected changes in interpersonal competence and loneliness that took place after the HEARTS class ended when students continued to engage in practice of new skills. In addition, the fact that the class took place during the COVID-19 pandemic when feeling lonely was universal, and opportunities to socialize were extremely limited, may have affected participants’ opportunities to practice interpersonal competence skills or decrease loneliness by engaging in new relationships. The null effect for loneliness also could have been due to the low reliability of the measure with the participants.

While this class represents a step forward for the field in that it sought to center the preferences and needs of autistic people, teach from a neurodiversity-as-strength perspective rather than a deficit model perspective, and to use an inclusive team teaching model, some of the participants’ comments are prompting further changes to HEARTS to make it less ableist. For example, the first time the class was taught teaching duties were divided between the teacher and the non-autistic teammate non-equitably in terms of “air time” or speaking time during class. In part, this was to accommodate a preference of one autistic co-teacher who felt anxious about being in a lead role the first time through the course. However, the teachers did not explain to the class why the teaching roles were divided up inequitably, and so students may have drawn the conclusion that the autistic teacher was valued less. To correct this problem, a subsequent iteration of HEARTS was primarily facilitated by the autistic co-teacher, and the non-autistic teacher is playing a secondary and supportive role.

Another problem that participants pointed out was that it felt to them like there was too much material to cover in six sessions. In part, the reason that students may have perceived that the teachers did not cover all of the material was that the Powerpoint presentations were purposefully loaded with extra content because, as the class was being delivered for the first time, the teachers weren’t certain which activities would be appealing and effective and which ones would need to be aborted in real time. Now that the class has been taught numerous times, the team has a better sense of how to streamline content and can protect students from seeing excess material in Powerpoint that is not covered by the teachers. It is also possible that because 46% of participants had co-occurring ADHD, sitting through a 90 minute class was challenging. Finding ways to accommodate the learning styles of those with ADHD could improve satisfaction. Lengthening HEARTS to include additional sessions may also be something to explore; the dosage (i.e., quantity, duration) of an intervention can influence outcomes (Baldwin et al., 2009), although multiple review articles contend that brief interventions may be important for accessibility and engagement purposes and that a higher number of sessions is not always predictive of better impact on participants (Baldwin et al., 2009; Kanter & Schramm, 2018; Niileksela et al., 2021; Robinson et al., 2020).

Our team has also given some thought to how we can avoid suggesting that non-autistic ways of communicating are preferential. There is one session that focuses on what are called “relationship boosters” and “relationship pitfalls,” which are akin to some basic social skills taught in other classes, such as the importance of asking questions and feeling curious about a conversational partner, turn-taking in conversations, being flexible about plans, and not over-apologizing. It is emphasized that students should not feel obliged to behave in ways that align with neurotypical (NT) cultural expectations, but that the information about what NTs tend to expect during social interactions is provided so that students can experiment—if they want to—with adhering to normative social rules when they feel it will benefit them. Offering such information in a way that affirms it is an option to engage in normative social behavior, and can often be strategic and beneficial, while simultaneously acknowledging it is not always possible, desirable or necessary to do so will continue to be a program goal.

It makes sense that some autistic people have strong, negative reactions to classes that contain elements of social skills training because historically, most social skills programs have been rooted in a deficits model perspective, and can be dehumanizing (Bottema-Beutel et al., 2018; Roberts, n.d.). Due to negative experiences being told that their behavior is not acceptable, some autistic people may find it triggering to participate in any type of class that promotes behavior change to improve relationships. This could be one reason why relatively few autistic young adults participate in such programming after high school (Connor et al., 2020; Turcotte et al., 2016). The HEARTS program is, therefore, a promising new option for the field because it is rooted in a neurodiversity perspective, is not offered in a clinical or medical setting—but through an autism advocacy organization, strives to offer information in an invitational and not prescriptive way, and does not discriminate against adults who are self-diagnosed as autistic. An inclusive, accepting, peer-delivered, client-centered class culture models the type of healthy relationship skills that HEARTS seeks to impart; strict rules about who is permitted to be in the class and acceptable participation styles (i.e., verbally or via chat; with camera on or off), would likely decrease the popularity of the class with autistic individuals and, we suspect, make it less effective.

Another consideration for future implantation of HEARTS is that the age range of 18–44 years old was somewhat arbitrary; as describe above, the age range was selected in an attempt to create a peer group of developmentally similar individuals who were wrestling with some of the same relationship challenges due to their stage of life. However, chronological age is not a perfect indicator of developmental stage or relationship experience, and so future iterations of HEARTS should perhaps be open to anyone age 18+ years old. Similarly, using a three-question social responsiveness screening may not be necessary. Although the class was intended to help people who are engaged in some level of socializing (as opposed to none), and the social responsiveness screening was designed to identify individuals who therefore might benefit most from HEARTS, it is possible that individuals who do not socialize and have no peer relationships could also benefit from participation.

Limitations and future directions

While the results of this feasibility study are encouraging, there are four limitations that should be taken into consideration. First, a larger randomized controlled trial (RCT) with a lengthier follow-up period is needed to evaluate the effectiveness of HEARTS. This pilot was limited by the lack of a control or comparison group. Second, the pilot test took place during the 2020 COVID-19 pandemic. Therefore, participants had few opportunities to put new relationship skills into practice in face-to-face interactions if they so desired, and were generally operating under extreme and unprecedented levels of stress. Positive effects on loneliness may have been dampened. A future RCT that takes place during a time when people are permitted to interact socially in person may identify even stronger positive effects. Third, in this sample the reliability of the measures of positive thinking and loneliness were not adequate. Both measures were developed for, and have primarily been used with, non-autistic samples. The development of new measures of positive thinking and loneliness that are valid and reliable for use with autistic adult populations would benefit the field and could strengthen future evaluations of HEARTS (Nicolaidis et al., 2020). Fourth, approximately one-quarter of this sample were self-diagnosed autistic individuals rather than diagnosed by a medical professional. The inclusion of self-diagnosed individuals is a strength of the research in that the diagnostic process is prohibitively costly and not widely available to adults, so restricting research to professionally diagnosed individuals biases samples towards those who have had more apparent autism symptoms from an earlier age, and those who are more trusting of medical professionals (Lewis, 2017)—which tends to be White, male, and socioeconomically advantaged people. However, the inclusion of self-diagnosed individuals also introduces the possibility that an unknown portion of the sample could be misclassified. Additional research that explores the pros and cons of including self-diagnosed autistic individuals in intervention research studies is warranted.