For children with autism spectrum disorder (ASD), the most impairing symptoms often extend beyond the core difficulties with social communication and repetitive behaviors and interests. Challenges with behaviors (including aggression, self-injury, hyperactivity, and ‘meltdowns') are highly prevalent (Hartley et al. 2008; Hill et al. 2014; Simonoff et al. 2008), as are anxiety symptoms (Simonoff et al. 2008; Van Steensel et al. 2013), affecting up to 84% of children with ASD (White et al. 2009). These co-occurring difficulties often persist over time (Buck et al. 2014; Simonoff et al. 2013), cause a high level of parenting stress (Lecavalier et al. 2006) and can significantly impact adaptive functioning and quality of life (Maskey et al. 2013). As such, they constitute important targets for early intervention.

Recent meta-analyses have pointed to significant benefits of behavioral parenting interventions for child disruptive behavior and caregiver stress in young children with ASD (Postorino et al. 2017; Tarver et al. 2019). For example, in the largest randomized control trial (RCT) to date (n = 180, age 4–7), the ‘RUBI’ individual parenting intervention [Research Units in Behavioral Intervention: Bearss et al. (2015)] led to superior reductions in disruptive behavior compared to a parent-education programme. With regard to parenting interventions for anxiety in children with ASD, the evidence-base is less established. A systematic review and meta-analysis points to the efficacy of Cognitive Behavior Therapy (CBT) for reducing anxiety in school-aged children with ASD (Ung et al. 2015), often closely involving parents in sessions (Wood et al. 2009). Emerging evidence suggests that interventions aimed directly at parents may also have a benefit (Cook et al. 2017), in line with findings in typically developing children (Cartwright-Hatton et al. 2011; Monga et al. 2015).

Parenting interventions in ASD to date have faced several challenges. First, existing programmes have addressed difficulties with either behaviors or emotions, not both. Targeting one domain neglects the evidence that symptoms of disruptive behavior and anxiety commonly co-occur in young people with ASD (Simonoff et al. 2008; Storch et al. 2012), and anxiety is often a key driver of behavior ‘meltdowns’ (Lecavalier et al. 2014; Rzepecka et al. 2011). Secondly, previous programs have not been ‘universal’ in their relevance for parents of children with ASD. Some have included only parents of children with clinical levels of disruptive behavior (Bearss et al. 2015) or anxiety (Wood et al. 2009); others have targeted only parents of verbal children or ‘high-functioning’ children with an IQ above 70 (Ung et al. 2015). Thirdly, most parenting interventions to date have been delivered individually (Postorino et al. 2017; Ung et al. 2015). This poses limitations for scalability, cost-effectiveness and also limits the social support and collaborative problem-solving that is well-documented in group settings (Williams et al. 2017).

Finally, it is notable that most studies to date have evaluated manualized approaches adapted from programs for typically developing children or those with behavioral or emotional difficulties. Examples include the Stepping Stones Triple P (SSTP) program (Tellegen and Sanders 2014; Whittingham et al. 2009b; Zand et al. 2017); the ASD-adapted Incredible Years group program (Webster-Stratton and Reid 2010; Williams et al. 2017), Parent–Child Interaction Therapy (Scudder et al. 2019) and ‘Coping Cat’ for anxiety (McNally Keehn et al. 2013). Consequently, parenting interventions to date, although theoretically informed in terms of the target symptoms, do not have a strong ASD-specific framework or underpinning concept.

Objective 1: Developing a New Group-Parenting Approach: Predictive Parenting

The current feasibility study aimed to develop a new parent-intervention, addressing behavior and anxiety difficulties within the same group program. It sought to bring together evidence-based strategies for children with ASD, including common components of functional analysis, behavioral management and CBT for anxiety. A key aim was to ensure the group has conceptual relevance for any parent of a child with ASD following diagnosis, regardless of their child’s level of language, cognitive ability or their current levels of anxiety or challenging behavior. It looks to support parents with current challenges, but also to prevent the likelihood of future behavior or emotional difficulties. A further aim was to ensure that the group had a framework with specific ASD-relevance. By developing an approach with strong ties to ASD symptomatology, cognitive style and conceptual theory, it sought to help parents understand why certain strategies are effective in targeting challenging behavior and anxiety.

Focus on Prediction

In clinical settings, the observation that children with ASD experience difficulties with predicting change often resonates with parents. Some researchers have queried whether a core difficulty with prediction might explain the disparate behavioral traits in ASD (Pellicano and Burr 2012; Van de Cruys et al. 2014). For example, individuals with ASD may prioritize current perception over past experience, struggling to use expectations to predict future situations (Gomot and Wicker 2012). This may create a world that they experience as ‘unpredictable’, in which the causes of events are unclear and predicting what will happen next is difficult and anxiety-provoking (Sinha et al. 2014). According to this idea, the key behavioral hallmarks of ASD are understandable reactions to, and methods of coping with, an unpredictable environment (see Table 1 for examples).

Table 1 Links between ‘prediction’ difficulties and ASD symptomatology
Full size table

To date, there is mixed and insufficient empirical evidence to suggest ‘prediction difficulties’ are a single underpinning impairment in ASD. However, in terms of clinical utility, it is an idea that provides a helpful organizing framework to bring together well-recognized behavioral strategies.

The current ‘Predictive Parenting’ program incorporates three main branches (Fig. 1):

  1. i)

    Learning to predict behavior more effectively: The first branch aligns with the evidence-based principles of functional analysis (Beavers et al. 2013), encouraging parents to identify the antecedents and consequences of their child’s behavior. By considering the ‘purpose’ of their child’s behavior, parents learn to target their strategies more effectively. By becoming ‘behavior predictors’, parents are supported to gain a greater understanding and empathy for their child’s disruptive behaviors. From the beginning, anxiety is considered as a possible setting event for behavior difficulties.

  2. ii)

    Making life more predictable: The second branch supports parents to increase predictability for their child and to prevent challenging behaviors associated with uncertainty, change and anxiety. This branch incorporates well-recognized approaches targeting behavioral antecedents, such as planning-ahead, increasing routine and using visual prompts; using predictable instructions; and improving functional child communication (Bearss et al. 2015; Knight et al. 2015; Whittingham et al. 2009a). There is also a focus on predictable consequences: emphasizing praise and rewards, as well as clear and predictable boundaries. Specific minimally-verbal elements in branch 2 include functional communication and the use of objects of reference (McLarty 1997) to increase the predictability of communication. In addition to well-known concepts of parental consistency and behavior management (Webster-Stratton and Reid 2010), this stage also includes ‘ASD-specific’ approaches such maintaining a predictably ‘low arousal’ environment (reducing sensory input, managing demands) (Bearss et al. 2018) and using ‘social stories’ to increase predictability in daily situations (Gray 2000; Barton et al. 2018). Understanding and managing autistic ‘meltdowns’ is also covered, with an increased emphasis on self-injury for minimally-verbal groups.

  3. iii)

    Helping children cope with unpredictability: It is impossible and undesirable to make life completely predictable for children with ASD. The third branch explores how to help children tolerate uncertainty. Drawing on established CBT approaches, it includes psychoeducation about anxiety, and the use of graded exposure to help children face their fears (Ung et al. 2015). In line with previous ASD programs, it emphasizes the use of sensory calming strategies, physically active coping skills, and relaxation techniques (Attwood 2004; Wood et al. 2009). Parents are encouraged to model anxiety coping skills, to reduce reassurance, and to increase emotional labeling (Cartwright-Hatton et al. 2011). The strategies in this stage are also in keeping with CBT techniques targeting ‘Intolerance of Uncertainty’ (Boulter et al. 2014; Rodgers et al. 2017). For example, parents are encouraged to gradually introduce periods of change or uncertainty during the day to develop their child’s flexibility. For parents of minimally verbal children, strategies in branch 3 have a strong emphasis on behavioral exposure strategies and managing parent responses, rather than a more cognitive CBT approach.

Fig. 1
figure 1

The three branches of Predictive Parenting

Full size image

The final part of the Predictive Parenting framework, covered throughout the course, relates to parent well-being and self-care. Parents are encouraged to be aware of their own stress symptoms and to take time to look after themselves in a variety of ways. The course incorporates aspects of ‘mindful parenting’, encouraging parents to ‘step back’ and respond to their child with awareness (in a predictable way), even in stressful situations (Bogels and Restifo 2014; Singh et al. 2006). A growing evidence base suggests this approach can reduce parental emotional reactivity, improve parent–child relationships (Cachia et al. 2016), and allow parents to more effectively manage behavior (Singh et al. 2006). To support the well-being of parents who may experience traits (or a diagnosis) of ASD, in line with the ‘Broader Autism Phenotype’ (Pickles et al. 2000), the course itself also includes a high level of structure and predictability, with a clear visual framework and concrete case examples. Parents are encouraged to set achievable goals and to prioritize strategies, with the aim of being a ‘good enough parent’ (Winnicott 1987).

Intervention Structure

The course comprises 12 weekly, 2-h sessions (see Palmer et al. 2019). Teaching methods incorporate didactic and interactive elements, as well as homework tasks to practice new strategies. If sessions are missed, parents are given the opportunity to discuss the session material over the phone or before the next session. Parents are supported to feed back to the group each week, allowing group problem-solving to increase autonomy in choosing and implementing strategies.

In the first instance, the groups were designed for 4–8 year-olds, taking into account challenges observed at this developmental stage and bringing together parents with similar experiences. Although the Predictive Parenting concept is universally applicable, separate groups were adapted and run for parents of verbal and minimally-verbal children (verbal n = 6; minimally verbal n = 6). Certain strategies were added or emphasized for parents of minimally verbal children (as described above), while others were removed or reduced (e.g. social stories; reward charts). This differentiation aimed to tailor content to the child’s level of ability, and to encourage greater group cohesion. Two hour-long individual telephone sessions (after sessions 2 and 9) allowed for further tailoring of group content for each family’s individual needs. The course is designed to be delivered by professionals with experience of (i) working with young people with ASD and their parents (ii) facilitating therapeutic and psychoeducational groups (iii) delivering behavior interventions.

Public and Patient Involvement (PPI) Panels

Two PPI panels (two parents of children with ASD and five adults with ASD) were involved in the development of the intervention. This is in keeping with recent drives for interventions to be ‘co-created’ with input from service users at each stage of development (Fletcher-Watson et al. 2018).

Following a presentation about Predictive Parenting, the PPI panels reflected positively on the concept of Predictive Parenting, commenting that the theme of ‘predictability’ was meaningful to them, and that the focus on anxiety was highly relevant to their lived experience. Panel suggestions led to modifications of group design, content, and materials. For example, participants provided ideas for improvement around time-out, reward systems and relaxation approaches.

Objective 2: Evaluating the Feasibility of Predictive Parenting

The effectiveness of the Predictive Parenting intervention is being tested as part of the Autism Spectrum Treatment and Resilience (ASTAR) study. ASTAR’s main objective is to reduce mental health and behavior difficulties in children with ASD and forms part of a larger project aiming to improve the outcomes for people on the autism spectrum (Improving Autism Mental Health: https://iamhealthkcl.net/). A feasibility study to test the acceptability and accessibility of the program was initially conducted (intervention n = 12; active psycho-educational control n = 12) (method and outcome measures are described in detail by Palmer et al. 2019). Inclusion criteria comprised being a parent/carer of a child with ASD (4–8 years old), with sufficient English to participate. Exclusion criteria included current participation in another behavioral parenting intervention, frequent epileptic seizures (> once a week), significant safeguarding concerns or visual/hearing impairment (parent or child).

To qualitatively assess parents’ views on the concept, content, and delivery of Predictive Parenting, semi-structured telephone interviews (n = 9 completers, lasting 40–60 min) were undertaken by an independent research team. All interviews were digitally recorded and transcribed verbatim. Analysis of the summarized data followed a structured and systematic approach, supported by MaxQDA qualitative analysis software (VERBIsoftware 2016). Phone interviews were conducted with parents who declined the intervention (n = 10) and parents who dropped out (n = 1). Parent satisfaction was also assessed quantitatively post intervention using a 6-item self-report questionnaire, created for the study. Therapist fidelity to the intervention manual was self-rated by the facilitators after each session. The fidelity form rated whether each aspect of content (between 8 and 14 items) had been covered by facilitators. Items were scored on a scale of 0–2 (0 = not covered; 1 = partially covered; 2 = fully covered) and a % fidelity rating was calculated for each session.

Results

Online Appendix 1 summarizes the demographic and descriptive characteristics of participating families. 11 of the 12 parents had a son with autism (mean age verbal group: 90.17 months (79–105); minimally verbal group: 84.00 months (49–110). Participating parents, from three South London boroughs, were diverse in terms of their socio-economic status; a quarter were of mixed / multiple ethnicity and one third had a household income of below £20,000. None of the parents in the minimally-verbal group were in paid employment, compared to 83% in the verbal group. Children in the minimally-verbal group presented with more severe ADOS scores and lower levels of adaptive functioning than those in the verbal group.

The intervention was delivered with strong fidelity to the manual. Average fidelity for each session was 97.5% in the ‘Verbal’ group (range 91–100%), 97.8% (91–100%) in the ‘Minimally verbal’ group. Retention was good. Only two participants dropped out, both from the verbal group (owing to childcare difficulties / work commitments). On average, parents attended 9 out of 12 sessions (range: 6–12, excluding dropouts).

Key themes from qualitative interviews are summarized in Table 2 (verbal n = 5; minimally verbal n = 4). Reports were largely positive, with parents reporting that the ‘Predictive Parenting’ concept was relevant and easy to grasp. Parents noted that the strategies were reflective of their existing parenting approaches, putting a framework around techniques they had developed instinctively. They reported having a greater awareness and understanding of their child’s behaviors and triggers and greater confidence in their ability to manage meltdowns and their child’s anxiety. They also noted feeling calmer, with less self-blame and stress. Immediate child impacts were less clearly described by parents. However, parents did note some positive changes, including reduced meltdowns, improved cooperation, greater independence and reduced frustration. Table 2 also summarizes several suggestions for intervention refinement (e.g. needing more strategies around self-injury, a slower pace to allow home practice, more ‘hands on’ activities and practice. Critiques were used to inform a subsequent pilot RCT (n = 62), in which the effect of Predictive Parenting on parent and child outcomes was compared to a psychoeducational attention control condition (Palmer et al. 2019). Analysis of the pilot trial is underway.

Table 2 Results from the independent qualitative interviews
Full size table

Parent satisfaction questionnaires (n = 9) showed high satisfaction with the intervention (mean 3.89/4 sd 0.31), with no significant differences between the verbal and minimally verbal groups. All parents reported that the group felt “very supportive” and the majority felt it helped them manage behavior and anxiety more effectively (mean 3.78/4 sd 0.42).

Discussion

Previous studies have suggested that parent-directed interventions can lead to significant improvements in both symptoms of disruptive behavior (Tarver et al. 2019) and anxiety (Ung et al. 2015) in children with ASD. This report describes the development and feasibility of ‘Predictive Parenting’, the first group parent intervention targeting difficulties with behavior and anxiety in the same program. This focus is needed, given the frequent co-occurrence, interplay and persistence of behavioral and emotional difficulties in this group (Simonoff et al. 2008; Storch et al. 2012). It was designed in consultation with a PPI panel of parents of young people with ASD and adults with ASD, using their lived experience to guide focus and content.

Predictive Parenting has been designed to have relevance for parenting any child with ASD, regardless of their level of language, cognitive ability or current difficulties with behaviors and emotions. It encourages a parenting ‘style’ that addresses current challenges but also seeks to prevent future difficulties with anxiety and behavior. The intervention draws together a number of existing parenting approaches within an ASD-specific framework. It draws on the clinical observation and the proposed theory that children with ASD often struggle with prediction (Pellicano and Burr 2012; Sinha et al. 2014).

This report presented the findings of a feasibility study (n = 12) of Predictive Parenting, carried out with two parent groups (one for verbal, one for ‘minimally verbal’ children). The results were promising in terms of parent attendance, drop-out rates, and quantitative satisfaction measures. Facilitators’ fidelity to the intervention manual was high, along with quantitative measures of parent-satisfaction.

In their qualitative feedback, parents reported that this unifying theme of ‘prediction’ allowed them to better understand their child’s behaviors and emotions. It aligns with approaches that many parents use instinctively, such as incorporating structure and predictability (O'Nions et al. 2018; Schaaf et al. 2011). In terms of content, participants reflected that they had found the ‘behavior prediction’ skill particularly helpful, especially for children with limited language. Strategies in branch 1 align closely with functional analytic approaches, which often form a key foundation of behavioral parenting approaches in typically developing children (Webster-Stratton and Reid 2010) and children with ASD (Bearss et al. 2018). Similarly, branch 2 brings together existing behavior management approaches that are well-evidenced and popular with parents of children with ASD (Bearss et al. 2015; Whittingham et al. 2009a).

In the final branch of Predictive Parenting, evidence-based emotion-regulation strategies for children with ASD (Weiss et al. 2018) (e.g. sensory strategies; emotional literacy; graded exposure) are framed within the context of difficulties coping with ‘unpredictability’ and the anxiety this can cause. Parents are well placed to help implement CBT strategies to manage anxiety, with a key role in reducing avoidance and accommodation, modeling coping, encouraging relaxation and encouraging relaxation and implementing exposure and reinforcement techniques (Thirlwall et al. 2013; Wood et al. 2009). Parent-mediated strategies targeting children’s ‘intolerance of uncertainty’, such as gradual exposure to change and unpredictability, may also have a beneficial impact on anxiety for children with ASD (Rodgers et al. 2017).

Following this feasibility study, adaptations were made in line with qualitative feedback and PPI panel recommendations. For example, groups for parents of minimally verbal children have been further refined, with greater emphasis on strategies such as functional communication, objects of reference (McLarty 1997); the immediacy of rewards or consequences; and managing dangerous behavior and self-injury. An even greater focus has been put on active tasks, coaching, and discussion, rather than didactic teaching.

As a feasibility trial, the current study was subject to several key limitations. First, the PPI panel was small, and weighted towards adults with ASD rather than parents. It would have been helpful to have a greater number of parents involved in the initial concept development. The participant sample size was also small and quantitative parent and child outcomes have not yet been explored in comparison to a control condition. The measures of fidelity and satisfaction reported here have not been independently validated. However, given the promising qualitative feedback to date, a pilot randomized control trial of Predictive Parenting is now underway (https://www.isrctn.com/ISRCTN91411078), exploring the efficacy of this program compared to an active control condition (Palmer et al. 2019). This will allow more quantitative investigation of the program’s effects on both parent measures (e.g. parenting behaviors and confidence) and child measures (e.g. disruptive behaviors and anxiety). The current groups have been designed for younger children (aged 4–8 years) in the first instance. However, the themes of the course are equally applicable in later childhood and adolescence, a time of significant physical, social and environmental change and unpredictability. Further research is currently underway to explore the application of the Predictive Parenting concept for older groups of children.