Evaluating a Treatment Package to Reduce Toe Walking and Improve Ankle Mobility in Children with Autism Spectrum Disorder: A Multi-Component Intervention

Abstract

This study evaluates the effects of a treatment package to reduce toe-walking (TW) in four male children with autism spectrum disorder, aged between three and six years, with persistent TW. The treatment package involved a combination of motor exercises, positive reinforcement, corrective feedback, and precision teaching. We used concurrent multiple probes across participants design, with RESA checks, to evaluate the effectiveness of the intervention. We measured the correct steps during probe sessions, the rate per minute of correct steps during the training, the ankle joint’s passive range of motion, and social validity, measured through caregiver questionnaires before and after the intervention. The results indicated a decrease in TW across all participants, with a beneficial effect on the participants’ gait patterns and overall physical mobility. It also underscored the potential for applying such an intervention to improve the daily lives of children with ASD. However, the study has several limitations, including not directly measuring generalization, the brief follow-up period post-intervention, and the absence of a component analysis to determine the differential effects of the intervention components.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40617-024-01035-8.

Although motor deficits are not part of the diagnostic criteria for Autism Spectrum Disorder (ASD), there is a consensus in the literature that individuals with autism frequently exhibit difficulties or motor atypicality (Mohd et al., 2021). An epidemiological study by Licari et al. (2020) reported that 35.8% of a sample of 2084 individuals with ASD also had a motor disability. The study of motor disabilities in individuals with ASD shows that these can become a significant barrier to the development of adaptive functioning (Bedford et al., 2016; Macdonald et al., 2013; Ohara et al., 2020). Motor atypia refers to unusual or atypical motor behaviors or movements that differ significantly from what is typically expected at a certain age or developmental stage. These can include variations in motor coordination, muscle tone, reflexes, and the execution of motor skills, which may not align with the conventional developmental trajectory. Motor atypias are often observed in individuals with developmental disorders, including ASD (Dewey et al., 2007; Provost et al., 2007). TW occurs when gait does not follow the heel-to-toe pattern. TW is not just limited to walking; individuals may also exhibit this behavior when standing and running (Valagussa et al., 2018). In cases where TW cannot be attributed to a specific medical condition, it is referred to as idiopathic TW (ITW; Williams et al., 2013). ITW is diagnosed in individuals who continue to walk on their toes beyond the age at which a heel-to-toe pattern is typically developed, which is between three to seven years (Shetreat-Klein et al., 2014). For a formal diagnosis of ITW, the TW behavior must persist for more than six months from its initial observation (Accardo & Barrow, 2015). TW has been found to be relatively prevalent among individuals with ASD (Wilder et al., 2020). Approximately 8.4% of individuals with ASD exhibit TW, compared to a prevalence rate of 0.47% in children with typical developmental patterns (Ming, 2007; Barrow, 2011). The severity of the diagnosis may influence the presence of TW in autism (Valagussa et al., 2022a, 2022b).

To date, the etiology of TW in autism is unknown; Weber (1978) stated that “toe-walking arises from the fixation of a normal transitory stage of development.” Other authors have proposed TW as a residue of a primitive gait pattern; others have suggested that it could be related to sensory disturbance or vestibular problems (Valagussa et al., 2022a, 2022b). TW is often observed in children with ASD due to sensory integration dysfunctions. These children may walk on their toes to manage sensory inputs and avoid uncomfortable sensations from the ground (Valagussa et al., 2018). Sensory-motor features should be considered in rehabilitation for TW in children and adolescents with ASD, as both groups exhibit sensory-related behavioral symptoms (Valagussa, Purpura et al., 2022a, 2022b). Persistent TW is associated with an increased risk of falls (Caselli et al., 1988). It has implications for aesthetic impact, functional abilities, quality of life, and musculoskeletal consequences, such as shortening of the posterior leg muscles or the Achille’ s tendon (Calhoun et al., 2011; De Oliveira et al., 2021; Pendharkar et al., 2008). The outcomes resulting from TW not only have medical consequences but also outcomes at a social level, with phenomena such as bullying, and affect the individuals’ motor and sports activities (Valagussa et al., 2022a, 2022b). Medical solutions for TW are available, but they are expensive and invasive (Perry et al., 2003). Since TW can emerge as a behavioral outcome in individuals with sensory processing sensitivities, professionals have targeted and implemented behavioral interventions to address underlying sensory integration challenges. However, research on the use of behavioral therapies, such as reinforcement-based techniques are needed (Hirst et al., 2019): despite extensive research, few studies have specifically focused on toe-walking in individuals with autism (Wilder et al., 2023). These interventions may involve movement, sensory perception, motor coordination, memory, attention, planning and organization of actions, expression of emotions, and behavior regulation (Barratt, 1959; Giangiacomo et al., 2022). Research has consistently shown that specific motor exercises can improve various aspects of walking. These improvements encompass faster walking speeds, greater stability, and better balance (Giangiacomo et al., 2022; Lichtsteiner et al., 2023; Martínez-Moreno et al., 2020). Given these benefits, employing targeted motor exercises could improve walking patterns in individuals experiencing gait and balance issues. Marcus and colleagues (2010) examined the effectiveness of a simplified habit reversal procedure incorporating a DRI procedure to reduce habitual TW in participants with ASD. The study introduced the use of GaitSpot Auditory Squeaker as part of the habit reversal procedure and found that it may be effective in reducing TW (Marcus et al., 2010); the limit of the study was, however, that treatment sessions were relatively few and individual sessions lasted only 10 min. Persicke and colleagues (2014) evaluated the effectiveness of the TAG™ procedure to reduce TW in a 4-year-old participant with ASD. The results showed a significant reduction in TW.

Hirst and colleagues (2019) aimed to demonstrate the effectiveness of using auditory feedback, rules, and reinforcement to decrease TW in a 5-year-old participant with no specific diagnosis but lower-than-average abilities. The treatment, which included differential reinforcement, auditory feedback, and verbal rules, effectively reduced TW. Still, the limitation of the study was that it relied in part on receptive language abilities. Many children who engage in TW may have other diagnoses that may limit their ability to engage in rule-governed behavior. Precision teaching (PT) has shown consistent benefits across various domains, including motor skills, by providing a clear and quantifiable measure of individual progress (Kapoor et al., 2023; Vascelli et al., 2020, 2023). PT offers a systematic approach to define and continuously measure behavioral characteristics precisely. PT utilizes the Standard Celeration Chart (SCC) not only to analyze behavioral data but also to underscore the efficacy of single subject design in demonstrating significant outcomes, thereby supporting decision-making processes to accelerate behavioral repertoires (Evans et al., 2021). To assess the fluency of targeted behavioral repertoires, performance is evaluated using the parameters of RESA’s components (Fabrizio & Moors, 2003); specifically, RESA—an acronym for retention, endurance, stability, and application (Fabrizio & Moors, 2003)—provides a comprehensive framework for evaluating the fluency of targeted behavioral repertoires. The history of RESA (Retention, Endurance, Stability, Application) begins with the work of Haughton (1980), who introduced an initial framework for evaluating fluent performance through the acronym REAPS (Retention, Endurance, Application, Performance Standards). This model focused on three main dimensions of learning: the ability to maintain a skill over time (retention), the capacity to sustain performance over long periods (endurance), and the application of skills to complex contexts (application) (Haughton, 1980).

Later, in 1992, Johnson and Layng refined this framework by distinguishing between two aspects that Haughton had originally grouped under endurance. They separated the ability to resist fatigue over time (endurance) from the ability to maintain performance in the presence of distractions (stability), thus creating the acronym RESA. This refinement allowed for a more detailed and precise evaluation of learned skills (Johnson & Layng, 1992).

The introduction of RESA provided a comprehensive framework for categorizing the outcomes of fluent learning and offered guidance for independently assessing each parameter. This approach facilitated the development of empirical methods to measure the ability to remember skills over time (retention), sustain performance without fatigue (endurance), maintain stability in distracting environments (stability), and generalize skills to new and untrained contexts (application) (Binder, 1996; Haughton, 1980; Johnson & Layng, 1992).

RESA has become an essential tool in behavior analysis and education, particularly in work with children with autism, where practical and functional learning is critical. This model has shifted the focus of educators and clinicians beyond mere accuracy, emphasizing the real-world utility, durability, and adaptability of learned skills (Fabrizio & Moors, 2003; Sundberg & Partington, 1998).

Achieving fluency requires meeting specific criteria in these areas: retention measures performance after a period without practice; endurance assesses sustained performance over extended periods; stability evaluates performance amid distractions; and application measures trained responses to untrained stimuli (Vascelli et al., 2024). In this study, we hypothesized that a package intervention comprising motor exercises, positive reinforcement, and corrective feedback would effectively decrease TW and enhance the range of motion in the ankle joints of young children with ASD, leading to improved walking patterns.

Method

Participants

The participants were four boys with ASD attending a weekly rehabilitative program at a behavior science-based rehabilitation center. All participants had a diagnosis of ASD according to DSM-5 criteria (APA, 2013) from local healthcare structures and exhibited TW (TW).

Luis, a 6-year-old boy, had basic motor skills appropriate for his developmental age but frequently walked on his toes, affecting his stability. He exhibited intellectual impairment and moderate language impairment, requiring moderate support in learning, communication, and movement. Socially, he interacted with peers but had challenges in sustained engagement.

Edward, a 5-year-old child, experienced motor coordination difficulties and walked on his toes, impacting his stability. He could correct his gait with adult guidance and had no sensory issues. He exhibited moderate language impairment without intellectual impairment, requiring minimal support in communication and motor skills. Edward showed social interest but had behavior issues related to waiting and frustration.

Paul, a 5-year-old child, had a consistent tip-toe gait, leading to instability and frequent falls. He tolerated foot contact only briefly while wearing shoes. Paul exhibited moderate intellectual impairment and severe language impairment, requiring significant support in learning, communication, and movement. He also had tactile, auditory, and visual hypersensitivity. Socially, he struggled with peer interactions due to his sensory sensitivities.

Andrew, a 4-year-old child, had significant motor, cognitive, and communicative difficulties, walking exclusively on his toes. We used the Portage (Glossop, 1989; Williams & Aiello, 2001), which offers personalized intervention planning and suggestions across five domains, to assess Andrew’s motor skills and praxis skills, or the ability to plan and execute coordinated movements. The Portage evaluation battery assesses specific areas such as motor planning, execution of movements, and coordination. The results indicated that his motor skills were equivalent to those of a 1.5-year-old. He exhibited severe intellectual and language impairment, communicating via augmentative and alternative communication (AAC) and requiring significant support in all areas. Andrew had substantial deficits in sensory processing and social interactions.

Participants were selected because of their pervasive TW, which affected daily activities, musculoskeletal structure, and compromised ankle joint range of motion. Reducing TW is clinically relevant for improving musculature and ligament flexibility. Previous studies (Nobile et al., 2011; Vilensky et al., 1981) documented specific reductions in ankle dorsiflexion and plantar flexion in children with ASD following interventions to reduce TW. Inclusion criteria also required regular attendance at the center and the absence of severe challenging behavior that would interfere with participation.

Setting and Materials

The study was conducted within the rehabilitative center during the hours of therapeutic sessions (Wednesday to Friday, between 10:00 a.m. and 3:00 p.m.). Participants attended the center two days a week, with each day’s session lasting 2 h. Observation and training sessions were conducted with other children in the room, each with their respective therapist.

During the pre and post baseline sessions, the participants were instructed to walk between two reference points located 5 m apart in the open area of a therapy room measuring 25 m². The setting comprised two visual targets (cones or chairs) that delineated the movement area for analyzing the spontaneous gait of the participants. The space surrounding the area destinated to baseline session was free of furniture and toys. The participant stood while the therapist positioned themselves to the side or behind the participant.

We conducted the training sessions in a section of a therapy room. The path was 6 m long, and the therapy room was free from external obstacles and toys. The therapist positioned themselves to the side or behind the participant. When present, a second observer would sit nearby. The motor path (Fig. 1) was constructed with cones and rods to create obstacles, undulating sensory surfaces, a soft mat, a seat, and three sensory tiles to assess walking patterns on different surfaces. The sensory tiles, measuring approximately 30 cm × 30 cm, were composed as follows: one with a rough surface made from kitchen sponges; the second with a smooth surface using plastic bags; and the third consisting of a transparent plastic bag filled with dried chickpeas. All these materials were arranged in sequence to allow participants to engage in a structured motor pathway designed to enhance their coordination, balance, and sensory processing skills. This setup aimed to systematically improve the participants’ ability to navigate different surfaces and obstacles, thereby promoting the development of motor planning, stability, and overall gait patterns.

Fig. 1.

The motor path

The materials used for the motor path had specific sensory characteristics and were utilized to foster the development of motor skills among children. These included tactile surfaces for foot placement, proprioceptive objects with various inclinations for walking, and vestibular challenges with obstacles for movement reorganization. We designed activities to promote motor stimulation, based on the developmental stages of autonomous walking in typically developing children. These included walking on different surfaces to promote the adaptability of walking patterns and sensory processing, overcoming obstacles to improve coordination and motor planning, and balance exercises to strengthen stability and posture. These activities have been specifically chosen to address the reduction of TW, which affects mobility and balance. All materials were free from toxic and volatile substances, possessed fireproof and CE certifications, and had been positively tested in accredited laboratories for resistance, toxicity, and safety. Additionally, they were designed to be easily washable and required simple maintenance. A stopwatch was employed during the training phase to measure the duration of training sessions. We used an orthopedic goniometer measuring 205 × 45 mm to calculate the degree of ankle plantar dorsiflexion. The instrument underwent periodic calibration sessions to maintain the consistency of the measurements. Participants’ performances were recorded using a video camera, and the number and frequency of steps were analyzed using pre-structured datasheets.

Measurement

The first dependent variable (DV1) was the number of correct steps the participants took while walking between two reference points at 5 m in the free part of a therapy room without the operator’s assistance. A correct step was defined as a step involving the correct sequence of placing the foot from the heel to the toe on the floor while wearing the shoes they wore during the therapy session. An incorrect response was defined as a step the participant engages in toe-walking, placing the toe on the floor before the heel or not completing the heel-to-toe sequence. A measurement session for DV1 consisted of a period of observation with variable duration, typically one to five minutes. The duration varied according to the time it took the participant to complete the task. Each session ended when the participant completed the target behavior. If, for example, the participant changed direction or stopped during the journey, then the measurement started again from the beginning. The second dependent variable (DV2) was the rate per minute of steps performed during the motor path. We designed it so that it could be completed in about a minute. We recorded the rate per minute of correct steps and errors. Each session included three opportunities for the participants to complete the motor path. The daily best timing was reported on the SCC. Additionally, we measured RESA’s components to assess the achievement of performance aims further. The third dependent variable (DV3) was the passive range of motion of plantar flexion and dorsiflexion of the ankle joint measured with a 360° two-arm orthopedic goniometer. We focused on dorsiflexion and extension (plantar flexion) of the tibiotarsal joint based on normative data that define a 20°–30° flexion ROM and 30°–50° extension ankle range of movement in extension. The gastrocnemius and soleus muscles and the Achilles tendon are mainly involved in the toe gait, frequently observed in children with ASD, making the tibiotarsal joint and flexion–extension movements in the transverse plane particularly relevant. The goniometer was positioned at the malleolus level in the neutral position of the foot, which coincided with the individual’s anatomical position in an upright stance. One arm of the goniometer was aligned with the orientation of the tibia. It remained stationary while the other arm moved along the axis of the tarsophalangeal joint, following the foot’s movement. During these measurements, participants were seated with their knees extended. This position allows for a more precise assessment of the ankle range of movement, minimizing potential sprains caused by instability or difficulty in balancing. The choice of seated posture with the knee extended for ROM measurements was guided by the representation of a critical phase of walking – the stance phase. The fourth dependent variable (DV4) was derived from a questionnaire developed by the first and authors. This tool collected caregivers’ insights on the toe-walking behavior intervention, focusing on its relevance, acceptability, and utility in the applied context. The questionnaire consists of ten questions with response options “yes,” “moderately,” or “no,” aiming to capture the caregivers’ perspectives on the effectiveness and applicability of the intervention measures. Appendix 1 lists the questionnaire used.

Data Collection

The experimenter (i.e., the first author) manually scored all sessions using a datasheet. For DV1, the datasheet included fields for recording the session’s date, and the number of correct and incorrect steps. Results were recorded on an equal intervals graph. For DV2, we also recorded the time needed to complete the motor path (duration) to calculate the rate per minute of correct and incorrect participants’ steps on the SCC, a graph that displays frequency on the y-axis across calendar days and provides both visual and quantitative analyses of behavioral phenomena and their relationship to environmental manipulations (Evans et al., 2021). Sessions took place two or three times a week. Data were collected for all sessions. Video recordings were utilized for calculating interobserver agreement.

The range of motion data was recorded at the beginning and at the end of the intervention for each participant on a decimal scale in a Word table. This table also incorporated normative values for the range of motion and documented the number of steps taken, distinguishing between correct and incorrect steps, and the timing of each step. This recording was used for assessing the participants’ progress, identifying areas requiring improvement, and monitoring their developmental trajectory. The results from the questionnaire were numerically transferred into a Word table for further analysis.

Experimental Design

We used a concurrent multiple probes across participants’ designs (Ledford & Gast, 2018) to evaluate the effectiveness of a combined motor intervention using a reinforcement-based procedure, to reduce TW. After one probe session, we introduced the first intervention tier for the first participants. The intervention ended after firmly reaching the participants’ aim (i.e., after four consecutive sessions with the rate per minute of correct steps equal to or above personal aim). We introduced subsequent tiers only after the intervention ended for the previous participant. We conducted probes roughly once a week for each participant.

Procedure

Pre-baseline Sessions

Before the intervention began, we assessed each participant’s specific aims, considering their developmental characteristics and motor abilities. This assessment involved measuring the performance of two groups of six children each, selected to match the developmental age of our participants: one group matched the developmental ages of Luis, Edward, and Paul, while the second group matched the developmental age of Andrew (1.5 years). The goal was to identify an achievable response level for each child, considering their motor skills. On the basis of the results obtained, we established the number of steps needed to walk between two reference points (DV1), and the performance aims to be achieved during the fluency task (DV2) by averaging the scores obtained. For DV1, we set a number of 18 steps for Luis, Edward, and Paul, and eight for Andrew. For DV2, we set the performance aim ranging from 18 to 20 steps per minute for Luis, Edward, and Paul, while for Andrew, with more limited motor skills due to their younger developmental age, the goal was set ranging from seven to eight correct steps per minute. This differentiation in performance aims was based on providing each child with a personalized aim that accounted for their specific abilities.

To identify stimuli with reinforcing properties, a formal preference assessment was not conducted, as each participant was allowed to freely select their preferred object from a box containing a variety of items commonly used by preschoolers, such as bubbles, tablets, animals, frisbees, and stuffed animals.

Baseline

We conducted baseline measurements for DV1 before, during, and after the end of fluency training. We conducted baseline sessions for DV1 before the fluency training and once a week during the intervention, one hour before the training, to avoid any practice effects. During the sessions, the therapist instructed participants to choose a preferred object and placed it at an endpoint 5 m from the starting point. The therapist then instructed the participants to walk from the starting point to the endpoint to retrieve the chosen object. The two points were arranged in a straight line, ensuring the number of steps required to cover the 5 m was consistent for all participants. If a participant met the target response criteria during baseline by walking from the first to the second reference point, they were allowed to retrieve the chosen object as a form of natural reinforcement. If the participant engaged in the number of target responses but had not arrived at the endpoint, they were instructed to continue until reaching the endpoint.

The therapist stood behind the child to guide him if he deviated owing to lack of attention or distraction. If the participant exhibited avoidance or escape behaviors, the trial ended. In this case, the therapist worked on the participants’ motivational operations and resubmitted the task. We included only complete sessions in the data collection to provide a comprehensive view of the participant’s behavior. Baseline sessions ended when the participants completed the task. We measured the ankle range of motion during the initial and final baseline sessions, following the guidelines described by Norkin and White (2016). Our protocol included standardizing the position of both the patient and the examiner, using calibration procedures for measuring instruments and repeating measurements to ensure consistency. Specifically, we used a two-arm orthopedic goniometer to assess ankle dorsiflexion and plantar flexion. The goniometer was positioned at the lateral ankle joint, with its arms aligned to the longitudinal axis of the femur and the metatarsal line. Each participant was seated, with their back resting against the back of the chair and their thigh fully resting on the seat; a second therapist stabilized the thigh at the knee level to isolate joints not involved in the measurement, thereby extending the leg on a flat surface to facilitate an accurate assessment of the ankle’s range of movement. The precise placement of the goniometer at the level of the malleoli, in a neutral position of the foot that coincides with the individual’s anatomical standing position, ensured the correct detection of the range of movement. One arm of the goniometer was aligned with the orientation of the tibia and held still. In contrast, the other arm followed the axis of the tarsal-phalangeal joint, thus monitoring the foot movement. We conducted three consecutive measurements for each assessment of dorsiflexion and plantarflexion, averaging the results to minimize measurement error.

Fluency Training

At the beginning of the session, the therapist asked the participants to select a preferred object and place it at the end of the motor path. The therapist instructed the participants to initiate the motor path. They first walked through three sensory tiles; the three sensory tiles to enhance their tolerance to different tactile sensations and promote balance adaptation (Karim & Mohammed, 2015; Laurie, 2022; Smith et al., 2015). Subsequently, the participants progressed to walking on a soft surface, which could be a cushioned or foam mat. This surface was chosen to challenge their balance capabilities, necessitating increased control and stability. After completing the walk on the soft surface, the participants advanced to walking on a wavy surface characterized by irregular patterns or a balance board. This segment of the training further tested their balance and coordination skills. Following the wavy surface activity, the participants encountered two obstacles that needed to be overcome: stepping over small barriers. This training component aimed to assess and improve the participants’ ability to navigate and negotiate obstacles successfully. Upon completion of the obstacle section, the participants walked freely toward a chair, indicating the end of the training session.

Timings began when the child started walking and ended upon reaching the chair. A correct response was defined as a step taken with the full sole of the foot on the ground, whereas an incorrect response involved toe-walking. Before initiating each step of the training, the therapist provided explicit instructions to the individual, such as “Keep your feet down.” The therapist provided praise, using phrases like “Good job!” following correct target responses, and offered corrective guidance for any incorrect behaviors; she positioned himself behind the participant and directed the child’s movement at the level of the legs or ankle by exerting posterior pressure so that the heel touched the ground. If the participant deviated from the path sequence, the therapist paused the stopwatch, took the child by the hand, accompanied him back, guided him to follow the path, and re-activated the stopwatch. The therapists identified the participants’ typical ways of expressing dissent, ensuring clear assent was obtained before applying manual force on their feet during the procedure. The training continued until each participant reached his specific aim for three consecutive sessions at least.

RESA

Once the participants reached their fluency aims and the data remained stable, the experimenters tested for retention, endurance, and stability. We conducted endurance and stability checks immediately after achieving the performance aims and retention checks about two weeks after the end of the training. The endurance check involved a practice session requiring participants to complete the motor path three times without stopping. Once they reached the end of the path, participants had to return to the starting point and then back to the endpoint again. We assessed stability by conducting a practice session in an environment with potential sources of visual and auditory distractions (e.g., toys, lights, sounds) present, and other users attending their therapy sessions. We conducted RESA sessions using the same methods as during the fluency training, recording the rate per minute of steps, and providing three opportunities for the target behavior to occur.

Social Validity

We administered a questionnaire to caregivers to evaluate the impact and perception of TW on the child’s daily life before and after the intervention. The questions asked caregivers about the extent to which they considered the TW behavior stigmatizing, whether there were any surfaces or situations in which it was most prominent (if so, the situation or surface should be reported), whether this behavior affected the participants’ activities, including play, and whether they considered it essential to work on the TW behavior. The questions were as follows: (1) Do you consider your child’s TW stigmatizing? (2) Does your child spend many hours a day on their toes? (3) Are there situations where TW is more prominent? (4) Are there surfaces where TW is more prominent? (5) Does the child show difficulty in letting their feet be touched? (6) Have you noticed an improvement in TW in the last two weeks? (7) Does TW affect your daily choices? (8) Have you ever given up activity due to TW? (9) Do you believe it is crucial to work on TW? (10) Have you considered doing something to improve your child’s TW? Each question had three response options (yes, moderately, no). The responses were numerically coded, with a score of 3 indicating “yes,” 2 indicating “moderately,” and 1 indicating “no.” The coding used was a quantitative indicator in which “yes” corresponded to the number 3, “moderately” corresponded to the number 2, and “no” corresponded to the number 1; the total score, therefore, was directly proportional to the severity perceived by the caregivers.

Interobserver Agreement and Treatment Integrity

We used 80% of the recorded observation sessions to calculate the exact agreement IOA (Vollmer et al., 2008). The mean percentage agreement for baseline sessions was 85% (range 81% to 98%), and during fluency training, the IOA was 97% (range 82% to 100%). Using this method, an “agreement” is scored if both observers count the same behavior instances. If they disagree, the interval is scored as a “disagreement.” The agreements are then divided by the total number of intervals and converted to percentages. We also calculated Total Duration IOA for 47% of fluency training sessions, with an agreement rate of 98% (range 95% to 100%).Two professionals were trained to observe the dependent variables to a level of accuracy of at least 90%.

To evaluate treatment integrity, the first author developed a checklist based on the necessary steps to implement the procedure. For the baseline phases, the specified checklist’s steps included: (1) ensuring correct placement of the start and end points at the appropriate distance, (2) accurately positioning the participant at the starting point, (3) initiating video recording, (4) verbally requesting the correct behavior, and (5) refraining from providing corrections. During the fluency training, the checklist’s steps were: (1) selecting the reinforcer, (2) positioning the participant at the beginning of the motor path, (3) initiating the timer and video recording, (4) providing verbal instructions with accompanying gestural cues for each step of the motor path, (5) reinforcing correct performance, (6) using the time delay strategy and providing assistance with physical guidance by standing behind and modeling the movement at the ankle level, (7) providing gentle redirection using verbal and gestural cues if a participant deviated from the expected path or sequence to guide them back on track without providing direct corrections, and (8) stopping the timer at the end of the path. To ensure consistency and accuracy in the implementation of the motor path, we considered several factors. First, the order of presentation of items was fixed for each session to maintain consistency, and this order was determined depending on the skill levels and needs of the participants. Second, a specific sequence was followed during each session to ensure that the participants completed the tasks in a structured and predictable manner. This approach was consistently applied to maintain the integrity of the training procedure.

The average Treatment Integrity for fluency training was 99% (ranging from 91 to 100%) for participant Andrew and 100% for participants Luis, Edward, and Paul.

To ensure the reliability of ankle range of motion measurements, two therapists with university training in motion analysis and measurement, typically entrusted to orthopedists and physiotherapists, conducted the assessments. We adopted a pre-study calibration process to harmonize their measurement techniques; we achieved a 90% agreement percentage during practice on individuals not involved in the study. The percentage of agreement between observers during baseline measurements was 99% (range 93–100%).

Results

We used visual analysis of graphical displays (Cooper et al., 2020; Horner et al., 2005; Lane & Gast, 2014) to identify the trend (progress over time), level (magnitude), and stability (bounce) of the data to evaluate participants’ performance on DV1 and DV2. Figure 2 depicts the number of correct steps performed by all participants during the baseline phases; Figs. 3, 4, 5, and 6 depict the rate per minute of correct steps during fluency training, and the results of retention, endurance, and stability checks. All participants increased the number of correct steps produced during baseline sessions after reaching their performance aim. The data relating to fluency training show that the rate per minute of correct steps increases, along with the reduction of incorrect steps, which decrease significantly (without, however, reaching values equal to 0), for all participants, except for Andrew, who, although decreasing the rate per minute of incorrect steps following the introduction of training, maintains a medium–high level of incorrect answers. Table 1 presents the range and mean values for the number of correct steps taken by participants during baseline measurements, specifically before, during, and after the introduction of fluency training. Before the fluency training, the level of correct responses during the baseline phases was low, with a stable trend across participants. However, after the training was introduced, the level of response remained low but showed improvement even before reaching the performance aims; the trend exhibited variability with multiple paths, indicating a therapeutic direction for improvement for all participants. Once the participants achieved their expected rate per minute aims, the baseline level of response improved further, with a consistently increasing and stable trend in the therapeutic direction. The percentage of non-overlapping data (PND) comparing the number of correct steps produced before and after achieving the performance aims was 100% for all participants.

Fig. 2.

Number of correct steps performed by all participants across phases

Fig. 3.

Results of fluency training and RESA checks for Luis

Fig. 4.

Results of fluency training and RESA checks for Edward

Fig. 5.

Results of fluency training and RESA checks for Paul

Fig. 6.

Results of fluency training and RESA checks for Andrew

Table 1.

Range and mean values for the number of corrected steps taken by participants during baseline measurements

Participant	Correct steps before fluency training		Correct steps during fluency training		Correct steps after fluency training
	Range	Mean	Range	Mean	Range	Mean
Luis	7	7	7 to 10	7.75	11 to 20	16
Edward	1 to 2	1.8	2 to 7	3.29	10 to 20	15.71
Paul	1	1	1 to 10	2.86	13 to 20	16.89
Andrew	0	0	0 to 4	1.09	6 to 8	7.5

Table 2 reports the number of fluency training sessions, the sessions needed to reach the personal rate per minute aims, celeration values (i.e., the change in the frequency of a behavior over time), bounce for performances (how rate of responding varies above and below the central trend line), calculated considering a 0.95 confidence interval, and endurance, stability, and application checks. The mean celeration value for correct steps across participants is X1.29, while the mean celeration value for errors across participants is /1.34. RESA checks indicate that Luis, Edward, and Paul have achieved fluent performance, as the values obtained are equal to or greater than the aims (except for Paul’s application check, which is slightly lower). However, the values obtained by Andrew are lower than its rate per minute aim.

Table 2.

Number of training sessions, session to aim, celerations, bounce values and RESA checks

Participant	Number of training sessions	Number of sessions to aim	Celeration Correct to aim	Bounce correct to aim	Celeration Errors to aim	Bounce errors to aim	Retention	Endurance	Application
Luis	16	10	X1.44	1.36	/1.63	2.43	20.00	19.78	20.00
Edward	19	16	X1.20	1.42	/1.43	3.84	18.00	19.00	18.35
Paul	27	24	X 1.25	2.05	/1.27	3.55	18.00	20.00	17.70
Andrew	27	23	X 1.25	1.76	/1.04	1.32	6.40	6.92	6.95

Table 3 presents data on joint range of motion. Concerning the impact of the intervention on the physical, structural component, the data showed an improvement in foot and ankle mobility following the intervention across participants. During initial measurements, Paul experienced hyperflexion of the plantar due to sensory intolerance to the goniometer and the touch of the foot. However, after the training, the participant’s tolerance to foot touch led to a clear improvement in the parameters related to foot mobility.

Table 3.

Joint ranges of ankle motion

Participant	Degrees of plantarflexion with extended knee (Initial measurements)	Degrees of plantarflexion with extended knee (Final measurements)	Degrees of dorsiflexion with extended knee (Initial measurements)	Degrees of dorsiflexion with extended knee (Final measurements)
Luis	25	30	5	8
Edward	20	30	5	9
Paul	10	30	5	9
Andrew	5	10	2	5

Social Validity

All four participants’ caregivers answered the social validity questionnaire (100% participation rate); the maximum score for each participant was 30 pt, for 120 pts. Before the start of the project, all four participants’ caregivers agreed (answering three corresponding to “yes”) that TW was stigmatizing and that an intervention aimed at reducing the behavior should be carried out; furthermore, concerning the question concerning the situations in which tip walking emerges most: the parents of Luis reported that it occurs in moments of intense emotion; the parents of Edward and Paul also reported an identical answer. The caregivers of Edwar and Paul reported an increase in TW when the two children walked barefoot and on surfaces such as grass or sand. About the questions concerning how much the toe-walking behavior affected the participants’ daily lives, the caregivers agreed that they answered 2 (i.e., “quite a lot”) but that it was not too limiting.

At the end of the training, the first caregiver scored 21 pts; the second 20 pts; the third 22 pts; and finally, the last caregiver 21 pts. In the post-intervention phase, there were no significant changes in the answers except for a few annotations for the surfaces on which the behavior emerges most: in fact, the caregiver of participant Paul reported that the surfaces remained unchanged but noted an improvement concerning the time spent on the toes and the angle of the foot, which is improved. The situations in which TW emerges with an emotional prevalence also remain unchanged.

The following was collected in detail: Luis scored 15 pts; Edward scored 20 pts; Paul scored 20 pts; Andrew scored 21 pts.

Discussion

In this study, we wanted to decrease TW during spontaneous walking and improve ankle joint range of motion in the gait of four young ASD through a treatment package combining motor exercises, positive reinforcement, and corrective guidance. The results of this study demonstrate that, for all participants, the toe-walking behavior reduced; as shown in Fig. 2, the number of correct steps in the baseline sessions increased for all participants. The effect of the intervention is replicated for all participants, suggesting the existence of a functional relation between the procedure used and the reduction of toe-walking. The results show that the effect is more pronounced for Luis, Edward, and Paul at the end of the intervention; they completed the baseline sessions by performing only the correct steps. Andrew maintained instances of toe-walking despite increasing the number of correct steps. It should also be noted that retention, endurance, and stability checks show performance declines, at least partly for Participant P and more significantly for Participant A; this suggests that the performance levels achieved after fluency training were not fluent for all participants.

The findings suggest that the intervention positively impacted the participants’ gait patterns and foot and ankle mobility: the training increased the joint range of motion in both plantarflexion and dorsiflexion, indicating improved ankle mobility.

Furthermore, the study assessed social validity by obtaining participant caregivers’ feedback. Caregivers acknowledged the stigmatizing nature of TW and supported the intervention aimed at reducing it. Questionnaire responses indicated that TW had some impact on the participants’ daily lives but was not overly limiting. After the second administration of the questionnaire, caregivers’ responses showed no significant changes, indicating the persistence of improvement in the participants’ behaviors.

This study has several limitations. First, a treatment package was implemented, which consisted of a combination of positive reinforcement, feedback, and a motor intervention. It is unclear to determine the differential effect of the components of the intervention on the target behavior. This is considered a limitation because understanding which components are most effective could lead to more targeted and efficient interventions. Future research should focus on conducting component analyses to determine the differential effects of the intervention components. Second, we did not measured variables related to balance, gait, and coordination. These are important aspects of motor performance that could influence the effectiveness of the intervention, and their absence limits the comprehensiveness of our findings. Future studies should incorporate these measurements. Third, we did not directly measure generalization in other significant settings; fourth, we conducted follow-up measurements only a few weeks after the end of the training. Generalization and long-term retention of skills are critical for assessing the sustainability of an intervention. Future studies could measure generalization in different settings than the one in which the training was conducted and conduct follow-up measurements after a more extended period from the end of the intervention. Fifth, participants did not reach a fluent performance: we observed a partial decay in performance during the RESA measurements. Fluent performance is essential for ensuring long-term maintenance of learned behaviors. Future studies may measure participants’ performance on toe-walking after reaching all the indicators of fluency. Sixth, the joint range was evaluated only during the first and last baseline sessions, lacking intermediate assessments, and remaining focused on structural rather than behavioral aspects; the social validity questionnaire lacked well-defined administration and scoring rules, resulting in a qualitative rather than a quantitative analysis. This limits the ability to track progress and make data-driven decisions during the intervention. Future studies should include more frequent assessments and develop standardized administration and scoring rules for social validity questionnaires. Seventh, the aims needed to be set to the correct mastery criteria. Accurate aim setting is crucial for achieving desired outcomes and ensuring participant success. Future studies could investigate a more accurate method of determining aims on a larger sample. Eight, to record instances of TW, we primarily relied on video recordings of participants’ performances; for future studies, a pedometer, an ankle-level gyroscope, heel position sensors could be employed to obtain more experimental and objective data on the number of steps, number of correct steps, and kinematic data related to the gait of children with ASD, such as walking speed and trajectory. Using more advanced technology could enhance the accuracy and reliability of data collection, leading to more precise findings. A starting point for future research could be integrating technology to have more accurate measurement methods and creating interactive and motivating psychomotor programs for participants. Ninth, the experimental design included the decision to introduce subsequent tiers of the intervention only after the previous participant met the mastery criteria. While this approach ensured that each participant achieved a certain level of proficiency before moving on, it may have created unnecessary delays in intervention delivery, potentially delaying the overall progression and impact of the intervention. This could limit the timely delivery of interventions to all participants. Future studies should consider introducing subsequent tiers into the intervention after the data shows a behavior trend in the desired therapeutic direction to promote the overall progression and impact of the intervention. Finally, the selected participants had heterogeneous characteristics and ages, although they all had the same diagnosis. Homogeneity in participant characteristics can provide more conclusive findings and enhance the generalizability of results. Future studies could measure the intervention’s effect on participants with more homogeneous functioning characteristics to draw more detailed and conclusive findings.

Conclusions

The prevalence of individuals with TW is exponentially increasing. Therefore, researchers and practitioners must start focusing on this aspect, investigating the phenomenon’s underlying causes and developing studies on rehabilitative approaches as alternatives to surgery. This study provides insights into the positive effects of a treatment package on toe-walking behavior and ankle mobility in children. Behavior analysts and therapists can use the procedures described.

Supplementary Information

Below is the link to the electronic supplementary material.

Data Availability

The data supporting the findings of this study are available within the article. All relevant data can be accessed upon request from the corresponding author.