Search Procedure

Electronic data records from the inpatient unit, available for all patients admitted between 1997 and 2020 (n = 630), were reviewed to identify patients who received a functional analysis targeting elopement as a behavior of clinical concern during their admission. The following keywords were used to search the patient record database: “elop” (to be inclusive of all variations of the word elopement), “run,” “bolt,” and “AWOL.” Participant inclusion criteria were: (1) having received an FA targeting elopement; (2) having received a treatment evaluation for elopement, regardless of treatment effects; (3) a SCED (e.g., a reversal or multiple baseline design) was used that could permit the demonstration of functional control, regardless of whether it was truly demonstrated; (4) interobserver agreement data (IOA) being available for a minimum of 20% of treatment sessions; and (5) having a diagnosis of autism. The search identified 41 individual patient cases where elopement was reported to be a target behavior. Application of the inclusion criteria resulted in a final sample of 14 individuals with a treatment targeting elopement (see Fig. 1 for a Participant Inclusion Flowchart with reasons for exclusion). Two treatment evaluations were completed for Participant 11 (P11) and three for P14 to target individual functions identified within FAs. This resulted in 17 treatment applications across 14 participants (Figs. 2 and 3 show graphical outcomes for P1-2, Supplemental Figs. 4–16 show graphical outcomes for P3-14). Treatment outcomes for P10 (Fig. 11) were also reported in Frank-Crawford et al. (2024).Footnote 1 For clarity, participants’ treatments will be referred to using their participant number (i.e., Tx 1–14), P11’s two treatments will be referred to as Txs 11(1) and 11(2), and P14’s three treatments will be referred to as Txs 14(1), 14(2), and 14(3).

Fig. 1

Participant Inclusion Flowchart

Fig. 2Fig. 3Participants

Participants were 14 children (13 males, 1 female) aged 5- to 21-years-old (M = 11.43, SD = 4.24) who were admitted for short-term treatment in an inpatient hospital unit for the assessment and treatment of severe challenging behavior displayed by individuals with IDD. Participants were typically referred for the assessment and treatment of multiple topographies of challenging behavior (e.g., aggression, self-injury) in addition to elopement, and were admitted for 3 to 8 months (M = 4.86 months, SD = 1.61). All participants had a previous diagnosis of autism and ID. A review of clinical records found that two had mild, three had moderate, four had severe, two had profound, and three had an unspecified ID (see Table 1). Although all participants received FAs and treatment evaluations for multiple topographies of challenging behavior, the present study reports only on the FA and treatment outcomes for elopement.

Table 1 Participant Demographics, Adverse Outcomes Due to Elopement, and Elopement FA ResultsSettingAssessments and Treatment Evaluations

The locations of assessments and treatment evaluations were chosen to facilitate opportunities to elope safely while minimizing the necessity to physically block elopement. Assessments were conducted in the same settings as treatment evaluations for P1, 3–5, 13, and 14, and in different settings for the remaining participants. Sessions with P1 were conducted in a large auditorium with closed, unlocked doors, allowing opportunities to safely elope from the treatment team without requiring physical intervention. Sessions for P3 were conducted in a classroom located off the unit that presented the opportunity to elope safely into a connected, enclosed room. Sessions were conducted in a conjoined session room for Ps 4–5, 13, and 14.

The FAs for P2 and P8–10 were conducted in a large open hallway on the treatment unit with a free-standing mat at the end of the hallway to create a visual barrier and prevent other patients from entering or exiting the area. Most treatment evaluation sessions for P2 and P12 were conducted off the hospital unit (see Generality Sessions). Treatment evaluations for Ps 8–10 were completed while walking around the unit without any mats obstructing their path, and as many doors as possible were open to allow increased opportunities to elope. P6. Assessments for P7 were completed in a conjoined treatment room, whereas the treatment evaluation was completed by walking around the unit with bedroom and activity area doors open. All sessions for P11 were conducted in a large open room that contained a couch, chairs, and a small table.

Generality Sessions

Generality of treatment effects were tested by moving treatment evaluation sessions from the original testing area to other areas on the hospital unit, new adults, application with parents, and out into the community (i.e., surrounding hospital public spaces, parks, restaurants, trips home, and stores). Generalization processes varied across treatments as a function of length of hospital duration and availability of parents to implement sessions. Generalization of Tx 2 began early after major initial reductions in elopement attempts were observed, though rates were continuing to vary between 0.1 and 0.9 rpm. Generalization sessions were conducted in a contained open tunnel system in the associated hospital, and then broadened to hospital hallways, cafeterias, courtyards, on community outings, and with parents once rates became more stable. All treatment evaluation sessions for Tx 12 were completed outside the treatment unit in surrounding buildings and the community to address elopement specifically in the context of transitioning. Generalization of Txs 1, 6, 8, 13, 14(2), and 14(3) was evaluated in the community (i.e., stores, park, restaurants) only after low, stable rates of elopement were observed in treatment sessions on the inpatient unit. Txs 3, 4, 5, 7, 9, 10, 11(1) and 11(2) were transitioned out of the initial treatment room or unit hallway in which the treatment evaluation was initiated, into common areas of the hospital unit. One generalization session was completed at the conclusion of the evaluation for Tx 14(1) to determine if elopement would occur in the community because it did not occur in the hospital unit. Generalization of Txs 14(2) and 14(3) was conducted with P14’s mother in the community and was scheduled based on when the mother’s availability to conduct sessions.

Measures

Elopement was generally defined as moving more than 3 ft from a caregiver without permission or leaving a confined area without a caregiver, consistent with previous definitions in the elopement literature (e.g., Lang et al., 2010). Definitions for P1, 4, 5, 6, and 14 included more specific qualifiers to their operational definition. Elopement was not recorded for P1 unless it occurred after 5 s walking independently beside the therapist. For P4, elopement also included attempting to or successfully opening the door to a room by grabbing the door handle. For P5 and P14, elopement was also scored when they placed at least 1 ft past the threshold of the door to exit the room in which all sessions occurred. For P6, elopement was defined differently across contexts. In an open area (e.g., hallway), elopement was defined as moving beyond 3 ft from a therapist. In a closed area (e.g., a room), elopement was defined as attempting to or successfully running towards or going beyond an open door. Rate, defined as elopement attempts or successes per minute, was the primary dependent variable across treatments except Tx 10, which tracked latency to elopement (the duration from session start to elopement attempt) and rate. To facilitate comparison across treatments, all data are reported as rates. We assessed the number of treatments that achieved an 80% or greater reduction in elopement, the inpatient unit’s standard minimum goal for challenging behavior reduction. In addition, treatment evaluations were conducted over the timeframe of a patient’s hospital admission, and discharge dates often prevented opportunities for treatment evaluations to continue until a 100% reduction was achieved.

ProceduresData Collection and Interobserver Agreement

Data from FAs and treatment evaluations were collected using BDataPro software (Bullock et al., 2017) on laptop computers or using paper data sheets. Data from all generalization sessions were collected using paper data sheets.

All assessments and treatments were implemented by behavioral treatment teams, comprised of one primary therapist and two backup therapists, under the direct, daily supervision of faculty-level board certified behavior analysts (BCBAs) with master’s or doctoral-level training in behavior analysis and extensive experience in the assessment and treatment of severe challenging behavior in the IDD population. Thus, all assessment and treatment components were selected based on the discretion and expertise of the supervising BCBAs and in consultation with the caregivers for each participant. All team members had intensive training in core treatment procedures (e.g., extinction/planned ignoring, least-to-most prompting, response blocking, DR), which were standardized by unit-wide protocols.

To promote high IOA, secondary observers were required to be trained to a 90% or higher agreement criterion with primary observers during FAs, treatment evaluations, and generalization sessions. IOA was collected for a minimum of 20% of sessions for all participants (M = 39.89, range = 20.00%–63.00%). The mean exact IOA of elopement per minute, across participants, was 96.8% (range = 91.81%–100%).

Functional Analysis

FAs were conducted using multielement (n = 12), reversal (n = 1), or combined reversal/multielement (n = 1) designs. All FAs measured rpm of elopement except for Tx 10, which measured latency to elopement after session start. Sessions were 5 or 10 min in duration. Across FAs, elopement received programmed contingencies, aggression and self-injury were blocked in the absence of verbal attention, and all other challenging behaviors were ignored. Conditions were selected for inclusion based on information gathered from intake paperwork and open-ended behavioral interviews completed with the parents by the BCBA at the time of admission. Within the FA, at least two test conditions were compared to one control (toy play) condition: attention, demand, ignore, alone, and tangible.

In the attention condition, therapists diverted their attention from the participant, and contingent on elopement, attention was provided in the form of verbal reprimands without response blocking for 30 s. In the tangible condition, the therapist and participant were present in an area with preferred items or activities out of the participants’ reach by at least 3 ft (Txs 1–2, 8–11) or in a connected room (Txs 3–7, 12–14) but within their line of sight. Contingent on elopement to the item or activity, the participant received 30 s access before being prompted to stop interacting with the item. In the demand condition, the therapist presented task demands (i.e., schoolwork or chores) to the participant using least-to-most prompting. Contingent on elopement, the therapist removed the task and ignored the participant for 30 s before presenting a new demand. The 30 s reinforcement interval for all social-reinforcement test conditions was removed from the total session time in calculating the rate of elopement. For the tangible and demand conditions, resetting procedures were implemented following reinforcement that consisted of using least-to-most prompting to guide participants back to the original location, except for P1 and P11 whose FAs were conducted in large open areas where items were re-presented to the participant in the location they were in following reinforcement. Resetting intervals were also excluded from rate calculations. In the ignore condition, the therapist was present in the area with the participant but ignored the elopement. The alone condition was implemented when the supervising BCBA considered it safe to leave the participant alone in a room to observe whether elopement continued when no social contingencies were in place. In the control (toy play) condition, the therapist and participant were present in the same area, the participant had continuous access to preferred items, activities, and the therapist’s attention, the therapist did not place any demands on the participant, and elopement was ignored.

One specific modification worth noting occurred in P3’s FA, which conducted all five tested conditions (attention, tangible, demand, ignore, toy play) in a divided room with two chairs in each section. In this FA, a tone was played every 40 s to signal to the therapist to physically guide P3 to sit in a chair on Side A. If P3 eloped to Side B, or continued to move out of the room, he was physically guided back to Side A to sit in a chair at the sound of the tone. If P3 was already seated in a chair on Side A, or had returned to Side A independently, he was physically guided to sit in a different chair on Side A at the sound of the tone. This procedure was used to control for the effect of physical attention across conditions.

A minimum of three series of all included conditions were completed for each participant, and sessions continued until clear differentiated responding was observed. The function of elopement was determined by supervising BCBAs using visual inspection procedures as outlined by Hagopian et al. (1997) or Roane et al. (2013) when fewer than 10 data points were available (see Table 1). For an exemplar FA graph from P7, see supplemental materials Fig. 17.

Competing Stimulus Assessment

A competing stimulus assessment (CSA; Ahearn et al., 2005; Fisher et al., 2000; for a review, see Haddock & Hagopian, 2020) was conducted with Ps 2, 3, 4, 6, 9, 10,12, and 13 to identify objects (i.e., toys, tablets, sensory items) that produced high levels of engagement and low rates of elopement. CSAs measured latency to elopement for P10 and rpm for all remaining participants. Items were selected for inclusion in the CSA based on behavioral observation, parent interviews, and clinical record review. One item was presented per session in a multielement design and levels of elopement were compared with a control condition in which no stimuli were present, except for P3 who was exposed to items in a pairwise format. Sessions were 5 or 10 min in duration, and the contingencies for elopement were the same as those used in the FA condition that produced differentiated levels of elopement relative to the control.

P6 received two CSAs: one to identify competing items while performing seated activities and a second to identify competing items during transitions. In the CSAs for P6 and P13, after observing low engagement rates during initial sessions, adult attention was provided contingent on the absence of elopement, which resulted in higher rates of engagement with competing stimuli. Synchronous attention was also provided for P2 but did not have a differential effect on engagement with competing items. Therefore, edible reinforcers were delivered every 5 s without an elopement attempt, and engagement with competing stimuli increased. Ultimately, 1 to 3 items identified through CSAs were added to subsequent treatment evaluations across participants.

Stimulus Avoidance Assessment

A stimulus avoidance assessment (Fisher et al., 1994) was conducted with Ps 2, 3, and 12. Prior to implementing the assessment, procedures were described to parents in detail, and consent was obtained. Parents were requested to complete a social validity rating scale to identify any procedures they would not find acceptable, which were then excluded. Procedures potentially assessed included a baskethold, a chair time-out, a visual screen time-out, contingent demands, and hands down (see Fisher et al., 1994, for a description of these procedures). Procedures approved for testing were presented across conditions in a multielement design. Sessions were 10 min and video recorded to ensure appropriate oversight and reliability coding. Contingencies mimicked those in the ignore condition from an FA, except that the targeted procedure was implemented noncontingently every 1 min for 30 s. To avoid adventitious reinforcement, the procedure was delayed by 5 s if the participant attempted to interact with the therapist. Following each session, participants were provided a 2 min break with preferred activities. The duration of negative vocalizations and avoidant behaviors, and frequency of elopement attempts, were recorded as dependent variables. After completing the assessment, a procedure was selected for inclusion in the treatment package if it was deemed socially acceptable by the patient’s family and was likely to reduce elopement based on rate of elopement attempts recorded during individual sessions.

Treatment Evaluations Experimental designs

Each participant received an individualized treatment evaluation for elopement. Of the treatment evaluations that were included, the following designs were implemented: 10 used a reversal (Txs 1, 4, 5, 7–13), two (Txs 2, 3) used a reversal with an embedded multielement design, one (Tx 6) used a multiple baseline across reinforcement types, and three (Txs 14[1–3]) were compared through a multiple baseline across functions design. A minimum of three sessions were conducted during baseline, in which elopement received reinforcement from the therapist conducting the evaluation if a social function was identified, or no contingencies when an automatic function was identified. Once the treatment was introduced, a minimum of two sessions were conducted before reversing back to baseline or introducing a new treatment condition. Session duration started at 5 min for seven treatments (Txs 4, 5, 7, 10, 11[1], 11[2], 13) and 10 min for 10 treatments (Txs 1–3, 6, 8, 9, 12, 14[1]–[3]). Case records indicated that session duration was extended for Txs 4 (15 min), 5 (5.5 min), 11(1) and 11(2) (30 min), and 8, though the final duration for Tx 8 was not reported.

Treatment components

Assessment results (e.g., FAs, CSA) and evidence from the behavior-analytic elopement literature were used to select treatment components. Antecedent and consequence-based procedures were used across treatments. Treatment components were added incrementally, starting with antecedent or antecedent and reinforcement-based procedures in 14 treatments and progressing to add additional treatment components if the initial treatment did not result in an 80% reduction in elopement. The exceptions to this were Txs 9, 10, and 12, which included response blocking in the initial treatment phase, which was faded out in the final treatment phase of Txs 9 and 12. In some cases, punishment procedures were added after antecedent and punishment procedures alone did not yield an 80% reduction in elopement. Punishment procedures had to be implemented by following clear pre-written implementation guidelines that were standardized for the hospital unit, and their use was closely monitored by supervisors. Individual punishment procedures are described below. Descriptions of individual components and how they were used are included below. For a tabular display of treatment components included in final treatment packages, see Table 2.

Table 2 Individual Intervention Components included in Final Treatment Packages Noncontingent reinforcement

Noncontingent reinforcement (NCR) was implemented as a treatment component for one treatment targeting an automatic function, four targeting tangible functions, one targeting an attention function, and one targeting multiple functions. Across treatments, NCR was delayed by 5 s if elopement occurred to avoid adventitious reinforcement. NCR involved a choice of multiple competing items identified through a CSA, on a fixed-time (FT) schedule. Items were available noncontingently while the participant was walking with the therapist (Txs 2, 9, 13) or while spending time in a large open space (Txs 11[1], 11[2]). NCR with moderately preferred items was used in Txs 10 and 14(2) to help participants tolerate when highly preferred items were unavailable. NCR schedules were faded in Tx 9 from FT-10 s to FT-30 s.

Hand Holding

Therapists would lightly hold the participant’s hand in Txs 7–10 and 12 while walking. In these instances, hand holding was not used as a form of response blocking, and treatment documentation noted that contingent on an attempt to pull away their hand, the therapist would let go. Hand holding was included in all treatment phases of Txs 7, 8, 10, and 12. Tx 9 added hand holding in the final treatment phase after finding that noncontingent reinforcement alone did not result in sufficient response reduction.

Differential reinforcement

DR was used in four treatments targeting an automatic function, two targeting a tangible function, and one targeting multiple functions. DR with a resetting interval was used for walking within arm’s reach of the therapist (Txs 1, 2, 8, 12) or waiting for a set period (Tx 5). Compliance with demands (i.e., schoolwork, chores) without elopement was reinforced with a token in Txs 3 and 4. The fixed interval DR schedules were thinned systematically to increase the interval of time participants were required to walk safely without eloping to receive reinforcement in Txs 2 (up to 1 min) and 8 (up to 5 min), as was the fixed ratio DR schedule for Tx 3 (continuous reinforcement to an FR10). A choice of reinforcers was provided in all treatments except Txs 8 and 12, where a single edible reinforcer was presented that was identified as the most preferred snack through a preference assessment.

Multiple schedules

Multiple schedules of reinforcement were used in one treatment targeting an automatic function, six targeting a tangible function, and one targeting an attention function. Components of the multiple schedules were signaled using a stimulus card (e.g., red/green) worn on a lanyard by a therapist while transitioning (Txs 1, 10). Txs 7, 11(1), 11(2), 14(2), and 14(3) used red/green board presented on a table during seated activities. Reinforcers were available contingent on a communicative response in the presence of the green stimulus and unavailable in the presence of the red stimulus and the contingencies were explained verbally at the start of each session. In Txs 1 and 5, a red/green card was used to signal that a resetting DR of walking within arm’s reach of the therapist (green, Tx 1) that was reinforced with a 20-s break (red, Tx 1), or waiting in a room without access to a preferred reinforcer (red, Tx 5) that was reinforced with access to the reinforcer (green, Tx 5). Tx 10 used a red/green board to signal the availability of opportunities to look at a specific object of interest. Requests to look at the object were honored when a picture icon representing the object was placed on the green side and ignored when it was on the red side.

Schedule thinning was implemented in Txs 7, 11(1), 11(2), 14(2), and 14(3) to increase the amount of time participants could tolerate the unavailability of reinforcers. Elopement attempts needed to be reduced by a minimum of 80% for at least one session to meet criteria for thinning to the next level across treatments and was decided at the discretion of the supervising BCBA. The duration of the extinction components was increased to 2 min in Tx 7, 10 min in Tx 11(1), 9 min in Tx 11(2), and 5 min in Txs 14(2) and 14(3).

Chained schedules

Chained schedules of reinforcement were used in two treatments targeting automatic functions. In Tx 3, if P3 placed tokens earned through DR of academic task completion, as described above, on a token board and handed it to the therapist, he received a small piece of food. This contingency was kept in place after introducing a choice in reinforcement and the choice to request his reinforcer using a functional communication card (see above). In Tx 4, the completion of an individual work task (e.g., matching a letter) was reinforced continuously with a token being placed on P4’s token board. Once P4’s token board had been completed twice, compliance without reinforcement was reinforced with a break for the rest of the session. Once consistent task engagement in the absence of elopement was observed, token reinforcement was thinned to an FR5 schedule and sessions were increased from 10 to 15 min. After successfully maintaining low rates of challenging behavior for an additional 10 sessions, the response requirement was increased to an FR 15 schedule, and the reinforcement period was increased to 5 min.

Extinction

Extinction was implemented in one treatment targeting an automatic function, six targeting tangible functions, two targeting attention functions, and one targeting multiple functions. Across all treatments, therapists actively observed to monitor and ensure safety while refraining from commenting on the behavior or showing an emotional reaction. In most cases, extinction was used in combination with an antecedent-based treatment. Extinction was trialed in Txs 3 and 12 but then replaced with a punishment procedure.

Functional communication training

FCT was conducted to teach the participant a communicative response in five treatments targeting a tangible function, two targeting an attention function, and one targeting multiple functions. Two treatments targeting an automatic function also incorporated general communication training, in which participants were taught a communication response to access competing items. Participants were most frequently taught to hand a picture card to the therapist to gain 30 s access to a preferred item (Txs 6, 7, 14(1)–14(3)) or attention (Txs 4, 11(2). Different approaches were used in Txs 10 and 13, where the reinforcer motivating elopement varied. In Tx 10, an omnibus mand to “go see” something in the environment was taught. In Tx 13, an initial attending response was taught to P13 to request therapist attention and access their communication device, which he then used to request reinforcers. Later, P13 kept the communication device by wearing it on a strap. Across all treatments that included FCT, a least-to-most prompting hierarchy was used to teach the communication response which occurred before adding FCT into elopement treatment evaluations to ensure mastery.

Delay tolerance training

In Tx 10, once a low and stable rate of elopement was established following FCT, a delay tolerance was introduced to teach P10 to tolerate waiting for reinforcement for up to 5 min. Then, firm denials of functional communicative requests to “go see” were introduced on a variable basis to prepare P10 for times in their natural environment when reinforcement would be unavailable.

Social skills teaching

Social skills teaching was used in one treatment targeting a tangible function, specifically access to preferred ways of playing activities. It was introduced with P14 before the third treatment phase to increase tolerance for not being able control how activities were played. Specific skills targeted were sharing with others, waiting for a turn, and accepting others’ ideas and behavioral skills teaching (modeling, rehearsal, role play with practice, and feedback; Parsons et al., 2013) was the instructional approach used. Once P14 successfully demonstrated targeted skills in over 90% of trials, this treatment component was incorporated into Tx 14(3).

Response blocking and harness use

Response blocking was included in one final treatment package targeting an automatic function, two targeting tangible functions, and one targeting an attention function. In Txs 2, 11(1), and 11(2), NCR and reinforcement-based procedures were trialed before introducing response blocking. In Tx 2, the treatment team introduced response blocking during the third treatment phase which produced modest decreases in elopement. Therefore, a harness with a long tether to limit how far the participant could elope prior to stopping the behavior replaced response blocking in the final treatment phase. Txs 9 and 12 began with the use of response blocking, but these were removed once consistent low rates of elopement were observed. In Tx 9, response blocking was introduced in the first treatment phase given that P9 was a large, 21-year-old adult whose elopement presented even higher safety concerns to self and others. Once stable and low levels of elopement were observed, response blocking was briefly removed, resulting in an increasing trend in elopement. In the third treatment phase, a prompt to hold an adult’s hand was added, with the adult letting go of P9’s hand contingent on an elopement attempt. This was used in combination with NCR to mitigate elopement attempts. Response blocking was included in the initial phase of Txs 10 and 12 to be consistent with behavioral procedures that had been occurring at home and school prior to admission. Response blocking remained in place for Tx 10, and in Tx 12, response blocking was replaced with an empirically identified punishment procedure (i.e., visual-screen time-out).

Baskethold

Following highly variable rates of elopement in four previous treatment phases employing reinforcement-based procedures, a 30-s baskethold was added during the final treatment phase for P3 targeting an automatic function. Here, a baskethold was implemented for 30 s by one or two therapists, after which P3 was prompted to return to the previous activity. The baskethold procedure remained in place as P3 was discharged. Baskethold procedures involved the therapist grabbing the participant’s forearms and crossing their arms in front of their own body, in a seated or standing position, to prevent elopement.

Visual screen time-out

A visual screen time-out, in which the therapist briefly held their hand or a screen (e.g., card) in front of the participant’s face but without touching the participant, to obstruct their view, was added to the final treatment phase of one treatment targeting an automatic function and one treatment targeting multiple functions. In Tx 2, three previous treatments were trialed relying on reinforcement procedures, with the third treatment phase also including response blocking. In the fourth treatment phase, a 30-s duration visual screen time-out was added after generalizing the treatment into the community and an increase in elopement attempts was observed. It was added into Tx 12 for a 1-min duration following one previous treatment phase that involved reinforcement-based procedures, hand holding, and response blocking. These procedures were initiated with the intent to phase out the number of punishment procedures included in the treatment package, however P12 was discharged prematurely before fading could be initiated.

Procedures for Testing Generality of Treatment Effects

Given the inherent concerns related to generality of treatment effects stemming from procedures implemented in a highly controlled setting, treatment teams followed several procedures to maximize opportunities to test for generality. First, lead therapists held weekly update calls with primary caregivers to describe current procedures being evaluated and gain their feedback on the feasibility of and buy-in for implementing these treatment procedures at home and in the community. Once consistently low rates of responding were observed and treatment packages were finalized, teams conducted generality sessions. Overall, generality sessions occurred in different settings (e.g., the community), and procedures involved a primary therapist running the session and one to two backup therapists who had previously established IOA with the lead therapist observing and collecting data using paper-based data collection sheets. Once treatment evaluations successfully reduced elopement attempts by 80% or greater and resulted in consistent, low rates of elopement, these same treatments were implemented by other therapists, direct care staff, and caregivers in the inpatient unit. Final treatment packages were also taught to community-based behavior therapists and teachers, if available. Supervising behavior analysts provided guidance to community-based treatment teams and families around systematic fading procedures to implement once consistent low rates of elopement were observed following discharge.

Case and Data Analysis

To characterize the sample, qualitative and quantitative data were collected from participants’ admission records regarding adverse outcomes associated with participants’ elopement (e.g., injuries, restrictions) prior to their admission. FA graphs were reviewed to determine functions of elopement identified by treatment teams. Authors reexamined FA graphs independently applying the Hagopian et al. (1997) or Roane et al. (2013) criteria and were in 100% agreement with identified functions across participants. To examine treatment evaluation outcomes, authors calculated the overall rate of elopement reduction for each individual treatment phase relative to baseline. To summarize elopement treatment effects across participants, the average response reduction was calculated across all treatment packages and for treatments grouped by function of elopement. Furthermore, for cases where treatment teams tested generality of treatment effects, clinical records were reviewed to characterize the location of sessions and types of caregivers involved, number of sessions conducted, and mean elopement rates across these sessions.