Abstract
Functional communication training (FCT) is a widely used and effective function-based treatment for problem behavior. The purpose of this article is to present two cases in which FCT was unsuccessful in reducing the occurrence of problem behavior displayed by two young children with an autism spectrum disorder. Both children received the same functional analysis plus FCT treatment package via telehealth that had proven to be highly successful for the other participants. The FCT package was conducted within tightly controlled single-case designs for each participant, which permitted subsequent analyses to determine why FCT was unsuccessful. These analyses suggested distinct reasons for the treatment failure for each child. Although the negative results of treatment appeared to be similar for both children, the specific reasons for treatment failure were highly individualistic and identifiable via the single-case analyses conducted. We present findings from both our initial and subsequent analyses and discuss the implications.
Keywords
The combination of functional analysis (FA; Iwata, Dorsey, Slifer, Bauman, & Richman, 1982/1994) plus functional communication training (FCT; Carr & Durand, 1985; Reichle & Wacker, 2017) has been shown to be effective across populations (e.g., Fyffe, Kahng, Fittro, & Russell, 2004; Rooker, Jessel, Kurtz, & Hagopian, 2013), behaviors (e.g., Fisher et al., 1993; Fyffe et al., 2004; Mace & Lalli, 1991), settings (e.g., Kurtz et al., 2003; Northup et al., 1994; Rooker et al., 2013; Wacker et al., 1998), and service delivery models (e.g., Lindgren et al., 2016). FCT is the most widely studied and successful function-based treatment for problem behavior (Tiger, Hanley, & Bruzek, 2008), with early studies demonstrating its effectiveness at decreasing problem behavior and increasing appropriate communication (e.g., Carr & Durand, 1985; Fisher et al., 1993; Wacker et al., 1998) and more recent studies demonstrating how to further improve its long-term effectiveness (e.g., Austin & Tiger, 2015; Falcomata, Wacker, Ringdahl, Vinquist, & Dutt, 2013; Greer, Fisher, Saini, Owen, & Jones, 2016; Wacker et al., 2011).
In 1992, Wacker and Berg began evaluating the effects of FCT on the problem behavior of young children with developmental disabilities when their parents conducted all procedures with coaching from a behavior consultant in their homes (Wacker et al., 1998). Initial results showed that problem behavior was reduced for all children within the first 3 months of the FCT treatment and continued to decrease for those who continued in treatment for up to 12 months. These results have remained consistent over the 18-year course of these projects, with the average percentage of reduction of problem behavior being 90% across 84 children (Wacker et al., 2017). In addition, similar results have been shown when these same procedures were conducted via telehealth with the connection occurring either between a hospital-based teleconsultation center and community-based clinics (Wacker et al., 2013a) or a teleconsultation center and the families’ homes (Lindgren et al., 2016).
Despite the overarching success of FCT, there have been cases in which FCT has been shown to be unsuccessful. For example, Wacker et al. (1998) showed that 15 of 28 participants had at least a 90% reduction in problem behavior. If a 90% reduction in problem behavior is the criterion for treatment success, then FCT was unsuccessful for 63% of participants in the study. Similar results have been reported by other research teams (e.g., Durand & Kishi, 1987; Fisher et al., 1993).
Durand and Kishi (1987) showed that FCT had little effect on the problem behavior displayed by two of five individuals with severe to profound intellectual disabilities. It was hypothesized that one reason for the lack of treatment success was the treatment fidelity of the staff as they had not previously implemented the treatment recommendations. This hypothesis was supported for one of these individuals (Kim) when treatment results changed from showing no effect at the 1-month follow-up to showing improvement at the 9-month follow-up, which was correlated with staff consistently implementing the FCT treatment. Similarly, Fisher et al. (1993) showed that FCT may not always be effective or effects may not always occur as quickly as had been shown in previous studies.
Although studies such as Fisher et al. (1993) demonstrate the limits of FCT in some cases, most do not highlight problem cases which in turn leads to the “file drawer problem” (Rosenthal, 1979), whereby studies that obtained nonsignificant results are unlikely to be disseminated. The purpose of this article is to present two cases in which FCT was unsuccessful in reducing the occurrence of problem behavior displayed by two young children with an autism spectrum disorder. Both children received the same FA plus FCT treatment package via telehealth coaching procedures that had proven to be highly successful for most of the other participants in the project (Lindgren et al., 2016) when conducted via telehealth. Thus, there were no differences in procedures that could account for the negative findings, and the fidelity of the procedures was monitored as described by Suess et al. (2014). Although the overall treatment effects were not successful, the analyses conducted within single-case designs suggested different reasons for treatment failure, which would lead to different follow-up treatments. In this article, we present findings from both our initial and subsequent analyses and discuss the implications for current practice and future research.
Method
Participants
Two children who were initially referred to a larger federally funded project (Lindgren & Wacker, 2011) participated in this study. The purpose of the host project was to conduct a randomized clinical trial of FCT delivered directly into the families’ homes via telehealth. Following the completion of an FA, participants were randomly assigned to either an immediate or delayed treatment group. Participants in the immediate treatment group received FCT immediately following the FA, and participants in the delayed treatment group received FCT after a 3-month wait period. To participate in the host project, children had to meet the following inclusion criteria: (a) were between the ages of 18 and 83 months at the time of the diagnostic evaluation, (b) diagnosed with an autism spectrum disorder, (c) engaged in problem behavior that had an identified social function, and (d) lived within the state of Iowa at the time of enrollment. All parents signed written consent to participate in the study using procedures approved by the institutional review board for the protection of human subjects at the University of Iowa.
For the present study, the inclusion criterion was that the child’s problem behavior was found to be unresponsive to the FCT treatment, which was defined as a lack of (a) an 80% reduction of problem behavior from baseline over three consecutive FCT sessions, (b) 80% compliance with task requests, and (c) independent requests at appropriate times within 25 weeks of treatment implementation. Liem and Sara were the only two participants within the host project whose problem behavior was unresponsive to FCT. Liem was a 2-year 10-month-old boy diagnosed with an autism spectrum disorder. He communicated using gestures, and problem behavior included self-injury (e.g., head banging), aggression (e.g., hitting, pushing, kicking), and property destruction (e.g., throwing work items and toys). Sara was a 6-year 11-month-old girl diagnosed with an autism spectrum disorder. She communicated using gestures, and her problem behaviors were aggression (e.g., hitting, kicking) and property destruction (e.g., throwing toys).
All FA and FCT sessions were conducted in the children’s homes by their mothers who received coaching via telehealth from a behavior consultant who was located at the University of Iowa Children’s Hospital (UICH). Liem’s mother had some college experience, was a stay-at-home mom, and had no previous experience conducting behavior analytic procedures. Her home was 296 miles away from the teleconsultation center. Sara’s mother had a master’s degree, was a stay-at-home mom, and had no previous experience conducting behavior analytic procedures. Her home was 64 miles away from the teleconsultation center at the beginning of the project and 562 miles away after moving out of state.
Settings and Materials
Telehealth center
The behavior consultants provided all coaching from the telehealth center located at the UICH. The telehealth center, as described by Suess et al. (2014), had four teleconferencing workstations equipped with a Windows-based PC, video monitor, webcam, and headset. Each workstation was installed with Skype™ to transmit and receive live audio and video input from the children’s homes. Video recording software (i.e., Debut) was installed on each workstation to allow the behavior consultant to record each session. A 6-s audio track was embedded on all video recordings to facilitate data coding. Data coders were able to adjust, as needed, the speed and volume of the video recordings using playback software (i.e., VLC Portable Media Player). Specific technological features of the telehealth center were described by Lee et al. (2015), and all parents in this study provided consent for the use of Skype™ for teleconferencing.
Participant’s homes
The behavior consultants assessed the parents’ equipment needs at the start of the project. Liem’s mother was provided with a Windows-based laptop that was equipped with Skype™ and a webcam. Sara’s mother was provided with a webcam and Ethernet cable. FA and FCT sessions were conducted in either the bedroom (Liem) or living room (Sara) of their homes. The parents placed the webcams in areas of the rooms to allow the behavior consultants to capture the child’s behavior on video throughout each session. Parents set up the rooms with play activities and demand materials before each telehealth visit. Demand materials were selected as those that were historically associated with problem behavior and included putting blocks in a bucket for both Liem and Sara. Boardmaker™ picture cards were used during treatment for both participants. A work picture card was used to signal when it was time to work, and a play picture card was used to signal when it was time to play.
Response Definitions, Observation System, and Interobserver Agreement (IOA)
Problem behavior was defined as self-injurious behavior, aggression, and property destruction. Specifically, self-injurious behavior was defined as any behavior that could result in tissue damage to the child (e.g., head banging, head hitting, and throwing self on the floor). Aggression was defined as any behavior that could result in tissue damage to another person (e.g., hitting, kicking, and pulling hair). Property destruction was defined as any behavior that could result in damage to property (e.g., throwing items, kicking items, and tearing paper). Task completion was defined as the child independently (i.e., without physical guidance) completing the demand specified by the parent. Mands were defined as the child independently (i.e., without physical guidance) requesting reinforcement following task completion during FCT. Manding for both participants was recorded when they pressed a BIGmack® microswitch with sufficient force to play a prerecorded message requesting playtime.
Trained data collectors independently coded child problem behavior using a 6-s partial interval recording system. The partial interval recording system was used to record discrete and continuous problem behaviors. Task completion and mands were coded using an event recording procedure. Task completion and mands were recorded as occurring either independently or following physical guidance. Ongoing review of participant data occurred on a weekly basis during team meetings that included discussions on the occurrence of problem behavior and parent fidelity during sessions. Based on these results and discussions, adjustments to the FCT treatment occurred.
IOA for problem behavior and mands was assessed using exact interval-by-interval comparisons. The number of agreements were divided by the number of agreements plus disagreements and then converted to a percentage by multiplying by 100. IOA for problem behavior and mands during the FA was collected during 36% of sessions for Liem (M = 97%; range = 94%–100%) and during 33% of sessions for Sara (M = 99%; range = 97%–100%). IOA for problem behavior and mands during the extinction baseline was collected during 36% of sessions for Liem (M = 94%; range = 93%–97%) and during 36% of sessions for Sara (M = 97%; range = 94%–100%). IOA for problem behavior and mands during FCT was collected during 31% of sessions for Liem (M = 97%; range = 93%–100%) and during 38% of sessions for Sara (M = 97%; range = 96%–100%).
IOA for task completion was assessed using a trial-by-trial comparison during escape and FCT sessions for both participants and the extinction baseline sessions for Sara. The number of agreements were divided by the number of agreements plus disagreements and then converted to a percentage by multiplying by 100. IOA for task completion was collected during 32% of FA and FCT sessions for Liem and 38% of FA, extinction baseline, and FCT sessions for Sara. IOA was 100% for both Liem and Sara.
Establishing operation (EO) and reinforcement analysis
When the FA and FCT data were reanalyzed retrospectively, the original 6-s partial interval data sheets were reevaluated to determine the occurrence of problem behavior during the EO and reinforcement periods. Problem behavior was scored as occurring in the EO period if it occurred within an interval in which the EO was established. Calculations of problem behavior during the EO consisted of totaling the number of intervals in which problem behavior occurred when the EO was in place and dividing it by the total number of intervals scored in the EO period. This calculation was then converted to a percentage by multiplying by 100. The same scoring and calculation procedures for the occurrence of problem behavior were used for reinforcement periods.
Two trained data collectors independently reevaluated the data for the occurrence of problem behavior within the EO and reinforcement periods across the FA, extinction baseline, and FCT conditions to assess for IOA. For Liem, IOA for problem behavior during the EO and reinforcement periods was assessed across 36% of sessions during the FA, 27% of sessions during the extinction baseline, and 28% of sessions during FCT. For Sara, IOA for problem behavior during the EO and reinforcement periods was assessed across 25% of sessions during the FA, 21% of sessions during the extinction baseline, and 39% of sessions during FCT. IOA was 100% for both Liem and Sara.
Treatment fidelity analysis
Treatment fidelity for both parents and behavior consultants was assessed during 21% and 23% of FCT sessions for both Liem and Sara, respectively. Treatment fidelity errors of the parent were defined as the parent failing to (a) instruct the child to complete the amount of work specified by the consultant, (b) require the child mand before receiving reinforcement, (c) provide reinforcement to the child when indicated, and (d) require the child to return to work when problem behavior occurred during reinforcement. Treatment fidelity errors of the behavior consultant were defined as failing to correct parent errors or correct parent errors at the appropriate time. Treatment fidelity errors for both the parent and behavior consultant were scored using an event recording procedure. For example, if a parent instructed the child to complete an extra work task during the EO period, and the behavior consultant did not correct this error, it was scored as an error for both the parent and behavior consultant.
IOA for treatment fidelity errors was assessed by conducting an exact trial-by-trial comparison for each of the four parent fidelity measures and for the single behavior consultant measure during FCT sessions. The number of agreements were divided by the number of agreements plus disagreements and multiplied by 100 to obtain a percentage agreement. For Liem, IOA of parent and behavior consultant fidelity was assessed across 21% of the FCT sessions evaluated and averaged 95% (range = 75%–100%) and 96% (range = 67%–100%), respectively. For Sara, IOA of parent and behavior consultant fidelity was assessed across 29% of the FCT sessions evaluated. Parent fidelity was 100% and behavior consultant fidelity averaged 75% (range = 0%–100%).
Experimental Design
Initial analyses of the host project
There were two sets of procedures that were always conducted as part of the host project (Wacker et al., 2013a, 2013b). An FA (Iwata et al., 1982/1994) was conducted within a multielement design to identify reinforcers maintaining problem behavior. Following the FA, a treatment evaluation was conducted within a nonconcurrent multiple baseline design across participants to demonstrate the effects of FCT on problem behavior.
EO and reinforcement analysis
To evaluate treatment failure, we reanalyzed the occurrence of Liem and Sara’s problem behavior during the EO and reinforcement periods during the FA and FCT evaluations. These evaluations were conducted within a multielement design to determine if differential frequencies of problem behavior occurred when the EO was in place versus when reinforcement was in place.
Procedures
All procedures described below were conducted as part of the host project.
FA
Prior to the FA, behavior consultants conducted an antecedent-behavior-consequence interview (Bijou, Peterson, & Ault, 1968) with the child’s parent to develop hypotheses regarding behavioral function. For all participants of the host project, including Liem and Sara, results of the interviews suggested that problem behavior was likely maintained by social reinforcers. During the FA, the parents conducted all FA sessions in their home with coaching provided by the behavior consultants via telehealth. FA sessions were counterbalanced and were 5 min in duration (50, 6-s intervals in which the child was in camera view). Session length varied slightly if the child left the camera view for the entire 6-s interval. The criteria developed by Roane, Fisher, Kelley, Mevers, and Bouxsein (2013) were used to identify the function for each child’s problem behavior assessed via the FA.
During the free play condition, the child had continuous access to play activities and parent attention. The parent refrained from giving demands during free play. There were no programmed contingencies for the occurrence of problem behavior. Three free play sessions were conducted prior to the FA to test the Internet connection and also to permit the children and their parents time to acclimate to the telehealth mode of service delivery.
During the escape condition, the parent presented a demand (e.g., “Put the block in the bucket.”). If the child engaged in problem behavior, the parent removed the demand and allowed the child to take a break for 20 to 30 s. Following the reinforcement period, the parent delivered another demand. Demands were presented every 30 s if the child completed the demand without engaging in problem behavior. If neither problem behavior nor compliance occurred with the demand, a three-step least-to-most prompting procedure was implemented. If the child engaged in problem behavior at any time during the three-step prompting procedure, he or she received a break for 20 to 30 s. If problem behavior did not occur, the prompting procedure continued until it was completed. When the 30-s time period for the demand elapsed, the next demand was presented.
During the tangible condition, the parent restricted the child’s access to high-preferred play activities (e.g., “We are all done playing with this toy.”). If the child engaged in problem behavior, the parent allowed the child to have access to the high-preferred play activities for 20 to 30 s. Following the reinforcement period, the parent again restricted access to the high-preferred play activities. High-preferred play activities were restricted for the duration of the session if no problem behavior occurred.
During the attention condition, parent attention was restricted while the child had access to play activities. If the child engaged in problem behavior, the parent delivered reprimands (e.g., “Do not throw the toys. I will play with you.”) for approximately 20 to 30 s. Following the reinforcement period, the parent again restricted her attention.
Extinction baseline
Extinction sessions were conducted to determine if problem behavior was resistant to brief periods of extinction and thus warranted more intensive treatment. In general, extinction sessions consisted of ignoring all occurrences of problem behavior while the relevant EO (identified in the FA) was in place. All sessions were 5 min in duration. Functional contexts targeted for treatment included tangible (Liem) and escape (Sara). The relevant extinction condition for each child is described below.
Tangible context (Liem)
Extinction baseline sessions for Liem were conducted once a month for 3 months as he was assigned to the delayed treatment group in the host project. Tangible extinction sessions began with the parent restricting access to high-preferred play activities, while less preferred play activities remained available. All problem behavior was ignored.
Demand context (Sara)
Extinction baseline sessions for Sara were conducted once every month for 3 months because she was assigned to the delayed treatment condition. Demand extinction sessions began with the parent directing the child to the work area. The demand used during the FA was presented every 30 s. Task completion produced praise, and problem behavior was placed on extinction. A three-step least-to-most prompting procedure, similar to the one described during the FA, was in place if the child did not complete the demand independently. All problem behavior was ignored.
FCT (general procedures)
Prior to initiating treatment, the parent and therapist determined the terminal treatment goal for each participant. For Liem, his mother wanted him to wait 5 min for preferred activities. During the wait period, Liem was required to complete a neutral demand as a means for giving him something to do during the wait period for preferred activities. For Sara, her mother wanted her to comply with 10 demands within a 5-min period.
FCT for both participants was conducted within a two-step chained schedule of reinforcement. During the initial link, the parent presented a work picture card along with a prompt to transition to the work area. After arriving at the work area, the parent presented the same demand used during the FA (Liem and Sara) and extinction baseline (Sara). The least-to-most prompting sequence was implemented if the child did not complete the demand independently. Problem behavior was placed on extinction. Task completion initiated a transition to the terminal link, which involved the parent presenting a microswitch or a play picture card to mand for playtime. Manding produced access to an enriched break with parent attention and play activities for 2 min. A response cost contingency was programmed during the enriched breaks. If problem behavior occurred during the enriched break, the parent restricted the child’s access to play activities and directed the child to return to work.
Each FCT session involved the completion of the two-step chain twice. If the child engaged in problem behavior and delayed completing the demand, the FCT session continued until the child completed the demand. The duration of the session was divided into 5-min blocks of time. Each 5-min block was recorded as a separate session.
Initially, Liem and Sara were required to place one block in the bucket during each of the two work activities during the 5-min session. Because of Liem’s persistent problem behaviors during FCT, we attempted several procedural variations (e.g., time-based work criterion, product-based work criterion, multiple free play sessions prior to FCT).
Results
The FA and FCT results of problem behavior for Liem and Sara are displayed in Figures 1 and 2, which represent the host project’s standard format for displaying the results of problem behavior. In addition, manding and task completion results are summarized in Table 1. Liem’s FA of problem behavior (see Figure 1, top panel) showed that problem behavior was maintained by access to tangible items. Specifically, problem behavior during tangible sessions was always elevated (M = 8%, range = 4%–12%), whereas it decreased to zero during the escape condition and rarely occurred in the attention and free play conditions. In addition, manding never occurred and task completion averaged 73.5% (range = 50%–100%).
Results of the functional analysis of problem behavior for Liem and Sara.
Results of FCT on problem behavior for Liem and Sara.
Summary of Manding and Task Completion for Liem and Sara During the Functional Analysis, Extinction Baseline, and FCT Conditions.
Note. FA = functional analysis; FCT = functional communication training.
Sara’s FA of problem behavior (see Figure 1, bottom panel) showed that problem behavior was maintained by access to tangible items and escape from demands. Across sessions, the occurrence of problem behavior continued to increase during both tangible (M = 6%, range = 0%–10%) and escape (M = 7.7%, range = 0%–18%) conditions. Except for one free play session, problem behavior did not occur during the free play and attention conditions. Similar to Liem, manding did not occur in any condition. Task completion averaged 16.8% (range = 0%–50%).
During the extinction baseline for Liem (see Figure 2, top panel), problem behavior occurred during an average of 14.5% (range = 6%–22%) of intervals, and manding never occurred. During the implementation of FCT, Liem’s problem behavior was variable during the first 15 sessions (M = 10.1%; range = 0%–28%), but then stabilized at zero for five consecutive sessions. During these sessions, Liem manded during all opportunities, except one, and his task completion increased to 100% during the last six FCT sessions. Following these seemingly successful results, problem behavior never restabilized at zero levels despite a number of procedural variations to encourage low levels of problem behavior. Overall, problem behavior remained variable throughout the rest of FCT (M = 9.4%, range = 0%–38%) and during free play sessions (M = 12.8%, range = 0%–58%). In addition, manding occurred during almost all opportunities provided and most tasks were completed independently. Thus, there was no correlation between problem and adaptive behavior.
For Sara (see Figure 2, bottom panel), problem behavior occurred during an average of 9.6% (M = 2%–14%) of intervals during the extinction baseline. Manding never occurred and task completion averaged 37.2% (range = 0%–66%). When FCT was implemented, results showed that problem behavior was variable with an increasing trend (M = 16.7%, range = 0%–40%). Although Sara’s manding and task completion improved slightly from the extinction baseline levels, overall results were variable and again not correlated with problem behavior.
EO and Reinforcement Analysis
Overall, the treatment evaluation results and our clinical observations discussed during weekly consultations for both Liem and Sara suggested that FCT was unsuccessful at reducing the occurrence of their problem behavior. To better understand why FCT was unsuccessful for these children, we formally reevaluated their FA and FCT data by determining whether their problem behavior occurred primarily in the EO, in reinforcement, or a combination of both.
Regarding the identified function of problem behavior during the FA (not displayed, but available from the first author upon request), Liem’s problem behavior primarily occurred during the EO (M = 10.5%, range = 4.3%–16%) rather than the reinforcement period (M = 2.7%, range = 0%–8%) of the tangible condition. Similar to Liem, Sara’s problem behavior primarily occurred during the EO of the tangible (M = 8.2%, range = 0%–17.2%) and escape (M = 10.9%, range = 0%–27.6%) conditions, which were identified as the maintaining conditions of problem behavior during the FA. Overall, these results further confirmed that the problem behavior displayed by Liem and Sara at the beginning of our evaluation was maintained by social reinforcement.
During the extinction baseline (see Figure 3, top panel), Liem engaged in problem behavior during an average of 14.8% of intervals (range = 6.1%–22.9%) during the EO. During FCT, problem behavior primarily occurred during the EO (M = 10.2%, range = 0%–41.9%). However, beginning in Session 26, problem behavior began to occur during the reinforcement period, which continued to occur intermittently until Session 78, whereby problem behavior occurred during almost all remaining sessions (M = 8.9%, range = 0%–26.5%). Thus, problem behavior was now distributed approximately equally between the EO and reinforcement periods, suggesting that it was no longer sensitive to the social contingencies. In addition, free play sessions, beginning in Session 80, were conducted and showed that problem behavior occurred on average during 12.8% of intervals (range = 0%–58%).
Results of the establishing operation and reinforcement analysis of FCT on problem behavior for Liem and Sara.
For Sara (see Figure 3, bottom panel), problem behavior only occurred during the EO of the extinction baseline (M = 9.8%, range = 2%–14.3%). Similarly, when FCT was implemented, problem behavior primarily occurred during the EO with an average of 21.5% (range = 0%–42.4%) of intervals. Problem behavior during the reinforcement period averaged 1.04% (range = 0%–10.7%). These results are distinct from Liem in showing that problem behavior continued to show a social function but was not responding to treatment.
Treatment Fidelity Analysis
Given that the results of the EO and reinforcement analysis yielded different patterns of responding for Liem and Sara, we evaluated whether these results might be correlated with treatment fidelity errors of either the parent, behavior consultant, or both. Results of this analysis showed that the treatment fidelity of the behavior consultant was 100% for both Liem and Sara. In relation to the treatment fidelity of the parents, results showed that treatment fidelity for Liem’s mother averaged 96% (range = 0%–100%), whereas for Sara’s mother, it averaged 45% (range = 0%–67%).
Discussion
FCT is a widely used (Tiger et al., 2008) and often successful treatment for problem behavior (e.g., Kurtz et al., 2003; Rooker et al., 2013; Wacker et al., 2017), with recent studies demonstrating that telehealth can be used to deliver FCT without degrading treatment effectiveness (Lindgren et al., 2016). In fact, the host project that the two children from the present study participated in demonstrated an average 97% reduction in problem behavior across 30 children (Lindgren et al., 2016). However, as previous research has shown, standard FCT programs are not always effective for reducing the problem behavior for all participants (e.g., Durand & Kishi, 1987; Fisher et al., 1993), and FCT may sometimes need additional treatment components (e.g., punishment) to be effective (e.g., Rooker et al., 2013). The results of the present study add to the overall results of FCT and specifically those reported by Durand and Kishi (1987) and Fisher et al. (1993) that FCT, as conducted in a two-step fashion via telehealth, may not always be an effective treatment for reducing problem behavior. To further improve the effectiveness of FCT, analyses of treatment failures are needed. Thus, in this study, we reexamined the data originally collected for two children whose behavior did not respond positively to FCT treatment. Although the initial analysis for both children showed a lack of success of treatment, the subsequent analyses showed specific reasons for the lack of positive effects.
Sara’s results showed that her problem behavior was under the control of social reinforcement contingencies, which was demonstrated when her problem behavior primarily occurred in the EO period and rarely occurred in the reinforcement period of treatment. These results suggested that her problem behavior should have been responsive to the FCT treatment. For this reason, we conducted the treatment fidelity analysis to formally identify if implementation problems were contributing to treatment failure because our clinical consultation and coaching during treatment sessions suggested ongoing treatment fidelity problems. Treatment fidelity results showed that Sara’s mother implemented treatment with an average of 45% fidelity across treatment sessions despite the behavior consultant providing ongoing and timely instructions and corrections. In general, Sara’s mother accurately and consistently instructed Sara to complete the appropriate amount of work, prompted her to mand following task completion, and provided her with praise and access to preferred items following the mand. However, she also often repeated task instructions, commented on Sara’s problem behaviors, failed to restrict high-preferred items during the demand, and failed to follow instructions given by the behavior consultant. For example, in one situation, the behavior consultant prompted Sara’s mother to remove Sara’s access to an item she was holding that was supposed to be restricted. However, Sara’s mother never removed Sara’s access to the item. Previous research on the effects of treatment fidelity on treatment gains has shown that when treatment fidelity occurs at less than 60%, treatment gains (i.e., reductions in problem behavior and increases in adaptive behavior) are less likely to be observed (St. Peter Pipkin, Vollmer, & Sloman, 2010). Thus, it is likely that poor, ongoing treatment fidelity from Sara’s mother contributed to the lack of success of the FCT treatment for Sara, and the behavior consultant was unable to correct these errors via telehealth consultation.
In contrast, Liem’s results showed a very different pattern relative to Sara’s results. Liem’s results suggested that his problem behavior likely became nonfunctional over the course of treatment. Initially, FA results suggested a social function, which was further confirmed by the EO and reinforcement analysis. Furthermore, during initial treatment implementation, problem behavior only occurred during the EO period, and it decreased to zero within the first 20 sessions of FCT, suggesting that at that point in time, his problem behavior was functional and the FCT treatment was effective. However, as we continued to implement FCT, problem behavior recurred and never consistently returned to zero. In fact, problem behavior was shown to occur equally in the EO and the reinforcement periods, suggesting that his problem behavior likely switched from being maintained by social reinforcement to being maintained by automatic reinforcement. Further confirmation of these results occurred during free play, in which he engaged in increasing levels of problem behavior over the course of treatment. In addition, the treatment fidelity analysis showed little to no errors by Liem’s mother or the behavior consultant suggesting that poor fidelity was likely not contributing to the negative treatment results. Thus, a different treatment was likely needed, perhaps based on a repeated FA to further show that the function of problem behavior had indeed changed.
Although the treatments for these two cases were unsuccessful, the results of this study advance our knowledge in two ways. First, on a clinical level, the results show that there continue to be conditions that undermine successful treatment, and these conditions may not always be easily addressed via telehealth for every person receiving behavior analytic services via telehealth. Wacker, Schieltz, Suess, and Lindgren (in press) discussed that one lingering question regarding the use of telehealth is the extent to which telehealth services can be effective in a field that relies heavily on the frequent use of gestures and modeling to train parents and caregivers on how to implement treatments such as FCT. With telehealth, the ability to gesture and model procedures to parents and caregivers is limited to a small set of options, thus requiring the behavior consultant to rely on vocal instructions. For Sara, her mother did not respond consistently to vocal directions from the behavior consultant, even though the behavior consultant consistently identified and provided instructions to address the observed treatment fidelity errors. Thus, the recommendation for her was to seek out in-person services as we believed that modeling and other parent training procedures were needed in order for FCT to be effective. That is, the behavior consultant could have directly assisted the mother with the extinction procedure, provided more hours of coaching per day, and/or conducted the treatment first to obtain initial effects before transferring it to the mother. Conducting these strategies in person may have resulted in changes of the effectiveness of FCT as a treatment for Sara as our secondary analyses of the EO and reinforcement periods, and of treatment fidelity, showed that the continued provision of telehealth consultation would not likely be successful. Follow-up studies continue to be needed to determine the conditions under which telehealth can and cannot be successful, and these should include analyses of treatment fidelity similar to Suess et al. (2014).
For Liem, the clinical issue involved the provision of treatment for behavior maintained by automatic reinforcement. We are not aware of any studies in the literature that have shown how to best assess (e.g., alone condition in the FA may be difficult to conduct) or treat severe problem behavior via telehealth when the behavior is maintained by automatic reinforcement. For Liem, problem behavior began to occur across the EO and reinforcement intervals of FCT and continued to occur during free play sessions. Treatment fidelity was assessed as adequate, suggesting either that his problem behavior emerged as being maintained by automatic reinforcement or that we missed identifying that function of behavior during the FA because we did not specifically evaluate an alone or ignore condition. The former appears to be a more parsimonious explanation of why FCT was unsuccessful as a treatment strategy for Liem because his problem behavior was shown to be persistent across reinforcement intervals and free play throughout the latter half of the evaluation, and the EO and reinforcement analysis of the FA suggested that problem behavior was initially maintained by social reinforcers. These results imply that a second FA that included an alone or ignore condition was needed to determine if, in fact, an automatic function was maintaining problem behavior. Thus, a repeat of the FA was needed followed by either a different treatment, or one that involved components (e.g., reductive procedures) not included in the current project’s procedures. Conceptually, we need to better understand changes in function of problem behavior, and especially if the function changes from social to automatic. Analyses of these types of changes would also likely benefit from in-person consultation, but of a very different type than for Sara. It is possible, however, that many of the same treatment issues would have developed for both Liem and Sara even if telehealth had not been the method of FCT treatment delivery.
The second way this study advances our knowledge is that the single-case analyses conducted permitted microscopic inspection of both the original treatment results and the subsequent results. These microscopic analyses were important because the original results only showed treatment failure, whereas the subsequent analyses showed why. These results add to the literature of other types of analyses such as repeated preference assessments, component analyses, and treatment fidelity analyses that demonstrate the conditions under which shifts in preference occur (e.g., Hanley, Iwata, & Roscoe, 2006), components of treatment are effective, neutral, or countertherapeutic (e.g., Millard et al., 1993), and types of fidelity errors that result in degradations of treatment effects (e.g., St. Peter Pipkin et al., 2010; Suess et al., 2014). By conducting single-case analyses such as these and those of the present study, we can begin to isolate other factors that may be related to treatment failure.
Demonstrations of unsuccessful treatment outcomes are important for informing the limits of the treatment. However, these demonstrations, albeit necessary, are most often descriptive (a treatment met or did not meet some criterion measure) and thus insufficient for advancing our science and informing our practice. To advance science and inform practice, it is equally important to evaluate the conditions under which the treatment was both successful and unsuccessful. Component analyses of successful treatments (e.g., Wacker et al., 1990) show which components of treatment packages are needed for each individual to experience reductions in problem behavior and increases in adaptive behavior. Secondary analyses of treatment failures can isolate the variables that are correlated with difficulties with the treatment, and can then be used to develop subsequent treatments. This development of subsequent treatments is the obvious next step to extend the current study and is needed to substantiate that the hypotheses related to treatment fidelity and changes in function were accurate and led to eventual identification of treatments that were effective in addressing problem behavior. Only by engaging in microscopic analyses such as the ones conducted in the present study can we begin to sort out why a treatment failed. Analyses such as these will then likely lead to adjustments to our treatment recommendations or developments of new treatments to address the problem behaviors displayed by individuals who are difficult to treat.
Footnotes
Authors’ Note
This information or content and conclusions are those of the authors and should not be construed as the official position or policy of, nor should any endorsements be inferred by the Health Resources and Services Administration (HRSA), the U.S. Department of Health and Human Services (HHS), or the U.S. Government.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by the Health Resources and Services Administration (HRSA) of the U.S. Department of Health and Human Services (HHS) under R40MC22644, Behavioral Treatment Through In-Home Telehealth for Young Children With Autism, US$926,743.