Operant vomiting is displayed by some individuals with developmental disabilities. Operant vomiting, which is under the control of operant contingencies,Footnote 1 differs from vomiting caused by physiological variables and is topographically and physiologically distinct from rumination (O'Brien et al. 1995). Frequent vomiting can result in health problems, including dehydration, weight loss, tooth decay, aspiration pneumonia, and gastrointestinal bleeding (e.g., Weiss 2002). Additionally, individuals (e.g., parents, teachers) working with people who exhibit operant vomiting are at risk for communicable diseases.

Though previous research has evaluated behavioral interventions to address operant vomiting (see Lang et al. 2011), there is a paucity of research using functional analyses (FAs) to assess operant vomiting (Beavers et al. 2013). Lang et al. (2011) reviewed 21 studies on behavioral interventions for operant vomiting and rumination. However, only five studies included a functional assessment prior to intervention. Three studies conducted an analogue FA (Iwata et al. 1982/1994) on rumination, which was found to be maintained by automatic reinforcement across all three studies. The remaining two studies assessed operant vomiting. One study used descriptive assessments (Lockwood et al. 1997) and the other used a combination of descriptive assessments and a series of extended no-interaction conditions (Baker et al. 2010) to identify the function of vomiting. Lockwood et al. (1997) hypothesized vomiting was maintained by escape from demands (i.e., vomiting occurred almost exclusively after the presentation of demands), whereas Baker et al. (2010) concluded vomiting was maintained by automatic reinforcement. Similarly, Wunderlich et al. (2017) conducted an FA (based on Iwata et al. 1982/1994) to evaluate self-induced (inserting hand into mouth) vomiting in a young child and found that vomiting was maintained by automatic reinforcement.

Results from the function-based treatment evaluation conducted by Lockwood et al., which included reinforcement for compliance, escape extinction, and antecedent manipulations, indicate that operant vomiting maintained by social contingencies may be amenable to treatment. Of course, in the absence of function-based assessment, it may be difficult to provide the functional reinforcer(s) which maintain vomiting, or otherwise identify antecedent manipulations or other consequent events that decrease its occurrence. Unfortunately, there is relatively little research on the use of FAs to evaluate vomiting, especially for cases in which vomiting is maintained by socially mediated reinforcers. The current study describes the implementation of an FA to identify the function(s) maintaining operant vomiting and a clinical treatment to address operant vomiting in a young male with high-functioning autism spectrum disorder (HF-ASD). Thus, the purpose of this study was to demonstrate the successful application of functional assessment and function-based treatment procedures to decrease operant vomiting. Additionally, this study highlights a number of important procedural issues and factors (e.g., mitigating potential medical concerns) directly related to assessment and measurement of operant vomiting.

Method

Participant and Setting

Sam was an eight-year-old male who was referred by his mother and school to a university-based applied behavior analysis clinic. Sam was diagnosed with autism spectrum disorder (ASD), bipolar disorder, attention-deficit/hyperactivity disorder (ADHD), oppositional defiant disorder (ODD), and obsessive compulsive disorder. He engaged in several topographies of problem behavior at school and home, including property destruction, aggression, elopement, and noncompliance. Sam was receiving behavioral services focused on decreasing aggression and property destruction and increasing compliance. Separate prior FAs for aggression and property destruction were conducted approximately one year earlier and indicated these topographies were maintained by escape from academic demands. His intervention for aggression and property destruction included functional communication training, differential reinforcement for compliance (using a token economy), and extinction, which had resulted in decreases in problem behavior and increases in compliance. Sam’s therapists had trained his teacher and paraprofessionals to implement his intervention. Approximately nine months after services began for these other behaviors (i.e., aggression and property destruction), Sam began engaging in vomiting and SIB, including self-choking, head banging, and head hitting (an FA indicated SIB was maintained by escape and adult attention). Sam would direct his vomit at therapists during and between sessions and vomited on his teachers and paraprofessionals at school. No consistent precursor responses to vomiting were observed. A medical doctor ruled out a physiological cause for Sam’s vomiting and provided approval for the assessment of vomiting. Sam’s doctor monitored his health during regular appointments throughout the intervention.

All FA sessions were conducted in the university clinic with Sam, a therapist, a paraprofessional (intervention) and observers (for IOA). During the FA, the table and chairs were removed because of Sam’s history of property destruction and aggression, as well as for sanitary reasons (i.e., so as to avoid cleaning emesis off of furniture after every session); Sam and the therapist would stand or sit on the floor to play with materials or complete work. During treatment, a table and two chairs were in the room. Materials included a stopwatch, Legos®, edibles, notebook paper and academic worksheets, and high-preferred leisure items.

Data Collection and Interobserver Agreement

The FA sessions were 5 min in duration. Primary and secondary observers used paper-and-pencil data collection to record the frequency of vomiting within each 10-s bin during the session. Frequency data were converted to rate (responses per min) by dividing the total number of responses by session duration. Intervention sessions were the duration of the school day (approximately 6 h). The primary and secondary observers (i.e., therapists, paraprofessionals) collected frequency data on the frequency of vomiting during 15-min intervals. Frequency data were converted to rate (responses per hour) by dividing the total number of responses by the hours in the day.

Vomiting was defined as any instance in which bile matter or body fluid (e.g., spit) crossed the plane of Sam’s mouth, excluding spitting in the absence of stomach contractions (e.g., Sam spit a piece of candy in the trash). Sam typically engaged in discrete instances of vomiting where he would vomit a small amount at one time with one instance lasting 1–2 s. Each occurrence of vomiting was recorded as a discrete response when there was more than 1 s in the absence of vomiting before the next instance.

Trained secondary observers simultaneously and independently collected data on vomiting during 24% of FA sessions. Data were collected by a secondary observer for at least one hour of the day on 44% of days during treatment. Using the mean count-per-intervals method, interobserver agreement (IOA) was calculated by comparing the two observers’ data during each interval (10-s intervals for FA, 15-min intervals for treatment), dividing the smaller number by the larger number within each interval to produce proportion for each interval. These proportions were summed, divided by the total number of intervals in the session, and converted to a percentage. Mean IOA on vomiting was 100% across FA sessions and 99% (range, 74–100%) across treatment days.

Functional Analysis

All FA sessions were 5 min in duration. Between eight and 14 FA sessions were conducted per day across three days. There was a 1 to 4-min inter-session interval between sessions and approximately a 10-min break mid-way through the sessions for one day. During the inter-session interval and break, Sam had access to food (e.g., crackers, candy) and water (regardless if and how many instances of vomiting occurred during the session) and therapists cleaned up any vomit. Therapists encouraged Sam to drink water to mitigate concerns about dehydration and ensure that he had sufficient stomach content to engage in vomiting. Therapists monitored Sam for any signs of adverse health effects from vomiting (e.g., facial coloring, energy levels, and for any signs of a hoarse voice or sore throat). Adverse signs were not observed; however, if they had been, sessions would have been discontinued for the day. The therapist wore a different colored shirt for each condition to help facilitate discrimination. A reversal design (Vollmer et al. 1993) was used, in which test and control conditions were conducted serially.Footnote 2

Control

During the control conditions, the therapist provided Sam with two high-preferred leisure items and stated, “Here are some toys that you can play with.” The therapist delivered attention on a fixed-time (FT) 15-s schedule and avoided placing demands. If Sam engaged in the target response (i.e., vomiting) or nontargeted problem behavior (e.g., aggression, disruptions), the therapist did not provide attention or comment on the behavior. If Sam engaged with the therapist or asked a question, the therapist played with Sam or answered his question.

Escape

During the escape condition, the therapist presented work materials (e.g., pencil, paper, handwriting sheets) and stated, “It’s time to do some work.” The therapist issued an academic directive (e.g., “Multiply eight by seven”). If Sam failed to engage in the correct response within 5 s or answered incorrectly, the therapist provided a model prompt (e.g., “eight multiplied by seven is 56”). The therapist provided nondescriptive praise following a correct response independently or after the model prompt. The therapist physically prompted the correct response following an incorrect response or no response after the model prompt. Following the target response, the therapist provided a break statement (e.g., “Okay, you don’t have to”) and removed the academic materials for 30 s. The therapist did not provide attention following appropriate behavior (e.g., break request) or nontargeted problem behavior.

Attention

At the beginning of the attention condition, the therapist stated, “I can’t talk right now, I need to do some work,” and pretend to fill out paperwork while oriented away from Sam. No toys were present. Contingent on the target response, the therapist provided brief (approximately 10 s) attention in the form of statements related to the target response. The therapist made statements of concern (e.g., “Are you okay?” or “Are you feeling bad?”) and statements directed related to vomiting (e.g., “That’s so gross!”) paired with physical attention (e.g., patting Sam on the shoulder). If Sam engaged in non-targeted problem behavior or attempted to recruit attention appropriately, the therapist did not respond or comment on the behavior.

Treatment

The treatment was conducted as part of an ongoing evaluation and data collection with Sam. During baseline, an intervention was in place to target Sam’s aggression and property destruction and to increase compliance. Because the data were collected as part of ongoing clinical services that were not for experimental purposes, as well as the considerable medical risks associated with operant vomiting, an AB design was used (see Lanovaz et al. 2019).

Baseline

The therapist provided choices of task order and reinforcers. Sam had noncontingent access to a break card that he could touch to receive a 1-min break from work—he could request a break at any time. Sam earned tokens (tallies on a laminated paper) for complying with directives and academic tasks (similar to those used in the FA; he earned one tally after approximately every 2–4 instances of compliance), asking for help with his work, or transitioning appropriately from trade-in/break times to work. Sam could trade-in his tallies at any point for edibles or tangibles from a menu—work paused during these trade-in periods. Sam could elect to trade-in fewer tallies for smaller amounts of edibles or tangibles or could save up tallies for a greater quantity. Each tally was worth one small piece of an edible or 30 s of access to a tangible leisure item (e.g., radio, iPad). The therapist would converse with Sam during his trade-ins. During baseline, the therapist delivered a brief statement of concern after Sam vomited and redirected him to vomit in a trash can or go to the restroom if he was not feeling well. Academic tasks were postponed to provide time for Sam to use the restroom, get some water, and to clean up the vomit.

Intervention

During the intervention for vomiting, the therapist implemented the choice opportunities, break card, and token-reinforcement procedures as described in baseline. Following vomiting, the therapist refrained from commenting directly about the vomiting and implemented a repeated prompting procedure. During the repeated prompting procedure, the therapist delivered an initial vocal directive. If Sam did not comply or engaged in vomiting, the therapist restated the directive and gestured or modeled the correct response every 10 s until Sam complied. Sam could not earn any tallies or take a break during the repeated prompting procedure. Once he complied with three consecutive directives, the procedure ended and his normal tally and break procedures resumed.

Results and Discussion

Figure 1 (top panel) shows the rate of vomiting across test conditions (escape and attention) and the control condition. During the control phases, Sam did not vomit. His rate of vomiting increased across sessions during the first (M = 0.9) and second (M = 1.2) escape phases. During the first attention phase, Sam’s rate of vomiting increased to moderate levels (M = 1.0). An increasing trend was observed across sessions during the second attention phase (M = 1.0). Responding during the FA suggested that Sam’s operant vomiting was multiply maintained by access to attention and escape from demands.

Fig. 1
figure 1

Rate of vomiting across FA sessions (top panel) and treatment days (bottom panel)

Full size image

Figure 1 (bottom panel) depicts the results from the treatment for vomiting. During baseline, Sam engaged in an average of 1.2 instances of vomiting per hour with variable rates across days (range, 0–8.4). Following the implementation of minimal attention for vomiting and the repeated prompting procedure, vomiting decreased to low levels (M = 0.06) that maintained over 100 days.

We replicated prior research showing FAs can be used to identify the function of operant vomiting (e.g., Wunderlich et al. 2017) by using a FA to demonstrate that vomiting was maintained by social reinforcement rather than nonsocial reinforcement. In the current study, a FA using a reversal design was used to show that operant vomiting was multiply maintained by access to adult attention and escape from academic demands. Thus, this study is among the first to demonstrate that vomiting may be under control of multiple social functions (similar to many other topographies of problem behavior; e.g., Beavers et al. 2013) but still be amenable to function-based behavioral intervention. Our intervention extends previous literature for the treatment of socially maintained operant vomiting. Although reinforcement procedures were in place for appropriate behavior during baseline, vomiting resulted in attention and avoidance of academic work, and occurred at relatively high levels. Once attention was minimized and repeated prompting was implementing following vomiting, operant vomiting decreased and maintained at very low levels.

Limitations and Future Directions

One limitation of the current study is that experimental control was not demonstrated during the treatment, though the AB design used within this study was selected given the serious medical nature associated with the target problem behavior (Lanovaz et al. 2019). When applying the dual-criteria method for visual analysis outlined by Fisher et al. (2003) to these data, the likelihood of obtaining these clinical outcomes by chance alone is extremely small (p < .001). Thus, given our use of structured criteria (Fisher et al. 2003) and the extended number of baseline and treatment sessions (i.e., more than six of each; Lanovaz et al. 2017), our AB design still reflects the considerable effect size of our intervention (Lanovaz et al. 2019) Nevertheless, future researchers should conduct more systematic evaluations of interventions to address operant vomiting. For example, the degree to which these treatment effects persist during treatment challenges (e.g., treatment integrity errors, changes in treatment contexts, removal of treatment contingencies) was not evaluated. Future research should examine and quantify the durability and long-term maintenance of this (and other) behavioral interventions to mitigate concerns regarding relapse, and eventually fade out unnecessary treatment components/contingencies (e.g., Nevin and Wacker 2013; Wacker et al. 2011).

Clinical Implications

When therapists hypothesize that vomiting is maintained by automatic reinforcement, extended no-interaction conditions may suffice to confirm that vomiting is not maintained by social variables (e.g., Baker et al. 2010). However, when practitioners hypothesize that operant vomiting is controlled by social reinforcement or multiple variables, an FA should be conducted to help inform treatment. Practitioners can consider using a reversal design as an alternative to multielement designs if carryover effects or discrimination fails produce undifferentiated outcomes. In addition, reversal designs allow for evaluation of rates across sessions within a phase. During Sam’s FA, we observed increasing trends of vomiting across phases, indicating reinforcement effects. Latency FAs (Thomason-Sassi et al. 2011) may also be useful if vomiting is difficult to reset (e.g., the individual reliably empties or mostly empties the contents of the stomach thereby making it difficult to vomit in subsequent sessions).

When conducting a FA on vomiting, clinicians should also consider other important factors, such as ensuring that clients are adequately hydrated/nourished between sessions to mitigate hydration concerns and to avoid altering motivating operations associated with depletions in stomach content/volume that could affect vomiting. Clinicians should consider the riskiness and potential for injury associated with the target behavior, including weighing the risk of potentially higher levels of vomiting during assessment or shaping up a response to occur in different situations (e.g., clinicians should not include a tangible condition unless indirect or direct observations suggest one is warranted). Clinicians should ensure they consult with a medical professional regularly when assessing and treating vomiting.