Escape Extinction Burst or Treatment Failure in Early Intervention
Lena is a 5-year-old autistic child receiving 2 hours per day of early intervention in a clinic. Baseline tantrums averaged 8 per day (range 6–11), typically 4–12 minutes, occurring mostly during table-top instruction and transitions. The BCBA wrote a plan combining functional communication training (FCT) with a break request and escape extinction: after a demand, the RBT keeps the task present and does not remove it until Lena completes one response; break requests are reinforced on a fixed-ratio 1 schedule. After 12 days, tantrums increased to 12–15 per day, duration up to 25 minutes, and Lena scratched an RBT hard enough to break skin. Direct observation shows the RBT often removes the demand after 30–45 seconds of screaming “to keep her safe,” and delivers the break card only after prompting. The parent wants to stop the plan. The BCBA has 3 days before authorization review.
1. What are the competing explanations for the worsening data, and how would you evaluate them before changing the plan?
Read the worked answer
The worsening data have at least four competing explanations: (a) an extinction burst; (b) failure of escape extinction because the RBT removes the demand after 30–45 seconds, which intermittently reinforces screaming; (c) FCT not established because the break card is prompted after escalation, so the mand is not under discriminative control; and (d) task difficulty or insufficient reinforcement. I would not call it a burst until integrity is measured. I would collect session-by-session treatment integrity (demand present, break delivered only after card, no escape after problem behavior), latency to mand, mand frequency, and aggression frequency/duration. If integrity is low, the data do not test the plan; the first intervention is RBT training. If integrity is high and aggression is increasing with a rising-then-falling pattern, then burst is plausible. Safety trumps the design: any skin break triggers review.
2. Propose a modified plan that maintains safety and is likely to be implemented with fidelity. Include measurement and decision rules.
Read the worked answer
Modified plan: stop nonremoval of demand in the high-risk context; instead use errorless demand fading with a break card available before the demand, reinforce mands immediately, intersperse easy tasks, offer choices, and teach tolerance for brief delays. Use escape extinction only in a low-demand, closely supervised context after the RBT meets 90% integrity on a checklist; do not use it during the current escalation. Train with BST: instruction, modeling, rehearsal, and feedback to criterion. Measure aggression, mands, latency, task completion, and integrity. Decision rule: if aggression is not below baseline after 8 sessions with ≥90% integrity, revise the hypothesis; if integrity is below 80%, pause and retrain. Obtain parent consent for the revision and document the safety rationale and the plan for thinning the break schedule.
3. How would you handle the parent’s request to stop the plan and the upcoming authorization review?
Read the worked answer
I would validate the parent’s concern and explain that the current data do not show that FCT failed; they show that the plan was not implemented as written and that safety is now the priority. I would present the modified plan, the rationale for stopping nonremoval in the high-risk context, and the criteria for resuming a stricter procedure. I would obtain informed consent for the revision and document the discussion. For the authorization review, I would submit baseline and treatment data, integrity data, the safety incident, and the revised plan with decision rules. I would not claim the original plan was effective, and I would not continue an unsafe procedure merely to preserve authorization. If the funder denies the revised plan, I would request a peer review and document the clinical necessity.