The Effectiveness of a Chatbot Single-Session Intervention for People on Waitlists for Eating Disorder Treatment: Randomized Controlled Trial.

Background: Early treatment is critical for improving eating disorder prognosis. Single-session interventions (SSIs) can provide short-term support to people on waitlists for eating disorder treatment; however, it is not always possible to access SSIs. We co-designed and developed a rule-based chatbot called ED ESSI (Eating Disorder Electronic Single-Session Intervention), which delivered an SSI and demonstrated its acceptability and feasibility. However, the effectiveness of ED ESSI is yet to be investigated.

Objective: This study aimed to investigate the effectiveness of an SSI delivered by ED ESSI. We examined the chatbot's impact on eating disorder pathology, psychosocial impairment, depression, anxiety, stress, and motivation for change or treatment in individuals aged ≥16 years on waitlists for treatment for all types of eating disorders.

Methods: This multicenter 2-armed randomized controlled trial included 60 people on waitlists for eating disorder treatment in the chatbot intervention group (n=30, 50%) or the control group (n=30, 50%). The ED ESSI chatbot guided participants through a 30-minute SSI of assessment and psychoeducation, while the control group received web-based information on the same core topics covered in the SSI. There were 4 time points: before intervention or baseline (time point 1 [T1]), after intervention within 72 hours of baseline (T2), 1 month after baseline (T3), and 3 months after baseline (T4). Eating disorder pathology (primary outcome) and psychosocial impairment, depression, anxiety, and stress (secondary outcomes) were measured at T1, T3, and T4, and motivation for change or treatment (secondary outcome) was measured at all 4 time points. Furthermore, the System Usability Scale was implemented at T2 for the chatbot intervention group only.

Results: ED ESSI induced significantly greater reductions compared to the control group in the primary outcome of eating disorder pathology (P=.003) and secondary outcomes of psychosocial impairment (P=.008), depression (P=.002), and anxiety (P=.040) over the 1- and 3-month time points, with small to moderate effects (Cohen d=0.15-0.53). Chatbot use also induced an increase in participants' confidence in their ability to change (secondary outcome) immediately after use (T2), with a moderate effect (P<.001; Cohen d=0.74). The chatbot was rated as "excellent" in terms of usability. A significantly higher proportion of participants in the chatbot group (28/30, 93%) entered treatment by 3 months upon the opportunity being offered to them, compared to the control group, with a moderate effect (21/30, 70%; P=.042; ϕ=0.30).

Conclusions: ED ESSI promptly induced improvements in eating disorder pathology, psychosocial impairment, depression, and anxiety, which were detectable at 1 month and maintained to 3 months. ED ESSI potentially represents an effective, accessible, and scalable form of early intervention for people aged ≥16 years waiting for eating disorder treatment. Further research is needed to determine the longer-term effectiveness of ED ESSI.

Trial Registration: Australian New Zealand Clinical Trial Registry ACTRN12623000680662; https://tinyurl.com/2h9v7hh7.