CoChat

A Cognitive Behavioral Therapy Chatbot (Otis) for Health Anxiety Management: Mixed Methods Pilot Study

Open access
Goonesekera, Yenushka, Donkin, Liesje · 2022 · JMIR Formative Research

The DOI matches its Crossref record; 62 of the 62 references it cites were found in a registry or an index; 22 of the 46 sources its claims cite were checked for support.

Abstract

Background: An increase in health anxiety was observed during the COVID-19 pandemic. However, due to physical distancing restrictions and a strained mental health system, people were unable to access support to manage health anxiety. Chatbots are emerging as an interactive means to deliver psychological interventions in a scalable manner and provide an opportunity for novel therapy delivery to large groups of people including those who might struggle to access traditional therapies. Objective: The aim of this mixed methods pilot study was to investigate the feasibility, acceptability, engagement, and effectiveness of a cognitive behavioral therapy (CBT)–based chatbot (Otis) as an early health anxiety management intervention for adults in New Zealand during the COVID-19 pandemic. Methods: Users were asked to complete a 14-day program run by Otis, a primarily decision tree–based chatbot on Facebook Messenger. Health anxiety, general anxiety, intolerance of uncertainty, personal well-being, and quality of life were measured pre-intervention, postintervention, and at a 12-week follow-up. Paired samples t tests and 1-way ANOVAs were conducted to investigate the associated changes in the outcomes over time. Semistructured interviews and written responses in the self-report questionnaires and Facebook Messenger were thematically analyzed. Results: The trial was completed by 29 participants who provided outcome measures at both postintervention and follow-up. Although an average decrease in health anxiety did not reach significance at postintervention (P=.55) or follow-up (P=.08), qualitative analysis demonstrated that participants perceived benefiting from the intervention. Significant improvement in general anxiety, personal well-being, and quality of life was associated with the use of Otis at postintervention and follow-up. Anthropomorphism, Otis’ appearance, and delivery of content facilitated the use of Otis. Technical difficulties and high performance and effort expectancy were, in contrast, barriers to acceptance and engagement of Otis. Conclusions: Otis may be a feasible, acceptable, and engaging means of delivering CBT to improve anxiety management, quality of life, and personal well-being but might not significantly reduce health anxiety. © Yenushka Goonesekera, Liesje Donkin.

Overview

CoChat Summary

## TL;DR

This mixed methods pilot study tested Otis, a Facebook Messenger CBT chatbot, for health anxiety management in 69 adults in New Zealand during COVID-19. Otis was feasible and acceptable and improved general anxiety, well-being, and quality of life, but it did not significantly reduce health anxiety.

## Key findings

  • Of the 69 participants who completed baseline measures, 35 (51%) completed the postintervention questionnaire and 29 (42%) completed the 12-week follow-up; 12 of 29 (41%) completed all 14 days.
  • Health anxiety decreased from baseline to postintervention and follow-up, but not significantly: SHAI-18 mean 19.93 to 19.31 at postintervention (P=.55) and 17.38 at follow-up (P=.08).
  • General anxiety improved significantly: GAD-7 mean 9.69 to 7.58 at postintervention (t28=3.30, P=.003) and 7.00 at follow-up (t28=3.26, P=.003).
  • Personal well-being and quality of life also improved significantly: ONS4 mean 5.70 to 6.31 at postintervention and 6.50 at follow-up (both P=.03), and quality of life improved at postintervention (P=.02) and follow-up (P<.001).
  • Mean overall user experience was 18.24/30 and overall acceptability was rated 8.24/10; the most common reason for abandonment was lack of time (14/29, 35%).

## Methods

  • Single-group, pretest-posttest mixed methods pilot study with semistructured interviews; participants were recruited via social media and email between May and July 2020.
  • Sample: adults aged 18+ in New Zealand with Facebook Messenger and internet access; final follow-up sample was 29 participants, predominantly female (25/29, 86%), mean age 37.4 years.
  • Intervention and analysis: a 14-day, primarily decision tree–based CBT chatbot (Otis) on Facebook Messenger; outcomes were measured at baseline, postintervention, and 12-week follow-up, analyzed with paired-samples t tests and 1-way ANOVAs, and qualitative data were thematically analyzed.

## Limitations

  • The paper states that the observational design limited causal inference, and results may have been confounded by unidentifiable variables such as psychological input outside the intervention and changes in COVID-19 restrictions.
  • The sample was mostly New Zealand European and highly educated, limiting generalizability.
  • The authors note low statistical power from the small sample size, and the text also suggests a possible floor effect because only 5 participants met the health-anxiety cut-off; these are evident in the paper’s discussion.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.2196/37877

Cites

62 of 62 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Avatar of Otis.
Figure 1 Avatar of Otis.CC BY · PubMed Central
Figure 2 Examples of conversations with Otis.
Figure 2 Examples of conversations with Otis.CC BY · PubMed Central
Figure 3 A conceptualization of the factors related to the engagement, acceptance, and feasibility of Otis.
Figure 3 A conceptualization of the factors related to the engagement, acceptance, and feasibility of Otis.CC BY · PubMed Central

Verification

Claims the paper makes

13 Valid · 6 Not valid · 15 Not checked · 89 Own results and interpretations

58 more claims · 3 with their cited sources checked

Create a free account to see all 123 claims and to verify the 3 references still open.

See all claims — free account
  • Probably not a substantial claim. 28
  • It may state two separate claims as one. 44

Create a free account to see these claims.

Citations

Cites

Research works printed but not linked: 3

  1. 2000, Diagnostic and Statistical Manual of Mental Disorders, 4th edition
  2. Ersoy, K, 2020, Social Sciences Research Journal, 9, 117
  3. RietzTBenkeIMaedcheAThe impact of anthropomorphic and functional chatbot design features in enterprise collaboration systems on user acceptance201914th International Conference on WirtschaftsinformatikFebruary 24-27, 2019Siegen, Germany

Cited by

See all 43

Metadata

Authors
Goonesekera, Y., & Donkin, L.
Affiliations
University of Auckland (NZ)
Corresponding author
Donkin, Liesje
Venue
JMIR Formative Research
Published
2022-10-20
Volume
6
Issue
10
Pages
e37877
ISSN
2561-326X (ISSN-L)
OpenAlex
W4296509688
Cited by
43
References
58
Citations by year
2026: 11 · 2025: 12 · 2024: 12 · 2023: 8
FWCI
7.07
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • JMIR Formative Research (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
Topics
  • Social Sciences › Psychology › Applied Psychology › Digital Mental Health Interventions
  • Social Sciences › Psychology › Clinical Psychology › COVID-19 and Mental Health
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Anxiety, Depression, Psychometrics, Treatment, Cognitive Processes
Keywords
health anxiety, cognitive behavioral therapy, chatbots, Otis, COVID-19 pandemic, digital mental health interventions, Facebook Messenger, decision tree, anthropomorphism, feasibility and acceptability, general anxiety, quality of life, well-being, intolerance of uncertainty, barriers to engagement

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-11 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.