CoChat

Influence of Topic Familiarity and Prompt Specificity on Citation Fabrication in Mental Health Research Using Large Language Models: Experimental Study

Open access

Published version of a preprint

The DOI matches its Crossref record; 19 of the 19 references it cites were found in a registry or an index; 21 of the 23 sources its claims cite were checked for support.

Abstract

Background: Mental health researchers are increasingly using large language models (LLMs) to improve efficiency, yet these tools can generate fabricated but plausible-sounding content (hallucinations). A notable form of hallucination involves fabricated bibliographic citations that cannot be traced to real publications. Although previous studies have explored citation fabrication across disciplines, it remains unclear whether citation accuracy in LLM output systematically varies across topics within the same field that differ in public visibility, scientific maturity, and specialization. Objective: This study aims to examine the frequency and nature of citation fabrication and bibliographic errors in GPT-4o (Omni) outputs when generating literature reviews on mental health topics that varied in public familiarity and scientific maturity. We also tested whether prompt specificity (general vs specialized) influenced fabrication or accuracy rates. Methods: In June 2025, GPT-4o was prompted to generate 6 literature reviews (~2000 words; ≥20 citations) on 3 disorders representing different levels of public awareness and research coverage: major depressive disorder (high), binge eating disorder (moderate), and body dysmorphic disorder (low). Each disorder was reviewed at 2 levels of specificity: a general overview (symptoms, impacts, and treatments) and a specialized review (evidence for digital interventions). All citations were extracted (N=176) and systematically verified using Google Scholar, Scopus, PubMed, WorldCat, and publisher databases. Citations were classified as fabricated (no identifiable source), real with errors, or fully accurate. Fabrication and accuracy rates were compared by disorder and review type by using chi-square tests. Results: Across the 6 reviews, GPT-4o generated 176 citations; 35 (19.9%) were fabricated. Among the 141 real citations, 64 (45.4%) contained errors, most frequently incorrect or invalid digital object identifiers. Fabrication rates differed significantly by disorder (χ22=13.7; P=.001), with higher rates for binge eating disorder (17/60, 28%) and body dysmorphic disorder (14/48, 29%) than for major depressive disorder (4/68, 6%). While fabrication did not differ overall by review type, stratified analyses showed higher fabrication for specialized versus general reviews of binge eating disorder (11/24, 46% vs 6/36, 17%; P=.01). Accuracy rates also varied by disorder (χ22=11.6; P=.003), being lowest for body dysmorphic disorder (20/34, 59%) and highest for major depressive disorder (41/64, 64%). Accuracy rates differed by review type within some disorders, including higher accuracy for general reviews of major depressive disorder (26/34, 77% vs 15/30, 50%; P=.03). Conclusions: Citation fabrication and bibliographic errors remain common in GPT-4o outputs, with nearly two-thirds of citations being fabricated or inaccurate. Reliability systematically varied by disorder familiarity and prompt specificity, with greater risks in less visible or specialized mental health topics. These findings highlight the need for careful prompt design, rigorous human verification of all model-generated references, and stronger journal and institutional safeguards to protect research integrity as LLMs are integrated into academic practice.

Overview

CoChat Summary

TL;DR

In six GPT-4o–generated mental health literature reviews, 35 of 176 citations (19.9%) were fabricated, and among the 141 real citations, 64 (45.4%) contained bibliographic errors. Fabrication and accuracy varied by disorder familiarity and, within some disorders, by prompt specificity.

Key findings

  • Across 6 reviews, GPT-4o produced 176 citations; 35 (19.9%) were fabricated. Of the 141 nonfabricated citations, 77 (54.6%) were fully accurate and 64 (45.4%) contained errors, most commonly incorrect or invalid DOIs (51/141, 36.2%).
  • Fabrication rates differed significantly by disorder (χ²₂=13.7; P=.001): major depressive disorder 4/68 (6%), binge eating disorder 17/60 (28%), and body dysmorphic disorder 14/48 (29%). Pairwise comparisons showed higher fabrication for binge eating disorder and body dysmorphic disorder than for major depressive disorder.
  • Overall fabrication did not differ by review type (general 17/102, 16.7% vs specialized 18/74, 24%; P=.21), but stratified analyses found higher fabrication for specialized versus general binge eating disorder reviews (11/24, 46% vs 6/36, 17%; P=.01).
  • Accuracy differed by disorder (χ²₂=11.6; P=.003), with results reporting major depressive disorder 41/64 (64%), binge eating disorder 26/43 (60%), and body dysmorphic disorder 10/34 (29%). Within-disorder review-type differences included higher accuracy for general versus specialized major depressive disorder reviews (26/34, 77% vs 15/30, 50%; P=.03) and higher accuracy for specialized versus general body dysmorphic disorder reviews (7/13, 54% vs 3/21, 14%; P=.01).

Methods

  • In June 2025, GPT-4o was prompted in new sessions to generate 6 literature reviews (~2000 words; ≥20 citations) on 3 disorders varying in public awareness and research coverage: major depressive disorder (high), binge eating disorder (moderate), and body dysmorphic disorder (low).
  • Each disorder was reviewed at 2 specificity levels: a general overview (symptoms, impacts, treatments) and a specialized review (evidence for digital interventions). All 176 citations were extracted and verified using Google Scholar, Scopus, PubMed, WorldCat, and publisher databases.
  • Citations were classified as fabricated, real with errors, or fully accurate. Fabrication and accuracy rates were compared by disorder and review type using chi-square tests, with stratified chi-square analyses for within-group differences.

Limitations

  • The paper states that findings may not generalize to other psychiatric disorders, specialized subfields, or academic disciplines; results are specific to GPT-4o; no prompt-engineering strategies were tested; and only one output per prompt was analyzed, so intraprompt variability was not assessed.
  • The paper also notes that low expected cell counts prevented formal modeling of interaction effects, relying instead on stratified chi-square tests.
  • Evident from the text: there is an internal inconsistency in the reported body dysmorphic disorder accuracy rate—the abstract states 20/34 (59%), while the Results section and Table 1 report 10/34 (29%).

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

Integrity signals

Registry

Crossref · 10.2196/80371

Cites

19 of 19 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

15 Valid · 1 Not valid · 2 Not checked · 63 Own results and interpretations

47 more claims

Create a free account to see all 81 claims and every source verdict.

See all claims — free account
  • It may state two separate claims as one. 18
  • Probably not a substantial claim. 18

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

4 of 4

Citations

Cites

Cited by

See all 40

Metadata

Authors
Linardon, J., Jarman, H. K., McClure, Z., Anderson, C., Liu, C., & Messer, M.
Affiliations
Deakin University (AU)
Corresponding author
Linardon, Jake
Venue
JMIR Mental Health
Published
2025-11-12
Volume
12
Pages
e80371
ISSN
2368-7959 (ISSN-L)
OpenAlex
W4416154989
Cited by
40
References
15
Citations by year
2026: 39 · 2025: 1
FWCI
41.17
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • JMIR Mental Health (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Figshare (submittedVersion)
Funders
  • Medical Research Council
  • National Health and Medical Research Council (1196948)
Topics
  • Social Sciences › Education › Academic integrity and plagiarism
  • Social Sciences › Library and Information Sciences › Academic Writing and Publishing
  • Social Sciences › Library and Information Sciences › scientometrics and bibliometrics research
Keywords
hallucinations, large language models, GPT-4o, bibliographic errors, prompt specificity, topic familiarity, mental health research, binge eating disorder, body dysmorphic disorder, major depressive disorder, digital interventions
MeSH terms
Large Language Models, Humans, Language, Mental Health, Biomedical Research

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 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (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.