Patient AI Trust Dynamics Before and After Orthopedic Consultation (ORTHO-OP-GPT)

Official title

Longitudinal Pre-Post Patient AI Trust Dynamics in Orthopedic Outpatients: A Mixed-Methods Observational Study With Matched Physician-Patient Dyads

Overview

Patients increasingly consult artificial intelligence (AI) chatbots such as ChatGPT for health information before clinical visits, yet the impact of an actual orthopedic consultation on patient trust in AI-derived information remains unknown. This prospective longitudinal observational study quantifies how a single orthopedic outpatient consultation modifies patient trust in AI chatbots, the concordance between AI-derived and physician-delivered information, and patient anxiety, using a paired pre-post survey design supplemented by a matched physician-side assessment. Adult patients (18 years and older) presenting to two orthopedic outpatient clinics in Cyprus complete a brief pre-consultation questionnaire (T0) capturing demographics, AI use patterns, prior AI consultation regarding the current complaint, baseline trust, expectations, and anxiety. Immediately after their consultation they complete a second questionnaire (T1) assessing concordance with physician advice, trust change, consultation facilitation, post-consultation anxiety, and future intention. The consulting physician completes a brief 30-second post-visit form capturing whether AI was discussed, the medical accuracy of AI-derived information conveyed by the patient, and the effect of the AI discussion on consultation duration. The primary outcomes are the paired within-patient change in AI trust between T0 and T1 and physician-patient concordance on AI versus physician advice. Target enrollment is 180 to obtain 150 paired completed assessments.

Detailed description

Background and Rationale: Cross-sectional surveys have documented increasing patient use of AI chatbots for health information seeking. However, no published study has assessed how an actual physician consultation modifies patient trust in AI in a paired pre/post design, nor has any study captured the physician perspective on the same encounter in a matched dyad. Routine clinical encounters may be the primary mechanism by which patients calibrate their trust in AI-derived medical information.

Setting and Population: Two university-affiliated orthopedic outpatient clinics in North Cyprus.

Procedures:

* T0 (pre-consultation, waiting room, approximately 5 minutes): 14-item self-report questionnaire. * Consultation: usual care. * T1 (post-consultation, departure, approximately 5 minutes): 10-item self-report questionnaire. * Physician form (post-consultation, approximately 30 seconds): 5-item brief assessment. * Patient and physician forms are linked by an anonymous Participant ID.

Statistical Analysis Plan: Paired t-tests or Wilcoxon signed-rank tests for paired continuous outcomes; McNemar test or Stuart-Maxwell for paired categorical outcomes; Cohen's kappa for inter-rater agreement (AI versus physician); multinomial logistic regression for predictors of trust shift. All analyses two-sided, alpha equals 0.05. SPSS version 28.

Data Management: Anonymous CSV stored locally, encrypted, retained for 5 years per institutional policy. De-identified participant-level data available upon reasonable request after publication.

No formal pilot study is conducted. Instead, the first 20 participants will be prospectively monitored for protocol feasibility (mean completion time, drop-out rate, item-level missing data) as an embedded running pilot.

Primary outcome measures

• Mean within-patient change in self-reported trust in artificial intelligence-derived health information, measured by a study-specific 5-point Likert item (T0.11) and a study-specific 3-level categorical change item (T1.4). [Time frame: Baseline (within 15 minutes pre-consultation in the orthopaedic outpatient waiting room) and immediately after the consultation (within 15 minutes of consultation exit, same-day index visit).]
• Patient-physician concordance on artificial intelligence-versus-physician medical advice agreement, measured by Cohen's kappa coefficient between a study-specific 4-category patient item (T1.2) and a study-specific 5-point physician-rated AI medical accu [Time frame: Immediately after the consultation (within 15 minutes of consultation exit), for both patient (T1.2) and physician (H2) forms; same-day index visit.]

Secondary outcome measures (6)

• Mean within-patient change in self-reported anxiety, measured by an 11-point 0-to-10 visual analogue scale anchored 0 = no anxiety and 10 = worst possible anxiety (items T0.14 baseline, T1.5 post-consultation). [Time frame: Baseline (within 15 minutes pre-consultation) and immediately after the consultation (within 15 minutes of consultation exit), same-day index visit.]
• Percentage of enrolled patients reporting pre-consultation artificial intelligence use for the current orthopaedic complaint, measured by a study-specific single-item yes/no question (T0.9). [Time frame: Baseline (within 15 minutes pre-consultation, same-day index visit).]
• Percentage of pre-consultation artificial-intelligence users whose physician independently confirmed that AI was raised during the consultation, measured by a study-specific yes/no physician item (H1). [Time frame: Baseline (T0.9, pre-consultation) and immediately after the consultation (H1, within 15 minutes of consultation exit), same-day index visit.]
• Percentage of consultations in which the physician reported that the artificial-intelligence discussion shortened, did not change, or prolonged the encounter, measured by a study-specific 3-category physician item (H3). [Time frame: Immediately after the consultation (within 15 minutes of consultation exit), same-day index visit.]
• Mean patient rating of how prior artificial-intelligence use facilitated the consultation, measured by a study-specific 5-point Likert item (T1.4b: 1 = much more difficult, 5 = much easier). [Time frame: Immediately after the consultation (within 15 minutes of consultation exit), same-day index visit.]
• Mean patient-reported future intention to use and to recommend artificial intelligence for health information, measured by two study-specific 5-point Likert items (T1.7 future use; T1.8 recommendation to a friend). [Time frame: Immediately after the consultation (within 15 minutes of consultation exit), same-day index visit.]

Eligibility criteria

Inclusion criteria

• Age 18 years or older
• Presenting to an orthopedic outpatient clinic for any consultation
• Able to read and respond to a Turkish-language questionnaire
• Provides informed consent

Exclusion criteria

• Inability to complete a self-report questionnaire (e.g., severe cognitive impairment, language barrier)
• Re-presentation within the same recruitment window (each patient is enrolled only once)
• Refusal of consent for either T0 or T1

Criteria are shown verbatim from the registry (in English). Final eligibility is always assessed by the study center.

Healthy volunteers: No

Study design

Observational model

Cohort

Study locations

Cyprus · 2 centers

• University of Kyrenia, Dr. Suat Gunsel Hospital - Orthopedic Outpatient Clinic — Kyrenia
• Near East University Hospital - Orthopedic Outpatient Clinic — Nicosia

Publications

• Gultekin O, Hirschmann MT, Arikan HI, Kilinc BE, Yilmaz B, Abul S, Inoue J, Kayaalp ME. Evaluating deepresearch and deepthink in total knee arthroplasty patient education: ChatGPT-4o excels in comprehensiveness, Deepseek R1 leads in clarity and readability of orthopedic information. Jt Dis Relat Surg. 2026 May 1;37(2):470-476. doi: 10.52312/jdrs.2026.2645. Epub 2026 Mar 17. PMID 41906842
• Gultekin O, Inoue J, Yilmaz B, Cerci MH, Kilinc BE, Yilmaz H, Prill R, Kayaalp ME. Evaluating DeepResearch and DeepThink in anterior cruciate ligament surgery patient education: ChatGPT-4o excels in comprehensiveness, DeepSeek R1 leads in clarity and readability of orthopaedic information. Knee Surg Sports Traumatol Arthrosc. 2025 Aug;33(8):3025-3031. doi: 10.1002/ksa.12711. Epub 2025 Jun 1. PMID 40450565
• Kahan R, Shen C, Wellborn P, Lauder A, Berchuck S, Javeed H, Pean C, Federer A. Artificial Intelligence in Triaging Patient Questions: An Evaluation of a Large Language Model for Distal Radius Fractures. J Am Acad Orthop Surg. 2026 Jan 1;34(1):e106-e115. doi: 10.5435/JAAOS-D-25-00456. Epub 2025 Aug 27. PMID 40896839
• Schepman A, Rodway P. Initial validation of the general attitudes towards Artificial Intelligence Scale. Comput Hum Behav Rep. 2020 Jan-Jul;1:100014. doi: 10.1016/j.chbr.2020.100014. Epub 2020 May 18. PMID 34235291
• Yildirim TO, Karaman M. Development and psychometric evaluation of the artificial intelligence attitude scale for nurses. BMC Nurs. 2025 Apr 22;24(1):441. doi: 10.1186/s12912-025-03098-6. PMID 40264200
• Norman CD, Skinner HA. eHEALS: The eHealth Literacy Scale. J Med Internet Res. 2006 Nov 14;8(4):e27. doi: 10.2196/jmir.8.4.e27. PMID 17213046
• Bafna Sherma N. Factors influencing patients' engagement with ChatGPT for accessing health-related information. Crit Public Health. 2024.
• Choudhury A, Shamszare H. Investigating the Impact of User Trust on the Adoption and Use of ChatGPT: Survey Analysis. J Med Internet Res. 2023 Jun 14;25:e47184. doi: 10.2196/47184. PMID 37314848

Identifiers

NCT: NCT07631585 · ORTHO-OP-GPT-2026

Primary sources (government registries)

• ClinicalTrials.gov (US NIH)

View this study on ClinicalTrials.gov ↗