CODEX Digest - 2.5.26

1) Upper extremity specialist puzzlement and misdiagnosis are more likely when patients interpret rather than describe their symptoms. (subscription required)

Crijns TJ, Mercado AE, Ring D, et al for the Science of Variation Group. Qual Manag Health Care. Epub 2025 Nov 14. 

When patients communicate symptoms to providers, misinterpretations can arise. This study compares the accuracy of orthopedic surgeons when given patient interpretations versus direct symptom descriptions.

2) Diagnostic safety in the policy landscape - a comparative policy document analysis of Australian and Aotearoa New Zealand vs US health care and research policy.

Dahm MR, Carey RM, Tucker L, et al. BMC Health Serv Res. Epub 2025 Dec 5.

Policy attention to diagnostic safety has been limited. This analysis reviewed policy documents from health quality organizations and emergency medicine colleges in three countries, comparing how diagnostic safety is addressed. Results show that recognition of diagnostic safety is increasing and highlight the importance of understanding national differences to guide future policy development and action prioritization.

3) Transforming clinical reasoning-- the role of AI in supporting human cognitive limitations.

Greengrass CJ. Front Digit Health. 2026;7:1715440.

AI has the potential to augment human reasoning by complementing and enhancing decision-making. This review examines cognitive factors, such as biases and heuristics, that can reduce diagnostic reasoning performance and discusses AI as a tool to ease them, while raising limitations and the need for prioritizing training of clinicians to recognize these issues.

4) What do clinicians edit in ambient AI-drafted clinical documentation? A qualitative content analysis. (This is a preprint that has not gone through peer review).

Guo Y, Hu D, Yang Z, et al. medRxiv. Epub 2026 Jan 6.

Ambient AI documentation is progressively being used to draft clinical notes, but how clinicians edit and revise the records are not well known. This preprint presents a qualitative analysis of real-world changes to AI-generated care session drafts, highlighting diagnostic, symptom, and lab test refinements made by physicians. The authors identify practical areas for improving and implementing these tools.

5) Prevalence of stroke and diagnostic performance of emergency MRI in acute isolated dizziness.   

Hu X, Liu S, Wu X, et al. Ann Clin Transl Neurol. 2025;12(12):2514-2522.

Stroke is often misdiagnosed in patients with acute isolated dizziness despite MRI and CT detection use. This Chinese retrospective analysis of a prospective cohort shows that while clinical features and emergency CT have limited value, MRI is highly accurate for stroke detection in these cases. Standardizing the use of MRI first in emergent situations may lower misdiagnosis risk.

6) 'ChatGPT can make mistakes' warnings fail: a randomized controlled trial. (subscription required) 

Kıyak YS, Coşkun Ö, Budakoğlu Iİ. Med Educ. 2026;60(2):138-142.

Warnings about AI fallibility are common in clinical settings, but their impact on user action is unclear. This Turkish study found that disclaimers posted with AI results did not change medical students' responses to AI diagnostic suggestions, suggesting simple warnings may not be enough to build trust in AI-assisted learning.

7) 'Diagnostic Excellence Measurement Challenges and Recommended Solutions.

National Quality Forum; 2026.

Effective measurement of diagnostic excellence is necessary to inform systems-level efforts to reduce diagnostic error. This report discusses primary challenges in measuring diagnostic performance and proposes practical strategies that center on equity, patient-reported measures, timeliness, and usefulness in active care environments.

8) Automating pending labs list into discharge summaries.

Nguyen KT, Aniemeka C, Feaster N, et al. BMJ Open Qual. 2025;14(4):e003622.

Lack of pending test communication during care transitions can contribute to diagnostic errors. This quality improvement project evaluated an automated mechanism for pending lab list generation to appear in discharge summaries. The results show that adding an automated link improved pending lab documentation, yet the successful communication of that information has yet to be determined.

​​​​​​

9) AI implementation in pediatric radiology for patient safety: a multi-society statement from the ACR, ESPR, SPR, SLARP, AOSPR, SPIN.

Shelmerdine SC, Naidoo J, Kelly BS, et al. Pediatr Radiol. Epub 2025 Nov 25.

AI could transform diagnostic radiology, but most solutions and guidelines are designed for adults, rather than children, opening the door to errors, harm, and waste. This multi-society statement outlines key aspects of adopting AI in pediatric radiology: regulation and purchasing, implementation and integration, interpretation and post-market surveillance, and education.

10) Hidden costs of diagnostic mistakes: a descriptive study of guilt, shame, and scapegoating among sonographers practising in the United Kingdom. (subscription required) 

Upeh ER, Hynes C, Eze CU, et al. Radiography (Lond). 2025;32(2):103268.

When mistakes occur in ultrasound practice, they emotionally affect the clinicians involved. This study finds that sonographers are directly affected by diagnostic errors and that current organizational support for them is ineffective. Prioritizing psychological safety, non-punitive reporting, timely debriefing, and empathetic counselling fosters staff wellbeing, retention, and patient safety.

11) Racial inequalities in eligibility and access to lung cancer screening: systematic review of United States studies.

Yambao Yang Y, McNamara C, Stuckler D. BMC Public Health. 2025;25(1):3978.

Disparities in cancer screening populations contribute to diagnostic delays. This review examines racial inequalities in US lung cancer screening eligibility and access after the release of guideline updates aimed at reducing disparities. Black and Hispanic Americans' eligibility were still lower than White Americans, highlighting the need for further reforms to improve eligibility and access for high-need groups.

12) Toward safer diagnoses: a SEIPS-based narrative review of diagnostic errors.

Yen C, Epling JW, Rockwell M, et al. Diagnostics. 2026;16(2):347.

Diagnostic excellence is recognized to be a systems-focused challenge that requires sensitivity to the interconnectedness of people, organizations, and environmental elements. This review uses the human-factors driven Systems Engineering Initiative for Patient Safety (SEIPS) model to analyze factors contributing to diagnostic error that consider human factors, clinical medicine, and systems safety improvement approaches.