CODEX Digest - 11.6.25

1) Clinical diagnosis is complex, but why? Identifying factors contributing to diagnostic complexity of infectious diseases in emergency medicine.

Dey S, Joshi S, Mishra PR, Khadilkar P. Proc Des Soc. 2025;5:941-949.  

Utilizing behavioral design methods can guide process changes and safety improvements. This article presents an assessment tool for estimating infectious disease complexity in emergency care. Tested in one clinical setting in India, the objective complexity score estimated by the tool matched the difficulty perceived by physicians.  

2) Case 28-2025: a 36-year-old man with abdominal pain, fever, and hypoxemia. (subscription required) 

Dhaliwal G, Hood CM, Manrai AK, et al.  N Engl J Med. 2025;393(14):1421-1434. 

AI is not yet capable of independent medical diagnosis. This fascinating comparison demonstrates how human experts and AI differ in clinical reasoning and information processing. While the human diagnostician in this case was correct and the AI “Dr. CaBot” was incorrect, the article provides insights underscoring the potential of AI as a diagnostic tool.  

3) Enhancing diagnostic reasoning in medical education through patient stories and illness scripts. (subscription required) 

Dudas RA, Bannister SL. Pediatrics. 2025;156(4):e2025072981. 

Clinicians compare patient presentations against standard disease characteristics or illness scripts to determine diagnosis. This commentary discusses how inviting and deeply listening to parent narratives can hone pediatric diagnostic skills. The article discusses teaching strategies like “think aloud” to enhance medical student ability to diagnose children.  

4) Socioeconomic inequities in pap test use among Black women in the United States: an intersectional approach. (subscription required) 

Fisher S, Agénor M. J Racial Ethn Health Disparities. 2025;12(5):3451-3459.

Racism operates in an intersectional manner with other forms of oppression to produce disparities in cancer screening and timely diagnosis. This study examines the intersection of educational attainment, employment, and income of Black women and cervical cancer screening. The research finds that​​​​ Black women​ with lower socioeconomic position​ had lower odds of Pap test use and calls for interventions to address a range of contributors to inequities to improve access to screening for this patient population. 

5) Clinical significance of a multicancer screening trial with stage-based end points.

Gogebakan KC, Lange J, Owens L, et al. JAMA Netw Open. 2025;8(10):e2536247.

A single test to detect multiple cancers early is being assessed on a national scale in the UK, which could lower mortality and guide screening policies. This modeling study derives population risk reduction estimates of late-stage disease and mortality at different screening timetables using the England Cancer Registry. Modeling predicts earlier detection and reduced mortality, adding to the importance of upcoming real-world studies. 

6) 10 Years After “Improving Diagnosis in Health Care” Where have We Been and Where are We Going.

Graber ML, Hoffer EP, Singh H, eds. Diagnosis (Berl). 2025:12(4):489-595. 

The publication of “Improving Diagnosis in Health Care” marked a watershed moment in the quest toward diagnostic excellence. This special issue 14 articles summarizes research and other efforts catalyzed by the report over the past decade and the continued efforts still required to achieve it. Topics covered include patient engagement, education, and AI.   

7) Real-world diagnostic performance and clinical utility of artificial intelligence-assisted interpretation for detection of lung metastasis on CT in patients with colorectal cancer.  (subscription required)

Jang S, Kim J, Lee JS, et al. AJR Am J Roentgenol. 2025;225(3):e2533063. 

Research is emerging that explores the diagnostic performance of AI in actual care settings. This Korean study evaluates how AI assistance affects radiologists' accuracy in identifying lung metastases on chest CT scans for colorectal cancer patients. The findings illustrate the practical value of AI in tracking the spread of cancer in individual patients and include clinical utility measures such as the diagnostic yield, the false-referral rate, and management changes after positive reports

8) How stroke imaging shapes malpractice risk.

Jeevika, Chaudhary SR, Roy S. Acad Radiol. 2025;32 Suppl 1:s114-s118.  

Delays in stroke imaging can lead to significant patient disability, family challenges, and clinician legal risks. This discussion is anchored in actual cases where lack of coordination, communication, or care escalation contributed to patient harm and malpractice claims. Strong teamwork and non-siloed communication support effective stroke diagnosis. 

9) Scalable screening for emergency department missed opportunities for diagnosis using sequential eTriggers and large language models. (This is a preprint that has not gone through peer review.)

Marks C, Gibney S, Stenson B, et al. medRxiv. Epub  2025 Oct 7.  

Emergency departments (ED) are hectic environments where diagnostic mistakes can happen. This preprint of a multisite study looking at eTriggers and using SaferDx shows that LLMs accurately ruled out cases, improved efficiency when paired with human review, and generated useful clinician feedback. LLMs show promise for identifying missed opportunities for diagnosis (MODs) and supporting quality improvement, offering a scalable way to enhance ED diagnostic excellence. 

**Adrian Haimovich will be featured in CODEX's next AI webinar.

10) Managing cognitive load and enhancing metacognitive learning in postgraduate training and practice. (subscription required) 

Ng IKS, Ko CJ, Lim TK. Diagnosis (Berl). Epub 2025 Aug 13.

Resident physicians experience mental workloads that can result in fatigue and poor decision making. This article examines how cognitive load affects post-graduate training and clinical decision making. It reviews interventions for managing cognitive load in clinical education and suggests strategies to foster adaptive thinking and professional development. 

11) If you provide them, they will come: an observational study of online pathology report access by patients at a large, academic, tertiary care hospital in Canada. (subscription required)

Ranot J, Noghani P, Rothwell D, et al. J Clin Pathol. 2025;78(9):636-640.  

Patient access to online test results could be used in diagnostic communication but may result in patient and family psychological harm. This Canadian study reports that over half of patient participants viewed pathology reports via portals, with usage varying by age and gender and result severity. Many received results before their doctor, highlighting the need for effective support mechanisms to help patients interpret their results.  

12) Rapid diagnostic stewardship and blood culture use in a pediatric medical center.

Vaugon E, Costales C, Assad Z, et al. JAMA Netw Open. 2025;8(10):e2535580.  

Lack of access to necessary testing equipment can delay diagnosis. This study tests a stewardship process limiting blood cultures drawn at an urban pediatric hospital during a supply shortage. The strategy reduced use of blood cultures without negative clinical impact, increased readmission rates, or longer lengths of stay.