CODEX Digest - 1.15.26

1) A caregiver's perspective: identifying "zebras"--listening to patients is a clinical necessity.

Bashe G. J Patient Exp. 2025;12:23743735251404807.

Rare disease diagnosis can be challenging even when communication is excellent. This piece shares the story of a patient whose rare disease symptoms were repeatedly missed or misdiagnosed, emphasizing the need to respect patient and family perspectives in the diagnostic process. Practical advice is offered to enhance care through recognition of the value of listening.

2) Reducing labeling errors in histopathology specimens: a quality improvement initiative on endometrial vs. endocervical specimens to promote patient safety in KK women’s and children’s hospital. (subscription required) 

Boricha YB, Zhao J, Ang FY, et al. Jt Comm J Qual Patient Saf. Epub 2025 Oct 28.

Laboratory errors can undermine diagnostic accuracy. This study from a Singapore OB/GYN center shows that simple, low-cost strategies—such as staff education, feedback, audits, and interface redesign—substantially improved diagnostic reliability and reduce specimen mislabeling in uterine biopsies, addressing a key area of diagnostic safety.

3) Artificial Intelligence in Medical Diagnostics. (purchase required)

Hirosawa T. Springer Nature; 2025.

AI and healthcare are increasingly interconnected. This book presents key concepts for both health professionals new to digital health and AI and developers seeking clinical insight. It covers machine learning in diagnosis, digital health records, and the ethical issues where technology meets medicine.

4) “Everything looks reassuring.”: managing uncertainty during the emergency department discharge process. (subscription required)

Krause SA, Steege LM, Pecanac KE. Patient Educ Couns. 2026;142:109365. 

Diagnostic uncertainty happens in the complex emergency department (ED) setting. This study recorded 40 interactions with 10 ED clinicians assesing how they communicate uncertainty upon discharge. Teams explained to patients why they were safe to go home, reducing concerns by ruling out acute conditions, and providing guidance for follow up actions to manage risk.

5) Diagnostic stewardship mechanisms in electronic test results management - a scoping review.

Li J, Thomas J, Baysari M, Georgiou A, et al. Int J Med Inform. 2025;207:106178.

Diagnostic process optimization contributes to excellence by supporting timely review and appropriate follow-up of test results. This review identifies factors supporting the diagnostic stewardship process in the electronic management of test results. Authors describe how non-technical interventions, such as protected time and performance feedback, are effective in an increasingly automated landscape.

6) Nurse-led visits reduce in-person referral from urgent care telehealth. Pediatr Emerg Care. (subscription required)   

Mason A, Whitt M, Skoglund D, et al. Epub 2025 Nov 3. 

Integrating nurses into diagnostic processes enhances timely, accessible care. This study assessed nurse-led virtual visits in pediatric urgent care, showing they can safely improve access and can reduce unnecessary in-person referrals, supporting value-based care.

7) Artificial intelligence powered mobile health apps for skin cancer detection: current challenges and a systems thinking approach for improved public health outcomes in low- and middle-income countries.(subscription required) 

Mukherjee S, Rao SR, Poddar A. Melanoma Res. Epub 2025 Nov 28.

Patient-facing AI tools are increasingly used worldwide for diagnosis. This review examines the risks and benefits of mobile health apps in low and middle income countries for skin cancer care, compares their challenges with those in high-income countries, and includes a causal loop diagram showing how multistakeholder actions can improve public health through safe, equitable AI-based mobile health app adoption.

8)  Estimating the long-term health impact and program cost-effectiveness of providing direct screening services to low-income, medically underserved patients through the Colorectal Cancer Control Program (CRCCP).

O’Leary MC, Koutouan PR, Mayorga ME, et al. Cancer Causes Control. 2025;36(10):1175-1195.

Public screening programs support early cancer detection. This modeling study examined US-based Colorectal Cancer Control Program (CRCCP) colonoscopy or stool testing data and found that screening reduces colorectal cancer disparities, delivers cost-effective services, and helps decision-making.

9) Symptom-Disease Pair Analysis of Diagnostic Error (SPADE) Research Tool User Guide Version 1.0. 

Agency for Healthcare Quality and Research. 2025.

The AHRQ Symptom-Disease Pair Analysis of Diagnostic Error (SPADE) method uses existing administrative data sets to identify the potential for diagnostic error. This tool provides background on SPADE, and how the research tool arrives at results and workflow considerations for implementing the tool.

10) Lessons in clinical reasoning - pitfalls, myths, and pearls: how search satisfying can keep eyes crossed. (subscription required)

Tashjian ME, Parker J, Edwards Mayhew RG, et al. Diagnosis (Berl). Epub 2025 Oct 22.

Problem lists are intended to support high-quality care, but are often long and outdated, leaving diagnoses remaining on lists far longer than clinically appropriate. The authors looked at community health center problem lists and if checking the "reviewed" button had any relationship to removal of short-term diagnoses.

11) First, do NOHARM: towards clinically safe large language models. (This is a preprint that has not yet gone through peer review.)

Wu D, Haredasht FN, Maharaj SK, et al. arXiv. Epub 2025 Dec 1.

Clinicians and patients are turning to AI for advice, but its safety is uncertain. This preprint shares a tool to measure harmful AI outputs. Severe harm occurred up to 22% of cases, predominatly from errors of omission. The results underscore the need for medical AI to be evaluated for safety rather than just accuracy.

12) Uncertainty-aware large language models for explainable disease diagnosis.

Zhou S, Wang J, Xu Z, et al. npj Digital Med. 2025;8(1):690.

Training an AI model to recognize and explain when a diagnosis is uncertain is difficult. This study introduces a Iarge language model-based diagnostic system called ConfiDx that was developed to identify uncertainty and clearly explain it using real clinical data. Results found that ConfiDx outperformed commercially available LLMs in diagnostic accuracy and recognizing diagnostic uncertainty alongside explanations for disease diagnosis, as measured by performance metrics and clinician review.