CODEX Digest - 6.5.25

Banner

Want this delivered straight to your inbox every Thursday? Subscribe now.

Here are this week's must-reads: 

Doing “detective work” to find a cancer: how are non-specific symptom pathways for cancer investigation organised, and what are the implications for safety and quality of care? A multisite qualitative approach. 

Black GB, Nicholson BD, Moreland JA, et al. BMJ Qual Saf. Epub 2025 Jan 29.  

Cancer diagnosis delay is common and costly. This three-year ethnographic study examined data from four National Health Service (NHS) trusts to explore non-specific symptom association with delay of diagnosis and timely referral. The findings outline team and system-level improvements to cancer diagnosis processes by reducing variability, improving generalist cancer knowledge, test-result management, and care continuity.  

Ignored, dismissed, and minimized: understanding the harmful consequences of invalidation in health care--a systematic meta-synthesis of qualitative research.

Bontempo AC, Bontempo JM, Duberstein PR. Psychol Bull. 2025; 151(4) 399-427.

Clinician invalidation or dismissal of patient concerns and symptoms lead to a lack of correct diagnosis and treatment. This systematic review examines qualitative evidence of patient experiences with invalidation. The analysis resulted in a conceptual model to identify the emotional, social, and physical problems associated with patient symptom invalidation. The authors highlight the importance of these findings for clinical learners as they develop empathetic skills when working with difficult-to-diagnose diseases.
 

Exploration of Foundational Terminology and Paradigms for Improving Diagnosis.

Cosby KS, Baron-Lee JM, White-Brown I, et al. Exploration of Foundational Terminology and Paradigms for Improving Diagnosis. Rockville, MD:
Agency for Healthcare Research and Quality; April 2025. AHRQ Publication No. 25-0043.

The language of diagnostic excellence is messy and contributes to a lack of shared mental models and collective understanding of diagnostic improvement work. This issue brief discusses problematic disalignment of terminology while recognizing the adaptive role of context and purpose in terminology use. 
 

Warning labels and positive labels for pulse oximeters.

Gerke S, Shachar C. ​​​​​​ JAMA Intern Med. Epub 2025 Apr 21.   ​​​​​​

Pulse oximeters have been called out for lack of reliability when used on patients with darker skin. This editorial discusses legal and regulatory actions positioned to affect pulse oximetry with particular emphasis on safety challenges associated with the over-the-counter market and calls for warning labels to raise awareness of the problem to minimize the potential for misinformation in this patient population. View related guidance.

Diagnostic error: have we made any progress? 

Hoffer EP. Am J Med. Epub 2025 Apr 26. 

The journey toward diagnostic excellence is challenged by human and system factors. This editorial provides a short summary of the history of the diagnostic improvement effort and argues that increased emphasis on reflection, second opinions, and technology are key to error reduction and patient harm. 

Evaluating acute stroke diagnosis using simulation scenarios. 

Liberman AL, Apley D, Zhu J, et al. Ann Emerg Med. Epub 2025 Apr 9. 

Stroke misdiagnosis and diagnostic delay are common and harbor great potential for patient and family economic, physical, and emotional harm.  This screening experiment studied simulated ischemic stroke presentations to the emergency room to identify factors affecting diagnostic accuracy, confidence, and timeliness. This research method demonstrated viability to examine clinician accuracy and diagnostic confidence while illustrating the role that distraction and availability of an incident witness played on effective stroke diagnosis. Diagnostic excellence researchers may consider this strategy in pursuing work examining causes for misdiagnosis of stroke in the emergency environment.  
 

Clinically undiagnosed diseases in autopsies: frequency and risk factors.

Maccio U, Meier CA, Reinehr M, et al.  Arch Pathol Lab Med. 2025;149(1):60-66

Autopsies are considered a gold standard in identifying diagnostic and treatment missteps. This retrospective analysis of 648 autopsies of hospitalized patients over the three-year period in Switzerland found that 98% held one undiagnosed condition, notably cardiac disease and cancers. Insights for frontline clinicians on underdiagnosis of distinct conditions could affect diagnostic strategies with patients presenting with distinct risk profiles.  

Generative Artificial Intelligence in Health and Medicine: Opportunities and Responsibilities for Transformative Innovation.

Maddox T, Babski D, Embi P, et al, eds. Generative Artificial Intelligence in Health and Medicine: Opportunities and Responsibilities for Transformative Innovation. National Academies Press; Epub 2025 Apr.

AI is a noted toolset to revolutionize medicine. This book examines the spectrum of both the promise and pitfalls of generative AI and large language models (LLM) across the biomedical and health landscape. The authors examine AI through the sociotechnical system context to provide overarching discussion points for health executives and technologists who are examining the potential impact of AI and LLMs on care processes, administrative tasks, and diagnostic support. 

**The lead author was featured at CODEX's May webinar on this publication.
 

Sociodemographic biases in medical decision making by large language models.

Omar M, Soffer S, Agbareia R, et al. Nat Med. Epub 2025 Apr 7.  

AI is derived from human-generated data that can contain biases, raising questions about how prejudices based on someone’s identity can show up in AI-generated medical evaluations. This study analyzing over 1.7 million AI-generated vignette responses, researchers found that race, gender, income, and housing status influenced evaluation and treatment recommendations—even when patients had the same health conditions. These findings raise the concern that LLM-generated advice could reinforce stereotypes and potentially lead to misdiagnosis and patient harm. 

**This was a recent CODEX Editor's Pick.
 

Leveraging artificial intelligence to reduce diagnostic errors in emergency medicine: challenges, opportunities, and future directions.

Taylor RA, Sangal RB, Smith ME.  Acad Emerg Med. 2025;32:327-339 

Diagnostic excellence in the emergency room is routinely tested due to production pressures, information gaps, and decision fatigue. This review explores the role that AI can play to improve diagnosis in emergency care environments. The authors highlight its potential to reduce cognitive load through information access, address biased decision making, and provide feedback in real time. The piece posits the impact of AI on diagnosticians and the patients they care for through the co-developed, practical, and fair implementation of AI. 

Towards conversational diagnostic artificial intelligence.

Tu T, Schaekermann M, Palepu A, et al. Nature. Epub 2025 Apr 9. 

The dialogue between patients and physicians is the primary mechanism for developing trust, gathering information, and reducing uncertainty. This randomized study examines one AI model in a simulated environment to support excellence in history taking, compassion, diagnosis, and communication. 159 clinician-provided cases were used to compare the conversational AI with actual primary care interactions to find higher diagnostic accuracy, information gathering efficiency, and empathic dialogue.  

Comparison of initial artificial intelligence (AI) and final physician recommendations in AI-assisted virtual urgent care visits.

Zeltzer D, Kugler Z, Hayat L et al. ​​​​​​Ann Intern Med. 2025;178(4) 498-506. 

Researchers are exploring how AI could help doctors make better diagnoses, but they need more data to understand its real impact. This retrospective cohort study determined that AI was more consistent in following treatment guidelines, while physicians performed better when patient symptoms changed over time or required a physical exam. The study highlights how embedded AI can support care by surfacing key information—like past labs or allergies—and help align treatment decisions, pointing to a future where AI works alongside providers.  

**This was a recent CODEX Editor's Pick.
 

About the CODEX Digest

Stay current with the CODEX Digest, which cuts through the noise to bring you a list of recent must-read publications handpicked by the Learning Hub team. Each edition features timely, relevant, and impactful journal articles, books, reports, studies, reviews, and more selected from the broader CODEX Collection—so you can spend less time searching and more time learning.

Get the latest in diagnostic excellence, curated and delivered straight to your inbox every week:

Subscribe Now

See past digests here