CODEX Digest 04.16.26

Shared decision making (SDM) is central to patient-centered diagnosis and treatment. This study analyses whether SDM affects prostate-specific antigen (PSA) screening rates across racial and ethnic groups and finds that less than half of eligible men received PSA screening, but among those, SDM reduces racial disparities.

Patient perspectives are important regarding emergency department diagnostic testing. This Australian qualitative study explores patient "expectations" of diagnostic tests in emergency departments. Patients expected empathetic communication and wise test selection, and value involvement in decision-making. Effective clinician-patient communication is key to diagnostic stewardship.

Technology enables remote access to specialists for rapid stroke care coordination, improving diagnostic excellence. This review outlines how virtual stroke networks connect professionals and deliver effective stroke care regardless of location. The authors summarize key components of the virtual networks, discuss AI's impact and possible integrations, address regulatory and ethical considerations, and propose a framework for future research and policy in stroke diagnosis and treatment.

Accurate laboratory testing is crucial for diagnosis, and identifying process weaknesses helps drive improvement. This retrospective study from Thailand found most clinical lab errors happen from incorrect test orders and specimen collection—91.1% caused by human factors. Almost all errors were harmless. To improve diagnostic excellence, healthcare organizations should adopt broad strategies targeting human-factor issues and optimize lab workflows.

Early detection of atrial fibrillation is key to effective care. This multicenter Hungarian study evaluated two smartphone apps for screening and found both had reasonably high diagnostic accuracy in real-world settings. These non-invasive tools may offer promising options for early rhythm screening and follow-up.

AI is increasingly used to support diagnosis and improve care efficiency. However, this retrospective descriptive cross-sectional study finds that only 4.4% of FDA-approved AI medical devices are labeled for pediatric use, with longer review time and limited clinical scope compared to adult devices. Pediatric tools are apt to focus on neurology diagnosis, while radiology and cardiology have fewer applications due to scarce high-quality imaging data caused by radiation concerns, posing ongoing challenges for pediatric-specific AI development.

The impact of AI misinformation on medical student decision-making has yet to be confirmed. This Chinese randomized trial ahead of print found that for novice medical learners, misleading AI explanations sharply reduced diagnostic accuracy while correct explanations offered no significant gain compared to controls. The study shows AI misinformation can be more harmful than helpful, underscoring the need to focus on developing learners' critical evaluation skills in medical education.

New human papillomavirus self-collection (HPV-SC) kits by mail can boost cervical cancer screening, but have low uptake. This qualitative study finds that diverse outreach and comprehensive educational materials for varied populations are essential for engaging patients in at-home screening and build trust in the home HPV-SC kits.

Overdiagnosis refers to the detection of conditions through screening that would not have caused symptoms or affected patient health. This review examines the extent, harms, and economic impact of lung cancer overdiagnosis from low-dose CT (LDCT) screening in high-risk groups, finding that LDCT marginally raises overdiagnosis and societal costs.

Little is known on potential patient impacts of screening-identified versus an incidental finding of cancer. This review found a small increase in short-term psychological impact in screening-detected cancers, but no strong evidence of psychological outcome differences between detection routes.

LLMs can create synthetic medical images (deepfakes), prompting concerns about misuse. This retrospective multi-national study evaluates 11 radiologists and LLMs in identifying ChatGPT-generated radiographs versus real ones. Radiologists and LLMs had low accuracy differentiating synthetic from authentic images. Training is needed to help physicians and LLMs recognize deepfakes and reduce risks.

Patient and family engagement (PFE) interventions help include patients and care partners in diagnosis, aligning goals to prevent errors and improve safety. This commentary about a recent systematic review of PFE discusses challenges in choosing appropriate PFE tools among many options and proposes a framework for teamwork-based shared mental models to guide the design, selection, and evaluation of PFE interventions.