CODEX Digest - 3.26.26

1) Disparities in MRI and TRUS for prostate cancer detection: A systematic review and meta-analysis of 94,020 cases.

Bahrami A, Tu LH, Ghanikolahloo M, et al. Curr Probl Diagn Radiol. 2026;55(1):148-169. 

Prostate cancer is the second leading cause of cancer death in men, with delayed diagnosis contributing to this issue. This analysis of North American data shows racial disparities in diagnostic imaging: White patients are more likely to receive Magnetic Resonance Imaging (MRI) for prostate cancer, while Black patients have less access to both MRI and Transrectal Ultrasound (TRUS). These findings suggest a need to address radiologic utilization and health policies to improve diagnostic equity.

2) Factors for patient trust and acceptance of medical artificial intelligence.

Bracic A, Spector-Bagdady K, Towle S, et al. JAMA Netw Open. 2026;9(3):e260815. 

AI is becoming more common in clinical care, but patient trust remains key to its adoption. A survey using diagnostic scenarios found that clinician involvement, transparent data use, and strong governance boosted patient trust and preference for AI-assisted care. Combining these factors could help AI reach its full impact in healthcare.

3) Pilot demonstration for physician compensation incentives to improve diagnostic specificity in clinical documentation.

Burstein DS, Schwartz N, Ganguly A, et al. Diagnosis (Berl). Epub 2026 Mar 5. 

Incomplete clinical notes degrade diagnostic communication. This pilot study finds that financial incentives can improve free text documentation for diagnoses needing two significant details, but not for those requiring three. However, more than half of patients lacked important documentation even after the intervention. 

4) From advice to action — real-world behavior of patients using an integrated diagnostic decision support system for navigating the health care system.

Cotte F, Lourenço FD, Pereira MP, et al. NEJM AI. Epub 2026 Mar 5. 

The impact of AI-supported symptom assessment and follow-up services on patient behavior remains unclear. This Portuguese study surveyed patients before and after using an AI-supported digital assessment pathway and tracked actual behavior through electronic health records. Results showed differences in intentions pre and post AI-support symptom assessment including a decrease in unnecessary emergency department visits, as judged by a physician panel. 

5) Physician-reported diagnostic uncertainty among children discharged from the pediatric emergency department with acute respiratory illness. 

Geanacopoulos AT, Drescher C, Kanaan J, et al. Pediatr Emerg Care. Epub 2026 Mar 2. 

Acknowledging diagnostic uncertainty to caregivers can undermine trust. This cross-sectional study examined how often physicians reported uncertainty when discharging children with acute respiratory illness from the emergency department and outlined related communication strategies and challenges. Nearly one-third of these discharges involved uncertainty. Further research should include clinicians and families to improve family-centered communication about uncertainty.

6)  An incidental finding unraveled.

Lee S. N Engl J Med. 2026;394(9):841-843. 

Incidental findings can affect patient wellbeing if effective communication doesn’t occur and resolution is delayed. This narrative commentary describes the psychological and care process journeys the physician-patient author transversed to resolve an incidental finding and the lessons learned from the experience she’ll now apply in her practice.

7) Taking a closer look: can an app improve diagnostic accuracy in urgent care? Cluster-randomized interventional trial DASI.

Noack EM, Antweiler K, Friede T, et al. PLOS Digit Health. 2026;5(2):e0001252.

Efficient medical history-taking is essential for accurate diagnosis in urgent care. This study tested a medical history-taking app in two German after-hours clinics and found its use did not improve diagnostic accuracy in urgent care settings. The authors note the study took place during the COVID-19 pandemic, which could have affected physicians' willingness to use an unfamiliar app.  

8) The first step in visual diagnosis: a study of novices developing the ability to distinguish normal from abnormal cases.

Oh S-Y, Burk-Rafel J, Reinstein I, et al. Adv Health Sci Educ Theory Pract. Epub 2025 Dec 22. 

Visual diagnosis training in medicine often lacks effective methods. This study  asked medical students and resident to interpret electrocardiograms and assessed how novices develop skills and confidence, finding cases varied from normal to abnormal with a difficult “confusability zone.” Participants' certainty improved slightly. The study recommends phased training: first focusing on abnormality detection, then advancing to detailed classification.

9) How a lack of visual fidelity in patient-generated images risks compromising care.

Payne R, Woywodt A, Wang Z, et al. Lancet Primary Care. Epub 2026 Feb 25. 

Patients use phones to provide images and videos while seeking a primary care diagnosis and referral decisions despite the potential for the images to be unreliable or distorted. This commentary argues that clinicians and health systems should implement safeguards and push for research, policy, and standards to ensure patient-provided visual data is accurate and safe. The authors outline actions at patient, clinician/service, system, and regulatory levels to improve patient image safety in care.

10) Ambient scribe in general practice: a multi-perspective before-after longitudinal mixed-methods study.

van Linschoten RCA, van Loon CM, Joanknecht L, et al. NPJ Digit Med. Epub 2026  Mar 2. 

Ambient scribes may help lessen physician workload and cognitive strain, but current evidence is limited and often excludes patient perspective. This multicenter Dutch study found ambient scribing can reduce documentation burden and aid communication, however, further research is needed to address problems identified that affect diagnosis like inaccurate summaries, reluctance in discussing sensitive topics, and impact on clinician reasoning.  

11) Interpretability of an FDA-authorized AI/ML sepsis diagnostic tool improved by SHAP values.

Watson GL, Staples G, Carver R, et al. JAMIA Open. 2026;9(1):ooag020. 

AI tools in healthcare can enhance patient care, but their lack of transparency makes clinicians cautious. This study examines whether standardized explanations added to an FDA-approved AI sepsis tool improved clinician understanding and trust. Clinicians interpreted explanations accurately 98% of the time, preferred having them, and gained insight into how sepsis risk scores were generated. Standardized explanations can make AI tools more transparent and encourage their adoption in patient care.

12) Practice pattern changes after adoption of diagnostic AI tool used in conjunction with cardiac imaging.

Zink A, Chernew ME, Neprash HT. Health Aff (Millwood). 2026;45(3):238-245. 

Reimbursement policies can influence use of diagnostic tests. This study looks at clinician adoption of Medicare-reimbursed CT fractional flow reserve (FFRCT) scans for coronary artery disease. After FFRCT adoption, CT scan orders increased while invasive heart tests decreased, raising per-patient diagnostic costs. The technology did not enable earlier diagnosis but was linked to more patients seen, more visits, and fewer heart-related complications.