CODEX Digest 10.29.26

As demand for adult ADHD evaluations grows, primary care clinicians face challenges distinguishing the disorder from other causes of cognitive difficulties. This commentary shares expert opinion to highlight the need for comprehensive assessment of childhood symptoms, functional impairment, and co-occurring conditions. It emphasizes clinician education and standardized guidelines to improve diagnostic accuracy while balancing missed diagnoses against inappropriate treatment.

Autopsies remain an important but underused source of information about diagnostic error, disease patterns, and causes of death. This commentary differentiates hospital autopsies from medicolegal examinations, examines their contributions to public health surveillance in the United States, and discusses changes in regulatory expectations that enabled the decline of their use. The authors suggest that strengthening autopsy practices can help identify missed diagnoses, improve diagnostic learning, and inform safer care.

Diagnostic incidents require investigations that identify system vulnerabilities including whether the response is proportional to the severity of the error. This UK-focused narrative review applies systems engineering principles to incident investigation in histopathology. It discusses approaches for understanding how work conditions, processes, and organizational factors contribute to errors. Although developed for pathology, the framework may inform broader diagnostic safety investigations.

Comparing AI-generated care recommendations to physician recommendations can reveal differences in proposed resource use, but lower estimated costs do not necessarily indicate better care. This study compared the costs of care recommended by 24 LLMs with those recommended by treating physicians and found the AI-recommendations cost three times as much per visit. Further evaluation should examine patient outcomes, omitted care, and the generalizability of the cost estimates associated with AI testing and referral suggestions.

Emergency clinicians must balance timely pulmonary embolism diagnosis against unnecessary testing. This observational study examines diagnostic decision-making and quality across encounters at 29 emergency departments (ED). By focusing on decision moments, it connects clinical reasoning and testing practices with diagnostic outcomes. The findings have implications for reducing missed diagnoses while avoiding overtesting. Its large clinical setting supports relevance, although results from one state may not generalize to all EDs.

Transitions between hospitals create opportunities for diagnostic information to be lost or misunderstood. This mixed-methods study examines diagnosis-related handoffs using an I-PASS informed process during inter-facility transfers to pediatric intensive care. It highlights the importance of communicating diagnostic reasoning, uncertainty, and pending information when children move between care settings. The findings offer practical implications for safer transfers, although the limited setting and exploratory design constrain generalizability.

Diagnostic safety concerns can remain hidden in routine primary care documentation. This Swedish quality improvement study combines structured medical record review and qualitative thematic analysis across 27 primary care centers to uncover latent record weaknesses that represent safety issues. Among 1,116 records, 32% contain at least one diagnostic safety concern, including undocumented reasoning, insufficient data collection, and failure to consider alternatives. The findings highlight opportunities for diagnostic reassessment, safety-netting, and learning from records.

Multicancer early-detection tests could change how cancers are identified, but earlier detection does not automatically improve patient outcomes. This UK randomized controlled Galleri trial evaluated whether screening with a multicancer early-detection test affected late-stage cancer diagnosis. The results find that adding a multicancer blood test to usual care didn't reduce late-stage cancers over three screening rounds, but longer follow-up is needed to see how results change over time. Its findings are relevant to screening policy, diagnostic pathways, and overdiagnosis.

As AI use becomes normalized in healthcare, maintaining clinicians' diagnostic skills is an important safety consideration. This commentary examines how AI use could contribute to clinical deskilling and discusses approaches to preserving independent reasoning. It highlights the need for thoughtful implementation, education, and oversight rather than assuming that automated assistance is always beneficial.

Improving diagnostic safety requires understanding when clinicians’ confidence matches their actual performance. This analysis combined a literature review with group discussion analysis involving 10 experts in clinical reasoning and diagnostic excellence. Experts emphasized that calibration varies with clinical context and recommend measuring meaningful decisions and contextual influences, rather than outcomes alone. The resulting proposed process offers practical direction for developing and evaluating real-world diagnostic calibration measures.

Emergency triage decisions can be affected by stigmatizing information, potentially compromising the equity and timeliness of care. This study examined how clinically stigmatizing information influenced large language model triage prioritization, using outcome-grounded scenarios. The findings indicate that some stigmatizing formulations can lead to harmful reprioritization. The study raises concerns about bias in AI-supported triage, although simulated evaluations cannot establish the frequency of these effects in actual clinical practice.

Assessing clinical reasoning requires more than a single examination or isolated performance measure. This educational methods study examines validity considerations in programmatic assessment, which combines evidence collected across multiple tasks and settings. It discusses how assessment design and interpretation influence conclusions about learners' reasoning abilities. The resultant three-stage framework is relevant to medical educators and diagnostic excellence initiatives, although it does not directly demonstrate improved diagnostic outcomes.