Neurology

PRO reporting in spinal RCTs is inconsistent and often incomplete

Across 42 spinal RCTs (8,669 participants), PROs were common but heterogeneous, using 25 different instruments across five domains. Only 27.2% of trials met high-quality PRO reporting standards by ISOQOL checklist criteria. Key reporting gaps included lack of instrument justification, unclear data-collection modalities, and inadequate handling of missing PRO data. Recommendation: standardize instrument selection and prespecify collection and missing-data plans when designing or interpreting PRO-based spinal trials. 1

LLM interventions show heterogeneous benefits and uneven reporting

Systematic mapping of 55 studies found most LLM deployments used human-AI collaboration for decision support (65.5%). LLM-assisted interventions reported improvements in psychological endpoints, but diagnostic accuracy in randomized trials varied from 0.65–0.88. Reporting quality was suboptimal (mean CONSORT-AI adherence 78.8%), with frequent omissions about data quality and error handling. Clinicians should require robust accuracy data, prespecified outcomes, and full reporting before adopting LLM tools in practice. 2

Rural residence linked to worse inpatient epilepsy outcomes

In 841,445 epilepsy admissions, patients from the most rural counties had higher adjusted odds of in-hospital mortality (OR 1.93), status epilepticus (OR 1.32), and prolonged stay (OR 1.29). Rural patients also had lower odds of receiving EEG (OR 0.88) and differing discharge patterns. Associations attenuated among privately insured patients, suggesting modifiable structural barriers rather than immutable geographic effects. Action: prioritize access to EEG and specialist care for rural patients and consider policy or system solutions. 3