Abstract
Background/Objectives: Statistical reporting guidance emphasizes estimation, analytical transparency and cautious interpretation. We assessed changes in statistical reporting and interpretation in clinical neurology journals between 2015 and 2025. Methods: We audited 200 original quantitative articles from 10 journals, sampling 10 articles per journal-year stratum. Thirteen reporting domains and five interpretation or consistency outcomes were evaluated. Journal-stratified Mantel–Haenszel odds ratios (ORs) and 95% confidence intervals were calculated, with Benjamini–Hochberg correction across 18 outcomes. Two external auditors independently rated 20 articles before full-dataset harmonization. Post hoc analyses included conditional and Firth logistic models and source-verifiability sensitivities. Results: Adequate effect-estimate reporting increased from 62.0% to 75.5% (OR 2.06), uncertainty reporting from 46.4% to 63.0% (OR 2.14), and exact P-value reporting from 68.5% to 85.3% (OR 2.55). Trend language was less frequent (OR 0.27), whereas causal overstatement increased from 15.0% to 33.0% (OR 2.69). No comparison met q<0.05 (minimum q=0.0532). Effect-estimate reporting was attenuated in the full-text subset. External agreement for effect-estimate, uncertainty and exact P-value coding was 68–71%; sensitivity coding agreement was 20% in five evaluable pairs. Conclusions: Selected numerical-reporting practices were nominally more favorable in 2025, while analytical transparency remained uneven and causal overstatement was nominally more frequent. These exploratory contrasts require cautious interpretation because coding reliability and source verifiability varied by domain.