Statistical Competence and Editorial Policies in Biomedical Publications

Area of Science:

• Biomedical Research
• Statistical Methodology
• Scientific Integrity

Background:

• Widespread misuse of statistical methods in biomedical research compromises scientific integrity and public health.
• Statistical errors stem from training gaps across the research lifecycle, not individual incompetence.
• Flawed analyses can result in misleading clinical guidelines and obscure true treatment effects.

Purpose of the Study:

• To highlight the pervasive issues in statistical methodology within biomedical research.
• To emphasize the need for enhanced statistical training and transparent reporting practices.
• To propose structural interventions at individual, publisher, and software levels to improve statistical quality.

Main Methods:

• Analysis of common statistical errors in study design, analysis, and reporting.
• Review of statistical software usage (SPSS, R/RStudio) and their impact on reproducibility.
• Examination of limitations in current peer-review processes and the rise of AI in statistical workflows.
• Discussion of publisher-level interventions like statistical editors and open data policies.

Main Results:

• Statistical errors are prevalent, impacting the reliability of biomedical findings.
• Current peer-review processes are insufficient to catch widespread statistical flaws.
• Software choice and AI integration present new challenges for reproducibility and transparency.
• Publisher-level interventions show promise for structural improvement.

Conclusions:

• Enhancing statistical quality in biomedical literature requires a coordinated effort from researchers, publishers, and software developers.
• Improved statistical education and adherence to reporting standards are essential.
• Implementing robust validation for AI tools and enforcing transparency are critical next steps.