Statistical review: frequently given comments updated

Area of Science:

• Rheumatology research
• Biostatistics in clinical studies

Background:

• Extensive experience in statistical review of manuscripts for rheumatology journals.
• Identified recurring issues in statistical methodology and reporting across numerous publications.

Purpose of the Study:

• To summarize frequent statistical review comments.
• To provide guidance on best practices for statistical analysis and reporting in medical research.

Main Methods:

• Systematic review of statistical comments from approximately 700 manuscripts submitted between 2006 and 2024.
• Categorization and synthesis of the most common recommendations provided to authors.

Main Results:

• Frequent issues include handling of missing data, covariate selection in regression, prediction model validation, and baseline value adjustments.
• Specific guidance is provided on dichotomizing continuous variables, appropriate statistical tests (e.g., Student t-test over non-parametric tests), and reporting means with standard deviations.
• Emphasis on reporting estimates with confidence intervals (CIs) and P-values, adjusting for multiplicity, and appropriate formatting of CIs and decimal places.

Conclusions:

• Adherence to these statistical recommendations can improve the quality and reproducibility of research findings.
• Clear reporting of statistical methods and results is crucial for scientific integrity in rheumatology and related fields.