Towards Good Statistical Practice. CORSTAN Validated Questionnaire for Assessing the Correctness of Statistical...

Area of Science:

• Biostatistics
• Medical Research Methodology
• Evidence-Based Medicine

Background:

• Systematic biases and incorrect data analysis pose risks to research methodology and internal study validity.
• Existing guidelines address data analysis and reporting, but a standardized tool for reviewers to assess statistical quality is lacking.
• Ensuring robust statistical analysis is crucial for reliable evidence-based medicine.

Purpose of the Study:

• To develop a formalized tool for assessing the quality of statistical analysis in scientific medical publications.
• To create a questionnaire that aids reviewers in evaluating statistical rigor.
• To enhance the reliability of published medical research through improved statistical review.

Main Methods:

• Development of the CORSTAN (CORrect STatistical ANalysis) questionnaire based on expert experience and existing guidelines (SAMPL, ICH E9).
• Internal validation involved two independent experts assessing 20 randomized controlled trials (RCTs).
• Statistical analysis of expert agreement was performed to confirm the questionnaire's reliability.

Main Results:

• The CORSTAN questionnaire comprises two parts: 10 general questions and 8 specific to RCTs.
• A stratification method for the risk of incorrect statistical analysis was proposed.
• Internal validation demonstrated substantial to almost perfect agreement among experts, indicating high reliability.

Conclusions:

• The CORSTAN questionnaire simplifies and harmonizes the statistical review process across institutions like journals and dissertation boards.
• It serves as a valuable aid for authors in manuscript preparation, ultimately improving publication and research quality.
• Future application and experience will guide further refinement of the questionnaire.