Transparency and reproducibility in data analysis: the Prostate Cancer Prevention Trial

Area of Science:

• Biostatistics
• Epidemiology
• Medical Informatics

Background:

• Reproducible research is a growing concern in statistics, particularly with complex datasets.
• Current standards often require sharing raw data and code, which is not always possible due to confidentiality or data messiness.

Purpose of the Study:

• To propose and illustrate a two-step strategy for ensuring reproducible research when raw data dissemination is impractical.
• To enhance transparency in the analysis of sensitive or complex datasets.

Main Methods:

• The proposed method involves presenting a summary table of the data.
• A formula is provided to estimate key quantities directly from the presented data table.
• This strategy was applied to data from the Prostate Cancer Prevention Trial.

Main Results:

• The Prostate Cancer Prevention Trial data was analyzed using the proposed reproducible research strategy.
• The analysis focused on the period prevalence of prostate cancer.
• The method demonstrated the feasibility of transparent analysis without raw data sharing.

Conclusions:

• A practical approach for reproducible research is presented, suitable for situations with data confidentiality or complexity.
• The strategy of using data tables and estimation formulas enhances scientific transparency and replicability.
• This method is valuable for critical research areas like prostate cancer prevention studies.