Stata Modules for Calculating Novel Predictive Performance Indices for Logistic Models.

Area of Science:

• Biostatistics
• Cardiovascular Disease Epidemiology
• Medical Informatics

Background:

• Cardiovascular disease (CVD) risk prediction is crucial for prevention, with numerous biomarkers emerging.
• Assessing the added predictive value of novel biomarkers requires robust statistical methods like discrimination, calibration, and reclassification indices.
• Limited availability of user-friendly software hinders the widespread adoption of these advanced model assessment techniques.

Purpose of the Study:

• To develop a user-friendly statistical software tool for researchers with limited programming expertise.
• To facilitate the implementation of novel methods for assessing the performance of risk prediction models.
• To enable quantification of the improvement in risk prediction offered by new biomarkers.

Main Methods:

• Developed a Stata command named 'addpred' for logistic regression models.
• The command calculates cut-point-free and cut-point-based Net Reclassification Improvement (NRI) and Integrated Discriminatory Improvement (IDI).
• Applied the command to real-world data from the Tehran Lipid and Glucose Study (TLGS) to evaluate the Framingham CVD risk algorithm.

Main Results:

• The 'addpred' Stata command is available for logistic regression models.
• Demonstrated the utility of the command in assessing the predictive performance improvement using family history, waist circumference, and glucose levels.

Conclusions:

• The 'addpred' Stata package simplifies the assessment of novel biomarker predictive capacity.
• This tool is expected to encourage the broader application of advanced statistical methods in biomarker research.
• Facilitates more accurate cardiovascular risk prediction and prevention strategies.