A linear model suitable for assessing graft patency in controlled clinical trials. SINBA Group

Area of Science:

• Cardiovascular Surgery
• Biostatistics
• Clinical Trial Methodology

Background:

• Assessing coronary artery bypass graft (CABG) occlusion is crucial for evaluating treatment efficacy.
• Current methods using occlusion ratios are affected by angiography timing, potentially underestimating cumulative occlusion rates.
• Variations in angiography timing between trials hinder accurate between-trial comparisons.

Purpose of the Study:

• To propose an alternative statistical analysis using logistic regression for CABG occlusion.
• To achieve asymptotically unbiased estimates of cumulative occlusion probability.
• To enable reliable comparisons of treatment efficacy across different clinical trials.

Main Methods:

• Logistic regression modeling was employed to analyze dichotomous occlusion outcomes.
• Patient data were modeled based on time of angiography, treatment type, and other covariates.
• The model estimates cumulative occlusion probability and the associated hazard function.

Main Results:

• The proposed logistic regression model yields asymptotically unbiased estimates of occlusion probability.
• This approach accounts for the timing of angiography, a key confounder in previous methods.
• Application to SINBA trial data demonstrated the model's utility.

Conclusions:

• Logistic regression offers a more robust statistical approach for analyzing CABG occlusion.
• This method enhances the accuracy of treatment efficacy evaluation and between-trial comparisons.
• The model provides a valuable tool for future cardiovascular clinical trials.