A multivariate model for the meta-analysis of study level survival data at multiple times

Area of Science:

• Medical Statistics
• Clinical Epidemiology
• Vascular Surgery

Background:

• Critical leg ischemia (CLI) poses significant challenges in vascular surgery.
• Accurate survival rate analysis is crucial for evaluating CLI treatments.
• Existing meta-analysis methods may not fully capture temporal survival trends.

Purpose of the Study:

• To develop and apply a novel multivariate model for meta-analysis of study-level survival data in CLI.
• To simultaneously model mortality rates at multiple time points for improved accuracy.
• To compare the new model's results against standard meta-analysis methodologies.

Main Methods:

• Utilized a meta-analytic dataset of 50 studies on CLI survival rates.
• Developed a multivariate model employing exact binomial within-study distributions.
• Enforced non-decreasing constraints on study-specific and overall mortality rates over time.
• Directly modeled probabilities of mortality at up to seven time points.

Main Results:

• The new multivariate model provides simultaneous analysis of survival data at multiple time points.
• The model incorporates constraints to ensure biologically plausible non-decreasing mortality rates.
• Presented I(2) statistics to quantify substantial between-study heterogeneity.
• Demonstrated comparison with standard methodologies, highlighting model advantages.

Conclusions:

• The developed multivariate model offers a robust approach for meta-analysis of longitudinal survival data in CLI.
• The method directly models clinically relevant mortality probabilities, enhancing interpretability.
• Accounting for time-dependent mortality and heterogeneity is vital for accurate CLI outcome assessment.