A Location-Scale Joint Model for Studying the Link Between the Time-Dependent Subject-Specific Variability of Blood...

Area of Science:

• Biostatistics
• Epidemiology
• Cardiovascular Research

Background:

• Cardio and cerebrovascular diseases (CVD) pose a major public health challenge due to high morbidity and mortality.
• Elevated blood pressure is a known risk factor, and emerging evidence suggests blood pressure variability may also be an independent predictor.
• Existing studies on blood pressure variability often have methodological limitations.

Purpose of the Study:

• To propose and validate a novel joint statistical model for analyzing repeated measures of a marker and competing events.
• To investigate the association between blood pressure variability and the risk of CVD and all-cause mortality.
• To demonstrate the importance of accounting for heterogeneous variance in risk prediction models.

Main Methods:

• Development of a joint location-scale model incorporating a mixed model with subject-specific, time-dependent residual variance and cause-specific proportional intensity models for competing events.
• The model allows event risk to depend on the current value and slope of the marker trajectory, and its variance.
• Estimation via maximum likelihood using the Marquardt-Levenberg algorithm, implemented in an R-package and validated by simulation.

Main Results:

• Application to a large clinical trial on stroke secondary prevention reveals that individual blood pressure variability is associated with the risk of CVD and death.
• The proposed model demonstrates improved goodness-of-fit and predictive accuracy compared to models that do not account for heterogeneous variance.
• The findings highlight the clinical relevance of incorporating blood pressure variability into risk assessment.

Conclusions:

• The developed joint model effectively captures the complex relationship between marker variability and competing risks.
• Blood pressure variability is confirmed as a significant independent risk factor for CVD and mortality.
• Accounting for heterogeneous variance is crucial for accurate risk prediction and clinical decision-making in cardiovascular health.