First and recurrent events after percutaneous coronary intervention: implications for survival analyses

Area of Science:

• Cardiovascular Medicine
• Biostatistics
• Clinical Research Methodology

Background:

• Composite endpoints and single-event analyses in clinical research can lead to substantial information loss.
• Recurrent event data analysis requires specialized statistical methods to avoid bias.

Purpose of the Study:

• To compare information loss in traditional statistical analyses versus alternative models for recurrent event data.
• To evaluate the impact of heart failure on outcomes following percutaneous coronary intervention using different statistical approaches.

Main Methods:

• Retrospective analysis of 4901 patients undergoing percutaneous coronary intervention (2010-2014).
• Construction of Cox models for composite endpoints, shared frailty models for recurrent events, and joint frailty (JF) models for simultaneous recurrent and terminal events.
• Evaluation of heart failure as a predictor for composite endpoints and death.

Main Results:

• Over 41% of patients experienced a readmission or death within one year; 60% of those with recurrent events had multiple readmissions.
• Heart failure was associated with an increased risk of the composite endpoint in all models (HRs ranging from 1.32 to 1.44).
• Heart failure was not significantly associated with death in the joint frailty model (HR 0.87).

Conclusions:

• Composite endpoints and first-event analyses result in significant information loss, particularly when recurrent events are common.
• Joint frailty models provide unbiased, event-specific hazard ratios for both recurrent and terminal events, improving analytical accuracy.
• JF models offer a more comprehensive understanding of disease progression and treatment outcomes in cardiovascular research.