The analysis of COVID-19 in-hospital mortality: A competing risk approach or a cure model?

Area of Science:

• Biostatistics
• Epidemiology
• Medical Statistics

Background:

• Competing risk models are common for in-hospital mortality, treating discharge as a competing event.
• However, discharge and death stem from the same disease process, making competing risk models potentially inaccurate.
• Cure models offer an alternative framework, viewing discharge as a sign of recovery or 'cure'.

Purpose of the Study:

• To propose and evaluate cure models for in-hospital mortality analysis.
• To compare cure models with traditional competing risk models.
• To investigate the impact of treatments on mortality and cure probabilities in COVID-19 patients.

Main Methods:

• Utilized mixture cure and promotion time cure models.
• Extended models to incorporate known cure status for discharged patients.
• Developed an Expectation-Maximization (EM) algorithm for the mixture cure model.
• Demonstrated the equivalence between competing risk and promotion time cure models.
• Applied models to a cohort of COVID-19 in-hospital patients with diabetes.

Main Results:

• The promotion time cure model indicated statin use improved overall survival.
• The mixture cure model revealed statin use reduced mortality in susceptible patients but improved cure probability only in older patients.
• Both cure models suggested treatments were more beneficial for older patients.

Conclusions:

• Cure models provide a more appropriate framework than competing risk models for analyzing in-hospital mortality when discharge signifies recovery.
• Findings highlight differential effects of treatments like statins based on patient age and susceptibility.
• Cure models offer valuable insights into disease processes and treatment efficacy in complex patient populations.