Propensity score matching for treatment delay effects with observational survival data

Area of Science:

• Biostatistics
• Epidemiology
• Clinical Research

Background:

• Treatment initiation in observational studies with survival outcomes is often time-dependent and influenced by covariates.
• When treatment is essential for the study population, the focus shifts to the causal effect of treatment delay.

Purpose of the Study:

• To propose a propensity score matching strategy to estimate the causal effect of treatment delay.
• To extend existing matching designs to accommodate time-varying covariates.

Main Methods:

• Utilized risk set matching to balance covariate distributions between early and delayed treatment groups at each time point.
• Extended Lu's matching design to incorporate time-varying covariates for treatment delay estimation.

Main Results:

• Simulation studies demonstrated that the matching-based analysis significantly outperformed conventional regression analysis (naive Cox model) in the presence of treatment delay effects.
• The method was successfully applied to assess the treatment delay effect of 17 alpha-hydroxyprogesterone caproate (17P) in recurrent preterm birth.

Conclusions:

• Propensity score matching with risk set adjustments is a robust method for estimating treatment delay effects in observational survival studies.
• This approach provides a more accurate assessment of treatment effects compared to traditional methods when treatment timing is a critical factor.