Causal mediation analysis with mediator values below an assay limit.

Area of Science:

• Biostatistics
• Epidemiology
• Clinical Trials

Background:

• Causal inference decomposes total effects into direct and indirect pathways.
• Biomarkers as mediators in clinical trials can be subject to assay lower limits.
• Assay limits complicate causal mediation analysis.

Purpose of the Study:

• To propose and evaluate methods for estimating causal indirect and direct effects with assay-limited mediators.
• To address the compounded problem of assay limits in mediation analysis.
• To apply these methods to HIV research.

Main Methods:

• Two novel estimation approaches: extrapolation and numerical optimization/integration.
• Methods rely on the Mediation Formula, applicable to various causal mediation analyses.
• A simulation study compared proposed methods against imputation with half the assay limit.

Main Results:

• The proposed methods provide accurate estimation of indirect and direct effects with assay-limited mediators.
• Simulation results demonstrate the performance of extrapolation and numerical optimization.
• Comparison highlights advantages over simple imputation methods.

Conclusions:

• The developed methods effectively handle assay lower limits in causal mediation analysis.
• These techniques enhance the interpretability of treatment effects in clinical trials.
• Application to HIV data demonstrates utility in estimating treatment effects on viral suppression via HIV RNA measures.