What question did this study set out to answer?

To assess how the accuracy of Cox's proportional hazards model and conventional logistic regression depends on risk set definitions and exposure characteristics in matched case-control studies with time-varying covariates.

December 8, 2003Statistics in Medicine

Evaluation of Cox's model and logistic regression for matched case‐control data with time‐dependent covariates: a simulation study

View Full Paper

Ask AI

Bookmark

Authors

KLKaren LeffondréUniversité de Bordeaux MAMichał AbrahamowiczUniversité du Québec à Chicoutimi JSJack SiemiatyckiInstitute for Prevention and Occupational Medicine

Discussion

Member takes

Implication

Simulation study reveals estimation biases between Cox and logistic models in matched case-control designs, indicating that conventional risk-set handling distorts time-varying exposure effects.

Key Points

To assess how the accuracy of Cox's proportional hazards model and conventional logistic regression depends on risk set definitions and exposure characteristics in matched case-control studies with time-varying covariates.
Generated simulated lifetime trajectories for a hypothetical population and sampled matched case-control sets across controlled exposure frequencies, ages at initiation, and durations.
Evaluated conventional logistic regression alongside two operational risk-set definitions of the Cox model, each adjusting for a fixed covariate and one or two time-dependent covariates (current exposure and exposure duration).
Conventional logistic regression produced odds ratios that deviated from the true hazard ratio depending on covariate type and effect magnitude, while struggling to untangle inter-correlated time-dependent covariates.
Both implementations of Cox's model exhibited systematic bias, yielding either substantial underestimation or moderate overestimation that scaled proportionally with the true effect size.

Cite This Study

Leffondré et al. (2003) studied this question.

synapsesocial.com/papers/6a0af8977e716524c8acba88 https://doi.org/10.1002/sim.1674

View Full Paper

Ask AI

Bookmark

Implication

Simulation study reveals estimation biases between Cox and logistic models in matched case-control designs, indicating that conventional risk-set handling distorts time-varying exposure effects.

Evaluation of Cox's model and logistic regression for matched case‐control data with time‐dependent covariates: a simulation study

Discussion

Member takes

Implication

Key Points

Cite This Study

Evaluation of Cox's model and logistic regression for matched case‐control data with time‐dependent covariates: a simulation study

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider