Instrumental variable estimation of the causal hazard ratio

Area of Science:

• Biostatistics
• Epidemiology
• Statistical Modeling

Background:

• Cox's proportional hazards model is widely used for time-to-event data.
• Unmeasured confounding factors can bias hazard ratio estimates in Cox models.
• Accurate estimation of exposure effects is crucial in observational studies.

Purpose of the Study:

• To develop a novel approach for estimating the causal hazard ratio.
• To address bias introduced by unmeasured confounding factors.
• To provide a consistent estimator within an instrumental variable framework.

Main Methods:

• Utilized a binary instrumental variable approach.
• Incorporated a no-interaction assumption in a first-stage regression.
• Developed a novel consistent estimator for the causal hazard ratio.

Main Results:

• Proposed the first consistent estimator for the causal hazard ratio in an instrumental variable setting.
• Derived the asymptotic distribution and variance estimator for the proposed method.
• Demonstrated the approach's validity through simulations and a data application.

Conclusions:

• The proposed instrumental variable method effectively estimates causal hazard ratios.
• This approach mitigates bias from unmeasured confounding in survival analysis.
• Offers a robust tool for epidemiological and biostatistical research.