Target Trial Emulation Using Hospital-Based Observational Data: Demonstration and Application in COVID-19

Area of Science:

• Epidemiology
• Biostatistics
• Health Services Research

Background:

• Observational studies frequently suffer from methodological biases when evaluating treatment effectiveness.
• Assessing treatment effectiveness in coronavirus disease 2019 (COVID-19) research is challenging due to limited survival data beyond hospital discharge.

Purpose of the Study:

• To emulate a target trial in a competing risks setting using hospital-based observational data for COVID-19 patients.
• To extend existing methodologies to address immortal time bias and time-fixed confounding biases in the context of limited survival data.

Main Methods:

• Emulation of a target trial using clone-censor-weight techniques on a cohort of 618 hospitalized COVID-19 patients.
• Extension of established bias-accounting methods to a competing risks framework.
• Application of cause-specific cumulative hazards and cumulative incidence probabilities for analysis.

Main Results:

• The target trial emulation framework was successfully extended to accommodate competing risks in COVID-19 hospital studies.
• The methodology simultaneously avoided immortal time bias, time-fixed confounding bias, and competing risks bias.
• A clinically justified grace period length ensured reliable and unbiased treatment effect estimation.

Conclusions:

• The extended trial emulation with competing risk analysis provides an unbiased estimation of treatment effects for COVID-19 patients.
• This approach allows for the interpretation of treatment effectiveness across all clinically significant outcomes.
• The method offers a robust solution for analyzing observational data with limited survival information.