Author correction to: "causal survival analysis under competing risks using longitudinal modified treatment policies"

Area of Science:

• Causal inference
• Longitudinal data analysis
• Survival analysis with competing risks

Background:

• The published manuscript contained errors in outcome and object definitions, impacting longitudinal modified treatment policy (LMTP) analysis.
• Longitudinal modified treatment policies (LMTPs) are a novel method for defining and estimating causal parameters related to treatment's natural value in longitudinal studies.
• LMTPs enable non-parametric definition and estimation of joint effects for multiple treatments over time.

Purpose of the Study:

• To correct errors in a previously published manuscript on LMTPs.
• To extend LMTP methodology to time-to-event outcomes with competing risks.
• To present identification results and robust, data-adaptive estimators for causal inference in complex longitudinal settings.

Main Methods:

• Correction of errors in the original manuscript's definitions.
• Extension of LMTP methodology to handle time-to-event data with competing events.
• Development of non-parametric, locally efficient, and data-adaptive estimators to minimize model misspecification bias.

Main Results:

• The corrected manuscript provides a robust framework for causal inference using LMTPs in the presence of competing risks.
• The proposed estimators are $$-consistent and utilize flexible regression techniques.
• Demonstrated application in estimating the effect of time-to-intubation on acute kidney injury in COVID-19 patients, accounting for death as a competing event.

Conclusions:

• The extended LMTP methodology provides a powerful tool for causal inference in longitudinal studies with time-to-event outcomes and competing risks.
• The data-adaptive estimators offer improved reliability by mitigating model misspecification.
• This work refines causal inference techniques for complex health outcomes, exemplified by the COVID-19 patient analysis.