Annotated Bibliography: Modern Causal Inference and Targeted Learning

This bibliography introduces key readings in modern causal inference, targeted learning, and real-world evidence generation. Each annotation summarizes why the paper is important and what it contributes to causal reasoning and applied biostatistics.

1. Causal Inference Roadmap and Target Trial Emulation

Dang et al. (2023) – A causal roadmap for generating high-quality real-world evidence.
Introduces the Causal Roadmap framework for structuring causal analyses in observational data. Emphasizes transparency, assumptions, and reproducibility in real-world evidence.

Gruber et al. (2023) – Evaluating and improving real-world evidence with Targeted Learning.
Applies the roadmap to re-analyze published results using TMLE, highlighting the link between causal identification and robust estimation.

Williamson et al. (2023) – An application of the Causal Roadmap in two safety monitoring case studies.
Demonstrates roadmap principles in practice for safety monitoring and outcome prediction using electronic health records.

Hernán & Robins (2016) – Using big data to emulate a target trial when a randomized trial is not available.
Defines target trial emulation, a cornerstone idea for translating causal inference principles to observational study design.

2. Estimand Specification in Clinical Studies

ICH E9 (R1) Addendum (2019) – Addendum on Estimands and Sensitivity Analyses in Clinical Trials.
Establishes the estimand framework to align clinical trial objectives, analyses, and interpretations.

Rufibach (2019) – Treatment effect quantification for time-to-event endpoints – Estimands, analysis strategies, and beyond.
Applies the estimand framework to survival outcomes, clarifying how censoring and non-proportional hazards affect effect interpretation.

3. Super Learner Ensemble Learning

van der Laan, Polley & Hubbard (2007) – Super Learner.
A foundational paper introducing ensemble learning via cross-validation for optimal prediction and causal estimation.

Phillips et al. (2023) – Practical considerations for specifying a Super Learner.
A practical tutorial on constructing and validating Super Learners, including library specification, cross-validation, and reproducibility.

4. Targeted Maximum Likelihood Estimation (TMLE)

Gruber & van der Laan (2010) – Targeted Maximum Likelihood Estimation: A Gentle Introduction.
Provides a clear step-by-step introduction to TMLE, integrating machine learning and influence function theory for efficient, doubly robust estimation.

5. Time-Dependent Confounding and Intercurrent Events

Petersen (2014) – Applying a Causal Road Map in Settings with Time-dependent Confounding.
Discusses longitudinal causal inference and how to handle time-dependent confounding through g-methods and TMLE.

Stensrud et al. (2019) – Limitations of hazard ratios in clinical trials.
Explains why hazard ratios can be misleading as causal measures and encourages absolute or survival-based contrasts.

Martinussen (2022) – Causality and the Cox Regression Model.
Clarifies the mathematical and conceptual limitations of hazard ratios, advocating more interpretable causal estimands.

6. Dynamic Treatment Regimes and Stochastic Interventions

Chakraborty & Murphy (2014) – Dynamic Treatment Regimes.
A comprehensive overview of adaptive treatment strategies and their estimation through sequential designs and reinforcement learning methods.

Kennedy (2019) – Nonparametric causal effects based on incremental propensity score interventions.
Introduces stochastic interventions that shift treatment probabilities incrementally, addressing positivity violations and improving policy relevance.