Causal Mediation Analysis of Non‐Mortality Outcomes With Follow‐Up Truncated by Death: Survivor Natural Direct and Indirect Effects
In medical studies, the death-truncation problem arises when subjects pass away between follow-up visits, leading to death-truncated non-mortality outcomes and mediators. Traditional measures in mediation analysis, such as natural direct and indirect effects, become inadequate for evaluating the mediator's role in the treatment's impact on the primary non-mortality outcome. To address this issue, this study introduces novel estimands-survivor natural direct and indirect effects-for causal mediation analysis involving death-truncated variables. By assuming exchangeability and utilizing a pretreatment surrogate variable, three estimation approaches are developed. The proposed method is illustrated using data from the National Emphysema Treatment Trial.
Authors
- An‐Shun Tai (ORCID: https://orcid.org/0000-0002-2583-0188)
- Tsung Yu (ORCID: https://orcid.org/0000-0002-5773-3665)
- Yu‐Cheng Chu
- Chen‐Hua Cho
- Sheng‐Hsuan Lin
- Yun‐Hao Chang
Institutions
- National Yang Ming Chiao Tung University (TW)
- National Tsing Hua University (TW)
- National Dong Hwa University (TW)
- National Cheng Kung University (TW)
Publication Details
- Journal
- Biometrical Journal
- Published
- 2026-11-17
- DOI
- https://doi.org/10.1002/bimj.70199
- Primary Topic
- Advanced Causal Inference Techniques
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- National Science and Technology Council
- National Science and Technology Council