pith. sign in

arxiv: 2412.06114 · v5 · pith:TG7JE5MFnew · submitted 2024-12-09 · 📊 stat.ME

Randomized interventional effects in semicompeting risks, with application to a hematopoietic cell transplantation study

Yuhao Deng , Rui Wang , Tao Zhang , Xiang Zhan This is my paper

Pith reviewed 2026-05-23 07:59 UTC · model grok-4.3

classification 📊 stat.ME
keywords semicompeting risksinterventional effectsmediation analysisright censoringtime-to-eventhematopoietic cell transplantationnonparametric estimation
0
0 comments X

The pith

The randomized interventional approach identifies direct and indirect effects on terminal events in semicompeting risks data with right censoring.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper extends the randomized interventional approach to decompose the total effect of a treatment on a terminal event into direct and indirect effects mediated by an intermediate event in semicompeting risks settings. Both the intermediate and terminal events are subject to right censoring. The key is to use a random draw of the intermediate event process from a reference distribution to identify the effects. Estimation is done with nonparametric maximum likelihood, and sensitivity analysis uses a latent frailty. The method is illustrated in a study of donor types in hematopoietic cell transplantation.

Core claim

We extend the randomized interventional approach to time-to-event outcomes, where both intermediate and terminal events are subject to right censoring. We present the identification formula for interventional effects.

What carries the argument

A random draw for the intermediate event process from a reference distribution, either marginally over time-varying confounders or conditionally given the observed history, that identifies the interventional effects.

Load-bearing premise

The random draw for the intermediate event process from a reference distribution identifies the interventional effects under the assumptions for the semicompeting risks setting with censoring.

What would settle it

If the identification formula does not recover the known effects in a simulation study where the data generating process matches the assumptions, the approach would be falsified.

Figures

Figures reproduced from arXiv: 2412.06114 by Rui Wang, Tao Zhang, Xiang Zhan, Yuhao Deng.

Figure 1
Figure 1. Figure 1: A graph illustrating the events. Red lines indicate the paths to deliver treatment effect on the inter [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Directed acyclic graphs (DAGs) for the counting processes. In the left figure, the counting pro [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Upper: Estimated cumulative incidence functions of death associated with Haplo-HCT, MUD [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Upper: Estimated cumulative incidence functions of death associated with Haplo-HCT, MUD [PITH_FULL_IMAGE:figures/full_fig_p014_4.png] view at source ↗
read the original abstract

In clinical studies, the risk of the primary (terminal) event may be modified by intermediate events, resulting in semicompeting risks. To study the treatment effect on the terminal event mediated by the intermediate event, researchers wish to decompose the total effect into direct and indirect effects. In this article, we extend the randomized interventional approach to time-to-event outcomes, where both intermediate and terminal events are subject to right censoring. We envision a random draw for the intermediate event process from a reference distribution, either marginally over time-varying confounders or conditionally given the observed history. We present the identification formula for interventional effects. We also discuss some variants of the identification assumptions. We estimate the treatment effects using nonparametric maximum likelihood estimation and propose a sensitivity analysis that incorporates a latent frailty. As an illustration, we study the effect of matched unrelated donor versus haploidentical donor on death mediated by relapse in a hematopoietic cell transplantation study with graft-versus-host disease (GVHD) as the time-varying confounder. We find that matched unrelated donor transplantation is preferable in terms of survival rates under the use of post-transplant PTCy GVHD prophylaxis for lymphoma patients.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 2 minor

Summary. The paper extends randomized interventional effects to semicompeting risks settings with right censoring on both the intermediate event (e.g., relapse) and terminal event (e.g., death). It defines interventional effects via a random draw of the intermediate event process from a reference distribution (marginal over time-varying confounders or conditional on observed history), presents an identification formula under stated assumptions (with variants discussed), estimates effects via nonparametric maximum likelihood, proposes a frailty-based sensitivity analysis for unmeasured confounding, and applies the method to compare matched unrelated donor vs. haploidentical donor transplantation on death mediated by relapse (with GVHD as time-varying confounder) in a hematopoietic cell transplantation study, concluding that matched unrelated donor is preferable under post-transplant PTCy GVHD prophylaxis for lymphoma patients.

Significance. If the identification and estimation hold, the work offers a useful extension of interventional effect decompositions to censored time-to-event data with semicompeting risks, enabling direct/indirect effect analysis in clinical settings like transplantation studies. The frailty sensitivity analysis strengthens robustness claims, and the real-data application provides concrete findings on donor choice. Strengths include targeting external identification assumptions rather than circular reductions and addressing censoring explicitly.

minor comments (2)
  1. [Abstract] Abstract states that an identification formula is presented but provides no explicit expression or key assumptions; adding a compact version of the formula (or reference to its equation number) would improve accessibility without lengthening the abstract unduly.
  2. [Identification and assumptions section] The description of the random draw for the intermediate process (marginal vs. conditional) is central; ensure the main text explicitly contrasts how each variant handles time-varying confounders and right censoring on both events.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive summary of our work and for recommending minor revision. The referee's description accurately reflects the manuscript's contributions to extending randomized interventional effects to right-censored semicompeting risks data, including identification, nonparametric estimation, frailty-based sensitivity analysis, and the hematopoietic cell transplantation application. No major comments were enumerated in the report.

Circularity Check

0 steps flagged

No significant circularity; derivation self-contained

full rationale

The paper extends an existing randomized interventional framework to semicompeting risks with right censoring on both events by stating an identification formula that draws the intermediate process from a reference distribution (marginal or conditional). Estimation proceeds via nonparametric maximum likelihood, with a separate frailty-based sensitivity analysis. No equation reduces a claimed prediction to a fitted input by construction, no load-bearing step collapses to a self-citation, and the central identification rests on externally stated assumptions rather than internal redefinition. The provided abstract and reader's assessment confirm the absence of the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review provides no explicit free parameters, axioms, or invented entities; identification relies on unstated assumptions about the random draw of the intermediate process and censoring mechanisms.

pith-pipeline@v0.9.0 · 5735 in / 1083 out tokens · 22019 ms · 2026-05-23T07:59:25.053192+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

28 extracted references · 28 canonical work pages

  1. [1]

    Deng, Y., Wang, Y., and Zhou, X.-H. (2024). Direct and indirect treatment effects in the presence of semicompeting risks. Biometrics , 80(2):ujae032

    work page 2024

  2. [2]

    D \' az, I. (2024). Non-agency interventions for causal mediation in the presence of intermediate confounding. Journal of the Royal Statistical Society Series B: Statistical Methodology , 86(2):435--460

    work page 2024

  3. [3]

    and Hejazi, N

    D \' az, I. and Hejazi, N. S. (2020). Causal mediation analysis for stochastic interventions. Journal of the Royal Statistical Society Series B: Statistical Methodology , 82(3):661--683

    work page 2020

  4. [4]

    D \' az, I., Williams, N., and Rudolph, K. E. (2023). Efficient and flexible mediation analysis with time-varying mediators, treatments, and confounders. Journal of Causal Inference , 11(1):20220077

    work page 2023

  5. [5]

    P., Jiang, H., and Chappell, R

    Fine, J. P., Jiang, H., and Chappell, R. (2001). On semi-competing risks data. Biometrika , 88(4):907--919

    work page 2001

  6. [6]

    Martin, A., Arora, M., Al Malki, M., Antin, J

    Gooptu, M., Romee, R., St. Martin, A., Arora, M., Al Malki, M., Antin, J. H., Bredeson, C. N., Brunstein, C. G., Chhabra, S., Fuchs, E. J., et al. (2021). Hla-haploidentical vs matched unrelated donor transplants with posttransplant cyclophosphamide-based prophylaxis. Blood, The Journal of the American Society of Hematology , 138(3):273--282

    work page 2021

  7. [7]

    R., Zhang, M.-J., Elmariah, H., Johnson, M

    Grunwald, M. R., Zhang, M.-J., Elmariah, H., Johnson, M. H., St. Martin, A., Bashey, A., Battiwalla, M., Bredeson, C. N., Copelan, E., Cutler, C. S., et al. (2021). Alternative donor transplantation for myelodysplastic syndromes: haploidentical relative and matched unrelated donors. Blood Advances , 5(4):975--983

    work page 2021

  8. [8]

    Huang, Y.-T. (2021). Causal mediation of semicompeting risks. Biometrics , 77(4):1143--1154

    work page 2021

  9. [9]

    Imai, K., Keele, L., and Yamamoto, T. (2010). Identification, inference and sensitivity analysis for causal mediation effects. Statistical Science , 25(1):51--71

    work page 2010

  10. [10]

    Kolb, H.-J. (2008). Graft-versus-leukemia effects of transplantation and donor lymphocytes. Blood, The Journal of the American Society of Hematology , 112(12):4371--4383

    work page 2008

  11. [11]

    and VanderWeele, T

    Lin, S.-H. and VanderWeele, T. (2017). Interventional approach for path-specific effects. Journal of Causal Inference , 5(1):20150027

    work page 2017

  12. [12]

    V., Symons, H

    Luznik, L., O'Donnell, P. V., Symons, H. J., Chen, A. R., Leffell, M. S., Zahurak, M., Gooley, T. A., Piantadosi, S., Kaup, M., Ambinder, R. F., et al. (2008). HLA -haploidentical bone marrow transplantation for hematologic malignancies using nonmyeloablative conditioning and high-dose, posttransplantation cyclophosphamide. Biology of Blood and Marrow Tra...

    work page 2008

  13. [13]

    Miles, C. H. (2023). On the causal interpretation of randomised interventional indirect effects. Journal of the Royal Statistical Society Series B: Statistical Methodology , 85(4):1154--1172

    work page 2023

  14. [14]

    H., Shpitser, I., Kanki, P., Meloni, S., and Tchetgen Tchetgen, E

    Miles, C. H., Shpitser, I., Kanki, P., Meloni, S., and Tchetgen Tchetgen, E. J. (2020). On semiparametric estimation of a path-specific effect in the presence of mediator-outcome confounding. Biometrika , 107(1):159--172

    work page 2020

  15. [15]

    Mittinty, M. N. and Vansteelandt, S. (2020). Longitudinal mediation analysis using natural effect models. American Journal of Epidemiology , 189(11):1427--1435

    work page 2020

  16. [16]

    Murphy, S. A. and Van der Vaart, A. W. (2000). On profile likelihood. Journal of the American Statistical Association , 95(450):449--465

    work page 2000

  17. [17]

    S., Wang, T., He, M., Hamadani, M., Finel Sr, H., Boumendil Sr, A., Glass, B., Castagna, L., Dominietto, A., et al

    Mussetti, A., Kanate, A. S., Wang, T., He, M., Hamadani, M., Finel Sr, H., Boumendil Sr, A., Glass, B., Castagna, L., Dominietto, A., et al. (2023). Haploidentical versus matched unrelated donor transplants using post-transplantation cyclophosphamide for lymphomas. Transplantation and cellular therapy , 29(3):184--e1

    work page 2023

  18. [18]

    M., Richardson, T

    Robins, J. M., Richardson, T. S., and Shpitser, I. (2022). An interventionist approach to mediation analysis. In Probabilistic and causal inference: the works of Judea Pearl , pages 713--764. Oxford University Press

    work page 2022

  19. [19]

    J., Hern \'a n, M

    Stensrud, M. J., Hern \'a n, M. A., Tchetgen Tchetgen, E. J., Robins, J. M., Didelez, V., and Young, J. G. (2021). A generalized theory of separable effects in competing event settings. Lifetime Data Analysis , 27(4):588--631

    work page 2021

  20. [20]

    J., Young, J

    Stensrud, M. J., Young, J. G., Didelez, V., Robins, J. M., and Hern \'a n, M. A. (2022). Separable effects for causal inference in the presence of competing events. Journal of the American Statistical Association , 117(537):175--183

    work page 2022

  21. [21]

    A., and VanderWeele, T

    Tai, A.-S., Lin, S.-H., Chu, Y.-C., Yu, T., Puhan, M. A., and VanderWeele, T. (2023). Causal mediation analysis with multiple time-varying mediators. Epidemiology , 34(1):8--19

    work page 2023

  22. [22]

    VanderWeele, T. J. (2011). Causal mediation analysis with survival data. Epidemiology , 22(4):582--585

    work page 2011

  23. [23]

    VanderWeele, T. J. and Tchetgen Tchetgen, E. J. (2017). Mediation analysis with time varying exposures and mediators. Journal of the Royal Statistical Society Series B: Statistical Methodology , 79(3):917--938

    work page 2017

  24. [24]

    J., Vansteelandt, S., and Robins, J

    VanderWeele, T. J., Vansteelandt, S., and Robins, J. M. (2014). Effect decomposition in the presence of an exposure-induced mediator-outcome confounder. Epidemiology , 25(2):300--306

    work page 2014

  25. [25]

    D., and Tai, B

    Xu, J., Kalbfleisch, J. D., and Tai, B. (2010). Statistical analysis of illness--death processes and semicompeting risks data. Biometrics , 66(3):716--725

    work page 2010

  26. [26]

    and Lin, D

    Zeng, D. and Lin, D. (2007). Maximum likelihood estimation in semiparametric regression models with censored data. Journal of the Royal Statistical Society Series B: Statistical Methodology , 69(4):507--564

    work page 2007

  27. [27]

    and Lin, D

    Zeng, D. and Lin, D. (2009). Semiparametric transformation models with random effects for joint analysis of recurrent and terminal events. Biometrics , 65(3):746--752

    work page 2009

  28. [28]

    and van der Laan, M

    Zheng, W. and van der Laan, M. (2017). Longitudinal mediation analysis with time-varying mediators and exposures, with application to survival outcomes. Journal of Causal Inference , 5(2):20160006

    work page 2017