EUVIMEDEUVIMED entdecken
Systematische ÜbersichtOpen Access · CC BY-NC-ND 4.0

Treatment switching in oncology evidence synthesis: a systematic review of meta-analytical practices incorporating time-dependent confounding, censoring bias, and patient-level variability

Edward Kurnia Setiawan Limijadi

Dialogues in Health · 2026-12 · 9 · 100346

DOI: 10.1016/j.dialog.2026.100346 ↗

Kurz gefasst

Systematische Übersicht in der Zeitschrift Dialogues in Health, 2026-12. Verfasst von Edward Kurnia Setiawan Limijadi. Beteiligte Einrichtung: Universitas Diponegoro, Indonesia. Die Arbeit nennt 23 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Treatment switching in the oncology trials introduced time-dependent confounding and informative censoring, which complicated the unbiased estimation of treatment effects in evidence synthesis.

Objectives

To systematically assess the meta-analytical practices that are used for adjusting treatment switching in oncology, with an emphasis on handling time-dependent confounding, censoring bias, as well as patient-level variability.

Methods

A systematic review was conducted using PRISMA 2020 guidelines. The databases searched systematically are MEDLINE, Embase, Cochrane Library and Web of Science, and the studies published between 2015 and the last search date were included (provide exact search date). The review was conducted with the reporting standards from the PRISMA 2020 statement, and because there was methodological heterogeneity, an analysis of the findings was conducted using a structured narrative synthesis. The eligible studies included randomized trials, simulation studies, observational studies, and health technology assessments, with the application of formal statistical methods to adjust for treatment switching.

Results

Six studies had met the inclusion criteria. Common methods involved rank-preserving structural failure time models (RPSFTM), marginal structural models (MSMs), and inverse probability of censoring weighting (IPCW). The modeling assumptions, the specification of the model, the switching mechanisms, the method of dealing with time-dependent confounders, and the method of dealing with censoring were shown to vary across the included studies and affect the results and estimates of the treatment effects. The reporting practices were varying, and there was limited integration into the meta-analytical frameworks.

Conclusion

Adjustment regarding the treatment switching is methodologically complex, having no universally optimal approach. Enhanced level of transparency, methodological standardization, along with the consideration of patient-level heterogeneity, is important to improve the validity of the oncology evidence synthesis.

Schlagwörter

Stichwörter der Autor:innen

And inverse probability of censoring weighting (IPCW)Marginal structural models (MSMs)OncologyRank-preserving structural failure time models (RPSFTM)Treatment switching

Autor:innen und Institutionen

  • Universitas Diponegoro, Indonesia.

Zitieren

APA

Limijadi, E. K. S. (2026). Treatment switching in oncology evidence synthesis: a systematic review of meta-analytical practices incorporating time-dependent confounding, censoring bias, and patient-level variability. Dialogues in Health, 9, 100346. https://doi.org/10.1016/j.dialog.2026.100346

Vancouver

Limijadi EKS. Treatment switching in oncology evidence synthesis: a systematic review of meta-analytical practices incorporating time-dependent confounding, censoring bias, and patient-level variability. Dialogues in Health. 2026;9:100346. doi:10.1016/j.dialog.2026.100346

BibTeX
@article{Limijadi2026,
  author = {Limijadi, Edward Kurnia Setiawan},
  title = {Treatment switching in oncology evidence synthesis: a systematic review of meta-analytical practices incorporating time-dependent confounding, censoring bias, and patient-level variability},
  journal = {Dialogues in Health},
  year = {2026},
  volume = {9},
  pages = {100346},
  doi = {10.1016/j.dialog.2026.100346}
}
Kostenlos starten →