EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

Inverse Probability of Treatment Weighting Using the Propensity Score With Competing Risks in Survival Analysis

Peter C Austin iD, Jason P Fine

Statistics in Medicine · 2025

DOI: 10.1002/sim.70009 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Statistics in Medicine, 2025. Verfasst von Peter C Austin und 1 weiteren Autor:innen. Beteiligte Einrichtung: ICES, Toronto, Ontario, Canada. Hauptthemen laut MEDLINE: Propensity Score. Die Arbeit wurde bisher 24-mal zitiert (Crossref) und nennt 68 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0). Gefördert von: CIHR, Canadian Institutes of Health Research.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Inverse probability of treatment weighting (IPTW) using the propensity score allows estimation of the effect of treatment in observational studies. We had three objectives: first, to describe methods for using IPTW to estimate the effects of treatments in settings with competing risks; second, to illustrate the application of these methods using empirical analyses; and third, to conduct Monte Carlo simulations to evaluate the relative performance of three methods for estimating time‐specific risk differences and time‐specific relative risks in settings with competing risks. In doing so, we provide guidance to applied biostatisticians and clinical investigators on the use of IPTW in settings with competing risks. We examined three estimators of time‐specific risk differences and relative risks: the weighted Aalen–Johansen estimator, an estimator that combines IPTW with inverse probability of censoring weights (IPTW‐IPCWs), and a double‐robust augmented IPTW estimator combined with IPCW (AIPTW‐IPCW). The design of our simulations reflected clinically realistic scenarios. Our simulations found that all three estimators tended to result in unbiased estimations of time‐specific risk differences and time‐specific relative risks. However, the weighted Aalen–Johansen estimator and the AIPTW‐IPCW estimator tended to result in estimates with greater precision compared to the IPTW‐IPCW estimator. In our empirical analyses, we illustrated the application of these methods by estimating the effect of statin prescribing on the risk of subsequent cardiovascular death in patients discharged from the hospital with a diagnosis of acute myocardial infarction.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Propensity ScoreHumansMonte Carlo MethodSurvival AnalysisComputer SimulationProbabilityObservational Studies as TopicModels, StatisticalMyocardial InfarctionRisk AssessmentHydroxymethylglutaryl-CoA Reductase Inhibitors

Stichwörter der Autor:innen

competing riskcumulative incidence functioninverse probability of treatment weightingpropensity scoresurvival analysis

Substanzen

Hydroxymethylglutaryl-CoA Reductase Inhibitors

Autor:innen und Institutionen

  • ICES, Toronto, Ontario, Canada.
  • Institute of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada.
  • Schulich Heart Research Program, Sunnybrook Research Institute, Toronto, Ontario, Canada.
  • Department of Statistics, University of Pittsburgh, Pittsburgh, Pennsylvania, USA.

ORCID-Profile: Peter C Austin

Förderung

  • CIHR
  • Canadian Institutes of Health Research

Zitieren

APA

Austin, P. C., & Fine, J. P. (2025). Inverse Probability of Treatment Weighting Using the Propensity Score With Competing Risks in Survival Analysis. Statistics in Medicine. https://doi.org/10.1002/sim.70009

Vancouver

Austin PC, Fine JP. Inverse Probability of Treatment Weighting Using the Propensity Score With Competing Risks in Survival Analysis. Statistics in Medicine. 2025. doi:10.1002/sim.70009

BibTeX
@article{Austin2025,
  author = {Austin, Peter C and Fine, Jason P},
  title = {Inverse Probability of Treatment Weighting Using the Propensity Score With Competing Risks in Survival Analysis},
  journal = {Statistics in Medicine},
  year = {2025},
  doi = {10.1002/sim.70009}
}
Kostenlos starten →