Sensitivity analysis of unmeasured confounding for adjusted survival curves using inverse probability weights

Chiba, Yasutaka (2009) Sensitivity analysis of unmeasured confounding for adjusted survival curves using inverse probability weights. Biomedical statistics and clinical epidemiology, 3 (3). pp. 135-140. ISSN 1973-2430

PDF portiere.pdf - Published Version Restricted to Repository staff only Download (711kB) | Request a copy

Abstract

Objectives. Kaplan–Meier survival curves adjusted for confounders have been presented based on the use of inverse probability weights. However, in observational studies, some confounders may not be measured. In these situations, it is important to evaluate the potential impact of unmeasured confounding quantitatively. Here, we propose a simple method for sensitivity analysis of adjusted survival curves using the inverse probability weights. Methods. We derived inverse probability weights, in which confounding risk ratios used as sensitivity parameters for unmeasured confounding were included. Using the weights with some plausible values for sensitivity parameters, a sensitivity analysis was performed. Results. The proposed method of sensitivity analysis was applied to observational study data. The results demonstrate that this new method can be applied to adjusted survival curves without complex computer programming. Conclusions. The proposed sensitivity analysis has the disadvantage of using sensitivity parameters as measures of risk as opposed to focusing on time to event. Nevertheless, the method is simple to perform and will aid researchers in evaluating the potential impact of unmeasured confounding on the Kaplan–Meier survival curves.

Item Type:	Article
Uncontrolled Keywords:	confounding risk ratio, epidemiologic methods, observational studies, potential outcomes.
Subjects:	500 Scienze naturali e Matematica > 570 Scienze della vita; Biologia > 570.15195 Statistica in Biologia (Classificare qui la Biometria, la Biostatistica) 600 Tecnologia - Scienze applicate > 610 Medicina e salute (Classificare qui la tecnologia dei servizi medici) > 610.7 Medicina - Educazione e Insegnamento; Medicina - Ricerca > 610.72 Statistica medica 600 Tecnologia - Scienze applicate > 610 Medicina e salute (Classificare qui la tecnologia dei servizi medici) > 614 Medicina legale; incidenza delle malattie; Medicina preventiva pubblica > 614.4 Incidenza delle malattie e misure pubblica per prevenirle (classificare qui l'Epidemiologia, l'Epidemiologia clinica)
Depositing User:	Emanuele Zinevrakis
Date Deposited:	07 Nov 2011 16:35
Last Modified:	07 Nov 2011 16:35
URI:	http://eprints.bice.rm.cnr.it/id/eprint/3507

Actions (login required)

View Item