Events

Seminar - Multistate approaches to model nosocomial pneumonia disease in intensive care units


9 March 2012

Presenter: Associate Professor Benoit Liquet, French National Institute of Health (INSERM)

Abstract:
The multistate models have become increasingly used to describe the occurrence of nosocomial infection in intensive care unit (ICU). In this presentation, we focus on the analysis of ventilator-associated pneumonia infections (VAP) in ICU by multistate models and suggest two novel approaches. We first interest on the estimation of the attributable mortality of VAP in a large multi-center cohort. Recently, a multistate model has been developed in order to take into consideration both the time-dependency of the risk factor (e.g., VAP) and the presence of competing risks (e.g., death and discharge) at each time point. However, this approach does not take into account the possible heterogeneity of the study population. We here extend the model including fixed covariates in the definition of attributable mortality. The methodology developed is applied to data on ventilator-associated pneumonia in 12 French intensive care units. Then, we propose a multistate frailty model to take into account that data come from different ICUs. The hypothesis of independent outcomes when observations are clustered into groups (or units) is not obvious, thus a flexible multi-state model with random effects is needed to obtain valid estimates. We show that the analysis of dependent survival data using a multi-state model without frailty terms may underestimate the variance of regression coefficients specific to each group, leading to incorrect inferences. Some factors were wrongly significantly associated based on the model without frailty terms. This result was confirmed by a short simulation study. We also present individual predictions of VAP underlining the usefulness of dynamic prognostic tools that can take into account the clustering of observations. Finally, we suggest a method that gives accurate estimates and enables inference for any parameter or predictive quantity of interest.

About the speaker: Benoit Liquet is an Associate Professor at INSERM (the French national institute of health) in Bordeaux. His PhD in 2002 was on semi-parametric model selection. His personal research interests include model Selection, dimension reduction and semi-parametric models, multi-state and survival models, and multiple testing.

Hosted by The George Institute for International Health


Time: 11am

Location: 341 George Street, Level 6, training room 1 (Wynyard)

Contact: Serigne Lo

Phone: 02 9657 0329

Email: 170e1e192d235b1c2908035e0122394d331f286f1742031b153a