Summary:
Survival analysis deals with the statistical analysis of the time to the
occurrence of an event like death, relapse, or graft failure. Survival analysis
takes into account censored data, which is important to estimate accurate
results. It is presented parametric, non-parametric methods and semi parametric
Cox proportional hazards model.
Often, in the disease or recovery process of a patient, multiple types of
events can occur. These different events may be mutually exclusive or may occur
sequentially. The first instance is called competing risks, the second a
multi-state model. Multi-state models generalize competing risks describing and
analyzing transitions of intermediate events.
This thesis’s goal is to assess whether and how nosocomial infections
(ventilator associated pneumonia-VAP, bloodstream infection and urinary
infection) acquired in the hospital affect to mortality and length of stay in
the intensive care units.
Multi state models are used to analyze the data of 294 ventilated patients. In
a multivariate model where infection, Apache score and duration of ventilation
were included we observe that the attributable mortality of VAP was 5,5% and
the change of length of stay due to VAP was 5.9±1.8 days.
Keywords:
Survival analysis, Multi state models, Competing risks, Ventilator associated pneumonia, R
