Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.491903
Title: Risk stratification modelling in post-operative abdominal aortic aneurysm patients
Author: Hadjianastassiou, Vassilis Georgiou
ISNI:       0000 0001 3523 6249
Awarding Body: University of Oxford
Current Institution: University of Oxford
Date of Award: 2006
Availability of Full Text:
Full text unavailable from EThOS.
Please contact the current institution’s library for further details.
Abstract:
Introduction: the aim of the project was to test the hypothesis that 'the principles of risk stratification modelling can be successfully applied in a combined group of both elective and emergency AAA patients in the immediate post-operative setting'. Methods: the applicability of two existing generic risk stratification models as predictors of in-hospital mortality was assessed in a combined group of elective and emergency post-operative abdominal aortic aneurysm patients, from 24 Intensive Care Units in the North-Thames. The better of the two models was chosen to develop a disease-specific risk stratification model (the APACHE-AAA) in this group of patients, using hierarchical logistic regression. The accuracy of this new model was compared to that of artificial neural networks and clinician predictions on the same patient population. The APACHE-AAA model was then externally validated in a patient population (from the Oxford/Lewisham Intensive Care Units) independent from the one used to develop it and the model's prognostic accuracy was also compared with existing risk stratification models (POSSUM or VBHOM based) advocated for use in abdominal aortic aneurysm patients. Results: The two generic risk stratification models did not predict outcome accurately when applied to the North-Thames population of patients. The disease-specific APACHE-AAA model subsequently developed, successfully predicted outcome in this patient population, as evidenced by all measures of internal validity such as callibration and discrimination properties and subgroup analyses. The APACHE-AAA and the corresponding artificial neural network models were found to be more accurate than Intensive Care resident doctors in quantifying prognosis. The artificial neural network model had inferior calibration properties to the logistic regression model. The APACHE-AAA model was successfully externally validated in the Oxford/Lewisham patient population. Existing POSSUM- and VBHOM-based risk stratification models did not model outcome accurately in this population. Conclusions: the principles of risk stratification modelling were successfully applied in post-operative abdominal aortic aneurysm patients. The APACHE-AAA model exemplifies the methodology required to formulate a national reference system for reliable assessments of the quality of intensive care, for evaluative research and in prognostication in this group of patients.
Supervisor: Not available Sponsor: Not available
Qualification Name: University of Oxford, 2006 Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.491903  DOI: Not available
Share: