The development, use, and understanding of severity of illness scoring systems has advanced rapidly in the last decade; their weaknesses and limitations have also become apparent. This work follows some of this development and explores some of these aspects. It was undertaken in three stages and in two countries. The first study investigated three severity of illness scoring systems in a general Intensive Care Unit (ICU) in Cape Town, namely the Acute Physiology and Chronic Health Evaluation (APACHE II) score, the Therapeutic Intervention Scoring System (TISS), and a locally developed organ failure score. All of these showed a good relationship with mortality, with the organ failure score the best predictor of outcome. The TISS score was felt to be more likely to be representative of intensiveness of medical and nursing management than severity of illness. The APACHE II score was already becoming widely used world-wide and although it performed less well in some diagnostic categories (for example Adult Respiratory Distress Syndrome) than had been hoped, it clearly warranted further investigation. Some of the diagnosis-specific problems were eliminated in the next study which concentrated on the application of the APACHE II score in a cardiothoracic surgical ICU in London. Although group predictive ability was statistically impressive, the predictive ability of APACHE II in the individual patient was limited as only very high APACHE II scores confidently predicted death and then only in a small number of patients. However, there were no deaths associated with an APACHE II score of less than 5 and the mortality was less than 1 % when the APACHE II score was less than 10. Finally, having recognised the inadequacies in mortality prediction of the APACHE II score in this scenario, a study was undertaken to evaluate a novel concept: a combination of preoperative, intraoperative, and postoperative (including APACHE II and III) variables in cardiac surgery patients admitted to the same ICU. The aim was to develop a more precise method of predicting length of stay, incidence of complications, and ICU and hospital outcome for these patients. There were 1008 patients entered into the study. There was a statistically significant relationship between increasing Parsonnet (a cardiac surgery risk prediction score), APACHE II, and APACHE III scores and mortality. By forward stepwise logistic regression a model was developed for the probability of hospital death. This model included bypass time, need for inotropes, mean arterial pressure, urea, and Glasgow Coma Scale. Predictive performance was evaluated by calculating the area under the receiver operating characteristic (ROC) curve. The derived model had an area under the ROC curve 0.87, while the Parsonnet score had an area of 0.82 and the APACHE II risk of dying 0.84. It was concluded that a combination of intraoperative and postoperative variables can improve predictive ability.
Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:uct/oai:localhost:11427/27057 |
Date | January 1995 |
Creators | Turner, John Scott |
Publisher | University of Cape Town, Faculty of Health Sciences, Department of Surgery |
Source Sets | South African National ETD Portal |
Language | English |
Detected Language | English |
Type | Doctoral Thesis, Doctoral, MD |
Format | application/pdf |
Page generated in 0.0316 seconds