In this thesis, measurement theory and the statistical evaluations of diagnostic test performance are applied to the measurement of injury severity. Relevant issues in traumatology, the rationale for measurement of injury severity and the importance of likelihood ratios (LRs) and Receiver operator characteristic (ROC) curve analyses are discussed. An assessment of the definition, scaling mechanism, reliability and validity of 22 severity instruments is organized into a reference guide. Data sources for this thesis include the literature, the Vancouver General Hospital trauma registry and the Pennsylvania State University trauma registry. / The LR and the area under ROC curves (A$ sb{ rm uc}$) are calculated from the best published evaluations of four triage instruments. The CRAMS, PHI, and RTI, which include anatomic information, are superior to the RTS, which contains only physiologic information. / A paired ROC curve analysis of the RTS and the RTI demonstrates that the performance of the RTI is equivalent to the RTS when mortality is the outcome; however the performance gain of the RTI over the RTS is 35% when major trauma is the outcome evaluated. / Explanations for differences in predictive validity are sought by evaluating scaling mechanisms, reliability and content validity. (Abstract shortened by UMI.)
| Identifer | oai:union.ndltd.org:LACETR/oai:collectionscanada.gc.ca:QMM.61337 |
| Date | January 1992 |
| Creators | Lett, Ronald |
| Publisher | McGill University |
| Source Sets | Library and Archives Canada ETDs Repository / Centre d'archives des thèses électroniques de Bibliothèque et Archives Canada |
| Language | English |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Format | application/pdf |
| Coverage | Master of Science (Department of Epidemiology and Biostatistics.) |
| Rights | All items in eScholarship@McGill are protected by copyright with all rights reserved unless otherwise indicated. |
| Relation | alephsysno: 001307469, proquestno: AAIMM80357, Theses scanned by UMI/ProQuest. |
Page generated in 0.2266 seconds