As healthcare and health systems become increasingly complex, expectations of what constitutes high quality care continue to evolve. Stakeholders now require contemporary and meaningful measures of system performance. As such, valid healthcare quality metrics are rapidly becoming essential for those providing and receiving healthcare to assess performance and motivate change. This thesis investigates the utility of quality indicators in trauma care. Multiple in-hospital indicators have been promulgated by various bodies for assessing quality of trauma care. The properties of ideal indicators have been widely documented however few published data have reported these properties for many trauma measures. The emphasis on trauma process measures (eg. time to interventions) highlights the need for indicators with known links to patient outcomes. This process-outcome link may be viewed as a measure of an indicator’s construct validity. As this property is unknown for many trauma indicators, this thesis focuses on the construct validity of a number of routinely utilised trauma indicators. In this thesis, the available in-hospital indicators proposed by The American College of Surgeons Committee on Trauma and additional indicators used in the Victorian State Trauma System were investigated for their relationships with patient outcomes. A small number of indicators were found to have statistically significant relationships with patient outcomes, however many indicators demonstrated counter-intuitive relationships, whereby high quality care was linked with poorer patient outcomes. These results suggested that links between indicators and outcomes may not be best measured using individual indicators for individual patients. Rather, a strategy for measuring patient outcomes at the hospital level may be needed. To combine multiple indicators into a single measure of hospital level performance, a number of composite methods were explored using two trauma registries. Three composite weighting schemes were employed. As composite measures are often used for provider ranking or benchmarking, the stability of hospital ranks between providers and over time was investigated. The composites were found to have moderate to strong correlations (0.76-0.99) however variability in composite hospital rankings existed, particularly for middle ranking facilities. The construct validity of each available indicator and composite score was investigated through the relationship with hospital level risk-adjusted mortality using Poisson regression models, risk adjusting for expected deaths using the TRISS formulation. Each composite measure demonstrated a significant association with mortality, with the mortality decrease across the middle 50% of each composite score ranging from 12.06% – 16.13%. These findings suggest that complex measures such as trauma composite indices may be better able to measure the interactions between processes within complex systems that influence quality of care. This thesis adds valuable insight into the use of indicators for assessing quality of care in trauma systems. The combination of individual indicators into composite forms appears to strengthen the construct validity of these measures. By demonstrating the process-outcome link for trauma composite indices, this thesis has identified a means of utilising process measures to assess hospital level performance that may become important for future public reporting and hospital funding schemes.
| Identifer | oai:union.ndltd.org:ADTP/234021 |
| Date | January 2008 |
| Creators | Willis, Cameron David |
| Publisher | Monash University. Faculty of Medicine, Nursing and Health Sciences. Department of Epidemiology and Preventive Medicine |
| Source Sets | Australiasian Digital Theses Program |
| Language | English |
| Detected Language | English |
| Rights | Full Embargo to thesis full text for 3 years after 2009. After this date access to thesis full text is restricted to institutional document delivery service requests as permitted by Section 51 (2) of the Australian Copyright Act 1968., This thesis is protected by copyright. Copyright in the thesis remains with the author. The Monash University Arrow Repository has a non-exclusive licence to publish and communicate this thesis online. |
Page generated in 0.0023 seconds