BACKGROUND: To investigate the association between a number of hospital level composite index methodologies developed from trauma indicators with inhospital mortality. METHODS: Data from January 2001 to December 2006 were extracted from the Victorian State Trauma Registry (Australia) and the Trauma Audit and Research Network (United Kingdom). Three composite methods were explored, including two denominator-based weight approaches and a factor analysis technique. The association between the composite measures and the count of inhospital mortality was investigated using Poisson regression models adjusting for expected deaths per hospital using the Trauma Injury Severity Score methodology. RESULTS: Composite scores were calculated per hospital, per year. The composite score was entered in statistical models as a raw score, and the mortality difference across the central 50% of the composite index was ascertained. In total, 9,218 patients were included and were distributed across 14 hospitals. Composite scores demonstrated an inverse relationship with risk-adjusted inhospital mortality. From the 25th to the 75th percentile of each composite, mortality decreased by 11.99%, 13.58%, and 16.13% (p < 0.05). CONCLUSION: Trauma composite indices demonstrate construct validity when used as measures of hospital level process and represent potentially useful methods of analyzing and reporting quality indicators.
BACKGROUND: To investigate the association between a number of hospital level composite index methodologies developed from trauma indicators with inhospital mortality. METHODS: Data from January 2001 to December 2006 were extracted from the Victorian State Trauma Registry (Australia) and the Trauma Audit and Research Network (United Kingdom). Three composite methods were explored, including two denominator-based weight approaches and a factor analysis technique. The association between the composite measures and the count of inhospital mortality was investigated using Poisson regression models adjusting for expected deaths per hospital using the Trauma Injury Severity Score methodology. RESULTS: Composite scores were calculated per hospital, per year. The composite score was entered in statistical models as a raw score, and the mortality difference across the central 50% of the composite index was ascertained. In total, 9,218 patients were included and were distributed across 14 hospitals. Composite scores demonstrated an inverse relationship with risk-adjusted inhospital mortality. From the 25th to the 75th percentile of each composite, mortality decreased by 11.99%, 13.58%, and 16.13% (p < 0.05). CONCLUSION:Trauma composite indices demonstrate construct validity when used as measures of hospital level process and represent potentially useful methods of analyzing and reporting quality indicators.