External Benchmarking of Trauma Center Performance

Authors:
Publisher website
Google Scholar
Abstract
The elderly injured have been identified as a population with unique needs compared with nonelderly trauma patients. We sought to determine whether trauma center (TC) performance is consistent across age groups and to assess whether aggregate evaluations of TC performance capture quality of care among the elderly. The recently launched Trauma Quality Improvement Program utilizes external benchmarking of TC outcomes to identify centers with above-average performance, with the goal of disseminating best practices. If variation exists in TC performance across age groups, such variation might significantly impact on the success of external benchmarking programs in improving quality of care. Study data were derived from the National Trauma Databank (2007), limited to level I and II centers and adults with moderate to severe injuries (injury severity score > 9). Separate logistic regression models were constructed to produce TC risk-adjusted mortality for both the young and the elderly (age > 65 years). Observed-to-expected mortality ratios were used to identify centers with above or below average performance overall, among the young and among the elderly. We identified 87,754 patients across 132 facilities; 25% were elderly. After adjustment for case mix, 9 centers were identified as above-average performers in the elderly population. Only 2 of these centers were also above-average performers among young patients. Overall, concordance for center performance across age strata evidenced poor agreement (κ, 0.23). In addition, aggregate assessment of center performance did not reliably identify high-performing centers for elderly patients. The use of outcome-based benchmarking harbors significant potential for trauma quality improvement. Evaluations of aggregate TC performance may not adequately reflect the care provided to the elderly injured. Elderly trauma patients may warrant special attention in the context of ongoing quality improvement programs.

This publication has 17 references indexed in Scilit: