Abstract
STUDY OBJECTIVE: The measurement of emergency department (ED) throughput as a patient-centered quality measure is ubiquitous; however, marked heterogeneity exists between EDs, complicating comparisons for payment purposes. We evaluate 4 scoring methodologies for accommodating differences in ED visit volume and heterogeneity among ED groups that staff multiple EDs to improve the validity and "fairness" of ED throughput quality measurement in a national registry, with the goal of developing a volume-adjusted throughput measure that balances variation at the ED group level. METHODS: We conducted an ED group-level analysis using the 2017 American College of Emergency Physicians Clinical Emergency Data Registry data set, which included 548 ED groups inclusive of 889 unique EDs. We calculated ED throughput performance scores for each ED group by using 4 scoring approaches: plurality, simple average, weighted average, and a weighted standardized score. For comparison, ED groups (ie, taxpayer identification numbers) were grouped into 3 types: taxpayer identification numbers with only 1 ED; those with multiple EDs, but no ED with greater than 60,000 visits; and those with multiple EDs and at least 1 ED with greater than 60,000 visits. RESULTS: We found marked differences in the classification of ED throughput performance between scoring approaches. The weighted standardized score (z score) approach resulted in the least skewed and most uniform distribution across the majority of ED types, with a kurtosis of 12.91 for taxpayer identification numbers composed of 1 ED, 2.58 for those with multiple EDs without any supercenter, and 3.56 for those with multiple EDs with at least 1 supercenter, all lower than comparable scoring methods. The plurality and simple average scoring approaches appeared to disproportionally penalize ED groups that staff a single ED or multiple large-volume EDs. CONCLUSION: Application of a weighted standardized (z score) approach to ED throughput measurement resulted in a more balanced variation between different ED group types and reduced distortions in the length-of-stay measurement among ED groups staffing high-volume EDs. This approach may be a more accurate and acceptable method of profiling ED group throughput pay-for-performance programs.