An Analysis of the New York University Emergency Department Algorithm’s Suitability for Use in Gauging Changes in ED Usage Patterns

Abstract

The Emergency Department Algorithm (EDA) developed at New York University uses administrative discharge data to distill hundreds of International Classification of Diseases-9 codes for emergency department (ED) visits into 4 categories, making it attractive to researchers and policy makers. The EDA has been used to analyze patterns of ED visits in a wide variety of locations and populations. However, there are concerns regarding the validity and use of the EDA for research and policy. To explain the findings of previous EDA users that it appears to lack sensitivity in detecting changes in ED utilization patterns. Mathematical simulation was used to analyze and explain the performance of the EDA in detecting differences in utilization patterns across hypothetical ED populations. Sensitivity analysis was used to illustrate the magnitude of changes in EDA outputs relative to changes in ED populations using a national sample of actual ED patients. The vast majority of possible EDA outputs are clustered so tightly as to show no significant change in outputs between different hypothetical populations. Sensitivity analysis shows that changes in EDA outputs are not nearly as great as the magnitude of the input differences across real-world populations. The EDA categorizes a very large variety of ED visits into a relatively small group of outputs. Its operating characteristics suggest that the EDA is insufficiently sensitive to changes in ED utilization patterns to be useful in assessing interventions to change them. This finding should caution potential users to consider the EDA's limitations before using it.