Venue: The Fuqua School of Business, Duke University, 1 Towerview Drive, Durham, NC 27708-0120
Presentation
Inducing Appropriate Incentives for Efficient Emergency Department Use
Rationale: Emergency Department (ED) use is usually more costly compared with regular physician visits. Inappropriate use of the ED has been cited as an important contributor to the increasing cost of health care. This has prompted a number of policy makers to implement strategies to discourage non-urgent ED visits that could be more appropriately addressed in a primary care setting.
Objective: A measure is needed, which is risk adjusted and can empower efficient prevision of quality care, to reward care providers who invest in a systematic approach to reduce avoidable non-urgent ED use.
Methodology: The study consists of two major steps. First, an algorithm is used to classify the ED utilization, i.e., to separate urgent ED visits from the avoidable ones. Second, a risk-adjusted measure is developed. To this end, two models are constructed and compared to evaluate the urgent ED visits. The first model includes only demographic (age/sex) information and the second uses both demographics and illness burden (diagnostic information) as explanatory variables. DxCG’s Diagnostic Cost Groups (DCGs) are used to account for the effect of illness burden on ED visits. In both cases, nonlinear and linear specifications are considered including ordinary least square (OLS) regression, Poisson regression, and negative binomial regression. In addition, because there is an excess number of non-ED users in the data, three more models are considered, namely, the hurdle model, the zero inflated Poisson regression model, and the zero inflated negative binomial model. The results show that the extra diagnostic information improves the explanatory power of the model and provides more information to identify providers who utilize the ED in a more efficient manner. Based on those results, a framework is proposed to induce incentives for efficient ED use.
Results: In the study population, 42% of ED visits are non-urgent and 35% of ED spending is avoidable, which implies that ED use can reduced significantly. The explanatory power of the second model, adjusting for age, sex and comorbidities, is ten fold higher than the demographics only model. Thus, a good pay-for- performance system should incorporate illness burden information to enhance the model performance.
Conclusions: The ED use can be more efficient if an appropriate reward system is available for primary care providers. In the absence of this, ED units may consider referring larger numbers of non-urgent patients to appointments at primary care facilities. This alternative would be viable only if the availability and coordination of primary care services is enhanced.