Controlling for endogeneity in attributable costs of vancomycin-resistant enterococci from a Canadian hospital
Decisions regarding the optimal provision of infection prevention and control resources depend on accurate estimates of the attributable costs of health care–associated infections. This is challenging given the skewed nature of health care cost data and the endogeneity of health care–associated infections. The objective of this study is to determine the hospital costs attributable to vancomycin-resistant enterococci (VRE) while accounting for endogeneity.Methods:
This study builds on an attributable cost model conducted by a retrospective cohort study including 1,292 patients admitted to an urban hospital in Vancouver, Canada. Attributable hospital costs were estimated with multivariate generalized linear models (GLMs). To account for endogeneity, a control function approach was used.Results:
The analysis sample included 217 patients with health care–associated VRE. In the standard GLM, the costs attributable to VRE are $17,949 (SEM, $2,993). However, accounting for endogeneity, the attributable costs were estimated to range from $14,706 (SEM, $7,612) to $42,101 (SEM, $15,533). Across all model specifications, attributable costs are 76% higher on average when controlling for endogeneity.Conclusions:
VRE was independently associated with increased hospital costs, and controlling for endogeneity lead to higher attributable cost estimates.