Search

By classifying hospital output into groups of patients with similar clinical characteristics and resource requirements, diagnosis-related groups (DRGs) are designed to be highly correlated with resource utilisation. Using a two-stage approach to control for variation within and between hospitals, we examine the ability of the diverse DRG systems in 10 European countries to explain variability in resource utilisation (costs or length of stay, LoS) for hospital patients undergoing surgical repair of inguinal hernia. Our national regression results suggest that DRGs are statistically significant in explaining cost/LoS variation in the absence of any other regressors and generally remain so in most countries when patient-level characteristics are added to the model. However patient-level characteristics, including those used in DRG assignment, are usually also statistically significant. In nine countries, where the number of relevant DRGs ranges from two (Poland) to seven (France), the inclusion of patient-level characteristics substantially improves model goodness-of-fit compared with that attained with DRGs alone. Only in Sweden is the converse true. If our analysis raises some concerns over the adequacy of DRGs to explain cost/LoS variation in inguinal hernia repair in nine of the 10 European countries, further research is required to consider whether future enhancements may be necessary. Copyright © 2012 John Wiley & Sons, Ltd.

By classifying hospital output into groups of patients with similar clinical characteristics and resource requirements, diagnosis-related groups (DRGs) are designed to be highly correlated with resource utilisation. Using a two-stage approach to control for variation within and between hospitals, we examine the ability of the diverse DRG systems in 10 European countries to explain variability in resource utilisation (costs or length of stay, LoS) for hospital patients undergoing surgical repair of inguinal hernia. Our national regression results suggest that DRGs are statistically significant in explaining cost/LoS variation in the absence of any other regressors and generally remain so in most countries when patient-level characteristics are added to the model. However patient-level characteristics, including those used in DRG assignment, are usually also statistically significant. In nine countries, where the number of relevant DRGs ranges from two (Poland) to seven (France), the inclusion of patient-level characteristics substantially improves model goodness-of-fit compared with that attained with DRGs alone. Only in Sweden is the converse true. If our analysis raises some concerns over the adequacy of DRGs to explain cost/LoS variation in inguinal hernia repair in nine of the 10 European countries, further research is required to consider whether future enhancements may be necessary.