Evaluation of cost and length of stay’s homogeneity per APR-DRG for oncological inpatient stays in 11 Belgian hospitals

Abstract

Objectives: A prospective fixed funding mechanism may be difficult in innovative areas such as oncology. This study aimed to estimate the length of stay (LOS) and cost of oncology hospital stays, to evaluate their variability and to identify drivers affecting the cost per All Patient Refined Diagnosis Related Group (APR-DRG)

Methods: Oncology stays (2012) were retrieved from 11 Belgian hospitals. Stays were classified per APR-DRG and severity of illness (SOI) combination. Data heterogeneity was assumed if the variation coefficient (VC) was > 1. Variance reduction was assessed by 1) outlier exclusion and 2) logarithmic data transformation. Cost drivers were investigated for selected APR-DRGs by ANOVA. Results: Mean LOS and cost of an oncology inpatient stay is 8,0 days and 6 231€. About 30% of APR-DRG and SOI combinations had a cost VC > 1. Both variance reduction methods
resulted in a VC < 1 in all combinations; only transformed data allowed to identify the cost drivers ‘hospital’ and ‘SOI’. Conclusions: Oncology stays have a higher cost if compared to non-oncology stays. The logarithmic data transformation reduced cost variability and was suitable to identify cost drivers, to check the feasibility of a prospective APR-DRG funding. The significant effect of ‘hospital’ on cost implies that an unadjusted lump sum per APR-DRG will under- or overestimate the real hospital costs.