PDG13 CAN THE PURCHASING PERFORMANCE OF HEALTHCARE INSTITUTIONS BE ASSESSED VIA THE DATA DECLARED FOR REIMBURSEMENT?

Abstract

French health authorities rely on information reported by healthcare facilities in E-PMSI to assess hospital activity and ensure their reimbursement. Some of these data are restored in open access (ScanSante database) and therefore can be used to improve purchasing performance through benchmark studies. Nevertheless, a previous study suggested obvious misreporting errors regarding expensive drug prices. We aim to explore the confidence that can be placed in these declarative data. This retrospective study assesses open data robustness based on reported annual quantities and spending of expensive drugs reimbursed in addition to the diagnosis related group tariff. We deduced the 2017 average purchasing prices (prices are confidential) for APHP (Paris hospitals, e-PMSI database) and for French hospitals formerly under overall allocation (ex-UOA, ScanSante database) and compared it to the national reimbursement price cap (RPC) (the oldest of the year, if so). The relative price differential (RPD%) should be negative or zero. Then, we compared (as an external validation) APHP monthly declared data in 2018 with accurate market data and with monthly RPC applicable in 2018. Results: APHP and ex-UOA hospitals consumed 216 expensive drugs of the additional list in 2017. Significant errors affected 25% of it (RPD>1%) that indicates data entry mistakes (RPD min: +1,01%; max: +152,2%; average:+7,3%). Reported biases concerned primarily immunosuppressors and antihemorragic agents. In 2018 in APHP, 48% of the declared purchasing price (n=222) were superior to the RPC and 35% were higher than the accurate market prices. Deviations were particularly explained by the price declaration of old contracts, repealed references and declared consumption unit errors. Conclusions: Despite the conservative methodology adopted, our study highlights the poor quality of the declarative data that limit their use. Improved data capture is a major challenge because of the financial consequences for institutions in terms of refunds granted.