Le projet CASSANDRE ou l’optimisation par diagnostic automatisé des documents cliniques

Abstract

Abstract. In January 2007, the Hopitaux Universitaires de Geneve (HUG) went from flat rate billing day basis to flat rate billing case basis. The system used until 2012 was the All Patient Diagnosis Related Group (APDRG) system. Since 2012, the system used is the Swiss DRG. Assuming that nowadays the hospital stay billing depends on the DRG used, and that this DRG is made on a short stay resume basis, we have identified three works areas to optimize the output document and therefore the stay billing. Those three works areas, that are part of the Cassandra project, are the improvement of the completeness of the output document, the optimization of the principal diagnosis choice and finally the creation of an essential comorbidities list. For the first area, we have implemented pathology detection algorithms, especially, for the dyskaliemies detection. In this case the potential gain is almost 2’000’000 CHF. For the second area, the principal diagnosis choice has to be done by having considered the diagnosis having concretely monopolized the most resources during the patient stay, in order to have a billing as fair as possible. Finally, for the third area, we have now identified 33 diagnoses that have an impact on the DRG choice and on the medical resources that those pathologies need. For those reasons, we have initiated a meticulous study regarding their clinical documentation and possible detection algorithms. The initial results being encouraging, we believe that those three works areas will help us to create tools, that once implemented in the coding chain will improve the coding quality at the HUG and especially will achieve a more accurate billing for the patients, the insurers and our institution.