A Process-Mining Model to Detect Adverse Postoperative Blood Transfusions

ImportanceErrors that threaten patient safety can cause patient harm, death, and rising health care costs. Manual process to expose adverse events (AE) increase time spend to detect them and increase costs of detection. ObjectiveThis study aims to make it easier and faster to expose AEs related with postoperative blood usages by process mining. DesignThese errors can be reported voluntarily by healthcare givers or exposed by Global Trigger Tool (GTT) determined by the Institute for Healthcare Improvement (IHI). With process mining Transfusion of Blood or Use of Blood Products (C1) cases were exposed in a data set. Actual life process was discovered and GTT C1 was searched as a process pattern in the discovered process. Instead of reviewing all cases manually, only detected cases were reviewed by patient safety subject matter experts. Setting and ParticipantsAnadolu Medical Center, Turkey was selected as the reference site for this quality improvement study. The data set includes 42,086 records, 2,870 cases and 20 activities for the period between October and December 2018. Main Outcomes and MeasuresWith the new process mining model, data was reduced to 2,704 records, 57 cases and 16 activities. 57 cases detected by the model were analyzed by the expert group and 10 of them are defined as AEs. Rate of C1 AEs per medical record is 1.0%. The rate of C1 AEs per medical record was between 1.3% and 8.3% in other research papers. Conclusions and RelevanceInstead of running the classic GTT model manually, only detected 57 patient files were analyzed. The new model 95% decreases time of experts who will review medical records to expose AEs.