| Abstract | During neonatal intensive care, blood components are often used in clinical situations where both their efficacy and safety lack solid justification. A practical system to continuously analyse actual transfusion practices is a prerequisite for improvements of quality in transfusion therapy. We hypothesized that such a system would reveal inappropriate variations in clinical decision making and offer a means for staff education and quality improvement and assurance. The study consisted of three 120-152-day periods (P I, P II and P III) between January 2000 and October 2001 and involved 543 new patient admissions (141 patients with birth weight < 1501 g) and 6227 days of patient care at a single tertiary level NICU. P I was a control with no intervention, P II was after technically introducing the computer system and, the last period, P III was after presenting and discussing the results of P I and P II at a staff meeting. Upon an order of platelet or fresh frozen plasma (FFP) unit from the blood bank, a computer-based audit system compared the last platelet count or prothrombin time [expressed as percentage of normal clotting activity, prothrombin time (PT-%)] to predefined criteria. In the case of exceeding the preset thresholds, the system required additional information and recorded the pretransfusion laboratory values for later analysis. Thirty-two per cent of platelet transfusions were given with pretransfusion platelet count >49 x 10(9) L(-1), and 60% of these transfusions (19% of all platelet transfusions) could not be clinically justified in retrospective chart review. There was no significant change in this practice from P II to P III. FFP transfusions were given with significantly different pretransfusion PT-% values during P II and P III. The proportions of FFP transfusions with pretransfusion PT-% > 49% were 7.8% and 0.9% during P II and P III, respectively (P < 0.0001). In chart review, none of the FFP transfusions with pretransfusion PT-% > 49% could be justified by clinical grounds. Inappropriate transfusions of both platelets and plasma remain a significant challenge for quality assurance of neonatal intensive care. Automated recording of pretransfusion platelet count and prothrombin time reliably identified the poorly justified transfusions and thus offered a practical resource-saving tool for quality assurance of transfusion in the NICU. A significant shift towards more appropriate use of plasma was demonstrated after implementation of the audit system. |