Optimal allocation of blood products by Godson A. Tetteh The high cost of collection and the short shelf life of apheresis platelets demand efficient inventory management to reduce outdates and shortages. Apheresis platelets are licensed for seven days, and blood centers are keen on knowing the consequences of various product collection and distribution strategies. To reduce outdates, inventory managers typically distribute the older units first, thereby following first-in first-out (FIFO) policy; however, hospital blood banks would prefer that the blood center issues out the freshest units first, equivalent to a last-in first-out (LIFO) policy. This study addresses the optimal distribution policy to achieve a desired outdate, shortage and average age of apheresis platelets.

A comprehensive literature review was conducted on previous models studied to efficiently distribute blood products. However, most of the research on blood inventory management has been restricted to the hospital blood bank level in terms of ordering policies and inventory levels. This study takes the approach from the perspective of the inventory manager at the regional blood center. The inventory manager needs a reliable forecast of the quantity and timing of future blood supply (collection from donors) and blood demand from hospital blood banks to make an effective decision on blood inventory control. A forecasting method is used in this study to predict collection and demand for Single Donor Platelets (SDPs), and solves the blood inventory problem using a heuristic method and a Linear Programming (LP) with a rolling horizon method to find the near optimal issuing policy, the expected average age, outdate rate, and shortage rate of a blood product from the perspective of the blood center. It is concluded that regional blood centers can distribute with a ‘mixed’ FIFO/LIFO strategy and not significantly affect outdates or ability to cover shortages. For the LP model with a rolling horizon schedule, the inventory manager at the blood center would have to use forecast windows of five to achieve good issuing policies. A simulation study comparing the heuristic method and an LP-based with a rolling horizon method indicated that LP models with forecast windows of five and heuristics methods with a ‘mixed’ FIFO/LIFO strategy can be used to optimize this inventory problem. ]