Dynamically allocating treatment when the cost of goods is high and drug supply is limited

Large-scale production of proteins by cell culture and subsequent purificat ion for use in novel medical therapies is a slow and complex process. Durin g the early phases of development of manufacturing processes, contamination and replication errors cause entire batches of material to be wasted. As a result, the cost of goods for large-molecule therapies in early clinical d evelopment can be significant, and the supply limited. When designing clini cal trials to test expensive biological compounds with limited supply, spon soring companies want to minimize the waste of drug, that is, to maintain s mall inventories of drug at the investigational hospitals. We must, however , weigh the benefits of smaller inventories against the costs of increased numbers of shipments to resupply when rapid enrollment causes shortages of drug. A well-planned randomization scheme may be able to balance these obje ctives. This paper demonstrates how a dynamic randomization algorithm can b e used to maintain smaller drug inventory at hospitals than a typical permu ted block randomization list plan, and how well it automatically restores b alance when the shortage of drug causes assignment of alternate treatments. Control Clin Trials 2000;21:44-53 (C) Elsevier Science Inc. 2000.