closely matches what public banks currently charge for a transplanted unit. If a national cord bank maintained an inventory of 150,000 units, the break-even fee could easily triple. Total costs range from $146 million for an inventory of 50,000 units to $210 million for an inventory of 300,000 units. Direct transplant costs account for at least 70 percent of total costs in all of the inventory scenarios.

Table E-7 displays the incremental cost per life year gained for each inventory level and scenario. For example, under scenario 1, an increase in the inventory from 200,000 to 300,000 units is associated with an additional cost of $23 million and a gain of 212 life years. The incremental cost per life year gained is $106,948 = ($23 million ÷ 212). The difference between scenarios 1 and 2 reflects the fact that in scenario 1 much of the gain in life years from expanding inventory is due to a shift toward better-quality cord blood matches among transplanted patients, rather than a greater absolute number of transplants. Failing to take account of how survival rates vary by match level ignores the survival gains from this first effect.

Equity between racial groups is an important motivation behind the establishment of a national cord bank. We examine the impact of increasing cord inventory on equity by comparing match probabilities and life years gained per patient between African Americans and Caucasians. For