Long Island's hospitals gave up hoarding blood for a shared algorithm

Blood is perishable — a unit of whole blood or red cells lasts 21 days — and both its supply (donations) and demand (transfusion requests) are random, with wildly different volumes at every hospital in a region. Through the 1970s, the standard U.S. practice was decentralized: each Hospital Blood Bank called its Regional Center every morning and ordered enough units to bring its own inventory up to a 'safe' level, typically 2-3 days of stock for common blood types and 7-10 days for rarer ones, competing with every other hospital in the region for the Center's limited supply.

A 1971 national survey found this local-safety-first system was outdating — discarding past the 21-day shelf life — 25% of the blood collected across the country, while hospitals simultaneously reported shortages frequent enough to postpone elective surgery. The two failures were happening at once, region by region, because every hospital acting to protect its own patients was, in aggregate, wasting the shared regional supply that protected everyone's patients.

Centralized scheduling equalized shortage and outdate probabilities across all hospitals, minimizing both without estimating their relative cost.

Each hospital's safety stock starves the shared pool, causing shortages elsewhere.

Central scheduling sees all inventories and demands, enabling region-wide optimization.

Equalizing shortage and outdate probabilities avoids disputed cost tradeoffs.

Routine scheduled deliveries replace daily negotiations, reducing waste and shortages.

When nodes hoard a shared perishable resource, local caution causes regional waste; centralize allocation to balance risks.

Run as a routine, centrally scheduled operation rather than 38 separate daily negotiations, the system, by its developers' own account, 'drastically reduced' both shortages and outdating in the Long Island region relative to the roughly 25% national outdating rate the decentralized approach had been producing. The work won Brodheim and Prastacos the 1979 International Management Science Achievement Award, the field's top prize for a practice application of operations research.