Exposure of Patients to Human Immunodeficiency Virus through the Transfusion of Blood Components That Test Antibody-Negative

The risk of transmission of the human immunodeficiency virus to recipients of blood transfusions exists chiefly during the period between the time a donor is infected and the time he or she has a positive blood test for HIV antibodies. Estimating the chance that blood will be donated during this period is an effective way to define the risk of HIV infection from transfusion. Using this approach, and employing data from over 17 million American Red Cross blood donations, we estimate that during 1987 the most likely number of units of blood infected with undetected HIV that were transfused was 131 (range, 67 to 227). For a patient, the odds of contracting HIV infection were 1:153,000 per unit transfused. A patient who received the average transfusion (5.4 units) had odds of 1:28,000. The risk has been decreasing by more than 30 percent a year. We estimate that donor-recruitment practices plus careful education and screening are eliminating 49 of every 50 donors likely to be HIV-positive and that testing is 92 to 97 percent effective, for a combined effectiveness of 99.9 percent. The risk of undetected infectious units can probably be further reduced by transfusing fewer units and units from fewer donors, recruiting more women and fewer men as new donors, and encouraging more frequent donations from donors who have been tested repeatedly. (N Engl J Med 1989; 321: 941–6.) PHYSICIANS, blood-bank managers, and the general public have expressed considerable concern about the risk that transfusion recipients will receive blood infected with the human immunodeficiency virus (HIV). 1 2 3 4 5 6 7 8 9 10 11 12 Recent estimates of the risk that seronegative blood components infected with HIV will enter the nation's blood supply appear to be deficient because they rely on older, less precise, or less comprehensive data, involve the use of relatively simple models, and are based on point estimates for certain key variables that can be reasonably calculated only within ranges. From data on American Red Cross donors and HIV tests we have generated new information . . .