Estimated risk of transmission of the human immunodeficiency virus by screened blood in the United States

Authors: Lackritz E, Satten G, Aberle-Grasser J, Dodd R, Raimondi V et al.
Source: New England Journal of Medicine. 333:1721-5. December 28, 1995.
Institutions: CDC; Orkand Corporation; American Red Cross.
Financial support: not specified.



This study estimates the risk of transmission of HIV from screened blood products, using data from the American National Red Cross obtained from regions across the United States. The data was gathered from 19 of the Red Cross' 44 regions and represents about 25% of blood gathered in the US during the study period.

Basic Approach

The authors assumed that most infective blood products would be due to donation during the "window period": after infection with HIV and before appearance of antibody detected by the screening process. The number of donations taking place during the window period depends on the length of this window period and on the incidence of HIV infection in the donating population. In addition, some infective units are passed by laboratory error (false negative screening tests). After summing these two quantities, the authors then subtracted an estimated number of units that would have been subsequently rejected on other grounds (such as abnormal LFT's). The final number yields the estimated number of infective units released to the public after screening.

Estimation Methods


During the study period, 4,119,095 donations were made in the 19 study regions, 74% of them by repeat donors. 68 donations from repeat donors were HIV seropositive, 250 from first-time donors. The overall incidence rate of HIV infection was estimated at 4.1 per 100,000 person years and varied by region (range: 0 to 19.7).

Using the calculations outlined above, the probability that a donation would occur during the window period was estimated at 1 in 360,000 for the 19 study regions and 1 in 450,000 when the results were extrapolated to all 42 regions. When the estimated laboratory errors were added and the estimated number of units that would have been discarded due to other tests was subtracted, the final estimation was 1 in 450,000 to 1 in 660,000 HIV postive donations accepted for processing.

Authors' Discussion

Extrapolating these results to the 12 million donations screened annually in the US, with each recipient receiving blood from an average of 5.4 donors, the risk of a recipient receiving HIV positive blood products would be between 1 in 83,000 and 1 in 122,000. This is lower than previous estimates because the authors assume a shorter, 25-day window period with contemporary ELISA assays, the donor pool has become safer and the data were obtained from regions across the US, some with a much lower incidence of HIV infection.

The authors note several limitations of the study, particularly the fact that the incidence of HIV among first-time donors could not be measured, only estimated and the fact that the measurement of the incidence of HIV infection among repeat donors assumes that all repeat donors who seroconvert after a negative donation return to make a seropositive donation.

Finally, the authors cast some doubt on the wisdom of a recent FDA recommendation that all blood donations be screened using a more sensitive p24 antigen assay. They state that, using their data, such an approach would only prevent the use of 4 to 6 infectious units each year, at considerable cost.


In my opinion, the major weakness of this study, mentioned briefly by the authors, is the fact that their estimate of the incidence of HIV among repeat donors assumes that every repeat donor who seroconverted during the study period would return to make a seropositive donation (and thus be counted in the calculation of incidence). Any donor who seroconverted but did not return for repeat testing was missed in the calculation of incidence. Since the probability of donation during the window period was calculated as the product of incidence times the length of the window period, if one half of the regular donors who seroconverted did not return to make a "seropositive donation", the probability would have been underestimated by at least one half.

Although it may be unlikely that such a high percentage of repeat donors would not return to donate after seroconverting, it is not unreasonable to assume that some significant number of these donors would be HIV-tested independently of the donation and not return if they became seropositive Also, the population of at-risk donors might well be less reliable in terms of returning for regular repeat donations than the donor population at large. At any rate, some sort of "guesstimate" at the magnitude of this number should have been made, or the problem discussed in greater detail.


Reader comments

April 11, 1996

Letters to the Editor about this article from the NEJM

Journal Club on the Web home page

Submit a comment about this article

Site Meter