Alternative test sequences for HIV screening of donated blood.

Human Immunodeficiency Virus (HIV) infection is extremely rare among volunteer blood donors. The highly sensitive Enzyme Linked Immunoassay (ELISA) test and the highly specific Western blot confirmation constitute the test sequence now used to minimize the possibility of transfusion associated HIV infection and to minimize the loss of donors due to false positive test results. The estimated operating characteristics for the test sequence permit the estimation of true infection rates which may be higher or lower than "observed" rates among subcategories of blood donors with progressively lower prevalence rates. The probability that a positive test result indicates true infection also declines with decreasing prevalence. The potential benefits of changing the test sequence so that complete HIV screening is implemented only for donations which are hepatitis-free include a reduction in the costs of Western blot testing and donor counseling, a reduction in the number of donors who use the blood bank inappropriately for personal HIV testing, and a more explicit recognition of the false positive problem when counseling donors.

Adjusting morbidity ratios in two communities using risk factor prevalence in cases.

Available data on cancer incidence for 1969-1971 showed statistically elevated rates for breast cancer in St. Louis Park, Minnesota, a community with creosote contamination of the water supply, when compared with the rest of the Minneapolis-St. Paul area taken as the reference population. In order to assess the effect of other known risk factors for breast cancer, 75 persons with breast cancer in each of the two populations were interviewed to obtain frequencies of known risk factors. An adjusted morbidity ratio in the two populations and an expected case rate in the exposed community were calculated from these frequencies, using relative risk values from the medical literature. The adjusted morbidity ratio was less than 1.0, and the observed rate was almost identical to the new expectation, although the age-adjusted rates alone had suggested a significant difference in incidence. This method makes use of relative risks from published studies rather than those associated with local cases and controls. It allows more refined evaluation of differences in cancer rates between communities than can be provided by age- and sex-specific calculations alone, and may allow use of available statistics in situations where cost, temporal considerations, or population size do not favor large new studies.