Recommendations for Prevention of HIV Transmission in Health-Care Settings [MMWR Vol. 36. No. 2S]

August 21, 1987 - Supplement

Serologic Testing for HIV Infection
Background
A person is identified as infected with HIV when a sequence of tests, starting with
repeated enzyme immunoassays (EIA) and including a Western blot or similar, more
specific assay, are repeatedly reactive. Persons infected with HIV usually develop
antibody against the virus within 6-12 weeks after infection.
The sensitivity of the currently licensed EIA tests is at least 99% when they are
performed under optimal laboratory conditions on serum specimens from persons
infected for >12 weeks. Optimal laboratory conditions include the use of reliable
reagents, provision of continuing education of personnel, quality control of procedures, and participation in performance-evaluation programs. Given this performance, the probability of a false-negative test is remote except during the first several
weeks after infection, before detectable antibody is present. The proportion of
infected persons with a false-negative test attributed to absence of antibody in the
early stages of infection is dependent on both the incidence and prevalence of HIV
infection in a population (Table 1).
The specificity of the currently licensed EIA tests is approximately 99% when
repeatedly reactive tests are considered. Repeat testing of initially reactive specimens
by EIA is required to reduce the likelihood of laboratory error. To increase further the
specificity of serologic tests, laboratories must use a supplemental test, most often
the Western blot, to validate repeatedly reactive EIA results. Under optimal laboratory
conditions, the sensitivity of the Western blot test is comparable to or greater than
that of a repeatedly reactive EIA, and the Western blot is highly specific when strict
criteria are used to interpret the test results. The testing sequence of a repeatedly
reactive EIA and a positive Western blot test is highly predictive of HIV infection, even
in a population with a low prevalence of infection (Table 2). If the Western blot test
result is indeterminant, the testing sequence is considered equivocal for HIV infection.
TABLE 1. Estimated annual number of patients infected with HIV not detected by
HIV-antibody testing in a hypothetical hospital with 10,000 admissions/year*
Approximate
Approximate number of
Beginning Annual number of HIV-infected
prevalence of incidence of HIV-infected patients
HIV infection HIV infection patients not detected
5.0% 1.0% 550 17-18
5.0% 0.5% 525 11-12
1.0% 0.2% 110 3-4
1.0% 0.1% 105 2-3
0.1% 0.02% 11 0-1
0.1% 0.01% 11 0-1
*The estimates are based on the following assumptions: 1) the sensitivity of the screening test
is 99% (i.e., 99% of HIV-infected persons with antibody will be detected); 2) persons infected with
HIV will not develop detectable antibody (seroconvert) until 6 weeks (1.5 months) after infection;
3) new infections occur at an equal rate throughout the year; 4) calculations of the number of
HIV-infected persons in the patient population are based on the mid-year prevalence, which is
the beginning prevalence plus half the annual incidence of infections.

13S