- Home>
- Specialty Care>
- HIV/AIDS>
- Summary and Comment
Limitations of Rapid Testing
The performance characteristics of a sequential rapid HIV-testing algorithm were poorer than anticipated in a Ugandan study, perhaps due to false-positive results from tests with "faint bands."
Rapid tests are often preferred for HIV screening because they are easy to perform, are relatively inexpensive, and provide highly accurate results in approximately 20 minutes. In many resource-limited settings, sequential rapid tests have become the standard for HIV diagnosis. To evaluate the accuracy of this approach, investigators examined testing data from 1517 men (age range, 15–49 years) who were screened for HIV infection as part of the enrollment process for two circumcision trials in Rakai, Uganda.
All the men provided blood samples that were initially tested with Determine HIV-1/2/O. Men with negative results were considered to be HIV-uninfected, and those with reactive results had their samples retested using HIV 1/2 STAT-PAK Ultra Fast. Patients with reactive results on both tests were told they were HIV-infected; those with discordant results had their samples tested a third time using Uni-Gold Recombinant HIV-1/2. All samples were later retested for quality control using two enzyme immunoassays (EIAs) and, in the case of discordant EIA results, Western blot confirmation.
Although sensitivity was high for the overall algorithm of three sequential rapid tests (97.7%), the specificity was lower than anticipated (90.4%). In a setting where the reported prevalence ranged from 11.2%, this specificity translated into a positive predictive value of 56.3% for the three tests combined. Reactive tests with "weak bands" (defined as fainter than the control band) were more likely than others to be falsely positive; when such tests were excluded from analysis, HIV prevalence in the remaining group was 14.3%, specificity improved to 99.6%, and positive predictive value improved to 97.7%. Among the 37 patients whose tests had weak bands, 94.1% were HIV-uninfected according to EIA and Western blot results.
Comment: This report is the first to describe the comprehensive performance characteristics of an HIV-testing algorithm that uses sequential rapid tests in a resource-limited setting. The combination of three tests reportedly had good sensitivity for chronic HIV infection, although test sensitivity for acute infection using this algorithm would likely be lower.
A potential advantage of sequential rapid testing is that a false-positive result on one test can be identified as such on a subsequent test. However, for this to happen, the condition that led to false positivity on the first test cannot be present in the second. Although such conditional independence is difficult to prove, the tests used in this study were perhaps not independent — in numerous cases, weak bands led to false-positive results on two of the three tests. Given these results, the criteria for positivity may need to be recalibrated to include band strength darker than the control band. Meanwhile, the manufacturers’ instructions for many rapid tests still claim that any band present should be read as reactive; thus, test results related to band strength should be interpreted with caution.
— Rochelle P. Walensky, MD, MPH
Dr. Walensky is Associate Director of the Program in Epidemiology and Outcomes Research at the Center for AIDS Research at Harvard Medical School and an Infectious Disease Specialist at Massachusetts General Hospital and Brigham and Women’s Hospital in Boston.
Published in AIDS Clinical Care July 16, 2007
Citation(s):
Gray RH et al. Limitations of rapid HIV-1 tests during screening for trials in Uganda: Diagnostic test accuracy study. BMJ 2007 Jun 1; [e-pub ahead of print]. (http://dx.doi.org/10.1136/bmj.39210.582801.BE)