Evaluation Of Algorithms Of Anti- HIV Antibody Tests

Research question: Can alternate algorithms be used in place of conventional algorithm for epidemiological studies of HIV infection with less expenses? Objective: To compare the results of HIV sero- prevalence as determined by test algorithms combining three kits with conventional test algorithm. S...

Full description

Bibliographic Details
Main Authors: Paranjape R.S, Kulkarni S.S, Chanda Sumit, Mehendale S.M, Jawadekar S.S, Pratinidhi A.K, Rodrigues J.J
Format: Article
Language:English
Published: Wolters Kluwer Medknow Publications 1997-01-01
Series:Indian Journal of Community Medicine
Online Access:http://www.ijcm.org.in/article.asp?issn=0970-0218;year=1997;volume=22;issue=2;spage=54;epage=57;aulast=Paranjape;type=0
Description
Summary:Research question: Can alternate algorithms be used in place of conventional algorithm for epidemiological studies of HIV infection with less expenses? Objective: To compare the results of HIV sero- prevalence as determined by test algorithms combining three kits with conventional test algorithm. Study design: Cross â€" sectional. Participants: 282 truck drivers. Statistical analysis: Sensitivity and specificity analysis and predictive values. Results: Three different algorithms that do not include Western Blot (WB) were compared with the conventional algorithm, in a truck driver population with 5.6% prevalence of HIV â€"I infection. Algorithms with one EIA (Genetic Systems or Biotest) and a rapid test (immunocomb) or with two EIAs showed 100% positive predictive value in relation to the conventional algorithm. Using an algorithm with EIA as screening test and a rapid test as a confirmatory test was 50 to 70% less expensive than the conventional algorithm per positive scrum sample. These algorithms obviate the interpretation of indeterminate results and also give differential diagnosis of HIV-2 infection. Alternate algorithms are ideally suited for community based control programme in developing countries. Application of these algorithms in population with low prevalence should also be studied in order to evaluate universal applicability.
ISSN:0970-0218
1998-3581