1046 
Comparison of HIV Incidence and Risk Factors for Recent Infection Based on Longitudinal Follow-up versus Cross-sectional cBED Assay Testing: A Study in Rural South Africa
Till Barnighausen*1, C Wallrauch1, A Welte2,3, T McWalter2,3, N Mbiazana1, J Viljoen1, N Graham1, F Tanser1, A Puren4, and M-L Newell1,5,6
1Africa Ctr for Hlth and Population Studies, Univ of KwaZulu-Natal, South Africa; 2Sch of Computational and Applied Mathematics, Univ of the Witwatersrand, Johannesburg, South Africa; 3Ctr for Excellence in Epi Modelling and Analysis, Stellenbosch Univ, South Africa; 4Natl Inst of Communicable Diseases, Johannesburg, South Africa; 5Africa Ctr for Hlth and Population Studies, Univ of KwaZulu-Natal, Durban; and 6Ctr for Pediatric Epi and and Biostatistics, Univ Coll London, Inst of Child Hlth, UK
Background: The cBED
IgG-Capture Enzyme Immunoassay (cBED assay) is increasingly used to estimate
HIV incidence in cross-sectional HIV surveys. We use data from a longitudinal
HIV surveillance in rural South Africa to measure the fraction of people with non-recent
HIV infection who are falsely classified as recently HIV-infected by the cBED
assay (the long-term false positive ratio) and compare cBED assay-based HIV
incidence estimates to longitudinally measured HIV incidence. We further
investigate whether risk factors for recent HIV infection identified in
cBED-assay based cross-sectional analysis agree with those found in
longitudinal incidence analysis.
Methods: Women aged 15 to
49 and men aged 15 to 54 years, resident in the surveillance area between 2003
and 2006, were eligible for HIV testing. We measure the long-term false positive
ratio in individuals with 2 positive HIV tests >360 days apart (n = 936). We
implement 4 different formulae to calculate HIV incidence using cBED assay
testing (n = 11,755) and compare the results to longitudinally measured HIV
incidence (224 seroconversions observed in 7686 person-years). To identify risk
factors for recent infection, we use logistic regression (for cBED assay data)
and survival analysis (for longitudinal seroconversion data).
Results: The long-term false
positive ratio was 0.0182 (95%CI 0.0106 to 0.0289). Using this long-term false positive
ratio, HIV incidence estimates (per 100 people per year) vary between 2.96 (95%CI
2.36 to 3.56) and 3.12 (2.49 to 3.75), depending on the formula used. Using a
long-term false positive ratio of 0.0500 based on previous studies, HIV
incidence estimates varied between 1.05 (95%CI 0.40 to 1.70) and 1.13 (0.43 to
1.83). The longitudinally measured HIV incidence was 2.80/100 people/year (95%CI
2.38 to 3.21), after adjustment to the sex-age distribution of the sample used
in cBED assay-based estimation. The risk factors for recent HIV infection
identified in cBED assay-based analysis agree well with those found in
longitudinal analysis of HIV seroconversion hazards.
Conclusions: In a rural
community in South Africa with high HIV prevalence the long-term false positive
ratio of the cBED assay is substantially lower than previous estimates. The
cBED assay performs well in HIV incidence estimation if the locally measured long-term
false positive ratio is used, but significantly underestimates incidence when a
long-term false positive ratio estimate based on previous studies in other
settings is used. The cBED assay method is a valid approach to investigate risk
factors for recent HIV infection.
|
