952
Comparison of HIV Incidence Estimates Derived from Laboratory Incidence Assays and Repeat Testing Data at HIV Testing Sites in San Francisco
Hong-Ha M. Truong*1,2, Hong-Ha M. Truong*1,2, T Kellogg3, B Louie3, J Klausner1,3, J Klausner1,3, J Dilley1, W McFarland1,3, and W McFarland1,3
1Univ of California, San Francisco, US; 2Gladstone Inst of Virology and Immunology, Univ of California, San Francisco, US; and 3San Francisco Dept of Publ Hlth, CA, US
Background: Concerns have been raised regarding whether the
BED immunoglobulin G capture enzyme immunoassay (BED-CEIA) may be
overestimating HIV incidence. We compared HIV incidence estimates derived from
BED-CEIA, Vironostika-LS (LS-EIA) and repeat testing
history among men who have sex with men (MSM) at HIV testing facilities in San Francisco.
Methods: MSM presenting for HIV voluntary counseling
and testing (VCT) at all anonymous testing sites in San Francisco from 2000 to
2003 (n = 11,856) and the municipal sexually
transmitted disease (STD) clinic from 2000 to 2004 (n = 9182) were evaluated for HIV infection. Blood specimens (n = 15,010) were screened for HIV
antibodies and positive specimens (n =
658) were evaluated for recent HIV infection using BED-CEIA and LS-EIA.
Self-reported HIV testing history was used to determine seroconversion,
defined as a current HIV+ test accompanied by a previous HIV–
test. Annualized HIV seroincidence estimates by
BED-CEIA were calculated using a revised formula by McDougal et al, which adjusts
for misclassification of long-term and recent cases and for missing data. Associations
were assessed using the χ2 statistical test.
Results: HIV incidence estimates are shown in the table.
Of 219 HIV+ cases identified at the anonymous testing sites, 69
cases were classified as recent infections by both assays, 8 cases by LS-EIA
only, and 26 cases by BED-CEIA only, yielding a substantial agreement with K = 0.68. Of 439 HIV+ cases identified at the STD
clinic, 134 cases were classified as recent infections by both assays, 21 cases
by LS-EIA only, and 14 cases by BED-CEIA only, yielding an almost perfect
agreement with K
= 0.82. Recent infection was associated with unprotected receptive anal
intercourse (p <0.001), sex with a
known HIV+ partner (p <0.001),
having >10 sex partners (p = 0.009),
having exchange sex (p = 0.042), Asian
ethnicity (p = 0.014), and amphetamine
use (p = 0.012).
|
HIV
Incidence Estimates (%)
|
Anonymous
Testing Sites
|
Municipal
STD Clinic
|
|
2000
|
2001
|
2002
|
2003
|
2000
|
2001
|
2002
|
2003
|
2004
|
|
BED-CEIA
|
4.00
|
4.06
|
4.46
|
2.26
|
3.36
|
2.36
|
3.83
|
3.74
|
4.01
|
|
LS-EIA
|
3.66
|
3.46
|
2.91
|
2.07
|
3.52
|
2.71
|
4.17
|
4.09
|
4.06
|
|
Repeat Test
|
1.90
|
2.57
|
1.93
|
1.78
|
3.28
|
3.01
|
2.97
|
2.78
|
4.11
|
|
|
|
|
|
|
|
|
|
|
|
Conclusions: HIV incidence estimates based on BED-CEIA and LS-EIA
results tended to be slightly higher than estimates using repeat testing data. There
was good concordance between the results from the 2 laboratory incidence
assays. Temporal trends of the 3 methods track each other fairly well.
|
