ELISA Specificity: 21.6%

Thread view:

Enable EMail Alerts

Start New Thread

Thread rating:

Alex - 23 Jun 2006 15:23 GMT

If any excuse was needed for using Western Blot in DHS surveys, this would be it.
I quote:

Predictive value of two commercial human immunodeficiency virus serological tests in cases with
indeterminate Western blot results.

Huang LJ, Liu CY, Chu SC, Wong WW, Lin YC, Liu WT, Chan YJ.
Division of Infectious Diseases, Department of Medicine, Taipei Veterans General Hospital and
National Yang-Ming University, Taipei, Taiwan.

"METHODS: From January 1, 2000 to December 31, 2003, a total of 565 serum samples collected from
individuals with a previous positive or borderline positive ELISA test for HIV at regular check-up
(281 samples) and a second group of individuals with known risk of HIV exposure or suspected
infection based on clinical presentation (284 samples) were tested for HIV infection by ELISA, PAT
and WB assay.

RESULTS: The result was positive for HIV infection and confirmed by WB assay in 197 samples (22.5%),
indeterminate in 127 samples (22.5%) and negative in 241 samples (42.6%). The sensitivity and
specificity of ELISA were 100% and 21.6% ."

Reply to this Message

DavidT - 26 Jun 2006 09:38 GMT

Which assay was used for the ELISA, do you know?

Reply to this Message

DavidT - 03 Jul 2006 11:51 GMT

> Which assay was used for the ELISA, do you know?

I realy am interested to see explanations for the results of this
study. Does anyone have any information on the assays, or why
specificity was reportedly so low?

Reply to this Message

Chris Noble - 05 Jul 2006 05:41 GMT

> > Which assay was used for the ELISA, do you know?
>
> I realy am interested to see explanations for the results of this
> study. Does anyone have any information on the assays, or why
> specificity was reportedly so low?

The study is available here.

http://jmii.org/content/pdf/v39n3p219.pdf

The low specificity reported appears to come from using a cohort of
people that had previously tested positive on an ELISA or borderline.
What was considered borderline was not defined. Selecting for people
with borderline ELISAs means that most are likely false positives. This
is different to testing the general population.

In any test there will be some people that are hard to classify. Most
will be either strongly positive or negative but some will be
borderline. Preselecting for borderline cases will lead to an apparent
low specificity.

They are also pushing a different test the PAT. This might be another
reason why they get apparently low specificity for the ELISA.

Chris Noble

Reply to this Message