Conclusions

1
NJHIV's Rapid Testing Program Keeps
False Positives in Check Evan M. Cadoff, MD1
Eugene G. Martin, PhD1 Sindy M. Paul, MD, MPH2
Sharon A. Holswade, MBA1 Jake Nelson1 Vivian H.
Shih, BA1 Dolores Van Pelt, MSN1 Nisha J.
Intwala, BS (MT)1 Franchesca N. Jackson, BS1
Gratian Salaru, MD1 1 UMDNJ-RWJMS, New Brunswick,
NJ and 2 New Jersey Department of Health and
Senior Services, Trenton, NJ.

• Figure 1
• OMT A New Form of Rapid Testing Introduced
• NJHIV participated in a CDC sponsored evaluation
  of OMT testing through May 2005.
• OMT testing was introduced at New Jersey testing
  sites in June 2005.

Abstract BackgroundFDA approval of rapid HIV
testing kits is partly based on sensitivity and
specificity claims. Recently, in San Francisco,
New York and Los Angeles, health care workers
confronted a spike of false positive results
(positive rapid HIV test with negative
confirmatory testing) on oral fluid (OMT). By
reviewing NJHIV quality assurance (QA) monitoring
data, we were able to assess the prevalence and
significance of false positive results in
NJ. MethodsThe NJHIV program began rapid testing
under a medical school QA program in late 2003.
The program has expanded to include 117 sites by
2005. A critical factor in the organization of
testing was the use of a centralized core staff
of clinical laboratorians who were responsible
for development of common, uniform policies,
procedures, training and recertification,
inventory control and validation, proficiency
testing, and a core communication hub
(www.njhiv.org) all part of a rigorous quality
assurance initiative. Data from all sites is
analyzed monthly. ResultsWe reviewed NJHIV data
since March 2005, including over 24,000 tests, of
which more than 5000 were OMT. Specificity of
fingerstick testing exceeded 99.9 . A slight
increase in the false positive rate (99.5
specificity) was noted when sites moved from
blood to OMT testing, but specificity remained
within manufacturers claims. Reviewing data from
each of our sites, each lot of reagent, each
month of testing and each specimen type, we did
not record an out of label FP spike as was seen
in some other states. ConclusionData routinely
collected in our QA program allowed rapid
assessment of HIV testing performance during the
introduction of oral mucosal testing. No increase
in false positives occurred beyond what would be
expected with the transition to oral fluid
testing. This is largely attributable to the
organization of the program.

• Methods
• NJHIV testing sites maintain logs of each test
  performed (Figure 3).
• Figure 3
• NJHIV testing log

• Results
• Data comparison among databases showed that data
  was complete in the NJHIV database for 22 of 23
  sites tested.
• After remediation of recordkeeping practices,
  NJHIV data at all sites is complete through June
  2006.
• One cluster of 3 simultaneous discordants was
  discovered at one site. Further analysis did not
  yield an explanation for these discordants.
• Consistent with the manufacturers product
  performance claims, there was a slightly higher
  false positive rate when using oral fluid than
  when using fingerstick whole blood specimens
  (Figure 5). Specificity occasionaly drops below
  manufacturers claims for individual months.
• Figure 5
• Specificity
• by month
• Analysis of discordant results did not show any
  other pattern when examined by month, site,
  reagent lot or sample type.

• Conclusions
• Our quality assurance system and databases
  allowed us to quickly evaluate our test results
  in response to reports of increased numbers of
  false positive tests in San Francisco, New York
  and Los Angeles.
• False positive OraQuick results in New Jersey
  were not an issue. Specificity remained within
  the claims of the manufacturer.
• The CDC presented NJHIVs data at national
  meetings (Figure 6) and included our findings in
  their decision not to change current rapid test
  recommendations.
• Figure 6 Slide from CDC presentation at CROI
  meeting, February 2006

• Background
• The NJHIV Rapid HIV Testing Program
• Built upon existing UMDNJ-Robert Wood Johnson
  Medical School, multi-facility,
  point-of-care-testing program.
• Centralized quality assurance process using
  pathologists, technologists and informaticians.
• OraQuick fingerstick whole blood testing began
  November 2003 at the first NJ licensed site a
  primary site in close proximity to the medical
  school. Currently there are 140 sites including
  fixed and mobile venues. (Figure 1)
• Testing activities are reviewed monthly by a
  medical technologist.
• Preliminary positive rapid tests are confirmed by
  Western Blot at the NJDHSS PHEL laboratory in
  Trenton.
• Quality Assurance Plan
• Management by a board certified Pathologist
• Supervisory control through site coordinators
• Central lab oversees
• Regulatory issues and proficiency testing
• Acquisition and validation of supplies
• Inventory control
• Common procedures and core policies
• Uniform administration at all locations
• Common training, certification of personnel,
  forms

Figure 4 Test records entered into database
which tracks frequency of testing by month,
primary site, satellite, reagent lot numbers,
sample type and test result.
Figure 4a Records selected for review based on a
single parameter. All oral fluid tests were
selected in the diagram to the left and are shown
here.
Figure 4b Records selected for review based on a
second parameter. Here, all discordant results
were selected from among the oral fluid tests in
the diagram to the left.