The 2007 HIV Diagnostics Conference was hosted jointly by the Division of HIV/AIDS Prevention, Centers for Disease Control and Prevention (CDC), and the Association of Public Health Laboratories (APHL) in Atlanta, Georgia, during December 5-7, 2007.  The purpose of the conference was to present and discuss the performance of nine HIV test algorithms proposed by APHL/CDC workgroups.  The goal of the conference was to gather information to assist in the development of guidelines for the use of assays in combination to aid in the diagnosis of HIV-1 and/or HIV-2 infection in community, clinical, and laboratory settings in the United States.  The conference was attended by over 270 professionals representing essential constituent groups, including the American Society for Microbiology (ASM), the College of American Pathologists (CAP), the Food and Drug Administration (FDA), the Department of Defense (DOD), the National Association of State and Territorial AIDS Directors (NASTAD), blood bank officials, and scientists from commercial, clinical and state and local public health laboratories.  In addition, the conference vendor exhibition space included representatives from nearly all HIV test manufacturers in the United States, which allowed participants to get more detailed information on the latest developments in HIV testing technology.  The two and one-half day meeting included 27 poster and 32 oral presentations that were solicited in response to call for data to evaluate the nine revised HIV test strategies developed by APHL/CDC HIV testing workgroups.  The conference was organized into eight individual sessions, during which oral presentations were given, and a closing session, which included a summary of the conference and a discussion of next steps.  The conference sessions allowed participants to present existing performance and other data that validate the proposed algorithms, identify gaps where more data are required, and develop a plan for research activities and other efforts necessary to finalize and disseminate updated HIV testing algorithms for the United States.

The eight individual sessions were:

  • Strategies for Routine Screening for Acute HIV Infection
  • Strategies for Targeted Screening for Acute HIV Infection
  • Laboratory-based Strategies Using Confirmatory Supplemental Tests or Combinations of Screening Assays
  • Laboratory-based Strategies Using Combinations of Screening Assays and Rapid Tests
  • Performance of Point of Care Strategies Using Combinations of Rapid Tests
  • Applications of Point of Care Strategies Using Combinations of Rapid Tests
  • Emerging Dilemmas: HIV-2 Confirmation and Diagnosis in HIV Vaccine Recipients
  • New HIV Diagnostic Technologies

The individual presentations from these sessions, additional data needs, and next steps to support the nine proposed alternative algorithms have recently been summarized in a report entitled, HIV Testing Algorithms: A Status Report, available at Contact with questions or comments about the Status Report, or to share information relevant to the data needs.

Closing session: Conference Summary and Next Steps

During the closing session, participants summarized the discussions that occurred at the conference and discussed the validity of the proposed algorithms, remaining data needs, and next steps. The closing session was led by the moderators of the point-of-care (POC) and laboratory workgroups: Bernard Branson, Berry Bennett, Michael Pentella, and Kevin Delaney.  During the closing session, the moderators posed the following six questions:

Do we want to eliminate Western blot testing as the mainstay confirmatory test?

What do settings want from point-of care testing?

What do clinicians want from laboratories; do we need to release preliminary positive immunoassay results?

What technology do we want?

What do we need to be concerned with in settings with different HIV prevalence?

What data do we want?

Do we want to eliminate Western blot as the mainstay confirmatory test?

Several presentations and discussions illustrated the limitations of the Western blot for confirmatory HIV testing. The Western blot is less sensitive during seroconversion than currently available screening tests, expensive, complicated and time-consuming to perform, its interpretation is subjective, and it produces a relatively high number of indeterminate results that do not resolve infection status. Despite these limitations, attendees were not ready to completely eliminate the Western blot as an option for HIV supplemental testing. Multiple attendees called for a reassessment of the interpretive criteria of the Western blot for serodiagnosis, and the removal of bands that do not represent HIV viral proteins from the criteria for an indeterminate result. The Western blot is useful for dried blood spot (DBS) testing, for differentiating stages of HIV infection, and for providing diagnostic information when applied to complicated cases and specimens with discordant results. In addition, there are advantages to using a lysate of native HIV proteins because it increases the likelihood of the detection of unusual or emerging HIV subtypes, compared to assays based on synthetic or recombinant proteins.  However, participants felt that alternatives to Western blot, including the use of multiple immunoassays and molecular testing methods, should be considered. Participants reminded the panel that currently some jurisdictions require that the Western blot be used for confirmation of HIV infection for case reporting and for HIV medical care funding eligibility. These requirements will need to be addressed if the Western blot is replaced as the gold standard for HIV confirmatory testing.  Participants suggested that if methods other than the Western blot are used for confirmation of HIV infection or referral of persons likely to be infected with HIV, the companies that manufacture Western blots may not have sufficient financial incentives to continue to produce them and ensure that they remain commercially available.  Return to Top 

What do settings want from point-of-care (POC) testing?

More data were presented on the performance and implementation of proposed POC testing algorithms than on the performance and implementation of laboratory testing algorithms.  Participants who perform POC HIV screening with rapid HIV tests expressed interest in improving the positive predictive value of screening in these settings.  Data presented from a POC setting suggest that conducting at least one additional rapid HIV test from a different manufacturer following an initial reactive rapid HIV test is an effective way to improve screening specificity and positive predictive value.  However, some participants expressed concern that there may be insufficient data about the performance of the POC algorithms in low prevalence settings, in which the proportion of reactive screening tests that are false-positive results will be higher and where there may be more opportunity to observe false-positive results  from the 2nd and/or 3rd tests used in the algorithm than in higher prevalence settings. Data are also lacking for pregnant women, who have previously been reported to be more likely to have false-positive HIV Western blot results and thus may be more likely to have false-positive results from multiple HIV screening tests.  Establishing the specificity of POC rapid HIV test algorithms among pregnant women is especially important because immediate treatment decisions may need to be made on the basis of a rapid test algorithm for HIV testing conducted during labor.  Investigators from San Francisco presented the quality assurance procedures they used for HIV testing as part of the POC algorithms.  Some participants were skeptical that such a system would be affordable and that sufficient oversight of the testing could be maintained in a larger, less centralized network of POC HIV test sites.  Participants also discussed the utility of using a third rapid test rather than two as part of a POC algorithm.  Because, overall, only a small number of clients will have false-positive test results, the importance of ruling out infection using multiple rapid HIV tests in a single visit needs further evaluation.  Return to top

What do clinicians want from laboratories; do we need to release preliminary positive immunoassay results?

Few HIV clinicians were present at the conference.  However, clinicians and representatives from clinical laboratories who were in attendance wanted to be able to report laboratory immunoassay screening results to patients in the same way that point-of-care rapid test screening results are reported, by giving the positive results of immunoassay tests to patients and informing them that confirmatory Western blot test results will follow.  The decision to provide preliminary positive immunoassay results may be depend on the context in which testing is conducted.  For example, expectations for a hospital or clinical laboratory where a physician can access additional clinical information and immediately order additional tests may be different from those for a large commercial laboratory where provision of additional specimen may require follow-up with the submitting physician. Therefore recommendations about providing positive immunoassay results should be permissive rather than proscriptive. Because the specificity of some laboratory-based immunoassays is lower than the specificity of some rapid tests, testing with a second immunoassay that is different from the initial immunoassay or using a rapid test as the second test may be necessary to improve specificity of testing. Using the results of two immunoassays to identify persons likely to be infected and allowing clinicians to make a definitive diagnosis was proposed as a preferred strategy. Clinicians could order a quantitative HIV viral load and CD4 cell count to establish a baseline profile for patients whom they believe are likely to be infected with HIV.  An undetectable viral load could identify persons for whom additional diagnostic testing is indicated.  Further evaluation of dual immunoassay algorithms in low prevalence medical care settings is needed to quantify how often persons who have false-positive results on each of two different tests occurs, and therefore how often this approach might be expected to result in unnecessary HIV viral load, CD4 cell count, and other laboratory tests being performed in persons who are not infected with HIV.  Data were presented that suggest that when two or more tests give false-positive results, the strength of the signal from each test might provide additional information. Guidelines for supplemental testing should consider the signal-to-cutoff ratio, as is currently done in testing for hepatitis C virus.  Follow-up testing may be the only way to resolve confirmatory dilemmas, and it is likely that using a confirmatory algorithm based on results of a single clinical specimen may no longer be the most appropriate HIV testing strategy. Return to Top

What technology do we want?

In response to this question, some participants requested more advanced technology, such as:

  • Point-of-Care (POC) HIV RNA tests
  • Better oral fluid antibody tests
  • HIV antigen/antibody combination tests
  • Rapid POC tests for multiple sexually transmitted diseases (HIV/syphilis or HIV/HCV)
  • Rapid POC confirmatory tests
  • HIV-2 confirmatory and viral load tests

Other participants requested revised indications for currently available tests, including:

  • RNA testing on specimens other than plasma (serum, dried blood spots (DBS), and possibly oral fluid)
  • Revisions to the Western blot interpretive criteria to reduce the number of indeterminate results
  • More tests that can be performed using DBS
  • More tests, with better performance, optimized for use with oral fluid specimens
  • More details on the antigenic components of the assays to allow for easier decisions about what tests to include in testing algorithms and what tests might be expected to provide accurate results in persons who have received an HIV vaccine
  • Changes to the labeling of screening tests to allow for more results than reactive and non-reactive, such as reporting and interpretation of quantitative immunoassay signal-to-cutoff results

International participants and others noted that the United States is far behind the rest of the world in the quality and sophistication of HIV diagnostic technology. Participants requested that CDC work with the FDA to make the approval process for new HIV diagnostic test technology easier, timelier, and less expensive. Participants felt that CDC should work with the FDA and HIV test manufacturers to facilitate submission and approval of new HIV tests to provide faster access to a larger variety of assays in the United States.    Return to top

What do we need to be concerned with in settings with different HIV prevalence?

In low prevalence settings, a higher proportion of reactive screening tests would be expected to represent false-positive results than in higher prevalence settings, but the total number of persons with reactive test (true and false-positive) results would be relatively small. Data were presented from testing of U.S. military personnel that showed the ability of an algorithm using two immunoassays to provide a high positive predictive value in a low HIV prevalence setting (0.06%). However, one of the two HIV tests used in this algorithm is no longer commercially available. Therefore the results from the military may not necessarily be replicated using currently available tests, which incorporate a limited number of highly sensitive and specific synthetic and recombinant HIV antigens.  Additional data are needed from pregnant women, a population with low HIV prevalence in the United States that has previously been reported to have nonspecific reactions to HIV antigens resulting in false-positive test results.  Questions remain about the frequency with which persons in low prevalence settings may have false-positive test results using multiple 3rd generation assays capable of detecting IgM and 4th generation immunoassays that detect both IgM and p24 antigen.  The utility of HIV rapid tests used in a POC test algorithm also needs further evaluation in low-prevalence settings.  Return to top

What data do we want?

This final question summarized the data presented, the applicability of the proposed test algorithms, and the data gaps that still exist.  No data on the performance of immunofluorescence assays (IFA) relative to other technologies were presented at the conference, which makes the role and utility of IFA in the various HIV testing algorithms impossible to characterize.  There were also limited data presented on the sensitivity and specificity of HIV-1/2 discriminatory assays relative to HIV-1/2 screening tests. Lack of a satisfactory HIV-2 confirmatory test makes the estimation of the numbers of false-negative and false-positive HIV-2 test results a challenge.   Testing with two different immunoassays (Laboratory algorithm 3: see HIV Testing Algorithms: A Status Report)seemed to be a promising strategy with the most need for further evaluation.  Representatives of large laboratories that conduct a high volume of tests wanted to address this with combinations of two automated immunoassays, including new assays that have recently been approved by FDA.  Because of the large physical size of these automated platforms, smaller laboratories and those seeking quicker turn-around times for results identified the need for more extensive evaluations of Laboratory algorithm 3 using a single immunoassay followed by a rapid test.  Because HIV-infected persons who are on effective antiretroviral therapy (ART) sometimes seek HIV serologic testing, there is also a need to evaluate strategies for use with ART-experienced individuals, and also the performance of assays in persons who have received pre-exposure or post-exposure prophylaxis with antiretroviral medications.  Based on the assumption that a manufacturer will receive FDA approval for a 4th generation screening test, the yield of acute infections identified from such a test relative to molecular methods (e.g., RNA or DNA assays) should also be evaluated.  The performance of molecular tests when used on serum or possibly on oral fluid also needs to be assessed to determine if such tests can be used for supplemental testing of specimens routinely sent for HIV serologic testing. Finally, several presentations described negative molecular RNA results in persons with reactive immunoassay and Western blot results.  The accuracy of RNA tests and the utility of quantitative HIV RNA and qualitative DNA viral load tests for confirmation of HIV infection also need to be further evaluated.  Return to top 

Additional thoughts from the conference moderators and other attendees:

There is a need to further evaluate appropriate messages and plans for referral of persons at each of the endpoints defined in the algorithms.  This will require feedback from HIV counselors, clinicians, and clients seeking HIV testing.  Participants from international settings noted that international sites have testing options which may be superior to those currently available in the United States.  Thus, any recommendations targeted to laboratories and others in the United States must point out the limitations and data gaps that result from the testing options available and make it clear that such recommendations only apply to HIV testing conducted within the United States.  Multiple participants encouraged CDC to take advantage of the momentum generated by the process to present a compendium of proposed HIV testing algorithms for public comment, revise if necessary based on these comments, and then publish updated HIV testing algorithms that include all currently available technologies.  Existing data gaps should be seen as ongoing areas for further research development and not barriers to moving forward.  Participants emphasized that the HIV diagnostics field is advancing, but even as new assays are being rapidly developed, they must await FDA approval.  As new technologies are approved, periodic meetings to review HIV diagnostics will be necessary, and all recommendations for revised HIV testing algorithms need to be flexible enough to allow for incorporation of new developments. Return to top