Conceived and designed the experiments: DK lS CM DPO. Performed the experiments: JW. Analyzed the data: DK KL lS CM DPO. Wrote the paper: DK lS CM DPO.
The authors have declared that no competing interests exist.
Concerns about false-positive HIV results led to a review of testing procedures used in a Médecins Sans Frontières (MSF) HIV programme in Bukavu, eastern Democratic Republic of Congo. In addition to the WHO HIV rapid diagnostic test algorithm (RDT) (two positive RDTs alone for HIV diagnosis) used in voluntary counselling and testing (VCT) sites we evaluated
2864 clients presenting to MSF VCT centres in Bukavu during January to May 2006 were tested using Determine HIV-1/2® and UniGold HIV® rapid tests in parallel by nurse counsellors. Plasma samples on 229 clients confirmed as double RDT positive by laboratory retesting were further tested using both WB and the Orgenics Immunocomb Combfirm® HIV confirmation test (OIC-HIV). Of these, 24 samples were negative or indeterminate by WB representing a false-positive rate of the WHO two-test algorithm of 10.5% (95%CI 6.6-15.2). 17 of the 229 samples were weakly positive on rapid testing and all were negative or indeterminate by WB. The false-positive rate fell to 3.3% (95%CI 1.3–6.7) when only strong-positive rapid test results were considered. Agreement between OIC-HIV and WB was 99.1% (95%CI 96.9–99.9%) with no false OIC-HIV positives if stringent criteria for positive OIC-HIV diagnoses were used.
The WHO HIV two-test diagnostic algorithm produced an unacceptably high level of false-positive diagnoses in our setting, especially if results were weakly positive. The most probable causes of the false-positive results were serological cross-reactivity or non-specific immune reactivity. Our findings show that the OIC-HIV confirmation test is practical and effective in field contexts. We propose that all double-positive HIV RDT samples should undergo further testing to confirm HIV seropositivity until the accuracy of the RDT testing algorithm has been established at programme level.
Since 2000, Médecins Sans Frontières (MSF), in conjunction with the local Ministry of Health, has run a comprehensive HIV programme in Bukavu, a city located in the conflict affected eastern region of the Democratic Republic of Congo (DRC). Bukavu is home to significant numbers of people displaced by the conflict that has plagued the region since 1996. In 2004, the HIV prevalence in Bukavu was estimated to be 3.1%.
Between January 2001 to May 2006, 13,678 clients were tested for HIV by nurse counsellors using rapid diagnostic tests (RDTs) performed according to the WHO HIV two-test algorithm
In late 2004, a group of patients from the MSF/MoH clinics who had initially tested positive for HIV on the basis of two positive RDT results were identified as having maintained a high CD4 count. They were retested with RDTs and shown to be HIV negative. The results were of great concern since a false-positive result may have significant adverse outcomes: stigma and discrimination, abandonment, domestic violence, exposure to unnecessary and potentially toxic medical treatment, and loss of confidence in the HIV programme. HIV counselling and testing practices were reviewed and supervision intensified; however the false-positive results persisted. The MSF algorithm was changed in December 2005 to introduce retesting of all VCT positive results and confirmation testing by Orgenics Immunocomb Combfirm HIV® (OIC-HIV) and/or western blot (WB). One of the study authors (JMW) observed that on visual reading of the rapid test result, some test lines were fainter than others raising the possibility that these weakly reactive tests may correlate with a higher risk of false-positive results.
In an observational study, we analysed 5 months′ routine programmatic data to compare the results obtained using the WHO 2-test algorithm with our new testing procedures under programme conditions. Our primary objective was to evaluate the performance of the OIC-HIV test compared to WB testing as a practical
Testing was performed on consecutive VCT clients, drawn from both community health education activities and sexually transmitted infection clinics during January to May 2006. Programme details and treatment outcomes have been described elsewhere
Clients were tested using Determine HIV-1/2® and UniGold HIV® rapid tests in parallel by nurse counsellors. The tests were performed in the presence of the client using plasma from EDTA venous blood samples collected and centrifuged immediately before testing. New gloves were used for each client. The specimen tube was labelled in front of the client using a unique client number. Manufacturers' instructions for test procedures were followed. Each test was interpreted by the counsellors as a positive result if both the test and internal control lines were positive. If both rapid tests were interpreted as positive, the client was informed that their result was positive but that a confirmation test would be done and that this might involve referral to an external laboratory. The individual was offered an immediate appointment in the HIV clinic, and accompanied to the clinic by the counsellor.
The counsellors further classified the results as strongly positive if the line had normal colour thickness and intensity or weakly positive if significantly thinner and weaker than normal positive results. If at least one of the two positive tests gave a weak result, the overall result was classified as weak positive.
The client was given a negative result if both tests were negative. If one test was positive and the other negative, the client was told the result was indeterminate and needed to be repeated in 6 weeks. All clients, regardless of the result, received routine counselling on risk reduction practices.
Testing was done by the same four counsellors during the study period. All counsellors had programme experience and received training from the laboratory supervisor before the study. During the study period, the counsellors were directly observed performing the tests on a weekly basis by one of the study authors (JMW).
Daily supervision and psychosocial support was given to counsellors. Counsellors saw a maximum of eight clients per day to limit fatigue and ensure quality of testing and counselling.
All double-positive results, whether weak or strong, were rechecked in the MSF reference laboratory using the same rapid test algorithm as that used in the VCT clinics. The laboratory control was performed unblinded on the same tube of plasma drawn by the counsellor. The laboratory technician further classified the results as strong or weak positives using the same criteria as the nurse counsellors. Samples from clients reported as negative by VCT were randomly selected for retesting in the laboratory.
All rapid test kits were stored according to manufacturers' instructions. Quality control was performed using a weak-positive sample prepared by the MSF reference laboratory on a randomly selected Determine and Unigold test each week and from every new box or pack.
Plasma samples confirmed as double-positive on RDTs by the laboratory were tested using OIC-HIV and/or WB (GeneLabs Diagnostics HIV BLOT 2.2®). OIC-HIV was chosen as it requires no specialised laboratory equipment and thus can be easily performed in a peripheral laboratory, takes less than 2 hours, is easy to interpret, and separately detects p24, p31, gp40, gp120, and p36 antibodies. In addition, the HIV-OIC test uses recombinant and peptide antigens as do the RDTs and therefore can be expected to have similar sensitivity. WB was chosen as it is the internationally recognised gold standard.
WBs were performed at the National Reference Laboratory, Kigali, Rwanda. The laboratory was blind to the OIC-HIV results and the strength of rapid test reactivity. OIC-HIV tests were performed blind to WB results by the laboratory supervisor in the MSF referral laboratory.
The WB results were interpreted according to Centers for Disease Control criteria
This is defined as a sample that gives a positive test result with two independent RDT tests, and classified as positive according to the WHO two-test algorithm, which is not confirmed by WB testing.
Analyses were performed using STATA Version 9 (STATACorp, Texas USA); 95% binomial exact CIs were used.
Informed consent was obtained verbally from all clients before testing. The study was reviewed and approved by the MSF Ethics Review Board.
There was no specific funding source. The study was funded as part of routine MSF operations.
2864 clients received voluntary counselling and testing at the two test sites during the study period. Approximately 60% of the clients tested were female. The majority of clients (70%) reported residence in Bukavu city while the rest came largely from the rural areas surrounding the city. The most common reasons given for testing were ‘to know their status’, to confirm results received elsewhere, or a concern prompted by repeated illness, a sexually transmitted infection, or a history of sexual violence.
The laboratory confirmed all 60 negative VCT rapid test samples selected for quality control as negative.
Repeat rapid tests were performed on 330 samples. There was 100% agreement between the VCT and laboratory rapid test results for positive/negative and strong/weak classification. All rapid test results presented here are laboratory results, not VCT results.
Rapid test results (n = 229) | OIC-HIV reactions | WB banding pattern | n | OIC-HIV result | WB result |
gp120+, gp41+, p24+, p31+ | ≥5 bands including gp120+, gp41+, p24+, p31+ | 199 | POS | POS | |
gp120+, gp41+, p24+ but p31− | ≥5 bands including p31+ | 1 | POS | POS | |
gp120+, gp41+, p31+, p24+ | ≥5 bands but gp120− | 3 | POS | POS | |
gp41+ p24+ | gp41+ p24+ | 1 | POS |
POS | |
gp120+ gp41+ | gp120+ p24+ | 1 | POS |
POS | |
NEG | NEG | 1 | NEG | NEG | |
NEG | p41+ | 2 | NEG | IND | |
p41+ | NEG | 1 | IND | NEG | |
gp41+ p24+ | gp41+ | 1 | POS |
IND | |
gp120+ gp41+ | gp41+ | 1 | POS |
IND | |
p24+ p31+ | gp41+ | 1 | IND | IND | |
7 | |||||
NEG | NEG | 9 | NEG | NEG | |
NEG | gp41+ | 7 | NEG | IND | |
gp120+ | NEG | 1 | IND | NEG | |
17 |
D+s = Determine HIV-1/2® strong-positive. D+w = Determine HIV-1/2® weak-positive. U+s = UniGold HIV® strong-positive. U+w = UniGold HIV® weak-positive. POS = positive. NEG = negative. IND = indeterminate.
The discrepancy between banding patterns in OIC-HIV and WB may have been caused by either laboratory/clerical error, or different patterns of cross-reactivity between the two tests.
Using Centers for Disease Control criteria
Using the manufacturer's interpretation.
Using the manufacturer's directions for interpretation of the OIC-HIV results, there was agreement between OIC-HIV and WB in classifying 227 of the 229 samples (99.1% [95%CI 96.9–99.9%]) as positive or negative/indeterminate (
Using the more stringent criterion for OIC-HIV positivity, requiring a minimum of two
Our findings demonstrate that two positive rapid HIV tests performed alone gave a 10.5% false positive rate compared to WB in our program in eastern DRC, and these findings strongly support the need for confirmation testing of double-positive HIV RDT test results. False-positive reactions in individual HIV serological tests have been widely reported for both RDT
Our study has the limitation that testing was performed under field conditions where it was not possible to follow up clients with negative or indeterminate WB results, or further confirm results by NAAT or p24-antigen testing. Therefore it is possible that some clients with positive RDT results and negative or indeterminate WB results may have been in early seroconversion. However a number of serological findings strongly support cross-reactivity as the major cause. The finding of a single gp41 band in 12/24 cases is not consistent with early phase infection and is more consistent with interference with non-HIV associated anti-gp41 as has been previously reported
The possibility that the false-positive results were a result of error in VCT is excluded because the results reported are based on the results obtained in the laboratory not during VCT. Although clerical error cannot be excluded it is unlikely given the high concordance between the results of the OIC-HIV testing performed in the MSF laboratory, and WB performed in a blinded manner in the Kigali laboratory. In addition, the false positives occurred using tests with different batch numbers, hence it is very unlikely that results were caused by test device error.
17 of the 24 samples that could not be confirmed by WB gave a weak reaction in one or both RDT tests. This is consistent with previously reported low specificity for weak reactivity in ELISA
For reasons of practicality and cost, WB and molecular biology (NAAT) confirmation testing is not available in most resource-limited settings. For this reason we evaluated the OIC-HIV test as an
However, using the manufacturer's recommended interpretation (2
Our study design did not allow an analysis of factors causing cross reactivity, although several possibilities exist. Firstly, there is a high degree of HLA polymorphism among Africans
A further possibility is that non-specific polyclonal B-lymphocyte antibodies formed in any early immune response may interfere with HIV serological testing
Our findings have important implications for scale-up of treatment in areas of low HIV prevalence. The PPV of the WHO two-test algorithm
We acknowledge the significant contribution of Nathan Makombe, Samuel Duyizere (Kigali Reference Laboratory), Gabriel Mulungano, Laurent Mbilizi, Dr Tina Amisi, Lievin Alimasi, Celine Kamba, Josephine Masonga, Alexis Balekege, Yvonne Tshikuju, Daniel Ntakalalwa, Alain Chitona, Dr Rebecca Adlington, Marja de Jong, and Corry Kik (MSF Bukavu).