Evaluation of nine HIV rapid test kits to develop a national HIV testing algorithm in Nigeria

Abstract

Background: Non-cold chain-dependent HIV rapid testing has been adopted in many resource-constrained nations as a strategy for reaching out to populations. HIV rapid test kits (RTKs) have the advantage of ease of use, low operational cost and short turnaround times. Before 2005, different RTKs had been used in Nigeria without formal evaluation. Between 2005 and 2007, a study was conducted to formally evaluate a number of RTKs and construct HIV testing algorithms. Objectives: The objectives of this study were to assess and select HIV RTKs and develop national testing algorithms. Method: Nine RTKs were evaluated using 528 well-characterised plasma samples. These comprised 198 HIV-positive specimens (37.5%) and 330 HIV-negative specimens (62.5%), collected nationally. Sensitivity and specificity were calculated with 95% confidence intervals for all nine RTKs singly and for serial and parallel combinations of six RTKs; and relative costs were estimated. Results: Six of the nine RTKs met the selection criteria, including minimum sensitivity and specificity (both ≥ 99.0%) requirements. There were no significant differences in sensitivities or specificities of RTKs in the serial and parallel algorithms, but the cost of RTKs in parallel algorithms was twice that in serial algorithms. Consequently, three serial algorithms, comprising four test kits (Bundi^TM, Determine^TM, Stat-Pak^® and Uni-Gold^TM) with 100.0% sensitivity and 99.1% - 100.0% specificity, were recommended and adopted as national interim testing algorithms in 2007. Conclusion: This evaluation provides the first evidence for reliable combinations of RTKs for HIV testing in Nigeria. However, these RTKs need further evaluation in the field (Phase II) to re-validate their performance.

FIGURE 1

Process for selecting HIV rapid test kits for development of interim national HIV testing algorithms. Commercial kits available in Nigeria were selected for evaluation singly based on WHO guidelines (Step 1). Of the nine kits, six were retained for inclusion in the algorithm testing exercise and three were dropped (Step 2). Serial and parallel testing algorithms were assessed for performance (sensitivity and specificity), cost and the country context (Step 3). Serial algorithms using four kits were selected and recommended as interim national guidelines in 2007 (Step 4).

FIGURE 2

Results from questionnaires administered to laboratorians conducting the evaluation. Respondents were asked to rate all nine kits based on the following criteria: ease of reading the reaction line (panel a); ease of interpreting the test results (panel b); ease of learning how to perform the test (panel c); and overall ease of using and running the kit (panel d). Scores ranged from very easy (1) to very difficult (5) for this set of four questions; panels a–d represent average scores. Respondents were also asked about the size of the packaging (panel e), with scores ranging from 1 (very bulky) to 5 (very compact); and about quantity of waste generated (panel f), with scores ranging from 1 (a lot of waste) to 5 (minimal waste). Panels e-f represent average scores.