Malaria rapid diagnostic tests (RDTs) detect Plasmodium parasites in blood by an antibody-antigen reaction on a nitrocellulose strip. Reactions are visible as cherry-red lines. Two-band RDTs are mostly designed to detect Plasmodium falciparum; they display a control line and a test line, which targets either histidine-rich protein-2 (HRP-2) or P. falciparum-specific parasite lactate dehydrogenase (Pf-pLDH). Three- and four-band RDTs display a control line and two or three test lines, one targeting P. falciparum specific antigen, another line targeting antigens common to the four species, such as pan-Plasmodium-specific lactate parasite dehydrogenase (pan-pLDH) or aldolase, and, in case of the four band RDTs, a third line which targets Plasmodium vivax-specific pLDH (Pv-pLDH).

RDTs are currently rolled out by national malaria control programmes in endemic settings as a tool for parasite based diagnosis in the scope of artemisinin-based combination therapy (ACT) [1]. In 2007, more than 70,000,000 tests were performed [2]. During the last few years, RDTs have technically improved and so-called one-step tests have replaced the older multistep tests. However, despite their apparent simplicity, they are not completely fail-proof [3,4].

Most malaria RDTs are marketed as kits that include material for 20 to 25 tests, such as lancets for finger pricking, test strips (available as dipstick, cassette or card box formats), transfer devices (pipettes, straws, capillaries or loops) and the buffer. Cassettes are usually individually packaged, and the number of lancets and transfer devices match the number of cassettes. All RDTs need a buffer to lyse the blood and to allow capillary flow along the nitrocellulose strip. Mostly, this buffer is supplied in a single bottle or dropper vial.

During field visits (particularly in Africa, ITM teams repeatedly observe availability problems of buffer vial: for instance, some cassettes were sent for testing in the ward, but the buffer vial did not return. To compensate, laboratory technicians took either a buffer vial from another kit (sometimes a kit of another brand), or used saline, distilled water (liquids for parenteral drugs dilution) and occasionally tap water as substitute liquids. Apparently, this substitution for the buffer seemed not to cause too much interference, as in all observed cases, there was enough background clearance and both control line and test lines were clearly distinguishable.

This phenomenon was also noted during a practical teaching session at the Institute of Tropical Medicine (ITM): students and staff were astonished to observe a visible HRP-2 line when the blood of one of the present authors (PG) was tested with a two-band RDT which was run with distilled water. PG had no symptoms of malaria, nor did he suffer from malaria in the recent past. When performing the RDT with the kit's dedicated buffer, there was no HRP-2 line visible. The HRP-2 line appeared upon retesting with distilled water and also when using saline and tap water as substitute liquids. As false-positive test line seemed to be the explanation, it was decided to explore this phenomenon.

EDTA-blood samples from five healthy subjects with no recent history of malaria were used. For all samples, the diagnosis of malaria was excluded by microscopy and species-specific PCR as previously described [5,6]. The presence of known causes of false positive RDT results such as the rheumatoid factor or antinuclear antibodies was ruled out [7-9] and none of the subjects had been manipulating mice during the past ten years, thereby reducing the probability of false positive results due to anti-mouse antibodies [10]. For most experiments, fresh samples were used; samples stored at -70°C were used in the case of delays of delivering of the RDTs. For comparison, all RDTs were run with their kit's dedicated buffer and two clinical samples, one infected with P. falciparum and another with P. vivax, at parasite densities of 36,140 and 3,600/μl respectively.

Malaria RDTs marketed as cassettes and folded card box were selected, and RDT brands commonly used in field settings were included. CE marking and FDA approval of the RDTs were recorded, as well as their presence on the World Health Organization (WHO) lists of RDT manufacturers and distributors complying with ISO13485:2003 or US FDA 21 CFR 820 production norms and their evaluation by the World Health Organization/Foundation for Innovative New Diagnostics (WHO/FIND) [11,12]. RDTs that were not on the local (Belgian) market were directly ordered from the manufacturer. In view of the wide lot-to-lot variations and the ever changing composition of RDTs, it was decided not to display the individual RDT brand and kit names, in line with previous comparative studies assessing characteristics of RDTs [13,14].

All RDT kits were used within their expiry date and had been stored at room temperature (maximum 25°C) before analysis. RDTs were assessed in the same run with the five subjects' blood and the following buffers: RDT kit's dedicated buffer, distilled water (Denolin, Brussels, Belgium), saline (NaCl 0.9%, Qualiphar, Bornem Belgium) and tap water from local supply. For each RDT and substitute liquid, the blood samples of all five subjects were used except when a particular RDT kit was finished. In addition, tests were run in the absence of blood, with distilled water as the substitute liquid.

All tests were performed according to the instructions of the manufacturer, except that samples were loaded with a pipette (Finnpipette, Helsinki, Finland) instead of the transfer device supplied by the manufacturer. Readings were performed by three readers at daylight assisted by a standard electric bulb, and within and not beyond the prescribed delay after application of the sample and buffer.

In case the control line did not appear, the result was considered as invalid and the test was repeated. RDT test lines were interpreted according to the manufacturers' instructions. In addition, test line intensities were scored into five categories: none (no line visible), faint (barely visible line), weak (paler than the control line), medium (equal to the control line) or strong (stronger than the control line) [6]. Observers were blinded to each others' reading. The results of the readings considered were based on consensus agreement [15].

The appearance, shape and crispness of the control and test lines and the clearance of the background were compared with those obtained with the RDT kit's dedicated buffer and the two Plasmodium-positive samples. Visible test lines observed in the malaria free subjects' samples will be further referred to as "false-positive test lines".

Inter-reader reliability was assessed and expressed as percentage agreements for all three readers and kappa values for each pair of readers. To assess test reproducibility, a sample from one subject was tested upon five occasions for all RDT brands and the three substitute liquids. Two RDT brands were not included in the reproducibility assessment because of shortage of tests.

The study was approved by the Institutional Review Board of ITM and by the Ethical Committee of Antwerp University, Belgium.

Thirty different RDT brands were selected. As two brands from the same manufacturer showed bad clearance of the background upon testing with Plasmodium-positive samples and another brand had a very low specificity, they were not included in the study. The final panel consisted of 27 brands (26 cassette and 1 card box format). Eleven (40.7%) of them had CE-mark compliance, one was approved by the U.S. Food and Drug Administration, 22 (81.5%) were included in the WHO list of RDTs adequate evidence of good manufacturing practice (GMP), 22 (81.5%) were evaluated by the WHO/FIND [11,12], and 9 (30%) are included in the list of malaria RDTs eligible for procurement by WHO [16]. The 27 RDT brands comprised two-, three- and four-band RDTs (Table 1) and all Plasmodium target antigens (HRP-2, Pf-pLDH, Pv-pLDH, pan-pLDH and aldolase) (Table 2).

When run with their kit's buffer, none of the RDT brands tested positive with any of the samples of the five malaria free subjects. Likewise, when tested with the two Plasmodium positive blood samples and the RDT kit's buffer, the expected test line results were observed.

When tested with a substitute liquid, there were eight invalid test results; they neither showed a control line upon repetition. Five of them occurred in a single RDT brand when assessed with distilled water. False-positive test lines were visible for at least one sample and substitute liquid in 20/27 (74%) RDT brands, at the following frequencies: saline: n = 16, distilled water n = 17 and tap water: n = 20 (Table 1). They occurred randomly among the samples of the five malaria free subjects, and occurred in at least three of them. For 10 RDT brands, all five blood samples tested positive with at least one of the substitute liquids (saline: n = 4, distilled water: n = 5, tap water: n = 7). Two-, three- and four-band RDTs were all affected, as well as all Plasmodium target antigens (Table 2).

For distilled water and tap water as the substitute liquid, the RDT strips with false-positive lines showed a clearance of background similar to those observed with the RDT kit's buffer. In the case of saline, the background was less clear. Overall, the colour and crispness of the control and false-positive test lines were similar to those obtained with the RDT kits' buffers and Plasmodium positive samples (Figure 1 and 2). The proportions of medium and strong line intensities for all substitute liquids combined were as follows: 26.7% (23/86) for HRP-2, 11.5% (3/26) for Pf-pLDH, 37.7% (20/53) for Pv-pLDH and 29.6% (26/88) for pan-pLDH. All false positive aldolase lines (n = 11) were either faint or weak. To the exception of better clearance of the background, the use of frozen sample did not influence the outcome of the results.

When tested with distilled water in the absence of blood, there was only a single invalid cassette, in which neither a control line appeared upon retesting. Control lines were observed in all remaining runs. Twenty-two false positive test lines were observed in 15 (55.6%) RDT brands and occurred in all Plasmodium targets (Table 3). Control and false positive test lines showed colour, thickness and crispness similar to those obtained with the RDT kits' buffers and Plasmodium-positive samples (Figure 1 and 2). The line intensities of false positives have however tended to be lower (Table 3).

The inter-observer agreement for positive and negative readings was high, with 96.4% overall agreement between the three observers and kappa values between 0.94-0.98 for each pair out of three observers. Inter-observer agreements were similar for all three substitute liquids. Discrepancies between line intensity readings were limited to one category of line intensity (e.g., line intensity read as weak by reader one and as medium (but not strong) by reader two).

At reproducibility testing, false positive test lines occurred in at least four out of five runs in approximately half of the Plasmodium targets lines affected (40.0% (10/25), 51.9% (14/27) and 51.6% (16/31)) for saline, distilled water and tap water respectively). In terms of RDT brands, they occurred in at least four runs in eight, eleven and nine brands for saline, distilled water and tap water respectively.

CE-mark and non-CE labelled RDT brands were equally affected by the buffer substitution (11 out of 20, versus four out of seven). In addition, there were 11/22 WHO-GMP listed RTD brands affected versus four out of five RDTs that were not listed, 12/22 FIND evaluated brands versus three out of seven, and five out of nine RDTs from the WHO procurement list affected versus 10/18. Although the false-positive test lines occurred randomly among the different RDT brands, there were certain RDT brands that were more affected than others.