The RDT assessment took place concurrently with a 42-day ACT efficacy trial that conformed to WHO guidelines, which has been described elsewhere 9. In brief, Benin is an African country in which malaria is highly endemic and seasonal, with peaks during the low (March to May) and high (September to November) rainy seasons. The study site consisted of two small towns, Allada and Sekou, in the south of Benin, 50 km north of Cotonou, the capital. Plasmodium falciparum accounts for 95% of the Plasmodium species in this region. Study approval was obtained from the ethics committee of the Benin National Malaria Program.

The RDT assessment was carried out between April and November 2007. After screening febrile children aged between six and 59 months, those confirmed with falciparum malaria, who met inclusion criteria for the ACT efficacy study were recruited after their guardians had provided informed written consent. Two rapid diagnostic tests were conducted, the PfHRP2-based Immunoquick^® Malaria (Biosynex, Strasbourg, France) and the PspLDH-PfLDH-based Optimal^® IT (Diamed, Cressier, Switzerland).

Children were randomized to receive either artesunate + amodiaquine (ASAQ), artemether + lumefantrine (AL), or sulphadoxine-pyrimethamine (SP). Most of the children were followed up to assess both anti-malarial efficacy and duration of false-positive RDT results, as defined by positive RDT and continued negative microscopy after anti-malarial treatment. For purposes of RDT assessment, follow-up was conducted by clinical examination and finger-prick blood samples on days 3, 7, 14, 21, 28, 35, and 42 post-treatment, or on any other day if the child was unwell.

Blood films and rapid tests were carried out on the same finger-prick blood sample. A thick blood film for microscopic detection of parasites was prepared and stained with 10% Giemsa. The thick films were read by experienced technicians. Asexual parasites were counted against 200-500 leukocytes and converted to the number of parasites per unit volume, assuming 8,000 leukocytes/μL blood 10. Gametocytes were reported after examining 100 microscopic fields. Slides were considered negative if no parasites were detected after viewing 100 microscope fields. Microscopists were unaware of patient treatment allocation.

Internal quality control included a blind second reading of a portion of the slides as follows: all slides taken during days 0 and 3, all positive slides after day 3, and 20% of negative slides. All day-28 and day-35 negative slides were reread. In case of a discrepancy, two technicians re-examined the slides together and mutually decided on the reading. External controls were conducted by a Benin Ministry of Health government reference laboratory in the provincial capital, Cotonou.

The RDT tests were obtained directly from the manufacturers and stored in their original packaging at room temperature. Package desiccant quality was checked before using the tests 11. All RDTs were labeled with patient ID numbers, carried out, and interpreted according to the manufacturers' instructions. All RDTs were read visually, and a control line was established for the purpose of validating the test. For cases in which the control line did not appear, the result was considered invalid and the test repeated. For the PfHRP2-based test, results were recorded 15 minutes after placing the strip into six drops (300 μL) of buffer. The presence of both the control and test lines indicated a positive result for P. falciparum; the presence of the control line alone indicated a negative result. Results were recorded after 20 minutes for the PfLDH-based test. For PfLDH, the presence of the test line indicated a positive result for P. falciparum, and the presence of the test line for PspLDH indicated a positive result for Plasmodium species, including P. falciparum. The technician recording the RDT result was unaware of the corresponding microscopy results. Internal quality control included an immediate blind second reading of 100% of the RDTs. In case of disagreement, two technicians re-examined the RDT together and decided on the reading.

Specificity, sensitivity, and predictive values of the RDT [with a confidence interval of 95% (95% CI)] were estimated using microscopy as the denominator for comparison. Sensitivity was defined as the percentage of positive tests among the total number of positive blood slides. Specificity was defined as the percentage of negative tests among the total number of negative blood slides. The positive predictive value (PPV) was defined as the percentage of positive blood slides among the total number of positive tests. The negative predictive value (NPV) was defined as the percentage of negative blood slides among the total number of negative tests. The percentage of false-positive (FP) tests was defined as the percentage of tests that remained positive during follow-up once the blood sample was negative by microscopy for asexual-stage parasites. The presence of gametocytes alone was not considered to be a positive microscopy result. Categorical variables were compared using the Fisher or X₂ exact tests, in accordance with the number of patients. A p value of <0.05 was considered statistically significant.

Between March and September 2007, 205 children presenting with fever and a P. falciparum parasitaemia above 1,000/μL were included in the study. Parasite densities ranged between 1,000 and 525,000 parasites/μl.

The children were divided into three groups, according to their treatment allocation: 81 were treated with the AL combination, 80 with the ASAQ combination, and 44 with the SP combination. During the six-week follow-up period, 80 children were followed to day 42 and 40 children were followed to day 28. Among the children included in the three groups, 72 were given additional treatment within 35 days, due to malaria recrudescence or re-infection (and were excluded from the RDT study after this second treatment), and 13 children were lost to follow-up. Detailed results of the efficacy study have been published 9. Seven gametocyte carriers were diagnosed during the follow-up (two in the AL treatment group, one in the ASAQ treatment group, and four in the SP treatment group.)

On day 0, similar sensitivities were observed with both RDTs. The PfHPR2-based tests were positive for 203 children (99%, [95% CI: 96.5-99.7]) and the PfLDH-based tests were positive for 204 (99.5%, [95% CI: 97.3-99.9]). PspLDH and PfLDH detection yielded similar results during the study. Since only children with P. falciparum malaria were included in the study, only PfLDH detection results are presented and compared with PfHRP2 detection.

Relatively few samples were positive by microscopy during the days following treatment. Thirty-five samples were positive on day 3, seven on days 7 and 14, 16 on day 21, 15 on day 28, 11 on day 35, and three on day 42 (Table 1). Sensitivity decreased with day of follow-up for both the PfHPR2-based test (from 100% on day 3 to 67% on day 42) and the PfLDH-based test (from 80% on day 3 to 67% on day 42) (Table 1). On the other hand, specificity increased during follow-up, going from 17% on day 3 to 95% on day 42 for the PfHPR2-based test and from 87% on day 3 to 100% on day 42 for the PfLDH-based test (Table 2). From day 3 to day 28 of follow-up, the specificity of the PfLDH-based test was statistically higher than that of the PfHRP2-based test (Table 2). Negative predictive values (except one) during follow-up were greater than 95% (Table 3) for both the PfHPR2- and PfLDH-based tests. On the contrary, positive predictive values were not satisfactory for either RDT (less than 80%), except for PfLDH-based tests carried out on day-28 samples (Table 4).

Figure 1 shows the proportion of false-positive results obtained during follow-up for PfHRP2 and PfLDH antigen detection as a function of parasite density at screening. Day-0 parasitaemia (number of parasites/μl) was stratified into low (1,000 - 10,000; n = 78), middle (10,000 - 50,000; n = 75), and high (50,000 - 525,000 (n = 52). The proportion of children with low parasite density was lower than the proportion with high parasite density from day 7 to day 35 for PfHRP2-positive antigen detection. Twenty-eight days after effective treatment, 8% (4/50; 95% CI 3.1 - 18.9) of the children with low parasite density had PfHRP2 false-positive results, compared to 42.9% (15/35; 95% CI 28.0 - 59.1) of those with high parasite density (p = 0.0004). In contrast, on any day during follow-up, the proportion of false-positive PfLDH antigen detection results was low and similar in all three groups (Figure 1).

Independent of the treatment received, PfHRP2 antigen persisted until day 35; indicating a high proportion of false-positive results with the PfHPR2-based test (Figure 2). The proportion of false positive results was unrelated to the treatment given. The specificity of PfLDH detection differed among treatment allocations. After SP treatment, there was no statistical difference between PfHRP2 and PfLDH persistence. After AL and ASAQ treatments, fewer false positive results were observed with PfLDH than with PfHRP2 from day 3 to day 28, (all p < 0.01). Moreover, after SP treatment, the proportion of false-positive results obtained with the PfLDH-based test was high until day 14 (Figure 2). The specificity of the PfLDH-based test was higher after ACT than after SP treatment from day 3 to day 14 (all p < 0.05). No correlation was found between the presence of gametocytes and the duration of RDT positivity, except on day 7, when three among the 13 results were false-positive (13/164) with the PfLDH-based test in gametocyte carriers.

Possible reasons for negative results included low parasite density and/or deficient in antigen parasites for both PfHPR2 and PfLDH detection. PCR-amplified exon 2 PfHRP2 fragments were obtained on day 0 from the two negative isolates with the PfHPR2-based test; therefore, the parasites were not antigen-deficient. False-negative results obtained during follow-up with both the PfHPR2- and PfLDH-based tests could be explained by low parasite densities, comprising between 16 and 800 parasites/μl.