Study area and design
This cross-sectional study was conducted in August 2011 in Bangui, capital of the Central African Republic, in the Hospital de l’Amitié, the Complexe Pediatrique and the Institut Pasteur of Bangui. Bangui is located beside the Oubangui River, north of the Democratic Republic of the Congo (geographical coordinates 7.00 N, 21.00 E). The climate is tropical, and rainfall peaks between April and November. The average temperature varies from 19°C to 32°C. The main malaria parasite is Plasmodium falciparum, and malaria transmission occurs throughout the year, with peaks at the beginning and the end of the rainy season, although no data are available on the intensity of transmission. Malaria accounts for more than 40% of morbidity in the country (Central African Republic Ministry of Health, 2010 annual report, unpublished data).
The Hospital de l’Amitié and the Complexe Pédiatrique are tertiary referral public health centres equipped with laboratories where thin and thick smears are analysed. The Institut Pasteur of Bangui is a private centre (International Network of Instituts Pasteur) for biomedical research, in which laboratory diagnosis for malaria is performed for patients referred by clinicians at health centres where this test is not available.
We analysed the performance of three RDTs: Paracheck™-Pf, SD Bioline malaria Ag-Pf and SD Bioline malaria Ag-Pf and /pan (Standard Diagnostics Ref 05FK60, Inc; Suwon City, Republic of Korea). SD Bioline Malaria Ag Pf and Paracheck™-Pf contain antibodies against P. falciparum-specific histidine-rich protein type 2 (PfHRP2), while Bioline Malaria Ag Pf/Pan contains antibodies targeting both PfHRP2 and lactate dehydrogenase specific to P. falciparum and other Plasmodium species (P. vivax, P. ovale and P. malariae). Blood samples from people attending each study centre for laboratory analysis were tested for the presence or absence of malaria parasites by microscopy and the RDTs. Samples from patients with negative microscopy (regardless of RDT result) were tested P. falciparum by nested PCR, on the assumption that false-positive results with RDTs are due to sub-patent parasitaemia .
Consecutive patients of all ages presenting at each study site were approached by the study technician for recruitment. Each patient’s medical card was checked, and those with a request for smear analysis were considered eligible if they were febrile (axillary temperature ≥ 37.5°C) or had a history of fever in the previous 24 h. They were included in the study if they gave written consent; for patients aged ≤ 18 years, informed consent was provided by a parent or guardian.
This project was reviewed and approved by the ethical committee of the University of Bangui. The Central African Republic Health Ministry also gave approval for this study.
Sample size estimation
For an estimated annual malaria rate in Bangui of 40% (Central African Republic Ministry of Health, 2010 annual report, unpublished data), an expected sensitivity of Bioline RDTs of 95%  and a sensitivity set at ± 2.5%, a sample size of 421 patients was estimated.
Selection of technicians
In each study centre, the laboratory team consisted of three technicians with university training, all of whom had been re-trained for diagnosing malaria in the laboratory by standard operational procedures . The technicians were also trained to perform and read the Paracheck™-Pf, SD Bioline malaria Ag-Pf and SD Bioline malaria Ag-Pf and /pan tests. At the Institut Pasteur, a fourth laboratory technician was designated to analyse all discrepant slides and to perform PCR.
Finger-prick blood samples were obtained for slides and for testing with the RDTs. Blood smears were air-dried, stained with 4% Giemsa and analysed under a light microscope (× 100 oil immersion) to detect asexual forms of P. falciparum. Parasite density was determined as the number of parasites per 200 leukocytes on the assumption of an average leukocyte count of 8000/μl of blood. A result was considered negative if no parasites were detected per 200 leukocytes. Each slide was read independently by two study technicians. In the case of a discrepant qualitative result (negative or positive), a third reading was done by the designated technician at the Institut Pasteur. All the laboratory technicians were blinded to the RDT results. The results of both microscopy and the RDTs were reported to the clinicians, who were advised to treat the patient for malaria if the results of these two analyses were discrepant.
The RDTs were performed by the third technician at the study centre following the manufacturer’s instructions.
For patients with negative microscopy, three drops of blood were collected on a piece of filter paper (Whatman®). The blood spots were air-dried and stored at 4°C in individual sterile plastic bags for PCR analysis.
Parasite DNA extraction and PCR assays were performed at the Institut Pasteur. The blood-impregnated filter paper piece was washed with distilled water and placed directly in a PCR tube containing the PCR reaction components. Genomic DNA was determined in an assay based on nested PCR for Plasmodium DNA .
Data were entered onto Excel spreadsheets and analysed with MedCalc®software (MedCalc Software, Acacialaan 22, B-8400 Ostend, Belgium).
The first step of our analysis was to determine the performance of the RDTs according to the falciparum density in samples found positive by microscopy. Density was categorized as ≤ 100, 101–200, 201–500, 501–1000, 1001–5000, 5001–50 000 and > 50 000 parasites/μl. RDT results were compared (paired proportions) with the McNemar test, at a level of significance (P) of 0.05.
The second step was to estimate the performance of the RDTs against the combined results of microscopy and nested PCR, expressed as true-positive (TP), true-negative (TN), false-positive (FP) or false negative (FN). The formulas used to calculate performance were TP/TP + FN for sensitivity, TN/TN + FP for specificity, TP/TP + FP for positive predictive value (PPV) and TN/TN + FN for negative predictive value (NPV). The results were interpreted with 95% confidence intervals (CIs). At baseline, the results of the comparison between the RDTs and microscopy were expressed as TP1, TN1, FP1 or FN1. Then, the analysis took onto account the results of PCR performed on samples that were negative by microscopy (classified as FP1 or TN1 in the RDTs). According to the PCR results, the results of the RDTs were designated as TP2 if PCR was positive in FP1, FP2 if PCR was negative in FP1, FN2 if PCR was positive in TN1 and TN2 if PCR was negative in TN1. Combination of the results of the RDTs for samples positive by microscopy and of PCR for samples negative by microscopy gave the values TP = TP1 + TP2, TN = TN2, FP = FP2 and FN = FN1 + FN2, which were used to calculate the performance of the RDTs.