The first study was located in Boende, Equatorial province, in the northwest of the country and the second study in Kabalo, Katanga province, in the southeast of DRC (Figure 1). Malaria is endemic in DRC with perennial transmission in Boende and seasonal transmission in Kabalo, with a peak between November and May. The ethics committee of the NMCP of the Democratic Republic of Congo approved the two studies. The ethics review board of Médecins Sans Frontières also reviewed the study protocol for Kabalo. Written informed consent was obtained from children's carers before enrolment.

The two studies were open-label. The study methodology was based on the WHO protocol for assessing efficacy of anti-malarials in areas of high transmission 6. In Boende, SP and AQ monotherapies were evaluated as well as in combination with artesunate (AS+AQ and AS+SP). In Kabalo, SP monotherapy, AS+AQ and AS+SP were evaluated. There were slight methodological differences between the two studies.

In Boende, thick blood films were stained with Giemsa 10% for 10 to 20 minutes and with Giemsa 5% for 25 minutes in Kabalo. Parasite density was determined as the number of parasites per 200 white blood cells (WBC), assuming a total WBC count of 8,000/μl. If less than 10 parasites were read per 200 WBC, the count was extended to 500 WBC. Microscopists reading the slides were blind to the treatment allocation. In the two studies, there was an independent double reading of all slides and resolution of discordant results (negative/positive/species or parasitaemia discordance above 50%) by a third senior reader. The third reading was taken as the final result. In addition, an external quality control of 10% of slides selected randomly from the screening, inclusion and follow-up slides was performed. From the study in Boende, slides were sent to the Institute of Tropical Medicine (IMT), Antwerp (Belgium). Slides from Kabalo were sent to the Amikuvu Regional Public Health Laboratory of Goma and the National Institute of Biomedical Research of Kinshasa (DRC). Haemoglobin was measured using the Lovibond technique (Assistant Co., Sondheim Rhon, Germany).

PCR genotyping to distinguish recrudescence from reinfection was performed in two different laboratories. For Boende, the analysis was performed at the ITM of Antwerp, Belgium using merozoite surface protein-1 (msp1) and merozoite surface protein-2 (msp2) P. falciparum gene loci 7. Infections were defined as recrudescent if at least one common band for both markers was observed in the presenting and recurrent infections. Recurrences were defined as reinfection if completely different bands were observed between the two infections in at least one of the two markers. For the study in Kabalo, genotyping was performed at the Epicentre laboratory at Mbarara University, Uganda according to a published method using three P. falciparum gene loci: msp1, msp2, and glutamate rich protein (glurp) 8. The glurp alleles were first compared between baseline and post-treatment samples. If they were different, the recurrence was attributed to a reinfection. If they were identical, then the msp1 and msp2 alleles were compared between the 2 samples. If there was a common band on one or two msp markers, then the recurrence was considered as recrudescence. If not, it was due to a reinfection.

In Boende, additional blood samples were collected from the children treated with SP as part of a multi-centre genotyping study to look for molecular markers of antifolate resistance. Analysis of mutations in codons 51, 59 and 108 of the dihydrofolate reductase (dhfr) and in codons 437 and 540 of the dihydropteroate synthase (dhps) genes of P. falciparum was performed in the ITM of Antwerp, Belgium using a real mutation-specific nested PCR and/or restriction digestions 910. The classification of samples was based on a published methodology 11. Each dhfr and dhps codon was characterized as 'wild type' (no mutation present), 'mixed' (both wild and mutant genotypes clearly present in the same infection), or 'pure mutant' (only mutant genotypes detected). The dhfr genotypes for each infection were categorized as follows: 'wild type': no mutation detected; 'single': infection involving parasites with a single mutation; 'double': infection involving parasites with a double mutations and 'triple': infection involving parasites with all three mutations detected. The dhps genotypes were defined as wild type; single or double mutations. Infections by parasites with a triple dhfr mutation and double dhps mutation were categorized as 'quintuple' mutation infections. These results have been reported in detail elsewhere 1213.

Data were entered in Microsoft Excel^® and analysed with Stata^® 10 (College Station, Texas, USA). Failure rates were estimated using Kaplan-Meier survival analysis, in which withdrawals, losses to follow-up and reinfections were censored on the last day seen. Patients with missing or indeterminate PCR results were censored on the closest prior visit when the malaria smear result was negative.

A total of 1,005 children were screened between September 2003 and February 2004, of whom 394 (39.2%) were enrolled (Figure 2). Three patients (0.8%) were lost to follow up. Baseline characteristics are presented in Table 1. Trends in fever, parasitaemia and haemoglobin over time are presented in Table 2. More than one quarter of patients treated with AS+SP or AS+AQ were parasitaemic on day 7. Of 68 children still parasitaemic on day 7, 3 had a fever and were classified as failures. The day 28 unadjusted failure rates were 55.5% [95% confidence interval CI: 46.0–65.4] for SP, 25.7% [95% CI: 18.3–35.4] for AQ, 52.2% [95% CI: 42.3–62.9] for AS+SP and 41.0% [95% CI: 31.3–52.3] for AS+AQ. PCR genotyping results are shown in Table 3. The end-point classifications after PCR adjustment are shown in Table 4. For the monotherapies, PCR adjusted failure rates were 35.9% [95% CI: 27.0–46.7] for SP and 18.3% [95% CI: 11.6–28.1] for AQ. Failure rates on day 28 after PCR-genotyping adjustment of AS+SP and AS+AQ were estimated as 24.6% [95%CI 16.6–35.5] and 15.1% [95%CI: 8.6–25.7], respectively. There were 2 (2.6%) ETFs in the SP arm and none in the other arms. Approximately half of all patients treated with SP, AQ or AS+SP required re-treatment by day 28. No child developed severe malaria after enrolment and there were no deaths. Three children were withdrawn from the study in the first week due to serious concomitant illness- pneumonia, dysentery and suspected meningitis.

Out of a total of 179 slides sent for external quality control, 41 (22.9%) had positive-negative discordant results during follow-up, 27 (65.7%) of them with very low parasitaemia (<100/μL). In three cases (1.7%), the discordance affected classification of the patient's outcome: patients with adequate therapeutic response were reclassified as ETF (two cases) and LPF (one case) based on the results of the external quality control.

The results of the molecular markers of antifolate resistance are presented in Table 5. There was 3/102, 2.9% [95%CI: 0.6–8.3] quintuple dhps and dhfr mutation, one with pure (1%) and two with mixed mutations, respectively.

A total of 880 children were screened between June and December 2004, of whom 207 (23.5%) were enrolled (Figure 2). During the study, investigators had to be evacuated from the site between October and December 2004 due to political instability. Twenty-seven patients recruited during this time were excluded from the analysis because of non-adherence to the study protocol. Baseline characteristics were similar between the groups (Table 1). Variations of fever, parasitaemia and haemoglobin over time are presented in Table 2. About 10% of patients in the AS+SP arm were parasitaemic on day 7. None had fever. The day 28 unadjusted failure rates were 33.2% [95% CI: 22.6–41.1] for SP, 16.6% [95% CI: 9.3–28.7] for AS+SP and 8.8% [95% CI: 3.8–19.8] for AS+AQ. PCR genotyping results are shown in Table 3. The final end-point classifications after PCR adjustment are shown in Table 4. The PCR adjusted failure rate for SP monotherapy on day 28 was 19.6% [95% CI: 11.4–32.7]. Corresponding failure rates were 0 [1-sided 95% CI: 5.8] for AS+SP and AS+AQ [1-sided 95% CI: 6.2]. There were eight (16.7%) ETFs in the SP monotherapy arm and none in the other treatment arms.

No child developed severe malaria after enrolment and there were no deaths. There was one serious adverse event – a child diagnosed with acute severe malnutrition during the first week. Of 182 slides sent for external quality control, 26 (14.3%) had positive-negative discordant results during follow-up; none of them affected patients' outcomes.