This randomized, open-label trial was conducted between October, 2006 and September, 2007 at two district hospitals within the Ashanti region of Ghana, the Child Welfare Clinic of the Agogo Presbyterian Hospital, Asante Akim North district, and the outpatient department of Agona Government Hospital, Afigya Sekyere district. At both sites, the predominant ethnic group is Ashanti, minorities are Fanti, Ewe, Ga and Dagomba.

The majority of the population subsists on small-scale crop farming, notably trading in general goods also plays an important role. The vegetation in the study area is partly tropical woodland, partly savanna. The area is classified as holoendemic and malaria transmission occurs perennial with seasonal peaks 14. The predominant species is P. falciparum and the principal vectors are mosquitoes of the Anopheles gambiae complex and A. funestus. According to annual reports of the District Health Directorate more than 20% of hospital consultations are attributed to malaria, and malaria-related mortality is reported to be around 25% in children less than five years of age. Ownership of insecticide-treated bednets (ITNs) in the region was reported to be only 1.6%, but subsidizing and education programmes are ongoing 15.

Children aged 6–59 months with measured axillary body temperature ≥ 37.5°C and/or history of fever within the previous 24 hours presenting at the hospitals were initially seen by the study physician. After obtaining capillary blood samples, thick and thin blood films were prepared, and Diff-Quik^® (Dade-Behring, Switzerland) stained thick films were microscopically examined for the presence of malaria parasites. Parents/guardians of slide-positive children with P. falciparum monoinfection, a parasite density between 2,000 and 200,000 asexual parasites/μl, and without signs of severe malnutrition (weight-for-age < -3 SD of the median of the WHO child growth standard) or any other severe underlying disease were asked to participate in the study if they were permanent residents in the area.

Further inclusion criteria were 1) absence of general danger signs or signs of severe malaria according to WHO 16, 2) no intake of antibiotics and no adequate antimalarial treatment within the previous 7 days, 3) absence of a history of hypersensitivity to any of the study drugs, 4) ability to tolerate oral therapy, and 5) informed consent provided by a parent/guardian. Before enrolment, the aims and procedures of the study were explained to the parent/guardian and his/her understanding was confirmed by interview before written or, in case of illiteracy, thumb-printed consent was obtained in the presence of an unbiased witness.

Children were randomly assigned to receive either ASAQ or AL. Before drug application a venous blood sample was taken to measure leukocyte counts as indicator for a bacterial co-infection. Treatment allocation was based on a computer-generated list with randomization of individuals in blocks of ten.

Co-blister packed ASAQ [Arsucam^® (50 mg AS/white tablet and 153 mg AQ/yellow tablet), Sanofi-Aventis, Paris, France] were given weight-adjusted [5 – <10 kg (AS 1/2 tablet + AQ 1/2 tablet), 10 – <21 kg (AS 1 tablet + AQ 1 tablet), and 21 – <40 kg (AS 2 tablets + AQ 2 tablets] once daily for 3 days, as recommended by the manufacturer.

AL [Coartem^® (20 mg artemether/120 mg lumefantrine fixed-dose combination tablets), Novartis, Basel, Switzerland] were administered twice daily as three-day, six-dose regimen according to body weight [5 – <15 kg (1 tablet), 15 – <25 kg (2 tablets), or 25 – <35 kg (3 tablets)], the second dose 8 hours after the first dose, according to manufacturer guidelines.

After enrolment, evaluation and consenting, participating children received the first dose of the individually allocated treatment (in sealed and numbered, opaque envelopes) directly observed by a study nurse in a separate cabinet, while study physicians were blinded to treatment allocation. Prior to administration the tablets were crushed and mixed with water.

Treated children were observed for 30 minutes and the dose was readministered if vomiting occured. Children who repeatedly vomited their first dose of study medication were referred for alternative management. Subsequently, the treatment schedule was explained to the parents, who were strongly encouraged to resume normal diet and give the study medication with fat-containing food or during nursing if the participating child was breastfed. The three-day treatment course was then completed at home.

Children received paracetamol for treatment of febrile symptoms. Those with haemoglobin concentrations of less than 10 g/dl received ferrous sulfate, folic acid and anthelmintic treatment (if appropriate) according to national guidelines.

Participants were asked to attend regular follow-up visits on days 3, 7, 14, 28. Except for day 3, postponement of one day was tolerated. Parents/guardians were encouraged to come to the clinic whenever a child felt sick, and to return immediately to the hospital in case danger signs and/or deterioration of the clinical status were observed during treatment, or in case a treatment dose was vomited.

Study participants were visited actively at home by members of the study team and taken to the clinic if absent on a scheduled visit day. On each visit a standardized medical history and medical examination Case Report Form was completed. Capillary blood was obtained by finger-prick for thick and thin blood films, a drop of blood was stored on filter paper (Whatman^®, Kent, UK) and a photometric measurement of haemoglobin (HemoCue^® Ltd., Derbyshire, UK) was performed. Full blood counts of venous blood samples were done with an automated haematology analyzer on day 0 (Kx-21N, Sysmex^® Europe GmbH, Norderstedt, Germany). Blood smears were stained with Giemsa and parasite densities were determined by counting asexual parasites per 200 white blood cells (or 500, if the count was less than 10 parasites per 200 white blood cells), assuming a white blood cell count of 8,000/μl. Smears were considered negative if no asexual parasite was seen after review of at least 100 high-power fields. Parasite species were determined by examination of thin films. For quality control slides of >10% of the enrolled patients were re-examined by a second microscopist who was blinded for the patient number, day of follow-up, and former microscopic results. After comparing the new with the original outcome classifications, no discrepancies were found. Children who did not tolerate oral therapy (repeated vomiting after initial study drug application), with mixed plasmodial infections, hyperparasitaemia (>200,000 asexual parasites/μl), or leukocytosis (>15,000 leukocytes/μl) were subsequently excluded from the analysis.

Treatment outcomes were defined according to WHO guidelines (2003) as early treatment failure (ETF): danger signs or complicated malaria or failure to adequately respond to therapy on days 0–3; late clinical failure (LCF): danger signs or complicated malaria or fever/history of fever in the previous 24 hours and parasitaemia on days 4–28 without previously meeting criteria for ETF; late parasitological failure (LPF): asymptomatic parasitaemia day 4–28 without previously meeting criteria for ETF or LCF; and adequate clinical and parasitological response (ACPR): absence of parasitaemia on day 28 without previously meeting criteria for ETF, LCF, or LPF 17. All patients classified as treatment failures received rescue treatment with quinine (10 mg/kg three times daily for 7 days).

The primary effectiveness outcomes were the 28-day risks of recurrent symptomatic malaria (ETF and LCF) and recurrent parasitaemia (ETF, LCF and LPF), unadjusted and adjusted by genotyping.

Recrudescent infections were distinguished from reinfections by nested-PCR of the variable block 2 region of the merozoite surface proteine-1 gene (msp-1, belonging to the three allele families K1, MAD20 and RO33) and of the polymorphic block 3 region of the merozoite surface proteine-2 gene (msp-2, 3D7 and FC27 allelic families), after extracting DNA from filter papers of paired samples (samples collected on day 0 and the day of failure) following the manufacturer's boiling protocol. Subsequently, amplicon lengths were analysed by capillary electrophoresis using a DNA analyzer (ABI Prism^® 3100 Genetic Analyzer, Applied Biosystems, CA, USA) with GeneMapper™ ID software (BioType, Dresden, Germany) as described elsewhere 14. A recurrent parasitaemia was categorized as reinfection if none of the msp1/msp2 amplicons in polyclonal infections had identical lengths. Accordingly, recrudescence was defined as the presence of at least one msp1/msp2 amplicon of identical length.

Secondary outcome measures were the change in the mean haemoglobin concentration from day 0 to day 28, the comparative tolerability, and the acceptance of the treatment regimen. The comparative tolerability was assessed by the risk of occurrence of an adverse event (AE), classified by means of the Division of Microbiology and Infectious Diseases pediatric toxicity tables 18 as mild (grade 1), moderate (grade 2), severe (grade 3), or serious (grade 4, SAE). A serious adverse event was defined as any AE resulting in death or in persistent or significant disability/incapacity, being life-threatening, requiring hospitalization or significant medical intervention to prevent serious outcome. For each AE, causality was assessed as recommended by the WHO 19.

Acceptance levels were assessed by asking the parents/guardians three standardized questions about the study treatment (after completion of the treatment course), i.e. a) how they appraised the treatment schedule, b) how they liked the packaging of the drug, and c) how they found the administration of the drug. Answers were scored with 1 point (difficult/bad), 2 (fairly difficult/fair), 3 (easy/good), or 4 (very easy/very good) points, and were compared using rank sum tests.

The sample size was calculated to get at least 90% power to detect a 10% difference of the risk of treatment failure (defined as the number of patients with clinical cure and clearance of asexual P. falciparum parasitaemia without subsequent PCR-confirmed recrudescence by day 28) between the two study arms at a 5% significance level. A cure rate of more than 95% in one of the two treatment arms was anticipated. Expecting a dropout rate of not more than 10%, the targeted number of participants in each study arm was 228.

Data generated during the trial were recorded in the Case Report Form (CRF) of the respective patient and timely double-entered into two separate databases (4^th Dimension^®, 2004 version 8.0.4, 4D, Neufahrn, Germany) with variable ranges and internal consistency controls before files were locked. Final analyses were done with Stata/IC™ version 10.0 (StataCorp, College Station, TX, USA).

Per-protocol analysis included all patients who matched all inclusion criteria, were properly randomized, had received the study drugs according to the protocol, and for whom data were available on the study end-point without protocol violation during the follow-up time. Intention-to-treat analysis comprised all randomized patients fulfilling the inclusion criteria without repeated vomiting after the first study drug administration. Characteristics of both study arms were compared by contingency (χ²) and nonparametric (Wilcoxon) tests, and for evaluation of the acceptance questionnaires Wilcoxon rank-sum tests were performed with scores allocated to the potential answers. Cox regression was used to calculate ACPR, LCF and occurrence of adverse events. P values < 0.05 were considered significant.

All clinical investigations were conducted according to the principles of the Helsinki Declaration. The protocol was approved by the Committee on Human Research Publications and Ethics of the School of Medical Sciences of the Kwame Nkrumah University of Science and Technology, Kumasi, Ghana, and the Ethics Committee of the Medical Association Hamburg, Germany. In addition to the immediate notification of SAEs, all AEs were reported on a monthly basis to the data and safety monitoring board (DSMB). An experienced clinical trial monitor audited the trial that was registered at http://www.ClinicalTrials.gov (NCT00374205).

The study started in October 2006 and was stopped in September 2007. At that time, a total of 776 children were screened and 246 were randomized (Figure 1). Adjusted and unadjusted treatment outcome was available for 217 children in the intention-to-treat (ITT) and for 199 children in the per-protocol (PP) analysis. There was no significant difference in baseline characteristics of the two treatment groups (Table 1). After initial randomization 9 children were excluded due to repeated vomiting of the study drugs (they subsequently received alternative therapy). According to the study protocol, another 20 individuals had to be excluded from analysis because of the detection of >200,000 asexual parasites/μl in the Giemsa-stained thick film, or of non-falciparum parasites, or of leukocyte counts >15,000/μl (Figure 1).

No ETF was observed. Adequate clinical and parasitological responses were similarly high in both treatment groups (Table 2). The per-protocol Kaplan-Meier survival curves are displayed in Figure 2. However, LCF rates by day 28 were significantly higher in AL-treated children than in ASAQ-treated children in both, the per-protocol analysis (17.5% and 7.3%, respectively; Hazard Ratio 2.41, 95% CI 1.00–5.79, p < 0.05; Figure 3) and in the intention-to-treat analysis (16.5% and 6.5%, respectively; HR 2.44, 95% CI 1.02–5.85, p < 0.05). A trend towards higher LCF rates was also seen after exclusion of reinfections (Table 2).