The study was conducted in two primary health facilities, Korle Bu Polyclinic (KBP) and Mamprobi Polyclinic (MP). KBP is attached to a tertiary referral hospital, Korle Bu Teaching Hospital (KBTH). The two facilities are situated in Ablekuma sub-district in Accra, Ghana. Both facilities serve a mixed, predominantly low socio-economic urban population. The study was conducted between October 2004 and December 2006.

Children presenting with a history of fever to the facilities were referred to the study team. A project physician examined the child to exclude concomitant illnesses. Subjects were recruited if a blood film was positive for Plasmodium falciparum, and if the study criteria were fulfilled.

Inclusion criteria for the study were, age six months to 14 years; signs and symptoms of acute uncomplicated malaria, including axillary temperature ≥37.5°C; confirmed P. falciparum mono-infection with parasite density between 2,000 – 200,000/μL; and willingness to comply with the follow-up schedule. Exclusion criteria were, symptoms or signs of severe malaria; obvious clinical evidence of chronic malnutrition or other severe disease; known intolerance or allergy to study medications; and reported treatment with any of the drugs under study one month preceding enrolment. Children aged five to 14 years were included to permit objective neurologic examination and audiologic assessment. Schoolchildren aged five to 14 years from two local primary schools were enrolled as controls for the audiometric study described below. Written informed consent was obtained from parents or guardians of all subjects. Ethical approval for the study was granted by the Ethical and Protocol Review Committee of the University of Ghana Medical School.

A computer-generated simple randomization scheme was prepared in advance. Allocated treatments were kept in sealed opaque envelopes. After completion of formal enrolment procedures, allocated treatments were administered by project nurses. All study personnel (except project nurses), were unaware of the assigned treatments.

AQ (Camoquine^®, Pfizer, Dakar, Senegal), 10 mg/kg body weight, single daily dose for three days, was administered with AS (Plasmotrim^®, Mepha Ltd, Aesch-Basel, Switzerland), 4 mg/kg body weight, single daily dose, for three days. AM-L (Coartem^®, Novartis Pharma AG, Basel, Switzerland; 20 mg artemether and 120 mg lumefantrine) was administered at 0 and 8 hours on the first day, and then twice daily for two subsequent days according to body weight: 5–14 kg, one tablet/dose; 15–24 kg, two tablets/dose; 25–34 kg, three tablets/dose; 35 kg and over, four tablets/dose.

Project nurses administered all treatments (except the evening dose of Coartem) in the clinic. Parents administered the evening dose of Coartem^® at home 2728. Children were observed for 30 minutes after each drug administration. Treatments were re-administered if the child vomited. Children who vomited the re-ad ministered dose were withdrawn. An extra Coartem^® dose was given to parents to re-administer at home if the child should vomit.

Subjects were followed-up on days 1, 2, 3, 7, 14 and 28 for clinical and laboratory assessments described below. Children who missed their day 14 or day 28 follow-up appointments were seen on days 15 or 29. After 28 days, children were visited at home every month for up to one year. During the monthly follow-up visits, parents or caregivers were asked about the child's well-being, behavioural and developmental concerns, as well as open questions about possible side event occurrence since the last visit. A study physician examined children during the follow-up visits. Follow-up was terminated if parents relocated. Each time a subject presented with an acute febrile illness during follow-up, thick films for malaria parasitaemia and a thorough examination was done, before and after the episode. All episodes of uncomplicated malaria diagnosed more than 28 days after treatment were treated with the same pre-assigned regimen as at randomization.

The primary end-point was the unadjusted and PCR-adjusted cure rates at days 14 and 28. Efficacy outcomes were classified as, adequate clinical and parasitological response (ACPR), early treatment failure (ETF), late clinical failure (LCF) and late parasitological failure (LPF) 32.

Secondary endpoints were proportion of subjects with fever and parasitaemia 24 and 48 hours after treatment, the total number of subsequent malaria attacks in each arm, and incidence of adverse events before and after 28 days.

All adverse events were graded by severity (mild, moderate, severe, life-threatening), and relationship to study medication (unrelated, unlikely, possible, probable, definitely related) 33.

The sample size was estimated on the assumption of a negligible (3%) risk of treatment failure with AM-L (because AM-L was new in the study area), and a 15% risk of AS+AQ treatment failure (because of prior extensive use of AQ monotherapy in the area). A sample size of 103 subjects was required to detect a 12% difference in cure rates between AS+AQ and AM-L with 95% confidence and 80% power. Categorical variables were compared with the chi square test with Yates correction or the Fisher's exact test. Continuous variables were compared with the student's t test or one-way analysis of variance. A logistic modelling (GENMOD™) procedure was used to compare the total number of treatments in the two arms for the year of follow-up. Risk differences and 95% confidence intervals were calculated with CIA™. Other analyses were done with Sigma stat (version 3.1; Systat software Inc., Richmond, California, USA) and SAS (SAS Institute, Cary NC, USA).

The primary efficacy outcome was analysed by a per-protocol (PP), and an intention-to-treat (ITT) method. In the ITT analysis, subjects lost to follow-up were included; these, as well as those withdrawn because of non-tolerance of oral medication, were considered treatment failures. In the PP analysis, only subjects with an assigned outcome were included. The follow-up time of subjects who did not complete one year of follow-up were censored at their last follow-up date. P values < 0.05 were considered significant.

A total of 227 subjects were randomized out of 846 subjects screened. Sixteen subjects who were correctly randomized did not complete 28 days follow-up. The reasons for non-completion of follow-up in the AM-L arm were, withdrawal due to persistent vomiting necessitating parenteral treatment (n = 2); culture positive co-infection with Salmonella typhi resulting in worsening clinical condition after parasite clearance (n = 1); and loss to follow-up (n = 5). The reasons for non-completion of follow-up in the AS+AQ arm were, withdrawals due to persistent vomiting (n = 2); withdrawal due to worsening clinical condition despite rapid parasite clearance in an infant who was subsequently admitted and managed clinically for suspected septicaemia (n = 1); and loss to follow-up (n = 6). The flow of subjects through the study (up to day 28) is outlined (Figure 1). There were no differences in demographic and clinical parameters between the two groups (Table 1). The proportion of subjects between the ages of five and 14 years was 50.4% (56/111) in the AM-L arm and 55.2% (64/116) in the AS+AQ arm (p = 0.56).

After one year of follow-up, there were 34 further episodes of symptomatic uncomplicated malaria in 93 subjects (incidence rate, 0.37 per year) in the AS+AQ group, and 28 episodes in 82 subjects (incidence rate, 0.34 per year) in the AM-L group (incidence ratio, 0.93). Similar results were obtained, when logistic modelling (p = 0.94), or survival analysis (p = 0.76) was used to compare incidence in the two arms. All subjects who received repeat treatment courses responded appropriately to treatment. The time to a repeat treatment episode (median; interquartile range) was longer in the AS+AQ (91 days; 53–193 days) arm than the AM-L (64 days; 33–88 days) arm. There were three sibling pairs among those who were treated for repeat infections. These siblings (each randomized to receive a different ACT) reported simultaneously with a repeat attack of uncomplicated malaria.

The "CVIET" (mutant) haplotype was present in 72.1% (75/104) of pre-treatment samples in subjects treated with AS+AQ. The effect of the "CVIET" haplotype prevalence on treatment outcome could not be evaluated because of the low number of subjects with treatment failure.

Neurological examination could be done for 168 children (AS+AQ, 92; AM-L, 76). Nystagmus was observed in two subjects (AM-L, n = 1, AS+AQ, n = 1) between days one and seven. A detailed history indicated that one of these two subjects (7-year-old male; AS+AQ treated) was admitted as a neonate into neonatal intensive care (NICU) because of probable birth asphyxia. This subject was still attendant at pre-school at seven years of age (which is unusual in Ghana), and demonstrated excessive emotional lability. An electroencephalography (EEG) done for this subject because of a history suggesting the possibility of a seizure disorder, reported the presence of "epileptiform loci and generalized cerebral dysfunction". The other subject with nystagmus (9-year-old male, AM-L treated), also had a history of NICU admission for macrosomia (birth weight 4.5 kg) and possible neonatal hypoglycaemia. This subject had poor academic performance (had repeated first and second grades in school) and possible cognitive impairment. A routine EEG done for this subject was normal. A positive Romberg's test (which may suggest sensory ataxia) was noticed on day 3 in an 11-year old female (AM-L). This child was not attending school (despite her age) and had a history suggestive of possible cognitive impairment. No other abnormal neurological findings were observed in the remaining children during the 28-days follow-up, monthly follow-up visits, or for subjects who received multiple treatments.

Serum alanine aminotransferase, aspartate aminotransferase and total bilirubin were initially (day 0) elevated, but there were no differences between the two treatment groups. The liver enzymes did not increase in response to treatment, but showed a trend towards normalization (reduction) as clinical symptoms improved (data not shown).

Within the first three days of treatment, complaints of dizziness were reported by 14 (12%) and eight (7.2%) subjects in the AS+AQ and AM-L groups, respectively (p = 0.31). Complaints of fatigue and excessive sleepiness were reported by five (4.3%) and four (3.6%) subjects in the AS+AQ and AM-L groups, respectively (p = 1.0). Pruritus was reported by three (AS+AQ, n = 2; AM-L, n = 1) subjects. A history of vomiting and nausea was part of the presenting complaint by 44.8% (52/116) and 40.5% (45/111) subjects respectively in the AS+AQ and AM-L groups on admission (p = 0.74). Between the first three days of treatment, 27.5% (32/116) and 12.6% (14/111) subjects respectively complained of vomiting and or nausea in the AS+AQ and AM-L groups (p = 0.01). Severe anaemia (haemoglobin 4.8 g/dl) occurred on day 14 in a 1.5 year old subject (AS+AQ). Severe anaemia and dizziness were considered unlikely to be drug related, but pruritus and excessive sleepiness were considered possibly drug-related. Other reported adverse events were mild in intensity, and overlapped with known malarial symptomatology. These were mostly classified as unrelated to study medications.