The protocol for this trial and supporting CONSORT checklist are available as supporting information; see Checklist S1
and Protocol S1
The study was based at Chileka health centre near Blantyre, Malawi, where malaria transmission is perennial, peaking during December to April. Between September 2003 and December 2005, children presenting with an illness suggesting falciparum malaria were screened. Inclusion criteria were: i) age ≥12 and <60 months, ii) weight ≥6 kg, iii) axillary temperature ≥37.5°C, iv) no history of treatment with an antimalarial, cotrimoxazole or a tetracycline antibiotic in the previous week, v) no features suggesting severe malaria or a concomitant illness, vi) haemoglobin ≥5.0 g/dl using Hemocue®, and vii) P. falciparum monoinfection with a parasite density between 2000 and 200,000 parasites per µl. Written informed consent was required from the parent of each child recruited. The study protocol was approved by ethics committees of the College of Medicine, University of Malawi and Liverpool School of Tropical Medicine. A data and safety monitoring board and local study monitor were appointed.
Treatment, Randomisation and Blinding
Children meeting all inclusion criteria on day 0 were recruited and randomised to one of four treatment groups. Randomisation was in blocks of 12 according to an off-site computer-generated code to assign patients equally to the four oral treatment groups: SP (25 mg/kg sulfadoxine and 1.25 mg/kg pyrimethamine as a single dose on day 0)+vitamin C 50 mg tablet (placebo) daily for 3 days; CQ (10 mg/kg on days 0 and 1, and 5 mg/kg on day 2)+SP; ART (4 mg/kg once daily for 3 days)+SP; or AQ (10 mg/kg daily for 3 days)+SP. In the case of children too young to swallow tablets, CQ syrup (50 mg per 5mls) and AQ syrup (50 mg per 5mls) were used (same doses as above). The other study drugs were not available as syrups and were crushed and given on a spoon with water if the child could not swallow a tablet. The different tablets were not identical in appearance or taste. A three-day supply of paracetamol (10 mg/kg) was given.
Each child was given a unique study number, assigned sequentially. A dedicated study ‘drug dispenser’ opened the corresponding randomisation envelope and directly observed all drug doses but was not involved in the assessment of children. All other members of the study team were blinded to the dispensing process and patients were uninformed of their treatment allocation for the duration of the study. Children were observed for 30 minutes after dosing. If the child vomited, a second dose was given. If vomiting occurred a second time, the child was withdrawn and treated with parenteral quinine.
Classification of Outcomes
Patients were assessed on days 0, 1, 2, 3, 7, 14, 28 and 42 and any other day if unwell. Blood was collected for parasite microscopy, storage on Whatman 3M filter paper and, at specified visits, for determination of the full blood count and biochemical parameters. Clinical outcome was assessed using the 2003 WHO therapeutic efficacy protocol for areas of intense malaria transmission 
. Participants were withdrawn if they failed to attend for follow up, withdrew consent or took a ‘banned’ drug i.e. all antimalarials, cotrimoxazole, doxycycline, tetracycline, chlorpheniramine and folic acid. Late clinical failures were treated with oral mefloquine (25 mg/kg). Severe malaria was treated with parenteral quinine in hospital.
The planned sample size of 100 evaluable patients per treatment arm was calculated to have 90% power to detect the difference between an “adequate clinical and parasitological response” (ACPR) rate of 80% with SP alone and 95% with combination therapies using the 5% significance level for each comparison with SP alone. The primary endpoint was the day 28 ACPR rate and the major analysis strategy for the primary endpoint was intention to treat (ITT). Patients with missing outcomes were all classified as successes in one analysis and then as failures in a separate analysis. Per protocol (PP) analysis was also done using polymerase chain reaction (PCR) corrected data to distinguish recrudescences from reinfections. When PCR analysis indicated that a post-treatment parasitaemia was a reinfection, the outcome was classified as a treatment success on that day, but excluded from subsequent analyses. If PCR was inconclusive, the case was excluded from the analysis.
Secondary endpoints included day 14 and 42 ACPR rates, time to fever resolution (axillary temperature≤37.5°C), time to parasite clearance, change in haemoglobin from day 0 to day 14 and the appearance of gametocytes by day 28 after treatment. We also compared adverse events (AEs) between the treatment groups, including self-reported AEs and laboratory AEs; rises in alanine transferase (ALT), total bilirubin, and creatinine between days 0 and 14.
Data were double entered and validated prior to the analyses. Data analysis was performed using Stata 8. Binomial regression was used to obtain risk differences between treatments and 95% confidence intervals. Fisher's exact p-values were reported. Tests of significance were performed using the 5% level to infer significance for the planned analyses. Pair wise comparisons between combination therapies were not planned and in these comparisons we adjusted the significance level to 1.7% (i.e. p<0.017) using Bonferroni's approach.
To look for evidence of selection of resistance mutations in the genes dhfr, dhps, pfcrt and pfmdr1, we compared the prevalence of resistance mutations in these genes before and after treatment, both within each of the four different treatment groups and between the four treatment groups.
Blood films were stained with Fields stain and parasite densities estimated from thick films by counting the number of parasites per 200 white blood cells (WBC) assuming a total count of 8000/µl. These parasite counts and haemoglobin (Hb) estimates using Hemocue® were used for screening purposes. In addition, on days 0 and 14, the full blood count was measured using a Beckman Coulter HMX and plasma ALT, total bilirubin and creatinine using a Vitros DTII dry biochemistry analyser. The actual WBC count from the coulter was subsequently used to calculate an accurate parasite count for the analyses. The presence or absence of gametocytes was noted on each blood film.
Parasite DNA was extracted from dried blood on filter paper. A nested PCR was used to distinguish recrudescent infections from new infections in all patients with parasitaemias appearing from day 12 onwards. The msp2
gene was amplified and size polymorphisms identified by gel electrophoresis using previously described methods 
. Parasites were classified as recrudescent if they shared any of the bands that were present on day 0 and as reinfections if they had no bands in common. Parasites that appeared before day 12, having initially disappeared, were assumed to be recrudescent. Nested PCR followed by mutation-specific restriction enzyme digestion was used to determine the prevalence of different alleles in the dhfr
genes in day 0 parasites and parasites appearing at any time from day 12 onwards after treatment. A detailed description of these techniques is available at http://www.medschool.umaryland.edu/cvd/plowe.asp
. We analysed the dhfr
gene for polymorphisms at codons 51, 59, 108 and 164; dhps
codons 437, 540 and 581; pfcrt
codon 76; and pfmdr1
codons 86 and 1246. Samples were analysed in a blinded fashion with respect to the treatment groups.