Nine RCTs were included in the review (n=1,480 children; range 37 to 576). It appeared that seven trials used artemether and two trials used arteether. Three trials reported adequate blinding. Seven trials reported adequate randomisation. Five trials reported adequate allocation concealment. Six trials reported use of intention-to-treat analysis.
Death: No statistically significant differences were found between the artemisinin derivative and quinine groups (RR 0.91, 95% CI 0.73 to 1.14; nine RCTs; I2=0%)
Neurological sequelae at recovery or discharge: No statistically significant differences were found in the risk of neurological sequelae between the artemisinin derivative and quinine groups (RR 0.94, 95% CI 0.75 to 1.17; six RCTs; I2=0%)
Coma recovery time and fever clearance: Overall there was no statistically significant difference between groups for coma recovery time (WMD -2.53 hours, 95% CI -5.58 to 0.52; RCTs; I2=55.3%) or fever clearance time (WMD -2.34 hours, 95% CI -6.97 to 1.28; six RCTs; I2=78.3%). Artemether showed a significantly shorter coma recovery time (WMD -3.50 hours, 95% CI -6.71 to -0.29; four RCTs, I2=54.4%) compared with quinine, while arteether did not.
Sensitivity analyses: The reviewers reported a number of sensitivity analyses; their results were reported as consistent with the main results.
Publication bias: The reviewers reported that the funnel plot was asymmetric, but that this could not be taken to indicate publication bias given the small number of included trials.