Pre-referral rectal artesunate in severe malaria: flawed trial

BACKGROUND: Immediate injectable treatment is essential for severe malaria. Otherwise, the afflicted risk lifelong impairment or death. In rural areas of Africa and Asia, appropriate care is often miles away. In 2009, Melba Gomes and her colleagues published the findings of a randomized, placebo-controlled trial of rectal artesunate for suspected severe malaria in such remote areas. Enrolling nearly 18,000 cases, the aim was to evaluate whether, as patients were in transit to a health facility, a pre-referral artesunate suppository blocked disease progression sufficiently to reduce these risks. The affirmative findings of this, the only trial on the issue thus far, have led the WHO to endorse rectal artesunate as a pre-referral treatment for severe malaria. In the light of its public health importance and because its scientific quality has not been assessed for a systematic review, our paper provides a detailed evaluation of the design, conduct, analysis, reporting, and practical features of this trial. RESULTS: We performed a checklist-based and an in-depth evaluation of the trial. The evaluation criteria were based on the CONSORT statement for reporting clinical trials, the clinical trial methodology literature, and practice in malaria research. Our main findings are: The inclusion and exclusion criteria and the sample size justification are not stated. Many clearly ineligible subjects were enrolled. The training of the recruiters does not appear to have been satisfactory. There was excessive between center heterogeneity in design and conduct. Outcome evaluation schedule was not defined, and in practice, became too wide. Large gaps in the collection of key data were evident. Primary endpoints were inconsistently utilized and reported; an overall analysis of the outcomes was not done; analyses of time to event data had major flaws; the stated intent-to-treat analysis excluded a third of the randomized subjects; the design-indicated stratified or multi-variate analysis was not done; many improper subgroups were analyzed in a post-hoc fashion; the analysis and reporting metric was deficient. There are concerns relating to patient welfare at some centers. Exclusion of many cases from data analysis compromised external validity. A bias-controlled reanalysis of available data does not lend support to the conclusions drawn by the authors. CONCLUSIONS: This trial has numerous serious deficiencies in design, implementation, and methods of data analysis. Interpretation and manner of reporting are wanting, and the applicability of the findings is unclear. The trial conduct could have been improved to better protect patient welfare. The totality of these problems make it a flawed study whose conclusions remain subject to appreciable doubt.