Rapid Responses to:

PAPERS: Hui Wen Fan, Luiz F Marcopito, João Luiz C Cardoso, Francisco O S França, Ceila M S Malaque, Ronnei A Ferrari, Robert David G Theakston, and David A Warrell Sequential randomised and double blind trial of promethazine prophylaxis against early anaphylactic reactions to antivenom for bothrops snake bites BMJ 1999; 318: 1451-1452 [Abstract] [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

The dangers of stopping a trial too early

Douglas E Ball   (9 June 1999)

The dangers of stopping a trial too early 9 June 1999
Douglas E Ball Send response to journal: Re: The dangers of stopping a trial too early	Editor The article by Fan et al1 is the second in as many months in the BMJ which investigates prophylaxis against adverse effects of snake antivenom. Earlier Premawardhena and coworkers in Sri Lanka concluded that low dose subcutaneous adrenaline prior to administering antivenom could prevent acute adverse reactions2. The Brazilian study1 showed no benefit in using promethazine as pre-treatment. However, perhaps more interesting than the conclusions is that both these clinical trials were stopped early due to interim analyses and provide examples of their use and pitfalls. The Sri Lankan group stopped recruitment halfway through their study since the analysis showed a significant reduction in all acute adverse effects when adrenaline was used as pre-treatment compared to placebo (p=0.0002). Unfortunately, in subanalysis of the severity of adverse reactions in the two groups, the differences in incidence of mild, moderate and severe reactions barely reached statistical significance due to the low power of the analysis (about 50%) and the small numbers. This is important since it is mainly the moderate and severe reactions which we wish to prevent. As it was, the trial was stopped due to a supposed benefit, but we cannot advise the use of subcutaneous adrenaline due to unconvincing evidence. In the Brazilian study, the interim analysis was in the form of sequential analysis. The use of this method, and to some degree the negative outcome of the study, prevents arguments of the nature used above. Interim analyses are often required on ethical grounds but they should be planned bearing in mind final data analysis and clinical significance. Too often emphasis is put on the ethics of continuing a trial where some patients may receive an apparently inferior treatment. However, stopping the trial may result in the evidence for the supposedly better therapy being insufficient to recommend policy changes - this too is unethical. Using and stating defined methodologies as by Fan and colleagues can help, but we must always bear in mind that statistics are there as an aid to reason, not as a replacement for it. The dangers of stopping a trial too early have been addressed recently by Pocock and White3 and they need to be fully appreciated by clinical researchers to prevent good research being condemned to the corridors of mediocrity. Douglas E Ball Drug Information Pharmacist and lecturer Drug and Toxicology Information Service Dept of Pharmacy University of Zimbabwe PO Box A178 Avondale Harare, Zimbabwe Klara Tisocki Lecturer Dept of Clinical Pharmacology University of Zimbabwe PO Box A178 Avondale Harare, Zimbabwe References 1 Fan HW, Marcopito LF, Cardoso JLC, França FOS, Malaque CMS, Ferrari RA, Theakston RDG, Warrell DA. Sequential randomised and double blind trial of promethazine prophylaxis against early anaphylactic reactions to antivenom for bothrops snake bites. BMJ 1999;318:1451-1452 2 Premawardhena AP, de Silva CE, Fonseka MMD, Gunatilake SB, de Silva HJ. Low dose subcutaneous adrenaline to prevent acute adverse reations to antivenom serum in people bitten by snakes: randomised, placebo controlled trial. BMJ 1999;318:1041-1043 3 Pocock S, White I. Trials stopped early: too good to be true? Lancet 1999;353: 943-944.