Statistics from Altmetric.com
It is by no means a new problem but does it have an answer? You are faced with a child who is desperately ill and may well die. You have a new potential treatment that has not been submitted to a controlled trial but preliminary reports suggest it might appreciably reduce the risk of death or morbidity. Do you go ahead with random assignment to new treatment or placebo, or do you do what you almost certainly would do if it were your own child and give the treatment? But you don’t know whether the treatment could in fact be harmful; of course you don’t, but in those circumstances you would probably be prepared to take the risk. So how do you prove new treatments and how do you get truly informed consent for randomised trials?
Two recent reports about treatment for severe meningococcal septicaemia illustrate the problem. The first (Giroir and colleagues; Lancet 1997;350:1439-43) concerned the use of rBPI21, a recombinant, N-terminal fragment of human bactericidal/permeability increasing protein, which binds endotoxin and kills meningococci. Twenty six patients were treated and one died. Between four and eight deaths might have been expected. Nevertheless it has been suggested that the treatment was given too late for it to be reasonable to suppose that rBPI21 had an appreciable effect on endotoxin or cytokine levels (commentary, Lancet 1997;350:1565-6). A multinational randomised trial has been set up. The second report, from Dublin, (Smith and colleagues; Lancet1997;350:1590-3) details the results of giving a mixture of three new treatments. Twelve patients were treated for meningococcal septicaemia with purpura fulminans and severe acquired protein C deficiency. All 12 were given intravenous protein C, 11 were given heparin, and nine had haemodiafiltration. No patient died although the predicted mortality was 57-80%. The authors call for a randomised trial.
I don’t know the answers. How do you obtain truly informed consent? Is it ethical not to tell parents about the results of the preliminary trials or do you simply say, no trial, no treatment? In a condition with high mortality would it be wrong to continue with a trial using historical controls? Supposing, for instance, the Irish trial were to continue uncontrolled and after 25, or 50, or 100 patients there were still no deaths and morbidity remained low, wouldn’t that be convincing, and wouldn’t it have avoided inevitable deaths in a controlled trial? But if the results became less clear cut wouldn’t that then provide an even stronger indication for a randomised trial? Perhaps somebody out there will explain.