Vitamin K is given to many babies born in the United Kingdom, but we still do not know if it has substantial hazards. Because the population exposed to vitamin K is very large even quite small hazards would involve many adverse events. It is therefore important to be able to put reasonably close bounds on the potential damage that vitamin K prophylaxis could cause. Past research has not allowed us to do this but a large randomised controlled clinical trial of vitamin K against no vitamin K, enrolling only infants at low risk of haemorrhagic disease, would do so. There is no question that vitamin K is a useful treatment in babies at highest risk of haemorrhagic disease: the question is whether the trend towards use of vitamin K in lower risk babies should be encouraged.

Vitamin K has been widely used for over 30 years as a prophylaxis against haemorrhagic disease of the newborn, but the first case-control study of its long term safety (in terms of the risk of cancer) was not published until August 1992. This suggested that vitamin K given intramuscularly to neonates might double the risk of childhood cancer.1 More recent Swedish and American studies have not shown similar effects.^2,3 Of course, all effective treatments carry some risk, and to assess whether widespread use of a particular treatment is warranted we must know both the risk of not treating and the risk of treating. Furthermore, if we can identify groups of neonates with more or less to gain from treatment we must know the risks within each such group. Randomised clinical trials are by far the most reliable method of assessing the balance of risks and benefits of treatments but a major difficulty is finding a form of clinical trial that is informative, …