Vitamin K is a fat soluble vitamin, and the development of a suitable commercial formulation for clinical use has long presented something of a challenge. Indeed, when it was first shown in 1939 that treatment with vitamin K could abolish symptomatic prothrombin deficiency in the first week of life,^1–,³ babies were generally given menadione, a water soluble analogue. A paper in the Lancet in 1944 generated widespread interest.⁴ It showed a five fold reduction in death from haemorrhage 2–8 days after birth once all babies were given 1 mg of oral menadione at delivery in Göteborg, Sweden, in 1940. A similar policy was soon widely adopted elsewhere even though many were unable to replicate these findings.^5,6 The argument, as Ethel Dunham put it in 1948, was that “since the vitamin does no harm and may do good, it is probably best to give it to all premature infants immediately after birth”.⁷

When this did not stop some babies from developing a bleeding tendency, or dying with an intraventricular haemorrhage, physicians started using larger and larger doses. Prothrombin levels are always relatively low at birth by adult standards, and remain so for some time, and it was (wrongly) thought that the low level seen at birth, and not just the further postdelivery drop, must be due to relative fetal vitamin K deficiency. By 1953 came a first report that high dose use could cause haemolytic anaemia,⁸ and by 1956 it had been established that this could, in turn, cause severe jaundice and even death from kernicterus after unbound bilirubin entered the brain.^9,10 The dose administered was cut back after that but, within five years, the water soluble product (Synkavit^™) was starting to be replaced by the natural, fat soluble, plant form of vitamin K (phylloquinone; vitamin K₁). This product did not seem to cause haemolysis,¹¹ and it is this product that has dominated the market in Europe and North America ever since, although menaquinone-4 (a member of the K₂ series) remains the main product still used in Japan.

Routine prophylaxis soon became the norm for every baby (not just every preterm baby) in some countries.¹² Intramuscular prophylaxis also became the route universally adopted, mainly because manufacturers never got round to licensing a product for oral use. It also became routine to give a 1 mg injection, even though this was a thousand times more than the dose of menadione needed each day,¹³ and 10 times the dose used in the only controlled trial of clinical efficacy ever conducted.¹⁴ Reluctance to advocate universal use persisted however, especially in the UK where most of the cases of kernicterus due to excessive dosage had been reported.¹⁰ There was also growing uncertainty as to just how common the condition really was. As a result it became increasingly common to only treat babies considered “at risk”—mostly preterm babies and babies having an operative delivery (a policy that gained increased credence and spread even more widely as a result of an influential editorial in the Lancet in 1978).¹⁵

Spontaneous bleeding, in the absence of trauma, was seldom seen during the next decade, either in units that opted for universal or for selective prophylaxis, and those cases that did occur were quickly spotted and controlled. Most presented with dark melaena stools, bloody vomit, nose bleed, blood stained urine, or bleeding from the umbilical stump, two to six days after birth. Circumcision before seven days brought many cases to light in cultures where this ritual remained common.¹⁴ Presentation was much the same in countries where prophylaxis is not generally available,^16,17 although here the babies risked death from blood loss.¹⁶ Then another paper appeared in the Lancet describing a resurgence of the condition in the UK not just in the first week of life but also in older babies.¹⁸ These were healthy children with a generalised bleeding tendency that responded promptly to treatment with vitamin K but where, typically, no abnormality had been suspected until there was a catastrophic intracerebral bleed two to 10 weeks after birth. It soon became clear that such problems were only being seen in breast fed babies who had never had even a few “complementary” feeds of bottle milk, and who had never received intramuscular vitamin K. Quite rapidly the previous policy of selective prophylaxis gave way to a policy of universal prophylaxis—the same policy as that long advocated in north America—although a growing number of units opted for oral rather than intramuscular prophylaxis for those babies who were well enough to be fed at birth.¹⁹ A survey showed that policy in the UK varied widely from unit to unit,²⁰ and a more recent retrospective survey has suggested that midwifery staff were often quite uncertain as to what unit policy actually was.²¹

Then, just as people were beginning to accept that universal prophylaxis was justified, not so much to manage the occasional case of early bleeding, but to prevent late bleeding in the breast fed baby, a paper was published from Bristol suggesting that babies given intramuscular vitamin K might be more likely to develop cancer (and particularly leukaemia) later in childhood.²² What made life even more difficult was the fact that the news broke in the Daily Mail almost two months before the definitive paper appeared in the BMJ in August 1992.

Different countries responded to the news in different ways. Several countries in Europe moved increasingly towards a uniform policy of oral prophylaxis.²³ The American Academy of Pediatrics reiterated its confidence in universal intramuscular prophylaxis,²⁴ and a small case control study from that country soon appeared suggesting that the risk, if real, was certainly not as big as the Bristol study had originally suggested.^25,26 Australia and New Zealand changed to universal oral prophylaxis and then reverted to intramuscular prophylaxis when cases of late vitamin K deficiency bleeding started to reappear.²⁷ The UK just got itself into a muddle. The Department of Health drew back from issuing any specific guidance,²⁸ although it had not shown any such reluctance some years earlier.²⁹ Indeed decades earlier it taken a much firmer line and made its own low cost multivitamin product available when presented with evidence that many young children were developing vitamin D deficiency rickets.³⁰ Fat soluble vitamin K could have been added to this product, but it was not. Their one pivotal pronouncement, after six years, was that parents should be allowed a choice between the oral and intramuscular options.³¹ The British Paediatric Association (the predecessor of the present College) recommended regular low dose oral prophylaxis³² but, since no licensed oral product existed in the UK, this advice was not widely followed. The Health Visitors Association, on legal advice, advised its members to avoid the issue altogether.³³ Midwives, who could give licensed products on their own authority,³⁴ were given the job of counselling parents, but not told what to say. Policy varied widely from unit to unit,²⁰ and cases of late bleeding continued to be reported quite regularly.³⁵ Other countries were able to assess the impact of policy on late bleeding,²³ but even this was difficult in the UK because policy varied so widely. Attempts to persuade Roche to apply for a UK licence for the oral product they already had a license to sell in the rest of Europe came to nothing. One small pharmaceutical company did start making something similar on request (Orakay^™) that the parents of breast fed babies could give their child at regular intervals after discharge,³⁶ but is has taken 10 years to assemble enough evidence of efficacy for the Medicines Commission to recommend the granting of a product licence.

Slowly, however, the situation started to become clearer, especially when the old term “haemorrhagic disease of the newborn” was replaced by “vitamin K deficiency bleeding” (or VKDB)—both because much bleeding in the newborn was not caused by vitamin K deficiency, and also because deficiency bleeding is not only seen in the first week of life.³⁷ Recent studies, using a standardised definition,³⁸ seem to show that the condition is not now very common, even in communities where prophylaxis is not yet available. It is certainly not nearly as common as some authoritative reports claim.²⁴ Two studies in Japan before the introduction of routine prophylaxis had suggested that one in every 6000 breast fed babies might sustain a late bleed when more than two weeks old.³⁹ The true risk of bleeding in the first week of life (the “classic” presentation) remains less clearly defined. It was nearly twice as common as late bleeding in breast fed babies offered no prophylaxis in the UK in 1988–89,³⁵ and three times as common in a recent Malaysian study.¹⁷ It has also become clear that oral prophylaxis can be as reliable as intramuscular prophylaxis (for further evidence relating to this statement see the web site: www.bmjpg.co.uk/books/neonatalformulary/chapters/vitk2comment.htm). What matters here is not so much the total dose given as the need for the dose regime to allow for the fact that natural body stores are low and turnover is rapid (a realisation that reinforces the suggestion that the solitary large intramuscular dose traditionally given at birth functions as a slow release “depot” store).⁴⁰ Even in adult life body stores are limited (∼1 μg/kg), and the turnover time is only 1–2 days.⁴¹ Low dose daily “drops” would be the most physiological option,⁴² but no commercial company has yet shown an interest in such a product.

Recent studies have certainly not provided any support for the belief that vitamin deficiency bleeding is commoner in preterm babies. Although prothrombin levels are lower than in term babies at birth, giving vitamin K does not cause a rise. Neither is bleeding commoner in babies undergoing operative delivery. However babies who are not fed at birth are certainly at increased risk, since all have low vitamin K stores at delivery, and milk provides their only source of vitamin K until bacterial activity in the gut starts to provide a secondary source. Bottle fed babies are at almost no risk because almost all these milks are artificially fortified. The babies of mothers on some anticonvulsants are also at risk, but such bleeding can occur at any time in the first 2–3 days of life.^43,44 The idea that cases presenting in the first day of life form a clearly distinct subgroup is ingrained,³⁷ but it gains no support from recent reports.^16,17

Further studies have also been done into the suggestion that intramuscular prophylaxis could be associated with a higher incidence of cancer in later childhood. The most informative of these were six studies that compared such children with others, matched both for date and either place or hospital of birth, who never developed cancer. A pooled analysis of these data, commissioned by the UK Department of Health in 1998, finally appeared last year: 2431 children developing cancer before 15 were compared afresh with 6338 controls matched for sex and year (but not place) of birth.⁴⁵ The resultant analysis confirms that solid tumours are certainly no commoner in children given intramuscular vitamin K at birth. The situation with regard to childhood leukaemia is less clear and, since almost every baby now gets prophylaxis in some form or other, is unlikely to be clarified by the collection of further data. The increased risk, if real, is small (unadjusted odds ratio 1.25; 95% CI 1.06 to 1.46), and could be due to the fact that those selected for prophylaxis (because of prematurity, operative delivery or the like) were already more at risk of later cancer for some unknown reason. Interpretation also depends on whether you believe that staff followed unit policy with regard to prophylaxis, as they claim, even where there is no proof of this in those records that do still exist (for further evidence relating to this statement see above website commentary). It has certainly proved difficult to prove that no risk exists. Only a controlled trial could ever resolve the residual uncertainty, and this would have to be quite unrealistically large.

For nearly 40 years now clinicians have been using a fat soluble form of vitamin K (phytomenadione) dispersed in a polyethoxylated oil or in polysorbate 80 with either propylene glycol (Konakion^™) or benzyl alcohol (AquaMEPHYTON^™), and giving this product uneventfully at birth to prevent vitamin K deficiency bleeding. However intravenous use in adults has occasionally been associated with severe anaphylaxis, possibly due to the polyethoxylated castor oil triggering histamine release.⁴⁶ As a result Roche finally brought out a new colloidal product, solubilised with lecithin and a bile salt (glycocholic acid), in 1996, and started to phase out their former product. Little was known about the new mixed micellar product (Konakion MM™) when it first became available, but it did seem to be better absorbed when given by mouth.⁴⁷ Roche therefore sought and obtained a licence, from the outset, for oral as well as intravenous and intramuscular use. This was something they never did with their earlier product in the UK, or in North America, even though it had been awarded a licence for oral use in Europe. However, it only has a license for the prevention of “haemorrhagic disease of the newborn.” Roche have not, as yet, claimed to have evidence that it prevents late vitamin K deficiency bleeding, so the administration of further doses after one week is, in some senses, currently an “off label” use.

A change in the formulation of a well established product might hardly seem to be a big issue. The new product has generated relatively little study since the early work required by the licensing authorities first appeared seven years ago, but two important papers in this issue of Fetal and Neonatal (see pp 109 and 113) have now established, with some precision, the effectiveness of oral administration.^48,49 It seems that, although the new product may be safer than the old product when given intravenously, its efficacy when given by mouth is no greater than that of other products. Neither do we yet know whether a single intramuscular injection provides the same long term protection as the product it has replaced, although this is currently under evaluation in Australia.²⁷