Vitamin K regimens and incidence of childhood cancer in Denmark
BMJ 1994; 308 doi: https://doi.org/10.1136/bmj.308.6933.895 (Published 02 April 1994) Cite this as: BMJ 1994;308:895
- J H Olsen,
- H Hertz,
- K Blinkenberg,
- H Verder
- Danish Cancer Society, Division of Cancer Epidemiology DK-2100 Copenhagen, Denmark
- Department of Paediatrics, University Hospital, DK- 2100 Copenhagen, Denmark
- National Board of Health, DK-1012 Copenhagen, Denmark
- Department of Paediatrics, Holbaek Hospital, DK-4300 Holbaek, Denmark
- Correspondence to: Dr Olsen.
A case-control study by Golding et al found that children who had received vitamin K intramuscularly at birth had a significantly greater risk of leukaemia than children who had received it orally or not at all.1 Draper and Stiller, however, questioned whether the relation was causal,2 and the finding was not verified in two recent case-control studies in Sweden and the United States.3,4 We compared the cumulative risk of cancer in Danish infants given vitamin K intramuscularly with that of infants who never received any form of manufactured vitamin K and infants whose mothers were given vitamin K during pregnancy.
Subjects, methods, and results
A national cancer registration system has been operating in Denmark since 1942.5 Regimens for administration of vitamin K to neonates have varied greatly over the 50 years of notification. No vitamin K was given to either newborns or mothers before 1955, we could define a non- exposed cohort as all children born in Denmark during 1945-54 (n=835 430).
Oral vitamin K was introduced for pregnant women in the late 1950s, and the practice was recommended by the Danish Board of Health in August 1960. All children born during 1960-9 were therefore included in the cohort with mothers who received vitamin K during pregnancy (n=797 472). A review of a random sample of 300 medical charts from one area of Zealand for 1960-3 (done by KB shortly after the birth of each child) showed that 3.3% of the children had received vitamin K intramuscularly.
Intramuscular injection of vitamin K to newborns became accepted practice in the early 1970s. From 1975, when less than 1% of Danish children were born at home, it can be assumed that nearly all children received vitamin K intramuscularly. We therefore defined the cohort that received vitamin K intramuscularly as children born in 1975-84 (n=586 378).
We identified all cases of childhood cancer in the three birth cohorts from the files of the cancer registry. The figure shows the cumulative risk for all types of cancer combined and for leukaemia and malignant lymphomas separately. The data are for children aged 1-15 years for the 1945-54 (no administration) and 1960-9 (maternal administration) birth cohorts and children aged 1-13 years for the 1975-84 cohort (intramuscular administration).
Cumulative risk of childhood cancer among children born 1945-54 (no vitamin K), 1960-69 (maternal vitamin K), and 1975-84 (intramuscular vitaminK).
The relative risk, defined as the ratio between cumulative risks at age 13 (intramuscular vitamin K v no vitamin K) was 1.00 (95% confidence interval 0.93 to 1.09) for leukaemia, 1.15 (0.97 to 1.36) for malignant lymphomas, and 1.29 (1.23 to 1.35) for all tumour types.
Comment
We found no trend in risk of childhood leukaemia over the three defined birth cohorts. The low increase in risk seen for all tumour types combined since the 1940s has been described before5 and is due mainly to continuous increases in risk of lymphomas in boys and of neuroblastomas in both sexes.
The comparison group in this study comprised children who were born before vitamin K was routinely given to the mother or child. Thus the trend in risk for childhood cancer, and for leukaemia in particular, seems to have been unaffected by the widespread use of intramuscular vitamin K since the early 1970s. Our findings are not consistent with those of Golding et al.1 The findings would agree if the prevalence of a strong risk factor for leukaemia were decreasing in parallel to the increasing use of intramuscular vitamin K. This, however, seems unlikely.
We thank B Carstensen for help with statistical testing and A Bautz for help with computing.