The number that comes out of two million births
The study covers 2,020,302 live births between 2003 and 2021 and follows the infants to 6 months of age. For 24,089 of those births there is no information that intramuscular vitamin K was given. In those infants the adjusted odds ratio is reported as 1.52 (95% confidence interval 1.27–1.81) for any bleeding diagnosis and 2.91 (2.13–3.96) for intracranial haemorrhage. The results were peer-reviewed and published in JAMA Pediatrics.[1]
In describing those ratios the sense of magnitude must not be lost. Intracranial haemorrhage is uncommon, and roughly tripling the odds of an uncommon event does not make it common. The published abstract gives no absolute event counts for the two groups. The most honest sentence available to a parent gives the number of infants in whom the risk rose, and that figure is missing. Substituting relative risk for absolute risk produces error in both directions here: the same incomplete sentence feeds both alarm and false comfort.[1]
The difference between drops and an injection
There is a second distinction in the data: infants given oral drops had more than twice the bleeding risk of those given the injection. Attributing that to absorption is plausible, since drops require several doses and doses can be missed. But in an observational study that inference has a competing explanation: families who choose drops may also depart from medical advice in other care decisions, and part of the measured difference may be carrying that. Separating the two does not need a larger dataset; it needs a follow-up in which the number of doses actually given is written down.[1]
There is also a direction. Non-receipt was at its lowest in 2006, at 0.66%; by 2021 it had more than doubled to 1.50%. That means the same relative risk lands on more infants every year. It is the least contested finding in the study and the most clinically useful one: whatever is disputed about how much any single infant's risk rises, the growth in the number of infants carrying that risk is a measured fact. The route to turning that number back runs through understanding why the injection is being declined; each year that pushes the rate under 1% again means a few infants with an infancy that passed without a bleed.[1]