Vitamin K Skipped: 1 Extra Bleed in 114 Newborns, Mostly Logged Early
Unshot newborns had more recorded bleeding diagnoses in every year from 2017 to 2025. The absolute numbers, the studies' limits and what to do after a declined shot.
Dr. Thomas Patterson treats newborns at some of the St. Luke's Health System hospitals in Idaho, and he described one recent day to ABC News in a single sentence: One day [this year], I had 10 babies and five of them went out of the hospital without vitamin K.
It fits a trend ProPublica documented on May 6, 2026. Across St. Luke's, 3.8% of families declined the shot in 2020 and 9.8% in 2025, and one hospital reached 20%.
On Thursday, Nov. 5, 2026, two studies put national numbers on it. A JAMA paper (doi:10.1001/jama.2026.21764) from Kristan Scott and colleagues at Children's Hospital of Philadelphia followed 5,240,780 infants. Babies with no recorded shot had a 47% higher adjusted risk of any recorded bleeding by six months, and more than double the risk of intracranial bleeding, meaning bleeding in or around the brain. A JAMA Pediatrics research letter (doi:10.1001/jamapediatrics.2026.5794) from Elizabeth Wolf, Steven Woolf and colleagues at Virginia Commonwealth University covered about 11 million infants and reported that diagnoses of vitamin K deficiency bleeding, or VKDB, rose 347%.
The direction is credible. The shot works, and a Swedish study of 2 million infants lands in the same range. But 47% and 347% count diagnosis codes, most of them logged in the first days of life, and in the JAMA data the codes rose in babies who got the shot too. They are not the chance of the dangerous late bleed at home.
Try setting a relative figure beside the absolute one. The paper's adjusted number needed to harm, how many newborns go without the shot for one extra bleed, is 114 over six months. The figure a parent needs sits after discharge: about 1 extra case in 1,800. Neither alarm nor a shrug fits that. A plan does.
Two papers, two ways of counting
The JAMA study looked back through Epic Cosmos electronic health records, the data a health system keeps on each patient, for babies born at 35 to 43 weeks from Jan. 1, 2017 through Dec. 31, 2025, in all 50 states. Of those who passed quality filters, 234,825 (4.5%) had no intramuscular vitamin K, a shot into the thigh muscle, on record. That share rose from 3.1% in 2017 to 6.8% in 2025.
"No shot" has a narrow meaning: no injection in the hospital's medication record during the birth admission. The study could not tell whether a baby got oral drops, or a shot after leaving. "Bleeding" is a composite of diagnosis codes, including intracranial hemorrhage, newborn hemorrhage and VKDB. Nobody measured a bleed; someone entered a code.
The authors list possible miscoding, no timing or severity for each bleed, hospital births only, and unmeasured confounding, meaning differences the adjustment missed. The authors also flag that clinicians may code bleeding more readily when vitamin K was withheld, which could inflate the estimates (Contemporary Pediatrics summarizes it as a limitation). (A risk ratio compares two rates, so 1.47 means 47% more.)
The VCU letter, summarized in VCU's release on the JAMA Pediatrics letter, counts diagnoses in a different pool: more than 2,300 hospitals and 50,000 outpatient settings. Injectable receipt there fell from 97.7% to 95.0%, a no-shot share of 2.3% and 5.0%. Do not blend those with the JAMA cohort's figures. VKDB diagnoses rose 347.1%, a 4.5-fold increase, not a quadrupling. The authors say association does not prove causality.
Now the part the headlines skipped: bleeding diagnoses rose in both groups.
The no-shot group rose faster: the yearly rise in the odds of bleeding was 8.7% for unshot babies and 5.6% for shot babies.
By Daybreak Wire's back-of-envelope arithmetic on the paper's endpoint figures, the overall rate climbed from 139.3 to 233.5 per 10,000. Had the unshot share stayed at 3.1%, the 2025 rate would have been 227.1. So babies moving into the no-shot group account for about 6.4 of the 94.2 extra, roughly 7%. About 93% of the rise happened inside each group.
That ignores the years in between, and it does not mean refusal is harmless. It does mean "bleeding rose because refusal rose" is not what this table shows. The authors say they cannot explain the trend:
"While reasons for this trend are unclear, it could reflect increased ascertainment or sicker maternal-infant dyads,"
Study authors, JAMA, as quoted by The Epoch Times
Ascertainment means how hard anyone looks for a bleed; dyads are mother-baby pairs.
Where the gap shows up, and where it doesn't
Here are the same babies through four yardsticks.
| Bleeding diagnosis, per 10,000 births | No shot | Shot | Adjusted risk ratio | Gap |
|---|---|---|---|---|
| Any, birth to 6 months | 288.0 | 185.9 | 1.47 | 102.1 |
| Any, after discharge | 10.3 | 4.8 | 2.11 | 5.5 |
| Intracranial, birth to 6 months | 33.0 | 13.2 | 2.43 | 19.8 |
| Death with a bleeding diagnosis, birth stay | 2.9 | 1.1 | 2.69 | 1.8 |
The "gap" column is Daybreak Wire's subtraction of the published rates (unshot minus shot); the rates and ratios are from the JAMA paper.
The after-discharge gap of 5.5 per 10,000 is a small slice of the 102.1 gap over six months (the periods overlap, so the pieces need not sum exactly). So the 47% is mostly a count of what was recorded before discharge, a window where clinicians may code more readily when no shot was given. That does not make it wrong. It makes it the wrong number to carry home.
The row to carry is after discharge, the window for late VKDB. The crude gap there is 5.5 per 10,000, about 1 in 1,800. The ratio climbs to 2.11 because the baseline is tiny. The absolute gap falls from about 1 in 100 for "any code" to about 1 in 1,800.
Still, no row in the table shows a ratio below 1.47, which is why the direction is hard to wave away. Sweden's study found adjusted odds of 1.52 for any bleeding and 2.91 for intracranial bleeding in babies with no record of a shot. Different system, similar ratios. That strengthens the direction, though neither design can show cause.
Never compare multipliers across rows. The 73 is the VKDB code during the birth stay, which the authors say is probably inflated. The 81 is the CDC's figure for late VKDB specifically. Each answers a different question.
The VCU letter needs the same care. Its 2025 VKDB rate of 107.4 per 100,000 is 1 in 931 babies, yet across the study period 87.8% of VKDB diagnoses came in the first week of life. The letter's coded rate, in a population 95% shot, is roughly 15 to 24 times the 1980s European late-VKDB range of 4.4 to 7.2 per 100,000 unshot babies, though that is not a like-for-like comparison. Woolf told Reuters the 2025 rate was approximately double 2024's. That could partly reflect coding, although the authors say other known risk factors did not rise.
The letter's stronger number: only 19.2% of VKDB cases had received the shot, while receipt was still 95.0% in 2025.
Why newborns need the dose
Babies are born with little vitamin K. Very little crosses the placenta, and newborn intestines lack the bacteria to make it. A mother's diet or supplements do not fix that, and breast milk is low in it, so exclusively breastfed babies stay at higher risk until they eat solids, usually between four and six months. Formula is fortified. One dose protects until then, by helping the liver make clotting factors.
Late VKDB, which strikes from one week to six months, is rare: the American Academy of Pediatrics (AAP) puts it at 1 in 14,000 to 1 in 25,000 babies without the shot; the CDC says under 1 in 100,000 with it. Before routine shots began in 1961, up to 1 in 60 babies developed VKDB in the first week.
It is worth preventing because of how it ends when it happens. Between 30% and 60% of late cases involve bleeding in the brain, and the CDC says 1 in 5 babies with VKDB dies. Dr. Scott Hadland of Mass General Brigham for Children told ABC News some survivors face cerebral palsy or a shunt for pressure around the brain.
The standard is one 1 mg dose in the thigh within six hours of birth, and the CDC says it can be delayed up to six hours for bonding. The AAP's parent guide to the vitamin K shot has the details.
On safety, a 1990s study suggested a childhood cancer link and later studies did not find it. The FDA's boxed warning came from allergic reactions in adults given high intravenous doses. It is a vitamin, not a vaccine.
Why not drops? Babies absorb oral vitamin K poorly and no oral product is FDA-approved in the US. Dosing needs repeat doses that can be missed, Dr. Jaspreet Loyal of Yale told PolitiFact. The JAMA editorial by Laura R. Kair of UC Davis and colleagues asks for an FDA-regulated oral product for families who firmly decline.
Doctors cite mistrust of pharmaceuticals, social media claims and plain unfamiliarity as reasons for refusal. David Hill, a Seattle pediatrician and AAP spokesperson, says he persuades about one in four initially refusing parents. Wolf ties it to the trust problems behind measles outbreaks: Similar to the recent measles outbreaks, we are seeing the real-world implications of growing mistrust of doctors and public health professionals.
If the shot was already declined
None of this calls for shame. The steps below come from pediatricians and agencies; the baby's doctor can confirm them.
First, tell every clinician, at every visit and in any emergency room, that the baby did not get vitamin K. Hill says that helps diagnosis. The CDC says late VKDB usually appears two to eight weeks after birth, and it can occur up to six months.
Second, know the signs. The AAP lists these, and says most cases come with no warning beforehand:
- bleeding from the umbilical cord or nose
- paler skin or pale gums
- bruising easily, especially around the face and head
- bloody stool, or black, dark, sticky stool
- vomiting blood
- a yellow tint to the whites of the eyes three weeks or more after birth
- seizures, irritability, excessive vomiting or sleeping too much
Hill, speaking to CIDRAP, describes what to treat as an emergency:
"If their baby is unresponsive or stops eating, or won't move one side of their body, or doesn't move one side of their face, they're having a medical emergency and have a very limited amount of time to respond."
David Hill, MD, Seattle pediatrician and American Academy of Pediatrics spokesperson
Third, do not count on substitutes: drops, delayed cord clamping, a mother's diet or breast milk.
Can the shot still be given later to a healthy baby? The CDC, AAP and Johns Hopkins pages do not say, so no protocol can be claimed here. Ask the pediatrician at the first visit whether anything further should be done. A Johns Hopkins explainer says vitamin K can be given as a last resort to stop active bleeding, but brain damage already done may be permanent.
Nobody is counting
The CDC does not track refusals, and it does not track VKDB bleeding. Federal officials told ProPublica's May investigation that VKDB has never been submitted for consideration as a notifiable condition. In June, Rep. Kim Schrier (D-Wash.) and Sen. Angela Alsobrooks (D-Md.) asked the CDC to start. An HHS spokesperson declined to tell Reuters whether it will.
ProPublica reports that deaths directly attributed to VKDB appear to be fewer than a dozen a year, while more than 700 newborns died of spontaneous brain bleeding in 2024, some portion perhaps involving vitamin K deficiency. Scott, the JAMA lead author, has said there is no clean data repository that would let anyone track this systematically. Dr. Robert Sidonio Jr., a pediatric hematologist at Emory University and Children's Healthcare of Atlanta, put it to ProPublica this way: If you don’t track it, you don’t document it,
and right now the federal government counts neither the refusals nor the bleeds.
