When is a bruise on an infant a “sentinel injury”?

Child Abuse Pediatrician Dr. Lynn Sheets of the University of Wisconsin is working with U.S. Senator Tammy Baldwin to enact a bill that would use taxpayer dollars to develop an “early detection” program for child abuse. Sheets proposes examining infants under the age of 7 months for child abuse if they have a bruise. Such an action would increase the already high rates of potential false abuse accusations.

This legislation is based on the belief that pre-mobile infants never get bruises, a belief that has not been proven by science. This legislation, supported by the American Academy of Pediatrics, would serve to get more and more children referred to child abuse pediatricians, thus increasing the importance of that sub-specialty.

Our justice system must protect the innocent, and maintain the principle of innocent until proven guilty. Allowing child abuse pediatricians – with their demonstrated bias towards over-diagnosis of abuse – to examine every child with a bruise is a threat to that principle of justice. Read the paper by Dr. Douglas Smith below to understand more of why this legislation is so dangerous.

Call or write Senator Baldwin, as well as your own U.S. senator to ask them to stop this legislation.

When is a bruise on an infant a “sentinel injury”?

Douglas M. Smith MD, PhD

There is a mantra among child abuse pediatricians, that “if they’re not cruisin’ they’re not bruisin’.” Because infants under 7-8 months of age cannot move on their own, pediatricians believe bruises on an infant are rare and raise a red flag for abuse.

Dr. Lynn Sheets, a child abuse pediatrician at the University of Wisconsin, now wants to spend 10 million federal dollars to try to prove that mantra.

Dr. Sheets claims[1] “One of the things we realized is if you just call it a bruise, everyone has bruises,” Sheets said. “Everyone thinks about it as a minor injury including the doctors, including child welfare. So we needed to change the way people are thinking about these minor injuries in young infants.” She continues by recommending, “Think about them differently. They’re warning injuries or important injuries. They’re sentinel injuries. That’s why I coined the term and it has caught on nationally and even internationally.”

Dr. Sheets has now gone so far as to propose a bill, which has been entered by Tammy Baldwin, US Senate. The doctor states the legislation “…will be helpful by creating a demonstration program to show how detecting these sentinel injuries can reduce physical abuse.”  The legislation requires Secretary of Health and Human Services to award grants for the program.

However the “no cruising no bruising” dogma has little basis in science, making such legislation not only a misuse of taxpayer dollars, but also dangerously unethical.  I want to explore the scientific basis and the implications of Dr. Sheets proposal.

Dr. Sheets bases her opinion on a study that she published in 2014.[2] When creating a study however, it is important to correctly word the hypothesis or question you are setting out to prove. Dr. Sheets designed her study to ask, “do abused babies have more sentinel injuries than non-abused babies?” This is different however, than asking “how many babies with a sentinel injury are being abused?” This distinction in wording may seem trivial, but it actually reveals a critical flaw in Sheets’ conclusion. Finding a correlation between bruising and abuse only aids in creating a rate of correlation. The predictive value of that correlation rate will vary with the prevalence of abuse in the population. The prevalence of abuse would be much higher in babies referred to a child abuse team than in a pediatrician’s office practice, which is where Sheets wishes this dogma to be utilized.

To better understand, one can consider some hypothetical numbers. If a study finds that 30% of 100 abused infants have a bruise, but only 5% of 100 non-abused infants have a bruise, this may seem like dramatic evidence that bruising signals abuse.

A very different picture emerges however, when one applies these same percentages to a more realistic population ratio. In real life, a general pediatrician will probably see far less cases of abuse than a child abuse pediatrician. If a child abuse pediatrician finds abuse in 50 of 100 patients, 15 of those babies (30%) will have a bruise and 2.5 (5%) of the non-abused children will have a bruise. In comparison, if a general pediatrician finds abuse in 1 of 100 patients, only .3 abused babies will have a bruise. But at the same time, the general pediatrician will also find 5 non abused babies with a bruise. One can calculate the positive predictive value of a general pediatrician declaring a baby abused on the basis of a bruise by dividing the rate of true positives by the rate of false positives plus the true positives: 0.3/0.3+5 = 6%. In other words, if a general pediatrician assumes a bruise means abuse, he or she will only be correct 6% of the time. In comparison, a child abuse pediatrician assuming a bruise meant abuse would see a positive predictive value of 15/15+2.5 = 86%

This is why Dr. Sheets should have focused on how many bruised children are abused, instead of comparing bruise rates among abused and non abused children, especially if she wishes to justify examining all babies with bruises for child abuse.

Even if the correlations Dr. Sheets discovered were of value however, there are additional issues with the study. In her paper, Dr. Sheets stated that bruises were discovered on 0-2.2% of “non-cruising” infants in well child physical examinations. She cited three references to support that conclusion. One was not an actual study but an opinion piece. The second study included 246 infants less than 9 months old, and only 3 had bruises (1.2%).[3] The third study found 11 bruises in 511 infants (2.2%) that were not yet cruising.[4]

These studies reached these numbers by excluding suspected abuse. There are two problems with this however. First, child abuse pediatricians often claim any lesion that doesn’t blanch when pressed is a bruise. Lesions such as petechial rashes and eczema can be misdiagnosed as bruises, thus inflating the number of bruises. Secondly, there is no objective, scientific way to determine whether or not a child has been abused. Dr. Sheets categorized cases as “abuse” cases if the case was determined to be abuse by a Child Protection Team. But since Child Protection Teams already believe in the mantra that non-cruising infants don’t bruise, they are already biased towards the conclusion Dr. Sheets is trying to prove. Children who received a bruise in a simple accident could have been placed into the “abused” category in Dr. Sheets’ study.

Dr. Sheets’ study essentially says, “Bruises prove the child was abused. Therefore, children with bruises should be placed in the ‘abused’ category of the study, and since many children in the abused category have bruises, therefore bruises prove the child was abused.” This circular reasoning self destructs because one cannot prove an assumption with the assumption itself.

One could object that the assumption is simply common sense. How could a pre-mobile baby get a bruise unless through abuse? This however, overlooks innocent accidents and medical conditions that can cause susceptibility to easy bruising. To assume bruises indicate abuse is dangerous, because it biases doctors to assume accidents don’t happen, or to ignore natural medical conditions for bruising.

One might also object that it is “better to be safe than sorry,” and that it is worth examining bruised children more carefully for abuse, even if is expensive. However, there are medical risks to examining for child abuse. A child abuse work-up includes taking x-rays of every bone in the body and in many hospitals, a CT scan of the head. The medical risk of skeletal x-rays is small, but the medical risk of CT scans is not insignificant. A radiologist from the University of Oregon put the lifetime cancer risk from CT scan at 1 in 1,000.[5] That may seem small, but imagine being a parent and knowing your child had not been abused. Would you be willing to take that risk?  Many parents in these situations do not have the luxury of that choice however. They are often coerced by the doctor or by Child Protective Services to allow the tests. Dr. Sheets’ study may actually have ethical issues, since research projects involving human subjects require informed consent.

There are 4 million babies born in the United States each year. If 1.2-2.2% of them have a bruise at any given point in time, that represents 48,000 to 88,000 babies. If all of those babies get a bone survey and a CT scan, that represents 10’s of millions of dollars for the costs of the tests, untold expenses to the families falsely accused because of false positive tests, and up to 88 excess cases of cancers per year.

In addition to the medical risk, requesting a work-up for child abuse creates the risk of a false accusation of abuse. The scans produced in child abuse work-up may lead to a doctor mis-diagnosing a fracture, or perhaps, discovering an old fracture and making biased assumptions as to their cause. As mentioned before there is also the problem that skin lesions such as petechial rashes and eczema can be misdiagnosed as bruises. The legal battles from false accusations of abuse are devastating to their victims. They lead to loss of finances, careers, reputations, friends, children, family, and even freedom. It is crucial that there is solid evidence to warrant such a serious investigation.

Another study by Dr. Harper and colleagues[6] asked the question in a different way. They took a database that was collected to study siblings of children who had been referred to the Child Protective Team for evaluation of physical abuse. The Harper study extracted data from the original cohort of children examined for abuse, specifically babies less than 6 months old who had been referred for isolated bruising. They then asked how many of them had been abused. There were only 50 children with a single bruise and only 14 of those (28%) had a final diagnosis of physical abuse. One should keep in mind that these were not all children that had a single bruise, but only children who were referred by another doctor to the Child Protection Team because they suspected abuse. In other words, this study fails to take a sample that reflects the population at large. There may have been many babies with a bruise, but were never sent to a child abuse pediatrician. A properly designed study would have covered a random sampling from the general population to get an accurate picture of how frequently bruises are from abuse.

Even with these limitations, I think what these studies actually show is that not every bruise is a sentinel injury. Instead of rushing to dangerous and costly assumptions, one must listen to the parent’s explanation, investigate, run non-invasive blood tests, and use the common sense of referring physicians before blindly doing bone surveys and CT scans. Hopefully, this approach will avoid some of the false allegations that can be so devastating to families.

Citations:

[1] Glauber, B. (2019, April 03). Children’s Hospital research aids Congressional push to help prevent infant abuse. Retrieved from https://www.jsonline.com/story/news/2019/04/03/childrens-hospital-research-aids-push-help-prevent-infant-abuse/3357387002/

[2] Sheets LK, et al “Sentinel Injuries in Infants Evaluated for Child Physical Abuse” Pediatr Clin North Am. 2014 Oct;61(5):923-35

[3] Sugar NF, Taylor JA, Feldman KW; Puget Sound Pediatric Research Network. Bruises in infants and toddlers: those who don’t cruise rarely bruise. Arch Pediatr Adolesc Med. 1999;153(4):399–403

[4] Labbé J, Caouette G. Recent skin injuries in normal children. Pediatrics. 2001;108(2):271–276

[5] Frequently Asked Questions. (n.d.). Retrieved June 16, 2019, from https://www.imagegently.org/Portals/6/FAQs-Medical Professionals.pdf

[6] Harper NS, et al “Additional Injuries in Young Infants with Concern for Abuse and Apparently Isolated Bruises” J Pediatr. 2014 Aug;165(2):383-388