Sentinel injuries: early clues of abuse

A four-month-old baby comes to a routine visit. The mother mentions, almost in passing, that he has a purple mark on his cheek and that he must have bumped it on the crib. The baby cannot even roll over yet. The doctor notes it, examines the rest, and the visit goes on. Weeks later, the same baby returns to the emergency department with a fracture.

This scenario appears often in studies on child maltreatment. Before the serious injury, many children had a small, visible, poorly explained injury. The literature calls this finding a sentinel injury. Recognizing the clue at the first contact is a task of clinical reasoning, and this text shows what the studies say about it, where the reasoning fails and what guidelines from several countries recommend.

Key points

• A sentinel injury is a small, visible injury, usually a bruise, in a baby who does not yet move around on their own.
• Many children who are victims of serious maltreatment had a sentinel injury before.
• Recurrence of maltreatment is frequent when the first suspicion goes uninvestigated.
• Team training programs increase identification and reduce new injury.
• Guidelines from several countries agree on imaging tests but differ on other points.
• The reasoning requires balance between not letting it pass and not accusing without basis.

What a sentinel injury is

The most widely used definition describes a sentinel injury as a small injury, usually a bruise or an injury in the mouth, in a baby who does not yet walk or crawl, for which there is no plausible explanation (Kim and Noh, 2023). The central idea is mobility. A baby who does not yet move around has few opportunities to get hurt alone, and so a mark on the body calls for an explanation.

Kim and Noh point out that these injuries are easy to ignore, because they seem trivial and almost always come with an apparently reasonable story (Kim and Noh, 2023). The injury itself rarely requires treatment. What matters is what it may announce.

The mobility rule

A phrase repeated in pediatric outpatient clinics sums up the reasoning: those who do not move around rarely get hurt. The study by Manan and colleagues takes up this maxim, known in English as "those who don't cruise rarely bruise", when analyzing bruises in infants (Manan and colleagues, 2022). The phrase helps fix the idea that the motor development stage changes the interpretation of the same finding.

In a child who already walks, a bruise on the shin is expected. In a three-month-old baby, a bruise on the ear, neck or trunk carries another weight. The text on signs, symptoms and history discusses how the same finding takes on a different meaning depending on the context.

In practice, the question about motor development must be asked directly and recorded. Asking whether the baby already rolls over, sits with support, crawls or stands holding on to furniture takes a few seconds and changes the reading of any mark on the body. The answer also helps assess the story told: a fall from the sofa described for a baby who cannot even roll over deserves more questions, while the same fall in a child who already pulls up on furniture is far more plausible.

Sites that draw attention

The studies pay special attention to bruises on the face, ears, neck and trunk, as well as injuries to the frenulum of the mouth (Kim and Noh, 2023). These sites are rarely hit in accidental falls and more often hit when there is abuse. The presence of several injuries at different stages of healing also raises suspicion.

This list of sites works as a clue, and the interpretation depends on the history, the age and the complete examination. Coagulation disorders, congenital marks and cultural care practices enter the differential diagnosis and must be considered before any conclusion.

Diseases that mimic maltreatment

Good reasoning when faced with a small injury always includes the question of what else could explain it. Coagulation disorders, such as hemophilia and von Willebrand disease, can produce bruises with minimal trauma. Congenital pigmentation marks, common in the lumbar region and on the buttocks, can be mistaken for bruises by those who do not know them. Rare bone diseases, such as osteogenesis imperfecta, increase the chance of fractures with minor trauma.

Traditional care practices, present in various cultures, can also leave marks on the skin. Recognizing these possibilities avoids hasty conclusions and guides the initial tests, such as coagulation assessment when bruises are the main finding. A complete differential diagnosis protects the family from an unfair suspicion and, at the same time, gives more strength to the suspicion when the alternatives are ruled out.

How often maltreatment repeats

The main reason to take the sentinel injury seriously is recurrence. In a cohort of 1361 children evaluated for suspected maltreatment, recurrence was 26 per 100 within one year and 40 per 100 within two years (Deans and colleagues, 2014). The authors identified factors associated with recurrence and argue that the first evaluation is an opportunity for protection (Deans and colleagues, 2014).

Another study followed 914 children evaluated for suspected maltreatment and found recurrence in 39 per 100 (Thackeray and colleagues, 2016). The repetition appeared both in children with confirmed suspicion and in those whose first evaluation was inconclusive (Thackeray and colleagues, 2016).

Why the first evaluation matters

Recurrence shows that the first contact is rarely the last. A child who has suffered an assault generally remains in the same environment, and the lack of any intervention keeps the risk in place. That is why the first evaluation carries a weight out of proportion to the size of the injury that prompted it.

That weight applies as much to the physician as to the nursing staff, the triage team and professionals from other services who see the baby. When each one assumes that someone else has already noticed the clue, the opportunity is lost. A clear flow of recording and communication among the team reduces this risk.

Earlier injuries that went unnoticed

Shanahan and colleagues looked back, using Medicaid data from four US states (California, Georgia, North Carolina and Texas). Among babies born from 2006 to 2010 with 12 months of continuous coverage, 4817 received a diagnosis of abuse or neglect before turning 1 year old, and 1473 of them, or 30.6 per 100, had a diagnosis of a previous injury (Shanahan and colleagues, 2022). Among the babies with maltreatment and a previous injury, 84 per 100 had attended at least one well-child visit before the injury (Shanahan and colleagues, 2022).

These numbers show that contact with the health service did happen, and that the opportunity to recognize the clue existed. The problem lies less in access and more in the interpretation of what was seen. The text on base rate in diagnosis helps to understand why a rare finding in a large population requires systematic attention.

Training the team changes the outcome

If the problem is one of recognition, it makes sense to train those who provide care. The TRAIN program, implemented in 6 hospitals in Ohio, in the United States, combined team training, evaluation protocols and case follow-up (Crichton and colleagues, 2023). The identification of suspicious injuries rose from 51 to 76 per month over the course of the intervention (Crichton and colleagues, 2023).

The most important effect appeared in new injuries. The proportion of children with a new injury within 3 months fell from 5.7 to 2.1 per 100, and within 12 months it fell from 6.5 to 3.7 per 100 (Crichton and colleagues, 2023). The result suggests that recognizing the clue at the first contact protects the child in the following months.

Prevention before the first injury has also been studied. In British Columbia, Canada, Sadler and colleagues compared hospitalizations for physical abuse in children up to 24 months before (1999 to 2008) and after (2009 to 2019) a universal nurse-led parenting program (Sadler and colleagues, 2024). The hospitalization rate fell by 30 per 100 after implementation, without statistical significance, and confirmed cases of abusive head trauma were excluded from the analysis (Sadler and colleagues, 2024).

What a program like this teaches

Training programs usually combine three elements. The first is teaching how to recognize the sentinel injury, with examples of locations and patterns. The second is offering an assessment protocol, with guidance on tests and on consulting child protection specialists. The third is tracking outcomes, so the team can see the effect of its own work.

That last element is often forgotten. When the team gets feedback on the cases it recognized, learning consolidates and the protocol stops being a bureaucratic obligation. The same principle appears in other areas of clinical reasoning, where feedback on the outcome is one of the most efficient ways to calibrate judgment.

More imaging when suspicion arises

Heyming and colleagues reviewed the emergency department records of a California pediatric hospital that adopted TRAIN in 2019, looking at infants under 6 months. Before the program, from January 2017 to September 2018, there were 12,812 visits, and 2.8 per 100 involved a sentinel injury (Heyming and colleagues, 2023). Among infants with a sentinel injury, the complete skeletal survey went from 17.1 per 100 in that period to 27.2 per 100 after the program, from October 2019 to July 2020 (Heyming and colleagues, 2023). New injury within 12 months showed no significant difference between the two periods (Heyming and colleagues, 2023).

The skeletal survey looks for occult fractures, which may be present even when the infant appears well. The increase indicates that suspicion led to a more complete investigation, which is the step after recognition. The text on clinical decision-making discusses how to choose tests when the harm of missing something is high.

Study	Population	Main finding	Source
Recurrence cohort	1361 children	Recurrence of 26 per 100 in one year and 40 per 100 in two years	(Deans and colleagues, 2014)
Follow-up cohort	914 children	Recurrence in 39 per 100	(Thackeray and colleagues, 2016)
Children's history	4817 infants with abuse before age 1	1473 with a previous injury, or 30.6 per 100	(Shanahan and colleagues, 2022)
TRAIN program	6 hospitals in Ohio	New injury within 3 months from 5.7 to 2.1 per 100	(Crichton and colleagues, 2023)
Guideline review	20 guidelines from 15 countries	41 per 100 of expected recommendations missing	(Blangis and colleagues, 2021)

What guidelines around the world say

Blangis and colleagues compared 20 guidelines from 15 countries on the assessment of suspected physical abuse in children (Blangis and colleagues, 2021). On average, 41 per 100 of the expected recommendations were missing from the documents (Blangis and colleagues, 2021). Only 8 guidelines, or 50 per 100 of those addressing the topic, included intraoral injuries and fractures among the warning findings (Blangis and colleagues, 2021).

There was agreement on one central point. All guidelines recommended skeletal survey, CT and MRI in the indicated situations (Blangis and colleagues, 2021). Divergence appeared in the description of warning findings, in the age cutoff and in the notification flow.

Why guidelines diverge

Each country organizes child protection in its own way, with its own laws, social services and notification flows. Clinical guidelines reflect these differences, and that is why the part about what to do after suspicion varies more than the part about how to examine. For the clinician, the practical lesson is to know your own service's protocol and to know whom to turn to.

The lack of recommendations in several documents also has a useful reading. Incomplete guidelines hand professionals decisions that could be standardized, and the individual decision is more subject to bias. A clear local protocol, reviewed with the team, makes up for part of this gap.

It is also worth remembering that guidelines age. New studies on recurrence, imaging tests and team training appear every year, and a protocol written long ago may leave out warning findings that are well recognized today. Reviewing the local document from time to time, with the participation of pediatricians, nursing, radiology and social services, keeps the flow up to date and helps the team know the path before they need it on a difficult shift.

Ethics between acting too much and acting too little

Bilson and Talia analyzed protection procedures in the United Kingdom and discussed the ethics involved (Bilson and Talia, 2023). The authors recall that an exaggerated reaction can bring harm to the family, with long investigations and unjust removal, and that an insufficient reaction leaves the child exposed to further abuse (Bilson and Talia, 2023).

This dilemma is similar to that of any diagnostic decision with serious consequences in both directions. The text on overdiagnosis discusses the human harm of excess, and the text on diagnostic uncertainty shows how to act with prudence when the answer is not yet clear.

What the pandemic showed

Lockdown during the COVID-19 pandemic changed children's contact with schools, daycare centers and health services. In France, a study of 178 children found twice as many cases of abusive head trauma during the lockdown period compared with previous years (Obry et al., 2023). In the United States, Sharma et al. described an increase in abusive trauma in May and June 2020 (Sharma et al., 2021).

These studies look at short periods and specific services, and comparisons between countries require care. Even so, they point in the same direction: isolation increased the risk for infants and reduced the chances that someone outside would notice the injury.

What these data teach

Reading these studies together reinforces a simple idea. When the child has less contact with adults outside the family, the sentinel injury has less chance of being seen. Each visit carries more weight, and the attention of the professional who provides care becomes one of the few protection barriers.

This lesson applies outside the pandemic. Small infants have few contacts outside the home, and well-child visits, vaccinations and trips to the emergency department are the most frequent opportunities to recognize a clue.

For services, this means thinking about who sees the infant at each contact. The vaccination room, emergency department triage and the well-child visit are places where the infant's clothing is usually removed and the skin is in view. Including skin inspection and the question about motor development at these moments increases the chances of recognizing a sentinel injury, and a simple path for communicating the suspicion to the medical team prevents the observation from being lost between one department and another.

What changes for the clinician

The studies allow building a reasoning script for the small injury in infants. The sequence below summarizes the most consistent points.

• Ask whether the infant already rolls over, crawls or walks, because mobility changes the interpretation (Kim and Noh, 2023).
• Examine the whole body, including ears, neck, trunk and mouth (Kim and Noh, 2023).
• Record the explanation given and check whether it is compatible with the developmental stage.
• Consider a skeletal survey and other tests according to the local protocol (Blangis et al., 2021).
• Remember that recurrence is frequent when the first suspicion goes without evaluation (Deans et al., 2014).

The list organizes the thinking, and referral follows each place's protection protocol. The clinician's role is to recognize and activate the network, and investigating the family is the job of specialized teams. Recording carefully, as in the text on organizing the reasoning in the medical record, protects the child and also the professional.

Where reasoning usually fails

Three failures appear often. The first is accepting the explanation without checking whether it fits the age. The second is examining only the area of the complaint. The third is letting the impression of the family guide the decision, for better or for worse.

This third failure is related to implicit bias in diagnosis. Stereotypes about social class, race or family structure can lead to suspecting some families too much and others too little. The same script for every baby reduces this risk. Affective bias also weighs in, because sympathy for a family or discomfort with the topic can lead to looking away.

How to talk with the family

The conversation with the family is one of the hardest parts of the process. The recommended tone is that of someone who wants to understand what happened and protect the child, and the assessment is presented as routine for injuries of this type in small babies. Explaining that the same approach applies to every baby with similar findings helps reduce the feeling of accusation.

The professional also needs to take care of their own reactions. Anger, pity or fear of making a mistake can get in the way of listening and recording. Writing down the explanation in the words used by the family, describing the injury precisely and avoiding conclusions in the medical record are simple precautions that help the specialized team do its work later.

Learning from cases

Discussing cases of sentinel injury in morbidity and mortality sessions helps the team recognize blind spots without blaming individuals. For those who want to practice with other pediatric scenarios, the text with clinical reasoning exercises for pediatrics offers cases to train on. Differential diagnosis with clues also helps to remember the diseases that mimic abuse.

Fictional cases built from these patterns are a safe way to train. They allow discussion of mobility, the locations of the injuries, the differential diagnosis and the referral without exposing any family, and they help the team agree on a common language for recording and communicating the suspicion. Over time, the question about mobility and the full skin inspection become part of every baby's examination, and the clue no longer depends on the memory of a single attentive professional.

How NEXO helps train this reasoning

In NEXO, the case of the day is fictional, with a real clinical presentation and up to six progressive clues, and you propose the diagnosis by searching for the ICD-10 code.

The limits of use of the game are described on the NEXO Economy page. If you research the teaching of clinical reasoning, the page for Researchers explains how to get in touch. Professionals who write or review cases find information in Curator and in Creators, and institutions can learn about the proposal on the partnerships page. Training with differential diagnosis and clues and dealing with diagnostic uncertainty complete this training.

Frequently asked questions

What is a sentinel injury?

It is a small injury, such as a bruise or an injury in the mouth, in a baby who does not yet move around on their own and for which there is no plausible explanation (Kim and Noh, 2023).

How often does abuse recur?

In 1361 children assessed, recurrence was 26 out of every 100 in one year and 40 out of every 100 in two years (Deans et al., 2014).

Does training the team make a difference?

It does. In the TRAIN program, new injury at 3 months fell from 5.7 to 2.1 out of every 100 after team training (Crichton et al., 2023).

Do guidelines from different countries agree?

They agree on imaging tests, and a review of 20 guidelines from 15 countries found 41 out of every 100 of the expected recommendations missing (Blangis et al., 2021).

Did the pandemic change the cases?

In France, abusive head trauma doubled during lockdown in a study of 178 children (Obry et al., 2023), and in the United States there was an increase in abusive trauma in May and June 2020 (Sharma et al., 2021).

Take-home messages

• A sentinel injury is a small injury in a baby who does not yet move around on their own.
• Many children who are victims of severe abuse had a previous injury recorded.
• Recurrence is frequent when the first suspicion goes without assessment.
• Training the team increases identification and reduces new injury.
• One same script for all infants reduces the influence of stereotypes.

References

1. Bilson A, Talia A. Bruises in Premobile Infants: A Contested Area of Research, Policy and Practice. Practice (Birm). 2023. PMID 38013882.
2. Blangis F, Allali S, Cohen JF, et al. Variations in Guidelines for Diagnosis of Child Physical Abuse in High-Income Countries: A Systematic Review. JAMA Netw Open. 2021. PMID 34787659.
3. Crichton KG, Spencer S, Shapiro R, et al. Timely Recognition of Abusive Injuries (TRAIN): Results from a Statewide Quality Improvement Collaborative. Pediatr Qual Saf. 2023. PMID 37051406.
4. Deans KJ, Thackeray J, Groner JI, et al. Risk factors for recurrent injuries in victims of suspected non-accidental trauma: a retrospective cohort study. BMC Pediatr. 2014. PMID 25174531.
5. Heyming T, Knudsen-Robbins C, Sharma S, et al. Child physical abuse screening in a pediatric ED; Does TRAIN(ing) Help? BMC Pediatr. 2023. PMID 36894913.
6. Kim HB, Noh H. Defining Sentinel Injuries of Suspected Child Abuse by Age Using International Classification of Diseases-10: A Delphi Study. Pediatr Emerg Care. 2023. PMID 37586374.
7. Manan MR, Rahman S, Komer L, et al. A Multispecialty Approach to the Identification and Diagnosis of Nonaccidental Trauma in Children. Cureus. 2022. PMID 36039273.
8. Obry S, Roman E, Tavernier E, et al. The monthly incidence of abusive head trauma, inflicted skeletal trauma, and unexplained skin lesion in children in six French university hospitals during the COVID-19 pandemic. Child Abuse Negl. 2023. PMID 36758374.
9. Sadler K, Rajabali F, Zheng A, et al. Impact of a Parent Education Program Delivered by Nurses and Health Care Providers in Reducing Infant Physical Abuse Hospitalization Rates in British Columbia, Canada. Can J Nurs Res. 2024. PMID 38115698.
10. Shanahan ME, Austin AE, Berkoff MC. Prevalence of Injuries Among Medicaid Enrolled Infants Prior to Child Abuse and Neglect. Child Maltreat. 2022. PMID 34319172.
11. Sharma S, Wong D, Schomberg J, et al. COVID-19: Differences in sentinel injury and child abuse reporting during a pandemic. Child Abuse Negl. 2021. PMID 33707071.
12. Thackeray J, Minneci PC, Cooper JN, et al. Predictors of increasing injury severity across suspected recurrent episodes of non-accidental trauma: a retrospective cohort study. BMC Pediatr. 2016. PMID 26772185.