Non-Accidental Trauma
- Non-accidental trauma (NAT), also called inflicted or abusive injury, is a diagnosis you make with your eyes and your suspicion — the images are the testimony.
- Certain fractures are "high-specificity" for abuse: classic metaphyseal lesions, posterior rib fractures, and fractures in non-ambulatory infants.
- The big-ticket emergency is abusive head trauma — subdural hemorrhage, often of differing ages, in a baby with no plausible story.
- The skeletal survey is the workhorse: a deliberate, protocoled set of films, not a single "babygram."
- The injuries that don't match the history are the whole point. A mobile-phone-flat baby does not spiral-fracture its own femur.
Some diagnoses are about a shadow on a film. This one is about a story that doesn't add up. Non-accidental trauma is the radiologist quietly noticing that the injuries in front of them could not have happened the way someone says they did — and then having the spine to say so. It is uncomfortable, it is high-stakes, and it is exactly the kind of thing we exist to catch.
The injuries that whisper "this wasn't an accident"
Most childhood fractures are gloriously boring: a toddler falls off a couch and buckles a wrist. What raises the hair on the back of my neck is a fracture in a baby who can't move yet. A two-month-old has not gone rollerblading. So when the mechanics of an injury require a level of force or coordination the child simply doesn't possess, the bone is telling you something the caregiver isn't.
A few patterns carry high specificity for abuse:
| Finding | Why it's suspicious |
|---|---|
| Classic metaphyseal lesion (CML) | A corner/bucket-handle chip at the end of a long bone, from violent pulling or shaking — not from a simple fall. |
| Posterior rib fractures | Need front-to-back squeezing of the chest, like adult hands wrapped around a tiny ribcage. |
| Fractures in a non-ambulatory infant | The kid can't generate the force; the story has to come from somewhere else. |
| Multiple fractures of differing ages | Healing happening at different stages means injury happened on different days. |
That last one is the radiology superpower. Bone heals on a timeline — fresh fractures look crisp, older ones grow fuzzy callus and remodel. Seeing injuries at different stages of healing is like finding tree rings: it tells you this wasn't one bad afternoon.
Abusive head trauma — the one that actually kills
If the skeletal findings are the paper trail, abusive head trauma is the emergency. The classic picture is subdural hemorrhage in an infant — crescent-shaped blood hugging the brain surface — frequently in a baby with no external bruise and no believable account of major trauma.
A baby with unexplained intracranial hemorrhage, retinal hemorrhages, and no plausible mechanism is a medical and protective emergency. Missing it can send the child straight back to the source of the injury.
Here, too, timing matters. Subdural collections of clearly different ages — some bright and fresh, some older and darker — point toward repeated events rather than a single accident. I treat any unexplained subdural in an infant as a question that demands an answer, not a footnote.
The skeletal survey: do it properly or don't do it
The single biggest technical pitfall in NAT is the "babygram" — one giant film of the whole baby at once. It looks efficient. It also hides exactly the subtle corner fractures you're hunting for, because the technique isn't optimized for any single bone.
A single whole-body "babygram" is not an acceptable skeletal survey. Subtle metaphyseal and rib injuries get washed out. The real survey is a dedicated, multi-image protocol — separate, well-collimated views of the skull, spine, ribs, and each limb.
Because these fractures are quiet, a follow-up survey a couple of weeks later is standard practice in many centers: injuries invisible on day one become obvious once healing callus shows up. Patience is a diagnostic tool.
When findings are subtle but the suspicion is real, a repeat skeletal survey after a short interval can reveal healing fractures that were invisible the first time.
Reading the images without losing your nerve
Describing the bones cleanly is half the battle — if your fracture language is shaky, it's worth a detour through how to describe a fracture. The other half is calibration. Plenty of normal infant anatomy mimics injury: open growth plates, normal developmental variants, and metabolic bone disease can all make a healthy bone look ominous.
So the honest stance is two-sided. We must not under-call a battered child back into harm — and we must not turn a family's worst week into a false accusation. That's why NAT is never one radiologist alone with a hunch. It's imaging plus the clinical exam, the history, the labs, and a child-protection team, all in the same room.
NAT is a clinical-radiologic diagnosis of teamwork. The radiologist flags injuries that don't fit the story; pediatricians, child-protection specialists, and sometimes the courts decide what it means. Our job is to describe accurately and raise the flag, not to litigate it.
The one thing to carry out the door
When the injury and the explanation don't match, believe the injury. A fracture in a baby who can't yet roll over, ribs cracked at the back, or blood around the brain with no story behind it — those are the images that should make you slow down, pick up the phone, and refuse to write "probably fine." Catching this is one of the few times a radiologist's read directly keeps a child safe.