PECARN Head CT Decision Rule

In the PECARN study, altered mental status (AMS) was defined as any of the following: agitation, somnolence (sleepiness beyond what is expected for the child’s age and time of day), repetitive questioning (asking the same question more than twice), or slow response to verbal communication. This is distinct from GCS — a child can have a GCS of 15 and still demonstrate altered mental status. For example, a 4-year-old who repeatedly asks “Where’s Mummy?” despite being told she is in the room, or a toddler who is unusually drowsy 30 minutes after a fall, meets the criterion.

In infants under 2, AMS may manifest as inconsolable crying, poor feeding, inability to be distracted by toys, or an abnormal interaction pattern. Parental input is essential — a parent who says “she’s just not right” or “he’s never like this” provides valuable clinical information that should not be dismissed.

In infants, the thin calvarium makes skull fractures more likely and also more easily palpable on clinical examination. A palpable skull fracture is defined as a bony step-off, depression, or diastasis (separation of suture lines) felt on careful palpation of the entire scalp. This requires a thorough, systematic examination — parting the hair, palpating through any haematoma, and running fingers along suture lines.

Key points for examination: palpation should cover the entire calvarium including parietal, temporal, and occipital regions; a boggy swelling alone is not sufficient — you must feel an underlying bony irregularity; open fontanelles should be assessed for fullness or bulging; and scalp haematomas can obscure fractures, so palpation through/around the haematoma is necessary. If there is any uncertainty, CT should be obtained — this predictor carries a ciTBI risk of approximately 4.4% when present.

Clinical signs of basilar skull fracture are high-risk predictors in children aged 2 and older. These signs include:

• Haemotympanum: Blood behind the tympanic membrane, visible on otoscopy. This is the most commonly identified early sign.
• Raccoon eyes (periorbital ecchymosis): Bilateral periorbital bruising not caused by direct facial trauma. Typically appears 1–3 hours after injury.
• Battle sign (mastoid ecchymosis): Bruising over the mastoid process behind the ear. Usually appears 12–24 hours post-injury and may not be present at initial assessment.
• CSF otorrhoea or rhinorrhoea: Clear watery fluid from the ear or nose. Can be confirmed by testing for glucose (present in CSF, absent in nasal mucus) or by the “halo sign” on gauze (clear ring around a bloodstain).

Note that Battle sign and raccoon eyes may not be apparent at initial presentation — their absence does not exclude basilar skull fracture. Haemotympanum is the sign most likely to be present early and should always be assessed with otoscopy in children with head trauma.

In children under 2, the location of a scalp haematoma is a significant predictor. Non-frontal haematomas — those in the parietal, temporal, or occipital regions — are associated with a higher risk of underlying skull fracture and intracranial injury than frontal haematomas. This is because the temporal and parietal bones are thinner in infants, and the underlying middle meningeal artery is more vulnerable in these locations.

Frontal haematomas (isolated forehead bumps) in an otherwise well child are extremely common and carry very low risk. The PECARN rule specifically distinguishes frontal from non-frontal location for this reason. When assessing, note the exact location, size (small vs. large/boggy), and whether the haematoma is stable or expanding. A large, expanding, or boggy haematoma — particularly in a temporal or parietal location — should raise concern. In children under 3 months, any non-trivial scalp haematoma warrants particularly cautious evaluation given the higher vulnerability of the immature skull.

The PECARN study defined “severe mechanism” using specific, reproducible criteria. These differ by age group:

Both age groups:

• Motor vehicle collision with patient ejection
• Motor vehicle rollover
• Death of another passenger in the same vehicle
• Pedestrian or cyclist without helmet struck by a motorised vehicle
• Head struck by a high-impact object (e.g., bat, golf club, brick)

Under 2 years: Fall from a height >0.9 metres (approximately 3 feet, or roughly the height of a countertop or changing table).

≥2 years: Fall from a height >1.5 metres (approximately 5 feet).

Note that “high-impact object” refers to objects with significant mass and/or velocity — not a soft toy or a light tap. Falls from standing height onto a hard surface in a walking toddler are generally not classified as severe mechanism unless there are other concerning features. Context matters: a fall from a shopping trolley onto concrete is different from a fall from standing onto carpet.

For children in the intermediate-risk group, the PECARN authors specifically endorsed a period of clinical observation (4–6 hours from the time of injury) as an acceptable alternative to immediate CT. The rationale is that ciTBI in this group is uncommon (~0.8–0.9%), and most children who initially have one intermediate-risk predictor will improve during observation.

During observation, monitor for: declining or persistently abnormal GCS, new or worsening vomiting (≥3 episodes), increasing headache severity, emergence of new neurological signs, progressive irritability or somnolence, and any sign of clinical deterioration. If any of these develop, CT should be performed. If the child improves and is well at the end of the observation period, discharge with written head injury advice is appropriate.

Factors that may favour CT over observation in the intermediate group include: age under 3 months, multiple intermediate-risk predictors present simultaneously, worsening symptoms during initial assessment, isolated or delayed presentation without reliable follow-up, physician concern about non-accidental injury, and strong parental anxiety that cannot be adequately addressed.

The two PECARN algorithms have different predictors for each age group, and applying the wrong algorithm will produce incorrect risk stratification. The under-2 algorithm includes palpable skull fracture and non-frontal scalp haematoma; the ≥2 algorithm uses signs of basilar skull fracture, vomiting, and headache instead. A 23-month-old should be assessed using the under-2 algorithm, while a child who just turned 2 uses the ≥2 algorithm. The age cutoff is exactly 2 years (24 months). This is the single most common implementation error in clinical practice and in published decision support tools.

The PECARN rule was derived for children with blunt head trauma presenting with GCS ≥14 within 24 hours of injury. It should not be applied to: children with GCS ≤13 (these children should receive CT); penetrating trauma; trivial mechanisms in well-appearing children (e.g., walking into a door frame, ground-level fall onto soft surface with no symptoms); children with known VP shunts or brain tumours; children with bleeding disorders; or presentations more than 24 hours after injury. Applying the rule outside these boundaries may result in false reassurance.

The PECARN rule predicts clinically-important TBI (ciTBI) — it does not predict whether any CT abnormality will be found. A child classified as very low risk has a <0.02% chance of ciTBI but may still have clinically insignificant findings on CT, such as a small linear skull fracture without associated intracranial injury. In the original study, approximately 5.2% of children with any TBI on CT did not meet ciTBI criteria. Clinicians should communicate this distinction clearly to families: “The risk of a serious brain injury that would change our management is extremely low” — not “There is no chance of any injury.”

In the under-2 algorithm, the criterion “not acting normally per parent” is an intermediate-risk predictor with real predictive value. Parents and regular caregivers know their child’s baseline behaviour better than any clinician. When a parent says “something isn’t right” or “she’s not herself,” this should be taken seriously and scored as positive, even if the child appears well on brief clinical assessment. Studies have shown that parental concern about altered behaviour has an independent association with intracranial injury in young children. Dismissing this predictor because the child “looks fine” to the clinician is a recognised source of missed injuries.

When observation is chosen over CT for intermediate-risk children, the observation period must be adequate — the PECARN authors recommend 4–6 hours from the time of injury (not from the time of presentation to the ED, which may be hours later). During this period, serial neurological assessments should be performed and documented. Sending a child home after a brief assessment with instructions to “watch for symptoms” is not equivalent to structured clinical observation. The child must be observed in a clinical environment where deterioration can be detected and acted upon. If reliable observation is not possible (e.g., late-night presentation, no available observation area, unreliable family situation), CT is the safer option.

The entire rationale for the PECARN rule is to reduce unnecessary exposure to ionising radiation in children. A single head CT delivers approximately 2–4 mSv of radiation — equivalent to 8–12 months of background radiation. Children are 2–3 times more radiosensitive than adults, and the younger the child, the higher the relative risk. The lifetime excess cancer risk from a single paediatric head CT is estimated at approximately 1 in 10,000. While this risk is small for any individual child, the cumulative population impact is significant when millions of paediatric head CTs are performed annually. This context is important when communicating with families — parents should understand both the benefits and the radiation risk to make informed decisions.