Learning from a missed posterior fossa tumour in the paediatric emergency department
Diagnosis in paediatric emergency medicine is rarely about a single data point. It is an exercise in synthesising imperfect information, under time pressure, in a system designed for flow rather than reflection.
The following case of an 8-year-old girl ultimately diagnosed with a posterior fossa brain tumour illustrates how diagnostic error is seldom the result of one mistake but rather the accumulation of barriers at multiple levels.
Drawing on the paediatric emergency medicine literature, this discussion frames learning across three themes: patient-level, clinician-level, and system-level barriers to diagnosis.
Hindsight is always 20:20.
One excellent resource is https://bettersafethantumour.com when suspicious about a space-occupying lesion in the paediatric patient.
Remember … Never forget to keep looking for zebras!
Patient-level barriers to diagnosis
Children are unreliable historians by necessity, not by fault. Pain, fear, and developmental stages all limit their ability to articulate symptoms clearly. In this case, the child identified abdominal pain as her primary complaint, physically reinforcing it by curling up and holding her stomach. Headache, neck pain, and vomiting—features that in retrospect pointed towards raised intracranial pressure—were intermittent and non-specific, and therefore deprioritised.
Avoid the trap of Zebra Retreat
Parental communication was also fragmented. The mother was distracted by caring for two younger children, limiting her ability to give a full history. As a result, she was unable to convey concerns voiced to her before outside of A&E. Her GP had concerns about the child’s headache, an optician had recently found nystagmus on examination of the child. The school had observed an uptick in repeated absences. Even the play therapist who had met the child and mother had noticed the child complaining of neck pain and discussed this with the mother as unusual.
Each piece of information alone might not have triggered an alarm; together, however, they formed a clear pattern that was highly suggestive of malignancy or serious neurological cause.
The literature consistently highlights how diagnostic accuracy in children improves when collateral histories are actively sought and integrated, yet this is often impractical in real-time emergency care.
We need to be mindful of potential sources of information and at the very least keep our history-taking open to their possibilities. It is easy to assume such information would come to the fore or be communicated through robust systems. Time and time again however this is often not the case.
The Patient and Family Micro-System
Symptom Masking: Children often localize distress to the abdomen, even when the pathology is neurological.
Fragmented Collateral: The “missing pieces” (optician, school, GP) existed but weren’t accessible.
Environmental Stress: A parent managing multiple children in a high-stress ER environment cannot always provide a linear, “textbook” history.
Clinician-level barriers to diagnosis
Clinicians do not practise in a cognitive vacuum.
Time pressure, cognitive load, and heuristics shape decision-making.
In this case, symptom attribution was influenced by a plausible and familiar diagnosis: constipation with associated abdominal pain and vomiting, supported by examination findings and past medical history.
This represents a classic example of anchoring and premature closure — well-described cognitive biases in emergency medicine.
Anchoring bias is the bane of all clinicians. In order to avoid this trap you must first acknowledge it exists and then ask yourself as the clinician, ‘if this isn’t the X diagnosis, what else could explain this presentation?’
Using the HeadSmart symptom card and decision support tools can help guide history taking when the diagnosis isn’t crystal clear.
The neurological examination was normal, symptoms were intermittent rather than progressive, and headache improved during observation. These factors reduced perceived risk, despite the known reality that posterior fossa tumours can present subtly, with fluctuating symptoms and normal early examinations.
The absence of “textbook” red flags contributed to reassurance. Retrospective awareness of the outcome highlights how hindsight bias can obscure the genuine uncertainty faced at the time.
The challenge is not to demand diagnostic perfection, but to recognise situations where diagnostic momentum should be slowed and uncertainty explicitly revisited.
The Cognitive Environment (The Doctor)
Heuristics: Using “constipation” as a working diagnosis is statistically logical but creates a cognitive blind spot. Here the certainty of features matching up with one diagnosis deterred looking for another.
The “Normal” Exam Trap: Did you know that 50% of children with brain tumours have a NORMAL neurological examination at diagnosis? (I didn’t either).
Fluctuating Symptoms: Improvement during observation is often misinterpreted as “recovery” rather than the natural waxing and waning of a posterior fossa pathology.
System-level barriers to diagnosis
System pressures also underpin many individual decisions. A crowded department, a three-hour wait and handover constraints all shaped care. The initial plan for admission and further psychosocial exploration was downgraded to home management due to flow pressures—an example of how system demands can alter risk tolerance.
The Macro – System
Flow vs. Reflection: The “Four-Hour Rule” or general department pressure creates a bias toward discharge.
Information Silos: There is a technical and systemic gap between community providers (opticians, schools) and the acute setting.
Communication failures were systemic rather than individual. There was no mechanism for integrating information from primary care, optometry, school, or allied health professionals into the emergency assessment. The play therapist’s observation of neck pain did not reach the treating clinician.
Such silos are repeatedly identified in patient safety literature as contributors to missed or delayed diagnoses.
Paediatric emergency medicine operates at the intersection of multiple systems; when those systems do not communicate, children bear the risk.
Key learning points
This case reinforces that missed diagnoses are rarely about ignorance, and more often about barriers. Actively seeking collateral information, acknowledging diagnostic uncertainty, and creating systems that support information sharing are critical safeguards. Brain tumours in children remain rare, but delayed diagnosis is common precisely because presentations are subtle and non-specific.
The task is not to scan every child with a headache, but to recognise when the story does not quite fit—and to pause long enough to ask what else might be going on.
Practical Takeaways for the ED
The “Wait and See” Caveat: If a child is discharged, the “safety net” advice must be specific. Instead of “return if worse,” use “return if the vomiting happens before breakfast” or “if they seem wobbly on their feet.
“Collateral Triage: If a child presents multiple times, the system should trigger a mandatory review of previous GP or specialist notes.
Cognitive Debiasing: Ask yourself: “If I couldn’t use the diagnosis of constipation, what else would explain this child’s behavior?”
In paediatric emergency medicine, diagnostic safety depends not just on clinical skill, but on curiosity, communication, and systems designed to support both.
