This review takes stock of where paediatric sepsis has moved. The Phoenix criteria shifted the definition away from inflammatory markers towards life-threatening organ dysfunction, conjugate vaccine programmes have changed which children present with invasive bacterial infection, and guidance increasingly stratifies treatment by severity instead of applying one bundle to everyone.
The gap the authors identify is where the deaths are. Many paediatric sepsis deaths still occur within the first 24 hours of referral to intensive care — meaning the outcome was substantially determined before the child reached the unit, in the emergency department, the ward, or the referring hospital. Early recognition remains the weak link: screening tools perform poorly in unselected febrile children, biomarkers are adjuncts rather than answers, and clinicians still depend on gestalt and on parental concern.
The authors expect progress to come from risk stratification rather than universal screening, with machine learning and phenotype-based approaches as candidates — both needing prospective validation. Until then the actionable point is unglamorous: the quality of the first hour on a general ward or in an emergency department matters more than anything that happens in intensive care, and parental concern remains one of the better-performing signals available.
- Treat a parent's statement that the child is not right as a clinical finding, and document it
- Apply the Phoenix criteria's focus on organ dysfunction rather than counting inflammatory signs
- Re-examine rather than re-score: screening tools miss children in the unselected febrile population
- Audit the interval from recognition to antibiotics and fluids on general wards, not just in intensive care
- Immunisation status changes the pre-test probability of invasive bacterial infection — ask for it every time
Why it matters
If most deaths are determined before intensive care admission, improving intensive care will not change them.
Don't overread it
A narrative review of the field, not new data — the machine learning and phenotyping approaches it describes are unvalidated.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for paediatrics, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free