Fever and fatigue: two of the five headline signs cannot be elicited from a baby
also called fatigue infant leukaemia · fever at presentation infant · symptom ascertainment infant · pre-verbal symptom reporting
Two of the five commonest presenting features of childhood leukaemia - fever at 53% and fatigue at 46% - come from studies of children who could say how they felt. A baby cannot report tiredness, and a fever in a baby is investigated as infection first, so both figures change meaning entirely when carried into the first year of life.
The same thing, for a scientist
Clarke et al. rank fever at 53% and fatigue at 46% among presenting features in children aged 0-18 [E1]. Both are ascertained differently in infants. Fatigue is a **reported symptom** with no infant equivalent: what is recorded instead is reduced feeding, hypotonia, increased sleeping and irritability, none of which is the same construct and none of which has a published frequency in infant ALL. Fever is **observable** but not specific: it is the commonest reason an infant presents to any acute service, so an infant with leukaemia and fever is investigated for sepsis, given antibiotics, and diagnosed with leukaemia only when the blood film is examined. A PubMed search for fever at presentation in infant leukaemia returned 8 records, all case reports or infection complications on title check, and none reporting a frequency [E2]. These are therefore not numbers awaiting an update; they are constructs that do not transfer, and an entry that carries 53% and 46% into an infant beat is reporting a measurement nobody made.
The two failure modes are different1 deeper
Fatigue: a construct that does not exist below about age three. The 46% figure is the proportion of children who, or whose parents, described tiredness. In an infant the observable proxies are: smaller feed volumes, falling asleep during feeds, reduced spontaneous movement, less social engagement, hypotonia. Each is a different measurement with a different reliability, and none is coded as 'fatigue' in any infant cohort. Pooling them under the older children's label would be an act of translation, not of measurement.
Fever: observable but non-specific, and it biases the diagnostic pathway. Fever in infancy is the single commonest acute presentation, so a febrile infant with leukaemia enters the sepsis pathway. Two consequences follow. First, the pre-diagnostic interval is lengthened by a round of antibiotics. Second, the leukaemia is often found on the full blood count taken for the fever, which means fever is over-represented among infants whose diagnosis was made early. That is an ascertainment effect on the frequency itself.
The search that found nothing
fever[tiab] AND (infant[ti] OR infants[ti]) AND (leukemia[ti] OR leukaemia[ti]) AND presentation[tiab] returned 8 records on 2026-09-19 2. Title-checking all eight: one case of ALL in a four-month-old, one visceral leishmaniasis in an infant with ALL, one disseminated Exophiala mycosis in infant MPAL, one infant T-ALL case report, one intracardiac thrombosis with COVID-19, one facial palsy as a presenting symptom, one parotid and renal enlargement case report, one invasive aspergillosis with bowel obstruction. Zero report the frequency of fever at presentation in infants. Six of the eight are infection complicating treatment, not fever at presentation, which is itself informative about where the literature's attention sits.
What infants do get measured on: infection during treatment
The infant literature has a great deal to say about infection - but after induction starts, not before diagnosis. In COG AALL0631 sterile-site infections during induction fell from 57.7% (15/26) to 19.5% (24/123) when induction was de-intensified and supportive care guidelines were added 3. In CCCG-ALL-2015 the rates across the whole of therapy were sepsis 44.8%, severe pneumonia 17.9% and invasive fungal infection 9.0% 4. Those are real infant-specific numbers with denominators, and they belong to concept-induction-mortality-and-infection-infant-all. They do not substitute for the presenting-fever frequency, and conflating the two would put a treatment-phase number where a diagnostic-phase number belongs.
| metric | construct validity of pooled presenting features in infants |
|---|---|
| population | infants <1 year |
features that do not transfer
[
{
"feature": "fatigue",
"pooled_pct": 46,
"reason": "reported symptom; no pre-verbal equivalent recorded in any infant cohort"
},
{
"feature": "fever",
"pooled_pct": 53,
"reason": "observable but non-specific; diverts to the sepsis pathway and is ascertained partly because it prompts the blood count that makes the diagnosis"
}
]infant infection rates during therapy — the numbers behind the chart
[
{
"cohort": "AALL0631 cohort 1 induction",
"sterile_site_infection_pct": 57.7,
"n": 26
},
{
"cohort": "AALL0631 cohort 2 induction",
"sterile_site_infection_pct": 19.5,
"n": 123
},
{
"cohort": "CCCG-ALL-2015 whole therapy",
"sepsis_pct": 44.8,
"severe_pneumonia_pct": 17.9,
"invasive_fungal_pct": 9,
"n": 134
}
]- What are the observable proxies for fatigue in an infant with leukaemia, and has anyone measured their frequency or inter-observer reliability?
- What proportion of infants with leukaemia are diagnosed on a blood count taken for suspected sepsis rather than for suspected malignancy? This determines how much of the fever frequency is ascertainment.
- Does the febrile presentation lengthen or shorten the pre-diagnostic interval in infants? Both mechanisms are plausible and neither has been measured.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.review
Fever was present in 53% and fatigue in 46% of children in a pooled meta-analysis of 33 childhood leukaemia cohorts aged 0-18 (n = 3084).
Clinical presentation of childhood leukaemia: a systematic review and meta-analysis. · Clarke RT, Van den Bruel A, Bankhead C, Mitchell CD, Phillips B, Thompson MJ · Arch Dis Child 2016What this evidence cannot show
Ages pooled 0-18. Fatigue is a reported symptom and the review does not state how ascertainment differed by age or who did the reporting. Fever is not defined by a temperature threshold in the abstract.
Evidence that argues against this (1)
- The review's own ranked list places recurrent infections at 49%, above fatigue, which suggests that in many included cohorts fever and infection were recorded as the same underlying event and are not independent features. Salzer WL et al. 2015
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
A PubMed search for fever at presentation in infant leukaemia returns 8 records, all of which are case reports or treatment-phase infection complications on title check, and none of which reports a frequency of fever at diagnosis in infants.
NCBI E-utilities esearch against PubMed: recorded queries and record counts, run 2026-09-19 · Q7-clinical-infant beat (this graph) · NCBI E-utilities 2026What this evidence cannot show
A narrow query restricted to infant and leukaemia in the title. Broader queries would retrieve more but with worse precision; the wider presentation query returned 17 records and also contained no infant cohort reporting sign frequencies. A frequency reported only inside a paywalled table is unreachable from any abstract-level query.
Evidence that argues against this (1)
- Six of the eight hits concern infection during treatment, so the query may be retrieving the wrong phase of illness rather than establishing that the diagnostic-phase number is absent. Leung AWK et al. 2024
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In COG AALL0631, de-intensifying induction and adding supportive care guidelines lowered sterile-site infections during induction from 15 of 26 infants (57.7%) to 24 of 123 (19.5%), p = 0.0002, with a significantly lower rate of Gram-positive infections.
Decreased induction morbidity and mortality following modification to induction therapy in infants with acute lymphoblastic leukemia enrolled on AALL0631: a report from the Children's Oncology Group. · Salzer WL, Jones TL, Devidas M, Dreyer ZE, Gore L, Winick NJ, et al. · Pediatr Blood Cancer 2015What this evidence cannot show
Sequential cohorts, not randomised: cohort 1 (n = 26) and cohort 2 (n = 123) differ in calendar time as well as in regimen, so secular improvement in supportive care is not separable from the protocol change. The small first cohort makes its 57.7% imprecise.
Evidence that argues against this (1)
- Because the two arms are sequential rather than randomised, the entire effect could in principle be secular improvement in paediatric intensive and antimicrobial care over the intervening years. Leung AWK et al. 2024
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Across the whole of CCCG-ALL-2015 therapy in 134 infants, sepsis occurred in 44.8%, severe pneumonia in 17.9% and invasive fungal infection in 9.0%, with a cumulative incidence of fatal infection of 13.5% (95% CI 7.7-19.3).
Outcome of infants with acute lymphoblastic leukemia treated with the Chinese Children's Cancer Group Acute Lymphoblastic Leukemia 2015 study protocol. · Leung AWK, Cai J, Wan Z, Qin J, Fang Y, Sun L, et al. · Haematologica 2024What this evidence cannot show
Whole-treatment rather than induction-only figures, so they are not comparable with the AALL0631 induction numbers without care. Single-country cohort with documented under-enrolment of high-risk infants. Read in the open-access full text at PMC11290514.
Evidence that argues against this (1)
- These rates are considerably higher than the induction rates reported after supportive-care intensification in AALL0631, and the authors themselves attribute this to supportive care rather than to disease, so it is a health-system measurement as much as a biological one. Salzer WL et al. 2015
- refutes →Fever and infection: the reason most of these babies see a doctor, and the reason some of them die
the 53% fever figure is ascertained in a population that can report and be asked; it does not transfer to infants
fatigue is a reported symptom with no pre-verbal equivalent
how presenting features are ascertained in the first year of life
the infection numbers that do exist are treatment-phase, not presentation-phase
queries A3-A6 return no infant cohort reporting sign frequencies