Introduction Lumbar puncture (LP) is recommended in febrile young infants evaluated for serious bacterial infection.

Purpose To assess the utility of age, fever, white blood cell (WBC) count, erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) as screens for performing LP in febrile infants without a source of infection.

Methods Retrospective study of all febrile infants (rectal temperature >38°C), aged 29–180 days, admitted between 2001 and 2007, for evaluation of fever without source. All infants underwent full sepsis evaluation including LP for cerebrospinal fluid (CSF) analysis and culture. CSF pleiocytosis was defined as >8 WBC/mm³ (Philadelphia criteria).

Results Of 328 infants enrolled, 65 (19.8%) had CSF pleiocytosis, including four infants (1.2%) with bacterial meningitis. There were no differences between the infants with and without CSF pleiocytosis, in relation to age (50 days; interquartile range (IQR) 38–83 vs 60 days; IQR 43–89; p = 0.96), height of fever (38.6°C; IQR 38.2–39 vs 38.5°C; IQR 38.2–39; p = 0.565), ESR (20 mm/h; IQR 10–36 vs 22 mm/h; IQR 10–40; p = 0.58) and CRP (0.5 mg/dl; IQR 0.3–1.9 vs 1 mg/dl; IQR 0.4–2.4; p = 0.17). Infants with CSF pleiocytosis had a higher WBC count (13 400/mm³; IQR 8800–18 700 vs 12 500/mm³; IQR 8400–18 100; p = 0.003). However, receiver operator characteristic curve analysis revealed that the WBC had only moderate ability in predicting CSF pleiocytosis (area under the curve 0.617, 95% CI 0.562 to 0.670) among febrile infants.

Conclusions The decision to perform LP in febrile infants 1–6 months of age should neither be based on the age of the infant nor the height of fever, nor rely on commonly used laboratory tests such as WBC, ESR and CRP.