No abstract

Aims To determine if previously well infants (<1 year) admitted with undifferentiated febrile illness with a negative blood culture (BC) at 36 h could have antibiotics stopped and be safely discharged if clinically well. Current policy is to wait for a 48 h BC result to inform discharge, though our local neonatal unit has been stopping antibiotics at 36 h if BC -ve and well. Reduction in admission length and cost are estimated. Method Retrospective case-note review of infants on whom a BC was taken at A+E admission for an 18 month period from October 2012. Infants with any significant previous illness were excluded. Forty BC +ve infants (notes missing in 6) and 807 BC -ve infants were identified. Time BC became positive was recorded (BacTec). Clinical and laboratory features suggestive of bacteraemia were identified. Using a case control study design all 34 positive, and 68 BC negative infants were compared. Two paediatric trainees (AM, GH) using a proforma, independently assessed whether BC negative infants could have been safely discharged at 36 h. Discordant (n = 3) results were moderated (JB). Results No significant BC became positive after 30 h of culture. 29/68 (43%) BC -ve children were assessed as being able to be discharged after 36 h, and mean admission length for these was 57 h. Given children are not normally be discharged between 21.00 and 09.00, we estimate children could have been discharged 10 h earlier on average, and for many avoided an additional night in hospital. In this Trust this could have saved 96 bed days (£47,000) per year, assuming timely discharge, and our retrospective assessments correct regarding safe discharge. Conclusion Many previously well, febrile infants may be safely discharged home following a negative blood culture at 36 h with significant potential benefits for both hospital and family. The next step will be to use a negative 36 h blood culture result to stop antibiotics in well infants and audit safety of this policy, with a view to facilitate earlier discharge.