Evaluation and treatment of neonates with suspected late-onset sepsis: a survey of neonatologists' practices.

OBJECTIVE To ascertain current diagnostic and treatment practices for suspected late-onset sepsis in infants in neonatal intensive care units (NICUs) and identify areas that may benefit from clinical practice guidelines. METHODS During June 2000, we conducted a multicenter survey of neonatologists and infection control professionals regarding practices related to late-onset sepsis in NICUs at children's hospitals participating in the Pediatric Prevention Network. RESULTS Personnel at 35 hospitals with NICUs completed surveys; 34 were infection control professionals, and 278 were neonatology clinicians, primarily attending neonatologists or neonatology fellows. At these facilities, coagulase-negative staphylococci (CoNS) were the most frequent blood culture isolate from infants with late-onset sepsis accounting for 54% of bloodstream infections. When late-onset sepsis was suspected, 83% of clinicians drew only 1 blood culture when no central venous catheter was present or when a central vascular was present with no blood return. Thirty-two percent obtained 1 or more C-reactive protein concentration determinations. Sixty percent of clinicians prescribed a vancomycin-containing regimen for a 900 g, 3-week-old infant with suspected late-onset sepsis. The presence of a central venous catheter or shock increased empiric vancomycin use. The presence of methicillin-resistant Staphylococcus aureus in the NICU did not increase vancomycin use, but a vancomycin restriction policy decreased empiric vancomycin use. Clinicians at an individual NICU tended to have similar empiric antibiotic-prescribing practices: in 29 (83%) of 35 centers > or =75% of respondents had similar practice with regard to prescribing a vancomycin-containing regimen for empiric therapy. Forty-seven percent to 85% completed a full course of antimicrobials when a single blood culture was obtained and grew CoNS, but a significantly lower percentage of respondents (22%-47%) completed a full course when 1 of 2 blood cultures obtained grew CoNS. Eleven percent of respondents removed an umbilical catheter at the time of suspected sepsis, but fewer than 5% removed a nonumbilical central venous catheter for suspected sepsis. Most (> or =61%) retained a nonumbilical catheter despite documentation of CoNS bacteremia. CONCLUSIONS Neonatologists varied in management of suspected late-onset sepsis, particularly that caused by CoNS. Procedures to prevent CoNS-positive blood cultures and to differentiate CoNS contaminants from pathogens are needed. For safely decreasing vancomycin use in NICUs, clinical practice guidelines should be developed, implemented, and evaluated. The guidelines should include optimal skin antisepsis and catheter disinfection before obtaining blood for culture, obtaining 2 blood cultures and using adjunctive tests and information to help differentiate contaminants from pathogens, and restriction on empiric vancomycin use.