Discussion

The significant variation in antimicrobial regimens for LOS highlights the lack of standardization of LOS management across Canadian and American NICUs. Initiation of vancomycin or an anti-staphylococcal penicillin for empiric gram-positive coverage was both center- and case-dependent.

The safety of anti-staphylococcal penicillins rather than vancomycin to provide empiric gram-positive coverage during LOS evaluation in centers with low MRSA prevalence has been demonstrated internationally, without evidence of increased duration of bacteremia, infectious complications or mortality. ^{Reference Magers, Prusakov, Speaks, Conroy and Sánchez5–Reference Romanelli, Anchieta, Bueno E Silva, de Jesus, Rosado and Clemente7} Despite this evidence, institutional practices have not consistently changed to reduce unnecessary empiric vancomycin use. This presents a major target for antimicrobial stewardship interventions given the high frequency of suspected LOS and its significant contribution to empiric vancomycin utilization. Vancomycin use has major pitfalls; multiple doses may be required to achieve therapeutic levels, frequent drug level monitoring is indicated, and risks include acute kidney injury and the development of resistant organisms. Vancomycin may also be less effective than beta-lactams for treatment of MSSA, an important consideration given that MSSA is more common than MRSA infections in NICUs. ^{Reference Mukhopadhyay, Sengupta and Puopolo3,Reference McDanel, Perencevich and Diekema8} There is precedent for level 4 (regional) NICUs successfully reducing vancomycin prescribing, in one case utilizing a multimodal approach of education, clinical pathway development, 48-hour antibiotic time-outs and prospective audit with feedback derived using a quality improvement framework. ^{Reference Hamdy, Bhattarai and Basu9}

The reasons for low uptake of anti-staphylococcal penicillin agents across NICUs remain unclear, but contributory factors may include lack of a neonatal society guidance document and hesitancy to change practice. Other factors include a perceived need to empirically cover for CoNS, a major cause of LOS which tends to be resistant to anti-staphylococcal penicillin agents, concern for MRSA and Enterococcus spp., and risk of chemical phlebitis with cloxacillin intravenous administration. Cefazolin has similar constraints, limited blood-brain-barrier penetration and less familiarity as empiric LOS therapy.

Despite concerns for CoNS in critically-unwell patients, lack of empiric CoNS coverage has not been associated with adverse outcomes, as CoNS have not been associated with fulminant sepsis, typically grow within 24–48 hours allowing for rapid adjustment in therapy if indicated, and may reflect blood culture contamination. ^{Reference Mukhopadhyay, Sengupta and Puopolo3,Reference Sánchez, Moallem, Cantey, Milton and Michelow4} A retrospective review of infants with CoNS bloodstream infection admitted to 348 NICUs between 1997–2012 showed no mortality difference between babies treated with empiric versus delayed vancomycin therapy. ^{Reference Ericson, Thaden and Cross10}

The high rate of empiric third and fourth generation cephalosporin prescriptions was unexpected and warrants further evaluation. In centers where these agents are used, one could argue that additional gram-positive coverage, such as with vancomycin, may not be needed.

This survey focused largely on empiric gram-positive coverage, potentially confounding interpretation if broader spectrum antibiotics were chosen for gram-negative considerations. Additionally, MRSA prevalence was unavailable to about one-third of respondents, potentially affecting reported vancomycin use. Survey distribution was mainly to academic centers, limiting generalizability. Practice variations within a NICU would not be captured in this survey, and the small sample size limited more in-depth statistical analysis.

This survey provides insight into the wide variation in empiric antimicrobial prescribing practices for LOS in Canadian and American NICUs, thereby serving as a basis to support antimicrobial stewardship programs in streamlining the approach to this issue. Reducing vancomycin use is an important antimicrobial stewardship target; prioritization of anti-staphylococcal penicillin agents for empiric LOS therapy in the NICU should be considered.