Antimicrobial Use Guidelines in the Setting of Multidrug-Resistant Gram-Negative Bacilli (MDR GNB) Resistance: Selecting a Second Gram-Negative Agent

Background: Guidelines for antimicrobial stewardship programs recommend that institutions develop antimicrobial (abx) use guidelines based on local microbiology and susceptibility (S) patterns. With considerable gram-negative (GN) resistance, the optimum empiric regimen is critical, but may not be readily apparent. At our institution, changing susceptibility patterns have brought into question our empiric broad-spectrum GN combination of choice (piperacillin/tazobactam + tobramycin). 

Objective: Reevaluate hospital-specific empiric abx guidelines based on local data to determine which agent or combination of agents would most adequately cover healthcare-associated GNB while minimizing overuse of broad-spectrum abx.

Methods: In a tertiary care academic center in New York, inpatient microbiologic data for 2009 were reviewed, including GNB S to GN agents. Most common GNB were evaluated for % S to each agent individually. Further evaluation was performed for isolates that were resistant (R) to primary broad-spectrum GN agents (piperacillin/tazobactam [P/T], cefepime [CPM], and meropenem [MERO]) to determine S to secondary agents (gentamicin [GENT], tobramycin [TOB], and levofloxacin [LEVO]). Descriptive statistics were performed.

Results: Most common GNB isolated from any source was E. coli (EC), followed by K. pneumoniae (KP), P. aeruginosa (PA), and E. cloacae (ENT). For all 4 GNB, MERO surpassed CPM which surpassed P/T in % S, with differences tending to be largest between P/T and CPM (p<0.001 for all GNB). Of the secondary agents, GENT was equivalent or superior to TOB except in the case of PA for which it was inferior (p<0.001). LEVO was inferior to GENT (p<0.001) for all GNB except ENT. When S to combinations of agents were evaluated for PA, the significant benefit of TOB over GENT was eliminated. Differences between % S of PA to P/T or GENT and P/T or TOB declined to 2% (p=ns). The difference between LEVO S compared to GENT also became less significant when evaluated in combination with P/T, although it remained lower for KP and PA (p<0.001, p=0.005 respectively).

% Susceptibility of GNB in Hospital Isolates, 2009

	P/T	CPM	MERO	LEVO	GENT	TOB		P/T or GENT	P/T or TOB	P/T or LEVO
E. coli (n=2607)	94	98	100	68	86	85		98	98	97
K. pneumoniae (n=966)	76	84	84	81	88	82		89	83	83
P. aeruginosa (n=593)	72	79	82	68	84	91		84	86	77
E. cloacae (n=251)	78	95	97	93	94	95		94	94	92

Conclusions: With significant local GNB resistance, hospital abx guidelines can be evaluated based on S to combinations of GN agents as well as individual agents. For our top GNB, the combination of P/T with GENT led to 84-98% S which may be a reasonable and superior alternative to CPM for empiric treatment of hospital-acquired infections due to MDR GNB. When significant renal dysfunction is present, LEVO may be used but may be suboptimal for PA and KP.