HCP Live
Contagion LiveCGT LiveNeurology LiveHCP LiveOncology LiveContemporary PediatricsContemporary OBGYNEndocrinology NetworkPractical CardiologyRheumatology Netowrk

Antibiograms for Optimizing Empiric Therapy: Use Them Wisely

Contagion, August 2018, Volume 3, Issue 4

An understanding of the limitations in their functionality and strategies for more appropriate applications are necessary for clinicians.

Antibiograms can be important tools that institutions and antimicrobial stewardship programs can use to develop empiric therapy pathways. In their simplest form, antibiograms present a cumulative susceptibility for a given organism to various antimicrobial agents over a time frame. This picture gives an idea of which agents are likely to sufficiently cover a pathogen and can subsequently drive empiric therapy recommendations for infections caused by those organisms. For example, in urinary tract infections (UTIs), empiric therapy primarily targets Escherichia coli. If an antibiogram shows that, at your institution, 70% of E coli are susceptible to ciprofloxacin but 93% are susceptible to nitrofurantoin, then nitrofurantoin makes much more sense as an empiric therapy recommendation for a UTI than a fluoroquinolone. Although it is true that this example suggests that nitrofurantoin is a better option than ciprofloxacin at targeting E coli, it is critical that clinicians appreciate the limitations of traditional antibiograms and, perhaps more importantly, strategies to improve their utility. The purpose of this review is to discuss strategies for improving the functionality of antibiograms and ultimately providing more appropriate empiric therapy.


One strategy that many institutions use to improve the functionality of their antibiograms is to analyze isolates from the hospital floor and the intensive care unit (ICU) separately. The logic is simple: Patients in ICUs represent a different population, having multiple risk factors for drug-resistant organisms, and are the population in whom the appropriateness of empiric therapy is most critical. Therefore, separating ICU isolates from floor isolates and directing empiric therapy for ICU infections, such as pneumonia, against the ICU isolates is pragmatic. Institutions most commonly do this for Pseudomonas aeruginosa and report their institutional antibiograms for this pathogen dichotomized into “ICU” and “floor” isolates. Institutions then often choose their empiric ICU pneumonia regimen to be the 2 agents with the best activity against this pathogen in isolates recovered from ICU patients. Although this approach is certainly preferred to the hospital-wide antibiogram, it is not without its own significant limitations. It assumes that for the treatment of pneumonia in the ICU, the only gram-negative pathogen of concern is P aeruginosa, and it fails to consider that the purpose of the second empiric agent is to provide activity when there is resistance to the first agent. Traditional antibiograms cannot address this issue because they do not assess cross resistance (ie, activity of the second agent in the setting of resistance to the first.)


Antimicrobial stewardship programs can overcome the limitations of traditional antibiograms through the development of an infection site—specific antibiogram to help determine the optimal empiric therapeutic regimen for a given disease state. The most common application of this is a combination antibiogram for the treatment of pneumonia in ICU patients. Instead of developing organism-specific antibiograms and using these to develop targeted empiric therapy recommendations based on the most likely pathogens, antibiograms are developed by including all isolates for a given site of infection over a time frame and assessing both the activity of monotherapy options and the likelihood of combinations having at least 1 active agent. Table 1 displays a sample site—specific antibiogram for all gram-negative respiratory isolates in an ICU over a given time frame. The first column consists of potential 1-drug options. As demonstrated in the table, different β-lactams were found to be active against 68% to 76% of gram-negative isolates in the ICU, making them unacceptable as empiric monotherapy regimens. As alluded to previously, in order to increase the likelihood of active empiric therapy, clinicians need to know what other drug is most likely to be active against a β-lactam–resistant isolate.

This is what makes combination antibiograms unique: They assess the activity of potential second agents (ie, fluoroquinolones or aminoglycosides) against β-lactam—resistant isolates. For example, of the 100 cefepime-resistant isolates listed in Table 1, 20 are susceptible to ciprofloxacin. Therefore 230 (210 cefepime susceptible and 20 cefepime resistant, ciprofloxacin susceptible) isolates are susceptible to at least 1 agent in the cefepime/ciprofloxacin combination. Repeating this process for all logical 2-drug combinations demonstrates that a β-lactam plus amikacin is the optimal gram-negative empiric regimen for hospital-acquired/ventilator-associated pneumonia in this ICU.

Although these data can provide critically important information for optimizing empiric therapy for pneumonia in ICU patients, it is also crucial for stewards to appreciate that what is good for one ICU might not be for another. For example, published data from the University of Michigan demonstrated that the optimal regimens, as determined by combination antibiograms, for the medical ICU (MICU) and surgical ICU were different.1


At the beginning of this article, we discussed how traditional antibiograms might be used to develop empiric therapy guidelines. The example provided indicated that an E coli antibiogram could be utilized to determine the optimal empiric regimen for UTIs given that up to 80% or more of UTIs are caused by E coli. Theoretically the same principle could apply to intra-abdominal infections (IAIs) given the frequency that E coli is isolated from these infections when aerobic organisms are present; however, there are 2 faults to this logic. First, E coli, although predominating, are not the only pathogens in these disease states. Secondly, the assumption that E coli in patients with UTIs and patients with IAIs are identical is inappropriate given that different patient populations present with these disease states and they have different risk factors for drug-resistant isolates.

To circumvent these issues, clinicians can develop syndrome-specific antibiograms, similar in many ways to the site-specific combination antibiogram described above. The syndrome-specific antibiogram takes all causative microbiology for a given disease state (eg, UTI or IAI) for a given time frame (eg, within 48 hours of admission to ensure that only community pathogens are included) and assesses what regimens will provide optimal coverage. Hebert and colleagues utilized this methodology to develop optimal empiric regimens for UTIs and IAIs at their institution.2 In their analysis, the investigators demonstrated that a one-size-fits-all method attacking E coli is insufficient for these 2 disease states and that the optimal empiric regimen, even from a purely aerobic coverage standpoint, differed between UTIs and IAIs. For example, ciprofloxacin was active against 62% of pathogens causing UTIs, whereas even when adding anaerobic coverage with metronidazole, it was only active against 37% of isolates in IAIs (P <.001). The authors described that the major driver of the difference was that the proportion of E coli in these disease states differed (~4-fold higher in UTIs), although differing susceptibilities of the E coli present in UTIs and IAIs could also play a potential role.


Combination antibiograms and syndrome-specific antibiograms offer significant advancements over traditional antibiograms that can be used by stewardship programs and clinicians to optimize empiric therapy. Furthermore, given that empiric therapy is syndrome based and not pathogen based, they are also more logical. However, clinicians should be aware that they still have significant limitations. For example, if we were to create a community-acquired UTI-specific antibiogram, as described in the above example, we might include all urine cultures obtained in the emergency department (ED) and within the first 48 hours of admission and develop an antibiogram to guide empiric recommendations. In reality, this is an overly simplistic grouping of isolates as doing so assumes that all of those cultures are created equally and that it is appropriate to group otherwise healthy community dwellers with those who have risk factors for resistant pathogens (eg, previous antibiotic exposure, recent hospitalization, presentation from a nursing home/long-term acute care facility). In fact, this is an inaccurate and inappropriate assumption and can lead to suboptimal empiric therapy recommendations.

To illustrate the concerns with this approach, Table 2 represents an internal analysis that was performed at Sinai-Grace Hospital in Detroit, Michigan, on all gram-negative UTIs presenting from the community over a 6-month period. The antibiogram was assessed first as all gram-negative organisms isolated over the time frame, similar to the above-described methodology, and then divided into those who presented from the community without risk factors for resistant pathogens, those classified as health care-associated (risk factors present for resistance such as recent hospitalization and previous antibiotic use), and those who came from nursing homes. The data indicate that if all isolates are grouped together, then multiple oral options (nitrofurantoin, ciprofloxacin) and more narrow-spectrum intravenous options (ceftriaxone and gentamicin) appear to have unacceptably high resistance rates that do not allow for routine empiric use. However, when split into different patient populations, the susceptibility differences— and, ultimately, the optimal empiric therapy recommendations as a function of patient type—become clearer. For example, even though nitrofurantoin and ceftriaxone appear to only be active against 80% and 85% of community isolates, respectively, when assessing the antibiogram as a whole, this is not the case when the antibiogram is stratified as a function of patient type. Once true community-acquired UTI patients (ie, those without risk factors for resistance) are separated out it becomes apparent that these agents are much more active (90% to 97%) and appropriate first-line therapies. Conversely, although tobramycin appears to be an appropriate first-line therapy, with activity against 90% of community isolates, a deeper dive demonstrates that if only assessing patients in the nursing home, susceptibility is much worse (60%) and thus tobramycin is an inappropriate empiric option for these patients.


The purpose of this article was to demonstrate the limitations of conventional antibiograms and suggest potential improvements that antibiotic stewards can make to further refine empiric therapy recommendations at their institutions. Clinicians should shy away from a one-size-fits-all mentality and incorporate flexibility into their guidelines. For example, a combination antibiogram of respiratory isolates from the MICU might mandate the empiric use of cefepime plus tobramycin for gram-negative coverage in septic patients with a suspected pulmonary source in that unit. However, when developing hospital-wide HAP guidelines, these MICU-based recommendations should not apply to non-ICU patients. Patients on the general ward often have less-resistant pathogens, are more hemodynamically stable, and usually have pneumonia as part of a differential diagnosis that includes both infectious and noninfectious etiologies. Although it is prudent to provide empiric antipseudomonal therapy to the right pneumonia patient, the risk : benefit ratio, of both toxicity and collateral damage, likely warrants avoidance of a second empiric gram-negative agent (ie, an aminoglycoside) in these more stable floor patients.

Similarly, although the antibiogram for nursing home— associated UTIs presented in this article suggests a very high rate of extended-spectrum β-lactamase (ESBL)-producing organisms, this does not mean that every nursing home patient who presents to the ED with mental status changes that are possibly secondary to UTI warrants empiric carbapenems. Risk stratification and the use of narrower-spectrum therapies while working up a stable patient is an important tool for an antibiotic steward. Importantly, however, the converse is also true and patient-specific factors need to be assessed for when broader therapy is warranted as well. For example, even if the empiric recommendation for pneumonia is cefepime, if a patient has a history of an ESBL-producing organism or just received 2 weeks of cefepime therapy for another infection, then a carbapenem is the more prudent empiric therapy choice for this patient. Ultimately, empiric therapy guidelines and patient-specific considerations will help clinicians balance the competing desires of appropriate empiric therapy and limit broad-spectrum antibiotic usage.

Dr. Pogue is an infectious diseases clinical pharmacist at Sinai-Grace Hospital and a clinical assistant professor of medicine at Wayne State University School of Medicine in Detroit, Michigan. He is a member of the Contagion® Editorial Advisory Board and an active member of the Society of Infectious Disease Pharmacists.


  1. Pogue JM, Alaniz C, Carver PL, Pleva M, Newton D, DePestel DD. Role of unit-specific combination antibiograms for improving the selection of appropriate empiric therapy for gram-negative pneumonia. Infect Control Hosp Epidemiol. 2011;32(3):289-92. doi: 10.1086/658665.
  2. Hebert C, Ridgway J, Vekhter B, Brown EC, Weber SG, Robicsek A. Demonstration of the weighted-incidence syndromic combination antibiogram: an empiric prescribing decision aid. Infect Control Hosp Epidemiol. 2012 Apr;33(4):381-8. doi: 10.1086/664768.