SCCM Resource Library-As Broad as We Need to Be: Empiric Antimicrobial Drug Selection and Clinical Decision-Making at the Bedside

Video Transcription

Thank you for joining me for this talk. As broad as we need to be, empirical antimicrobial drug selection and clinical decision-making at the bedside.  I'm Mojda Hevner, I work as a clinical pharmacy specialist in the medical intensive care unit at University of Maryland Medical Center.  The question of empiric antibiotic drug selection comes up quite frequently for us, so I truly enjoyed putting together this talk.  In this brief overview, I will discuss the epidemiology of drug resistance in the ICU and the clinical implications of drug-resistant infections on our clinical decision-making.  We will review the empiric antibiotic recommendations that are provided by international guidelines for managing the sickest of our infected patients.  In walking through this, we'll identify patient risk factors for drug-resistant infections, including the decision to empirically cover for fungal infections.  I will lead us through these objectives by setting up the epidemiologic landscape, delving into patient-related factors, and then putting it all together into a review of programs  that have been established to systematically identify optimal empiric antibiotics in the ICU population.  My colleague, Dr. Forhan, will then go further into new options for treating multi-drug-resistant  infections, and Dr. Maves will take us to the final frontier of biomarkers and rapid molecular tests that facilitate optimal empiric antibiotic selection.  Let's start off with an introduction of an unfortunate patient case that likely at least some of us have encountered at some point.  This patient comes in as a transfer from an outside hospital. It's a bit of a whirlwind, and the team is trying to get all the pertinent history, microbiological  data, and imaging, but at the moment, very little information is available. Unfortunately, he is rapidly deteriorating when he arrives at your unit.  He receives broad-spectrum antibiotics for septic shock with piperacil and tazobactam and vancomycin.  The vancomycin is infused first, so the piperacil and tazobactam is not started until about two and a half hours after he arrives.  After a few hours of trying to stabilize the patient, you discuss as a team and decide it may be prudent to administer a dose of aminoglycoside, just in case.  When blood culture results and sensitivities eventually come back, it turns out the patient had grown a multidrug-resistant pseudomonas that was resistant to both drugs you gave  him for gram-negative activity. We missed the boat, but what went wrong? It's vital before we dive too deep into the specifics of empiric antibiotic prescribing  to understand the global prevalence of infection in ICU patients. Point prevalence studies have been conducted at national and regional levels and internationally  over time, and we have a strong sense of infection in ICU patients and associated outcomes. The most recent and most robust international point prevalence study was conducted during  a 24-hour period in 2017, taking place in 88 countries across 1,150 centers. The most highly represented countries in this study were China, the United Kingdom, and  Brazil, and I've highlighted some of the other highly represented countries in this study as well.  We may know this as the EPIC-3 study, Extended Study on Prevalence of Infection in Intensive Care 3.  About 70% of the 15,165 patients were receiving at least one antibiotic in the study day, and 50% were receiving antibiotics therapeutically.  44% of the patients had acquired their infection in the community, with about 35% acquired  in the hospital, followed by 22% in the ICU. ICU length of stay was 1.5 times longer for patients with hospital-acquired infections  and threefold higher for ICU-acquired infections compared to those with acquired infections in the community.  Mortality trended up in the same direction, although the difference between the groups was not quite as dramatic there.  In patients with at least one positive microorganism identified, about two-thirds had a gram-negative  bacteria, about one-third of patients had gram-positive bacteria, and less commonly, fungi were isolated in 16% of patients, with a minority of patients having anaerobes, viruses,  parasites, or other bacteria, or mixed flora. ICU length of stay was longest for patients with fungal infections, followed by gram-negative organisms.  Not surprisingly, patients with fungal infections had the highest ICU mortality at about 32.4%.  Genetic of infection is also quite important when understanding the epidemiologic landscape.  Consistent with previous studies, respiratory tract infections represented nearly two-thirds of patients.  The abdomen, bloodstream, and genitourinary system were other common sites of infection in this study. On to the next important question.  What is the prevalence of resistant bacteria in ICU patients? Well, the answer is not so straightforward because it depends on geography.  In North America, a study of critically ill patients with pneumonia reported a rate of about 14% with MDR infections.  Comparatively, a study of critically ill patients with bloodstream infections worldwide showed a rate of about 48%, which included a 20% rate of extensively drug-resistant organisms  and 0.5% pan-drug resistant. It is important also to note that in the last study, there were stark differences seen between countries.  As an example, Australia reported a rate of 8% MDRs, while some countries in Eastern Europe, including Turkey, Greece, Croatia, and Serbia had rates of up to 80%.  One thing is clear across the board. Gram-negatives seem to predominate the MDR landscape. Among gram-negative bacteria, organisms that are most likely to be multi-drug resistant  include extended-spectrum beta-lactamases, pseudomonas, acinetobacter, and cenotrophomonas.  Gram-negative resistance includes vancomycin-resistant enterococci and methicillin-resistant Staphylococcus  aureus, but their incidence appears to be decreasing, as seen by the trend with increasing gram-negative resistance.  We all know by now that administration of inappropriate initial antibiotics is associated with increased mortality, which is at the forefront of our minds when we make these  decisions. This also suggests an association with longer hospital stay and greater hospital costs.  At the same time, overly broad therapy, especially overly broad therapy that is prolonged, can  lead to selective pressure for nosocomial infections with multi-drug resistant or fungal organisms as well as consequential Clostridium difficile infection.  It's important as well to consider the adverse reactions associated with antibiotic agents, especially in a particularly vulnerable host.  Overtoxicity, ototoxicity, cardiac, and CNS side effects come to the top of my mind with  drugs and drug classes like vancomycin, aminoglycosides, polymyxins, macrolides, quinolones, and carbapetums.  Overly broad empiric therapy can also lead to increases in drug expenditures. So how do we make sure we get it just right?  Well, in order to hit the mark, we need to focus on specific factors that will help us make the most targeted decision.  The patient factors that are the most important, how sick is the patient? Is this septic shock? If so, we have even less capacity to make a mistake by choosing empiric therapy that  is too narrow. I will dig deeper into the specific recommendations from the surviving sepsis campaign guidelines in the next slide.  We should also always think even early on about suspected site of infection using infectious disease society guidelines wherever possible.  By identifying a likely site or source, we can better identify empiric therapy based on the most common pathogens.  We can also ensure that we are choosing agents that have activity and concentrate well at that suspected site.  It's also extremely important to look into the patient's unique microbiological history. Has the patient had previous infections with unusual or resistant pathogens?  In general, looking back at six months is most helpful for this guidance, but it may be prudent to be aware of the past five years of infectious history.  If the patient has a history of resistant organisms, especially gram-negatives that are MDRs, it may be necessary to utilize two agents with specific mechanisms of action  that are different in order to increase the chances that at least one of the agents has sensitivity.  As usual, of course, we will want to know if the patient has had severe allergies to any antibiotics in the past and choose appropriately or monitor the patient if we deem there's  a low risk for cross-reaction. And finally, environmental factors should always be on our radar at an institutional or even ICU level.  We should consider local resistance patterns through antibiogram data or from other sources.  This is especially relevant when there is limited information available about the patient, as in the case vignette I presented earlier that we'll want to consider.  The surviving sepsis campaign guidelines for management of sepsis and septic shock focus quite a bit on the timing of appropriate antibiotic therapies.  Based on the scope of this talk, I'll bring your attention to any pertinent mentioned considerations for empiric agents rather than the timing, although I think we all know the  ideal timeline for administration of appropriate antibiotics is really one hour of recognition.  When a patient is in shock, we have little time to waste and little time to obtain data to guide our decisions.  Therefore, the surviving sepsis campaign recommendations are consistent with the general approach we  just reviewed of considering the patient's specific risk for unusual or resistant pathogens.  As I mentioned previously, we are most likely to be dealing with a gram-negative infection in the ICU.  The guidelines suggest dual gram-negative coverage empirically for sepsis and septic shock when the patient is considered high risk.  Factors that can guide the decision include proven infection or colonization with an MDR organism in the past year, local prevalence of antibiotic-resistant organisms, hospital  acquired as opposed to community acquired, broad-spectrum antibiotic use in the past 90 days, concurrent digestive system decontamination, travel to a highly endemic country in the  past 90 days, and hospitalization abroad within the preceding 90 days. An additional consideration is for duplicate gram-negative coverage, especially if the  beta-lactam option is considered to have less than 90% susceptibility for local pseudomonas strains.  Additional guidance on the approach to MDR gram-negatives has been provided by the Infectious Disease Society of America recently on the three most concerning groups of resistant  gram-negatives. The IDSA guidance focuses on extended-spectrum beta-lactamase-producing enterobacterioles  or ESBLE, carbapenem-resistant enterobacterioles, pseudomonas aeruginosa with difficult-to-treat resistance.  They note that the dramatic rise of ESBLE infections in the United States in the past five years has been primarily due to increased community-acquired infections.  This is an important consideration as we may see increasing resistance in community-acquired  infections relative to hospital-acquired infections, which is actually different from what we've seen previously in trends.  With regard to empiric antibiotic therapy specifically, the guidance is largely consistent with what we've already discussed.  However, the recommendation is to look at antibiotic exposures in the past 30 days and consider coverage with gram-negative agents from a different class that offer similar  spectrum of coverage. For example, if the patient was exposed to a course of PIP4-cell and tezobactam two weeks  ago for a ventilator-associated pneumonia, consider a choice of maripenem empirically for new pneumonia.  This guidance also provides a suggestion to consider whether the patient is immunocompromised when choosing empiric antibiotics. Specific guidance beyond that is not provided.  Let's shift gears back to the surviving sepsis campaign guidelines and now look at the recommendations for empiric gram-positive coverage.  The guideline points out that only 5% of culture-positive infections in ICU patients end up being MRSA  and risk factors to consider for MRSA infection would include MRSA infection or colonization previously, recent intravenous antibiotics, a history of recurrent skin and soft tissue  infections or chronic wounds, presence of invasive devices, hemodialysis, recent hospital admission, and severity of illness.  Therefore, if a patient has septic shock, you may want to consider empiric vancomycin with a rapid de-escalation once data becomes available.  Among undifferentiated patients with pneumonia or sepsis, broad-spectrum coverage, including  an anti-MRSA agent, was associated with actually higher mortality, especially in patients without MRSA colonization or history.  This highlights the importance of peeling back gram-positive coverage quickly if it  is started empirically and resulting microbiologic information suggests that it may not be indicated.  The surviving sepsis campaign also provides recommendations that patients with high risk for fungal infection may be considered for empiric antifungal therapy.  As many of you likely know, the empiricus randomized clinical trial, which looked at  mycofungin empirically for patients with ICU-acquired sepsis based on Canada colonization, multiple organ failure, failed to show a difference with mycofungin versus placebo.  However, this study did not look at patients who may be at higher risk. The guidelines highlight an example of a patient who has febrile neutropenia but does not defer  VESS after a few days of broad-spectrum antibiotics. It would be absolutely reasonable to consider empiric antifungal therapy in this patient  given risk factors, exposure, and clinical course. The guidelines suggest using empiric therapy in patients with prior colonization or infection  with fungi, prior exposure to prophylactic or therapeutic antifungal therapy, while considering patient-specific factors for fungal infection.  I'd like to highlight in particular the risk factors for Canada sepsis that they provided,  which include Canada colonization, beta-D-glucan, neutropenia, immunosuppression, severity of  illness, length of stay in the ICU, invasive devices, GI surgery, renal failure, hemodialysis, burn injury, and prior surgery.  So perhaps future studies can use these risk factors in combination with the inclusion criteria in the empirica study to identify a narrower population that can have the greatest  benefit from empiric antifungal therapy. Other important considerations for empiric antifungal therapy in sepsis would be to think  about risk factors for endemic yeasts and mold in certain patient populations, notably  patients with solid organ transplantation, stem cell transplantation, and high-dose corticosteroids may require even broader coverage for these types of invasive fungal infections.  The Surviving Sepsis Campaign makes no specific recommendations for empiric antiviral coverage in sepsis and septic shock.  However, a point prevalence study indicated that viruses are rarely the cause primarily of sepsis and make up less than 4% of infections in this population.  The question that frequently arises is to what degree the primary viral infection is  causing sepsis versus a new secondary bacterial infection leading to the severity of the patient's presentation.  Similar to empiric fungal coverage, there are certainly patient populations who may be particularly vulnerable to viral infections.  We do know that viruses can cause a clinical presentation of sepsis, especially in the early course of disease and especially in a vulnerable immunocompromised host.  We've discussed a lot of the epidemiological and clinical aspects of the decision, but  I also think it's worth mentioning the psychological drivers behind empiric antibiotics as well.  The decision, especially in the setting of septic shock in a patient who is rapidly crashing,  is made under high stakes and requires a balance between antibiotic stewardship and not wanting to miss the boat and cause the patient harm from undertreatment.  By understanding how decisions are made, we can consider improving the process through education or development of practice frameworks that drive optimal choices.  This is a bigger question too of patients who exhibit traditional signs of infection that may be caused by other conditions, but for now we're focused on the decision of how  broad to go and not even the decision about whether to start antibiotics. A survey was conducted among ICU practitioners, which included physicians at all different  training levels, consultants such as infectious disease physicians, and non-physician clinicians such as ICU pharmacists.  A few themes rose to the top of the survey, and there appear to be four main factors driving the decision for empiric antibiotics.  Clinician capability, which in and of itself includes experience, training, knowledge, and skills of the practitioner.  Past experiences, so a positive experience in which a patient recovered versus a negative experience involving a lawsuit or being scolded or a patient that died.  Medical context, meaning the decision is being made during the daytime workday versus off hours.  Is there input available from other team members such as the pharmacist or the ID consultant? And policy and practice norms. Does the hospital have a guideline?  Do they have antibiotic restrictions? The result is the prescribing decision. Based on where the clinician falls in each of these parameters, they may either be weighted  more heavily on the side of prescribing antibiotics more aggressively or conservatively. In reviewing this framework, you can see how a program can be implemented to drive decisions  in a desired direction by providing education to boost clinician capability, implementing additional support, or by beefing up the policy and practice norms.  But either way, these can be factored into programs that provide decision support capabilities into existing workflows.  Now that we discussed how decision support abates the psychological dilemma, let's talk more about these frameworks and how they can be built into our existing workflow.  Quite generally speaking, decision support tools can be embedded via different mechanisms into the decision to start antibiotics in the first place.  So is this really infection or is this something else? More in line with the scope of this talk, they can help us decide which antibiotics  to start, and then even more specifically, how to dose those antibiotics based on patient specific factors.  Several studies have shown improved outcomes associated with using decision support tools for empiric antibiotics, but I will highlight a few that stood out to me as particularly  notable. As early as 1994, a study compared antibiotics suggested by automated antibiotic consultant computer decision support tool to standard of care prescribing by physicians.  The tool was seamlessly connected to the patient's information, which made it easy for the physicians  to use and resulted in most of them saying that they would recommend the program to others.  The antibiotic consultant recommendation and appropriate antibiotic regimen was appropriate 94% of the time compared with 77% of the time when physicians made the decision on their  own. Of note, this was not an ICU study, but it was one of the first programs and therefore important that we mention it here.  And then next in 1999, some of the same investigators studied the impact of their computer assisted  decision support specific to the shock trauma intensive care unit at the same hospital. In this iteration of the study, physicians followed computer suggested regimens for 46%  of patients, but went with the suggested dose and interval for 93% of those orders. In a pre post evaluation, the computer assisted approach resulted in improved drug selection,  reductions in adverse drug events and costs. Fast forward to 2021 and a group in Toronto published some interesting data on their quality  improvement efforts for antibiotic prescribing. The intent of their work was to facilitate de-escalation of therapy within the empiric  antibiotic window in hopes of potentially improving the adequacy of antibiotic coverage.  This group implemented a pharmacist led prospective audit and feedback intervention that provided  early suggestions to the prescribing clinician when a patient's empiric antibiotic therapy  did not align with the patient's microbiologic history for MRSA colonization, previous ESBLs, and other resistant gram negatives.  The pharmacist had access to an antimicrobial stewardship database that integrated patient data to identify the bug drug mismatches.  Patients throughout the hospital were eligible for evaluation through this initiative, so it was not localized to the critically ill patients there.  The result was improved prescribing, shorter time to concordant therapy, and an increase in the percentage of patients receiving appropriate therapy based on their culture results.  Another iteration of this study was specific to patients with gram negative bacteremia. These patients were understandably sicker and more in line with the scope of our talk today.  With a similar pharmacist led and computer database bug drug mismatch approach to the previous study, the results of this focused initiative showed a 1.7 fold likelihood of  patients to have their empiric therapy de-escalated and a higher proportion of patients on the narrowest adequate therapy at the time that cultures were finalized.  Let's revisit the patient case earlier, but with a modified scenario that many of us have also likely encountered before.  The outside hospital mentions in the transfer handoff that the patient has completed a course with piperacillin-tazobactam for hospital acquired pneumonia due to pseudomonas about  two weeks ago. Based on that information, you decide to initiate meropenem empirically along with vancomycin given that the patient is high risk for gram positive resistance too.  The patient grows multi-drug resistant pseudomonas that was fortunately sensitive to meropenem.  Shock resolves in about 24 hours and vancomycin is quickly peeled off based on rapid molecular diagnostic tests.  The patient is liberated from the ventilator in about six days and ends up going to a pulmonary rehab facility after two weeks at your hospital.  This is the beauty of getting it right with empiric antibiotics. So in summary, we are seeing increasing bacterial resistance in the last few decades and this  necessitates broad spectrum empiric antibiotic therapy, especially for the sickest patients.  However, providing overly broad empiric therapy can also result in patient and societal harm. So it's really important for us to find that perfect balance.  Empiric antibiotic therapy should be patient specific and consistent with local microbiological landscapes which we can find from antibiograms.  And decision support frameworks can support clinicians in making these decisions about empiric antibiotics and help us to find that optimal balance between stewardship and patient  centered care. Thank you so much for joining me and I hope you stay on with this session and continue with Dr. Forehand and Dr. Mabes' talks as well.

Video Summary

In this video, Mojda Hevner, a clinical pharmacy specialist, discusses the importance of empiric antibiotic drug selection in managing infections in the intensive care unit (ICU). She highlights the global prevalence of infection in ICU patients and the impact of drug-resistant infections on clinical decision-making. The video emphasizes the need for targeted antibiotic therapy based on patient risk factors, such as previous infections and antibiotic exposures, and the importance of considering local resistance patterns. Recommendations from international guidelines, such as the surviving sepsis campaign, are discussed, including the use of dual gram-negative coverage and empiric antifungal therapy in high-risk patients. The video also explores the psychological drivers behind the decision to start empiric antibiotics and the use of decision support tools to improve prescribing practices. Overall, the video stresses the need for a balanced approach to empiric antibiotic therapy that considers both efficacy and antibiotic stewardship.

Asset Subtitle

Infection, Pharmacology, Quality and Patient Safety, 2022

Asset Caption

Increasing rates of antimicrobial drug resistance in the community and in the ICU imperil our ability to care for patients with sepsis and septic shock. In this session, we will review the clinical features that clinicians can use to guide rational antimicrobial drug use, new and emerging diagnostic tests that help us make the right choice of drugs more rapidly, and the data for novel antimicrobial agents, including beta-lactam/beta-lactamase inhibitors and tetracycline derivates, that may support the care of infected patients in the ICU.

Meta Tag

Content Type Presentation

Knowledge Area Infection

Knowledge Area Pharmacology

Knowledge Area Quality and Patient Safety

Knowledge Level Intermediate

Knowledge Level Advanced

Membership Level Select

Tag Antibiotics

Tag Pharmacokinetics Pharmacodynamics

Tag Evidence Based Medicine

Year 2022

Keywords

empiric antibiotic drug selection

infections

intensive care unit

patient risk factors

antibiotic stewardship

clinical decision-making

	Society of Critical Care Medicine 500 Midway Drive Mount Prospect, IL 60056 USA Phone: +1 847 827-6888 Fax: +1 847 439-7226 Email: support@sccm.org	Contact Us About SCCM Newsroom Advertising & Sponsorship DONATE	MySCCM LearnICU Patients & Families Surviving Sepsis Campaign Critical Care Societies Collaborative	GET OUR NEWSLETTER
© Society of Critical Care Medicine. All rights reserved. \| Privacy Statement \| Terms & Conditions The Society of Critical Care Medicine, SCCM, and Critical Care Congress are registered trademarks of the Society of Critical Care Medicine.