Prescribing antibiotics is an essential component of initial therapy in sepsis. Early antibiotics are an important component of therapy, but speed of administration should not overshadow the patient-specific characteristics that determine the optimal breadth of antimicrobial therapy. Cultures should be drawn before antibiotic therapy if it does not significantly delay administration. Combination antibiotic therapy against gram-negative infections is not routinely required, and combination therapy involving vancomycin and piperacillin/tazobactam is associated with an increase in acute kidney injury. Emergency practitioners should be aware of special considerations in the administration and dosing of antibiotics in order to deliver optimal care to septic patients.
Key points
- •
Although early retrospective studies found decreased survival associated with each 1-hour delay in antibiotics, prospective studies have not validated these findings; the optimal time benefit of antibiotic delivery within the first 6 hours is not known.
- •
Inappropriate initial antibiotic therapy is associated with an increase in mortality; it is appropriate to start broad-spectrum antibiotic therapy that provides coverage of the most likely pathogens.
- •
The use of 2 antibiotics to double-cover gram-negative infections is not routinely required, especially if empiric therapy involves an antipseudomonal penicillin, cephalosporin, or carbapenems.
- •
Patients who receive both vancomycin and piperacillin/tazobactam may be at greater risk for acute kidney injury.
- •
The loading dose of antibiotics is the same in patients with and without renal dysfunction. Subsequent doses need to be adjusted in patients with renal dysfunction.
Introduction
The timely use of appropriate antimicrobials is a cornerstone therapy for patients with sepsis syndromes. Recent publications have sparked debate regarding how the selection and timing of antimicrobial therapy affect the outcomes of patients with severe sepsis and septic shock. The selection of empiric antibiotics for septic patients in the emergency department (ED) likely plays a significant role in patient mortality. Practitioners need to consider many patient-specific factors when tailoring an antibiotic regimen to a patient’s’ clinical presentation. Attention should be directed toward administering the selected antimicrobials in a timely manner. However, recommendations about the timing of administration are lacking: the Surviving Sepsis Campaign (SSC) guidelines have been criticized for their lack of timing advice founded on feasibility trials. Many EDs now stock empiric antimicrobial regimens within the confines of the department rather than in a central pharmacy to enhance the speed and appropriateness of initial therapy. These empiric antimicrobials are often chosen according to local susceptibility patterns and antibiograms. Regimens for appropriate coverage vary according to the suspected disease process, so speed and breadth need to be weighed against the need for a thorough diagnostic work-up to localize the source of infection. The addition of antiviral and antifungal coverage to antibacterial therapy must be considered in certain at-risk patients. Patient-specific characteristics such as renal function, weight, and allergies necessitate antibiotic substitution or dosing adjustments for many critically ill patients.
Introduction
The timely use of appropriate antimicrobials is a cornerstone therapy for patients with sepsis syndromes. Recent publications have sparked debate regarding how the selection and timing of antimicrobial therapy affect the outcomes of patients with severe sepsis and septic shock. The selection of empiric antibiotics for septic patients in the emergency department (ED) likely plays a significant role in patient mortality. Practitioners need to consider many patient-specific factors when tailoring an antibiotic regimen to a patient’s’ clinical presentation. Attention should be directed toward administering the selected antimicrobials in a timely manner. However, recommendations about the timing of administration are lacking: the Surviving Sepsis Campaign (SSC) guidelines have been criticized for their lack of timing advice founded on feasibility trials. Many EDs now stock empiric antimicrobial regimens within the confines of the department rather than in a central pharmacy to enhance the speed and appropriateness of initial therapy. These empiric antimicrobials are often chosen according to local susceptibility patterns and antibiograms. Regimens for appropriate coverage vary according to the suspected disease process, so speed and breadth need to be weighed against the need for a thorough diagnostic work-up to localize the source of infection. The addition of antiviral and antifungal coverage to antibacterial therapy must be considered in certain at-risk patients. Patient-specific characteristics such as renal function, weight, and allergies necessitate antibiotic substitution or dosing adjustments for many critically ill patients.
Timing of antimicrobial therapy
Sepsis has been defined as “life-threatening organ dysfunction due to a dysregulated host response to infection,” so it makes intuitive sense that the earlier antimicrobial therapy is instituted, the better outcomes patients will have. Kumar and colleagues found that each hour’s delay in antimicrobial administration was associated with a mean decrease in survival of 7.6%. Their multicenter, retrospective study included 2154 patients with hypotension as the start-time marker for septic shock. A second retrospective evaluation, this time of the SSC database, also showed increased in-hospital mortality with each hour’s delay in antibiotic administration. Similarly, a retrospective observational study in pediatric intensive care unit (ICU) patients showed an increased mortality risk with each hour’s delay from sepsis recognition to antibiotic administration. Note that time-to-intervention studies provide information primarily on correlation, not causation. Given the inherent limitations of retrospective studies and the complex variables that can confound time-to-intervention studies, caution is warranted when interpreting the results.
Prospective studies have failed to validate an increased risk of mortality with delayed antibiotics, as long as they are administered within 6 hours after the diagnosis of sepsis. A systematic review and meta-analysis found no significant mortality benefit of administering antibiotics within 3 hours after ED triage or within 1 hour after shock recognition in patients with severe sepsis and septic shock. These data do not suggest that early antibiotic administration is not important, but that the exact time of maximum benefit is yet unknown. Because sepsis is a complex spectrum of illness, many factors affect the risk of death and the length of stay in an ICU. The arbitrarily assigned markers of time to antibiotic administration that are currently used as quality metrics might not be supported by the evidence that emerges from future studies.
Appropriate antibiotic selection
After appropriate cultures are obtained, prompt initiation of broad-spectrum empiric antibiotic therapy is essential. Individualizing therapy in the ED is difficult, especially when empiric antimicrobials must be chosen without culture data. Reports suggest that 10% to 40% of initial empiric antimicrobial therapy is inadequate. Antibiotic selection should be driven by multiple factors, including the suspected site of infection, local susceptibility patterns, and patient-specific factors. The suspected site of infection suggests common potential pathogens and thus indicates the antibiotics that can achieve an adequate concentration at the site. Local susceptibility patterns, guided by institutional antibiograms, aid in the identification of antibiotics with the highest likelihood of coverage for suspected pathogens. Patient-specific factors include organ function, infection history, antibiotic exposure history, surveillance cultures, and allergies. There is little margin for error in patients with severe sepsis, so it is appropriate to start broad-spectrum antibiotic therapy that provides coverage of the most likely gram-positive and gram-negative pathogens. Recommendations for empiric antibiotic regimens for septic patients are presented in Table 1 .
| Suspected Source | Regimen | Comments |
|---|---|---|
| Sepsis of unknown origin |
| A broad-spectrum β-lactam antibiotic should be administered before the anti-MRSA coverage because of its faster infusion times and broader coverage of potential pathogens |
| Sepsis – Suspected pulmonary source |
|
|
| Sepsis: suspected meningitis |
| |
| Sepsis: suspected urinary source |
| Fluoroquinolones (eg, ciprofloxacin and levofloxacin) should be avoided if local antibiogram shows significant resistance to Escherichia coli (threshold >10% per IDSA guidelines) |
| Sepsis: related to central line | Treat as above for sepsis of unknown origin, tailor antibiotics based on blood culture Gram stain | |
| Sepsis: intra-abdominal source | Treat as above for sepsis of unknown origin; ensure anaerobe coverage is included in the regimen (eg, piperacillin/tazobactam, a carbapenem, or add metronidazole) |
Inappropriate initial antibiotic therapy is associated with an increase in the mortality. A single-center study of patients with bacteremia found a 34% difference (28% vs 62%) in the mortality among patients given inappropriate antibiotics on the first day of therapy and those given the right antibiotics. In one of the first observational studies on the adequacy of antibiotics specific to patients in the ICU, the mortality was significantly higher in those who received inadequate antimicrobial therapy initially. Further studies have confirmed these findings in patients with gram-negative sepsis, severe sepsis, and septic shock.
Although it is important to consider the antimicrobial stewardship principle of using the most narrow-spectrum agent possible for an infection, this practice does not apply in the management of sepsis until culture data are available. However, it is still important to use very broad-spectrum agents such as the carbapenems judiciously, reserving them for patients who have a high likelihood of multidrug-resistant (MDR) infections and in communities in which local susceptibility patterns warrant them.
A classification of infections as health care–associated infections had been used to describe patients at higher risk for MDR organisms. The most recent guidelines have removed this category due to its lack of ability to accurately describe patients who required broad spectrum antibiotic coverage. A lack of consensus regarding the power of these risks exists, and recent studies have delineated even more risk factors, including an immunocompromised state, hospitalization during the previous year, previous antibiotic therapy, age greater than 60 years, and Karnofsky index score less than 70. Because these risk factors might be overly broad in identifying patients with resistant organisms, Shorr and colleagues designed a clinical score that can be used to assess ED patients’ risk of harboring a resistant pathogen ( Table 2 ). In a cohort of 977 patients, resistant organisms, defined as methicillin-resistant Staphylococcus aureus , Pseudomonas aeruginosa , and extended-spectrum β-lactamases, were isolated 46.7% of the time. The risk score was higher in those with a resistant organism (median 4) than in those without a resistant organism (median 1) ( P <.001). A score greater than 0 had a high positive predictive value of 84.5% for resistant organisms. In addition, not all risk factors for MDR organisms are equivalent in their prediction of pneumonia caused by resistant pathogens in the community. Hospitalization in the preceding 90 days and residence in a long-term care facility were independent predictors of infection with a resistant pathogen in an observational prospective cohort of patients from the community who were hospitalized with pneumonia. Most of the studies using the health care–associated infection classification are limited to respiratory and bloodstream infections. Of note, the health-care associated pneumonia designation was removed from the updated hospital-acquired pneumonia and ventilator-associated pneumonia guidelines as there is increasing evidence that many patients defined as having HCAP are not at high risk for MDRA pathogens and do not account for underlying patient characteristics that are also important determinates for risk of MDR pathogens.
| Risk Factor | Point Value |
|---|---|
| Recent hospitalization (within 90 d) | 4 |
| Presenting from long-term care facility | 3 |
| Chronic hemodialysis | 2 |
| Admission to ICU within 24 h of ED evaluation | 1 |
Another important factor in the selection of empiric antibiotic therapy is the patient’s reported allergies. Between 15% and 20% of patients report an allergy to β-lactam antibiotics. Patients’ self-report of antibiotic allergy has been associated with antimicrobial resistance, increased length of stay, ICU admission, increased costs, and even death. To optimize therapy, a thorough allergy history should be documented, because some so-called reactions to antibiotics are frequently diagnosed inaccurately as allergies. The risk of cross reactivity with cephalosporins, particularly third- and fourth generation, and carbapenems is very low, so the risk/benefit of giving a septic patient potentially suboptimal therapy such as a fluoroquinolone versus a β-lactam with a low risk of cross reactivity should be considered carefully.
Empiric antifungal coverage is not indicated for most patients, because fungal infections are typically diagnosed late in the course of hospitalization. The mortality associated with candidal infections can reach as high as 60%. Risk factors for invasive candidiasis are categorized as host-related factors (eg, immunosuppressive disease or therapy, neutropenia, age, solid organ transplant) and health care–associated factors (eg, catheter use, total parenteral nutrition, recent surgical interventions, use of broad-spectrum antimicrobial drugs). The Candida Score developed by León and colleagues uses 4 variables for diagnosing probable candidal infection in non-neutropenic hosts: multifocal candida colonization (1), surgery (1), receipt of total parenteral nutrition (1), and clinical signs of severe sepsis (2). A score greater than 2.5 is associated with a greater than 7-fold increase in the likelihood of a documented candida infection. Notably, there was no association between the presence of a central venous catheter and candidal bloodstream infection. The combination of infrequent need for antifungal therapy and delayed culture results leads to delayed treatment and a high mortality among patients with candidal infection. Using the Candida Score in at-risk patients might assist in deciding whether fluconazole or an echinocandin should be ordered preemptively for a critically ill patient.
Irrespective of the conflicting data on time to antibiotic administration, the choice of antibiotics is vital. The use of appropriate antibiotics is associated with a lower mortality and a shorter ICU length of stay. Broad-spectrum antibiotics should be initiated as early as possible. Delays are common, with risk factors including not being seen by an emergency physician, not considering the diagnosis of sepsis initially, and delay of therapy while waiting for diagnostic tests to be performed. Several groups have implemented strategies to remove specific barriers to the timely administration of appropriate antibiotics. Kalich and colleagues implemented an antibiotic-specific sepsis bundle and reported a significant improvement in the initiation of appropriate initial antibiotic therapy for severe sepsis in the ED. Adding appropriate antibiotics to unit-based cabinets also reduced order-to-administration time for first doses.
The role of cultures
Obtaining appropriate blood or tissue cultures before initiating antibiotic therapy is important in identifying the causative organisms. The SSC recommends obtaining cultures before the start of antimicrobial therapy if it can be done without delaying therapy more than 45 minutes (grade 1C). In practice, 2 culture sets, each containing an aerobic and anaerobic culture bottle, should be drawn from 2 sites. Although the sepsis guidelines recommend that the volume of blood drawn into culture bottles should be greater than or equal to 10 mL, other infectious disease guidelines suggest that 20 to 40 mL of blood should be drawn, because the volume collected is directly proportional to the yield of pathogens. Skin antisepsis can be achieved with tincture of iodine or chlorhexidine gluconate, and the aerobic bottle should be filled first.
Fungal infection can cause delays in growth and difficulty in identification of organisms with routine blood cultures. In addition to cultures, the SSC gives a moderate recommendation for obtaining a 1,3B-D-glucan or antimannan antibody assay when fungal infection is suspected, noting that false-positive results are caused by colonization and advising that the utility of this test in critical care settings needs further investigation. Blood culture yield in sepsis syndromes is variable based on the underlying source of infection. Table 3 shows rates of positive blood cultures according to suspected site of infection.
