Patients will commonly present with symptoms suspected to stem from an infectious etiology and will require empiric antibiotic therapy while awaiting results of cultures and other testing. Empiric antibiotic selection can be accomplished using an escalation or a de-escalation strategy. In an escalation strategy, relatively narrow-spectrum antibiotics are initially prescribed. Antibiotic coverage is later expanded to cover a broader range of pathogens based on culture results or if there is no clinical improvement. De-escalation strategies initially prescribe very broad-spectrum antibiotics to provide empiric coverage for a wider range of organisms. Antibiotics are then narrowed based on culture and susceptibility results and clinical improvement, or discontinued if another cause for critical illness is found. In either strategy, once culture results are obtained, antibiotics are tailored to provide adequate coverage of the pathogen and the narrowest possible spectrum.

The need for appropriate cultures cannot be overemphasized. Every attempt should be made to obtain blood cultures prior to initiation of antimicrobial therapy (1), as well as preantibiotic specimens from other possibly infected sites (e.g., cerebrospinal fluid [CSF], urine, lower respiratory tract, pleural fluid). Viral studies and fungal antigen testing should be considered in the appropriate patient. While a respiratory viral diagnosis may not prompt additional therapy, identification could allow for antibiotic deescalation depending on the clinical context (2,3); however, for critically ill children diagnosed with a viral respiratory infection who do not have airway cultures, the possibility
of a secondary bacterial infection should be considered as coinfection is not uncommon (4).

Although aggressive antibiotic use in the PICU can often be justified, it is not without potential negative consequences involving both the individual patient and the larger community. Antimicrobials account for the more commonly seen drug-associated adverse events, ranging in severity from rash to anaphylaxis and including organ injury (renal or hepatic impairment) and drug-drug interactions. Overuse can also lead to antibiotic-associated diarrhea/colitis and superinfection with drug-resistant bacteria and fungi. Additionally, perturbation of resident flora within the individual’s microbiome may impact susceptibility to later infections (5). On the larger community level, there is concern for increasing antimicrobial resistance, which has been associated with antimicrobial exposure (6,7).

The rise of antibiotic-resistant bacteria has made empiric antibiotic selection more challenging. Organisms once only seen in nosocomial infections are now being acquired outside of the hospital and even without contact with the healthcare system. Community-acquired methicillin-resistant Staphylococcus aureus (CA-MRSA) is an increasing problem and is presenting as increasingly invasive infections. Gram-negative bacilli also pose a treatment challenge, particularly those with potential antibiotic resistance, such as Pseudomonas aeruginosa and extended-spectrum β-lactamase-producing species. In adults, there is awareness of these healthcare-associated infections, but there is much less appreciation for this type of infection in children, and indeed the pediatric literature discussing healthcare-associated infections often evaluates only hospitalacquired infections (19,20,21).

Given the importance of early appropriate empiric antibiotics, there are several approaches that have been proposed to ensure coverage even for antibiotic-resistant pathogens. First, one could prescribe very broad-spectrum antimicrobials to all children admitted to the PICU with suspected infection (22). This approach would make it highly likely that the empiric coverage would, indeed, cover the vast majority of pathogens that might be cultured. This would, however, expose many patients to the toxicities of broad-spectrum antibiotics and may increase drug costs. A second approach would be to stratify children based on their risk for potentially resistant pathogens, similar to what is recommended in adult guidelines. Those with minimal risk could be treated with narrowerspectrum agents targeting community-acquired/commonly susceptible organisms, and children felt to have increased risk for potentially resistant organisms could be empirically treated with broader-spectrum antibiotics. Commonly used risk factors for potentially resistant pathogens include:

A regimen for these patients might include MRSA coverage as well as coverage for gram-negative pathogens, including P. aeruginosa. For adults, there are published consensus guidelines that can aid risk assessment and antibiotic decisions (Table 88.1) (23,24,25). Unfortunately, no guidelines currently exist for children, and it is unknown whether application of adult guidelines is appropriate for the pediatric population.

Empiric antiviral or antifungal coverage will likely not be necessary for most PICU patients. However, patients with recent exposure to antibacterials, indwelling central venous catheters, or immune compromise may be at higher risk for fungal disease (26,27). Empiric antiviral therapy should be considered for critically ill patient with suspected influenza (28,29) and may also be appropriate in cases of encephalitis or after solid-organ or hematopoietic stem cell transplant (HSCT), or in the critically ill newborn where herpes simplex infection is suspected.

One of the most important tools intensivists have to aid empiric antibiotic selection is a local, unit-specific antibiogram summarizing the antibiotic susceptibilities of all pathogenic bacteria identified in their PICU (Fig. 88.1). Understanding which organisms are prevalent locally and which antibiotics are most effective against these organisms can enhance antibiotic decisions even after a patient’s risk for particular organisms has been determined. Knowledge of local unit and community microbiology can be useful in developing a unitbased ICU empiric antibiotic strategy. For example, units with high rates of Stenotrophomonas maltophilia may wish
to include coverage of this organism for all patients at risk for healthcare-associated infections.

To further address the issue of resistant bacteria, combination antibiotic therapy is a strategy that has emerged. This approach has been shown to improve mortality in critically ill adults with community-acquired pneumonia (6, 23,30). There are guidelines and evidence to support combination empiric therapy in HCAP, VAP, sepsis, and septic shock (31,32,33,34,35,36,37). The best approach to ensure coverage of resistant gram-negative organisms appears to be an antipseudomonal β-lactam or carbapenem combined with an aminoglycoside (AG) rather than fluoroquinolone (38). Importantly, once a pathogen is identified, de-escalation to monotherapy appears to be safe and preferred, provided there is clinical improvement (31,39,40). Another strategy that may be helpful to address antibioticresistant bacteria in an empiric antibiotic protocol is antibiotic heterogeneity (e.g., cycling) (41,42). In this approach, antibiotics for specific organisms (e.g., gram-negative bacilli) are rotated such that there are time periods where a particular antimicrobial is the preferred agent for all patients and other periods where that drug is restricted and another is preferred. Cycling does not improve antibiotic appropriateness but rather attempts to decrease the development of resistance. Antibiotic cycling has been proposed as a potential component of an antimicrobial stewardship program (43) and has mixed evidence as to its benefits in both gram-positive and gramnegative infections (44,45,46,47,48). At this time, recommendations cannot be made for or against cycling.

A final approach to resistant bacteria focuses on discontinuation. Specifically, it is important to consider the use of shorter antibiotic courses for infections in the ICU. Studies in adults suggest that shorter antibiotic courses, perhaps ≤7 days, may be as effective as longer courses while theoretically decreasing pressure for the development of antibiotic resistance, particularly in the setting of gram-negative infections (33). There is currently active ongoing research examining the use of biomarkers to safely shorten treatment courses without negatively impacting outcomes.

Another special situation that deserves mention is the infant or child admitted to the PICU with a viral respiratory infection. While a respiratory viral diagnosis may not prompt additional therapy, identification could allow for antibiotic deescalation depending on the clinical context (2,3); however, consideration of bacterial coinfection in critically ill children with viral respiratory infections is crucial. Serious bacterial infections in infants with bronchiolitis are quite uncommon outside of the ICU, and a recent Cochrane Review found no evidence to support antibiotic therapy for children with bronchiolitis across the board (49). However, in critically ill children with respiratory syncytial virus (RSV) bronchiolitis requiring mechanical ventilation, coinfection with bacterial pathogens may occur in 40%-50% of cases (4,50,51). Additionally, bacterial coinfections, particularly with S. aureus, are frequently seen in patients with influenza requiring ICU admission (52,53). For these reasons, cultures, including lower respiratory cultures, are justified from all children requiring intubation and mechanical ventilation for viral lower respiratory infections, along with empiric antibiotic coverage until culture results are known (54).

Specific laboratory markers of infection would be helpful as decisions are made regarding antimicrobial therapy. Identifying markers that are sufficiently specific for bacterial infection in the individual critically ill patient has been challenging. Markers, such as CRP and procalcitonin, have shown promise, but current knowledge is insufficient for such data to provide clear direction with treatment decisions (59,60). CRP values show significant overlap in bacterial, viral, and fungal infections. Absolute values may be less helpful than trends over time (61,62). Decreasing CRP levels may allow for more confidence in therapeutic decisions, such as discontinuation of antimicrobial therapy. Procalcitonin may be elevated not only in relation to bacterial infection but also in noninfectious conditions (63). Procalcitonin levels have been used as part of an antimicrobial stewardship program, and its use has been shown to reduce antibiotic use without increasing patient morbidity/mortality (64).

Concerns about increasing bacterial resistance and decreasing availability of new antimicrobials have resulted in the development of antimicrobial stewardship programs in many healthcare facilities. The primary goal of these programs is “to optimize clinical outcomes while minimizing unintended consequences of antimicrobial use, including toxicity, the selection of pathogenic organisms (such as Clostridium difficile), and the emergence of resistance” (43). A traditional approach to antimicrobial stewardship is to restrict access to targeted antibiotics or classes, perhaps by requiring preauthorization for use of specific antimicrobials. The prescriber must then justify the use of specific antimicrobial either prior to or within a short period of time after its initiation. Although such an approach has been shown to decrease use of specific antibiotics, it is not always clear that overall antibiotic use is reduced. Additionally, there are concerns with prescriber autonomy and the development of an adversarial relationship between prescribers and the antimicrobial stewardship team.