A high suspicion for VAP followed by rapid diagnosis and treatment is critically important. Zilberberg and colleagues found that among nearly 400 patients alive at 48 h with HCAP, inappropriate empiric antibiotic therapy was associated with a significant increase in mortality (30 % versus 18.3 %, p = 0.013; OR 2.88 95 % CI 1.46–5.67 in multivariable logistic regression). Treatment escalation did not change the risk of death in this single-center study [3]. Unfortunately, treatment is often delayed. In one study among 107 patients, 30.7 % of patients had their therapy for VAP inappropriately delayed, defined as ≥ 24 h passing between VAP onset and providing the appropriate antimicrobial treatment. A delay in writing the antibiotic orders was the primary reason for delay in therapy in 75 % of cases [4].

Patients with VAP should universally initiated on therapy for (1) MRSA (for example, with vancomycin or linezolid) and for (2) resistant gram-negatives, such as Pseudomonas aeruginosa, Klebsiella pneumoniase, and Acinetobacter species. Treatment options could include: antipseudomonal cephalosporins (cefepime or ceftazidime), antipseudomonal carbapenems (imipenem or meropenem), β-Lactam/β-lactamase inhibitor (piperacillin-tazobactam). For patients for whom combination therapy is considered (see Evidence Contour below), addition of an antipseudomonal flouroquinolone or an aminoglycoside should be considered. The dominant pathogens in one’s local ICU should also contribute to decision making for appropriate choices of therapy but should be guided by the overall principles of the ATS/IDSA guidelines, as demonstrated by the IMPACT HAP collaboration [5, 6].

In addition to MDR risk factors, appropriate antimicrobial therapy should consider the patient’s risk factors for: (1) extended-spectrum beta-lactamase-producing Enterobactereriaceae; (2) Legionella; and (3) anaerobes. If ESBL Enterobactereiaceae is suspected, a carbapenem should be used. Concerns about Legionella should prompt use of a macrolide or fluroquinolone over an aminoglycoside. Some providers would treat patients with recent aspiration events for anaerobes, using clindamycin, β-Lactam/β-lactamase inhibitors, or a carbapenem.

For all other patients for whom the suspicion of VAP is low, appropriate therapy should be guided by the patient’s risk factors for multidrug resistant organisms. In the absence of risk factors for MDR organisms, the ATS/IDSA guidelines recommend antibiotic therapy that targets Streptococcus pneumonia, Haemophilus influenza, Methicillin-sensitive Staphylococcus aureus, and antibiotic-sensitive enteric gram negatives: ceftriaxone or levofloxacin, moxifloxacin, or ciprofloxacin, ampicillin/sulbactam or ertapenem [2]. While not all patients with HCAP have MDR organisms, distinguishing between the two may be difficult with recent residence in a nursing home or hospitalization for more than 48 h in the past 3 months appearing to increase the patient’s risk the most [7, 8].

Patients with ventilator-associated pneumonia should have the duration of antimicrobial therapy guided by type of organism. In a study of 401 patients using a randomized controlled design, there was no difference in mortality in the arm treated for 8 days versus 15 days, although patients with Pseudomonas spp. had higher rates of recurrence [9]. A subsequent meta-analysis demonstrated patients with lactose non-fermenting gram-negative bacilli had nearly a 2-fold increased odds of recurrence with shorter therapy courses [10]. However, more recent systematic reviews and the ATS/IDSA 2016 guideline [2] has challenged this distinction. The current recommendation for all VAP, including non-fermenting gram-negative bacilli, is to treat for a short course (7 or 8 days) [11].

It is critical to de-escalate antimicrobial therapy when a specific pathogen has been identified, or when cultures are negative at around 72 h. This helps prevent over-use of antibiotics and the development of resistance. Observational data provide a strong safety signal. In a study of surgical patients, neither mortality (34 % versus 42 %) nor recurrent pneumonias (27 % versus 35 %) differed between patients with VAP who underwent de-escalation versus those who did not [12]. Among 398 patients with VAP in Kollef et al., de-escalation of therapy occurred for 22 % of patients. These patients had a lower mortality rate (17 %) than those patients who underwent escalation (43 %) or who had not change to their regimens (24 %) [13].

Another possible guide to safely de-escalate antibiotic therapy may be procalcitonin levels. In subgroup analyses of the PRORATA trial, investigators found that patients with ventilator-associated pneumonia assigned to the study arm (where antibiotics were discontinued after procalcitonin levels reached <0.5 μg) had 3.1 fewer days (95 % CI 0.7 days – 5.6 days) than those patients assigned to the control arm, without a difference in mortality found in the overall study [14]. Other studies that have looked at procalcitonin to guide therapy for undifferentiated septic shock or in broader settings have replicated that mortality does not appear to be affected when procalcitonin is used to guide therapy, although the findings on duration of antibiotics is more heterogeneous [15, 16]. Current recommendations, however, are to continue to use clinical evidence rather than biomarkers [2].

Not all fevers are pneumonia, even in ICU patients with radiographic infiltrates. If patients are not improving at 48–72 h and respiratory cultures taken before antibiotics are negative, be vigilant for other causes of fever (such as central line infections, etc.) and for complications of pneumonia (such as empyema). This scenario should also prompt reconsideration of the potential presence of resistant pathogens, and it may warrant consultation with infectious disease specialists.

In all patients with suspected VAP, obtain an endotracheal aspirate for culture at minimum. Whether to pursue bronchoscopic sampling (or other invasive techniques) is more controversial. Endotracheal aspirates are very sensitive – a negative result is quite helpful because it has a high negative predictive value. Positive results can be harder to interpret. In one study, in 52 episodes of pneumonia, endotracheal aspirate was found to have a sensitivity of 97.7 % and specificity of 50 % as compared with protected brush specimen [17]. Other studies have employed the Clinical Pulmonary Infection Score (CPIS) with a cut-off of 6 as a noninvasive method of identifying patients with VAP, using autopsy findings of pneumonia as the gold standard (Table 29.1) [18]. Fabregas et al. found a score of greater than 6 had a sensitivity of 77 % but a specificity of 42 % [19]. Conversely, bronchoscopic sampling may be less sensitive but is more specific for pneumonia. Randomized controlled trials are mixed. An RCT of 413 patients found no benefit to invasive sampling in unadjusted analyses, but did after adjustment for baseline factors [20]. A more recent RCT of 740 patients found no benefit to bronchoalveolar lavage over endotracheal aspirate [21]. Our practice is to perform immediate endotracheal aspirate in all patients with suspected VAP, but to reserve bronchoalveolar lavage or protected brush for selected cases.

The current recommendation from the ATS/IDSA is for coverage with either (1) 15 mg/kg of vancomycin every 12 h with a target serum trough between 15 and 20 mg/kg OR (2) 600 mg of linezolid. One major prospective trial of 1184 patients, however suggested that linezolid may be superior to vancomycin. In this study, 46 % of patients treated with vancomycin had cultures persistently positive for MRSA, while only 17 % of patients treated with linezolid did. At 60 days, however, there was no difference in mortality rates, although nephrotoxicity did occur at greater rates with vancomycin [22]. As research in this space continues to evolve, linezolid may be a particularly good option among patients with renal failure, although current guidelines suggest either therapy for treatment.

Coverage with a second agent for gram-negative bacilli may be warranted based on local microbiologic patterns and was recommended by the 2005 recommendations for VAP treatment by the ATS/IDSA [2]. Current recommendations from the 2016 update are to prescribe 2 antipseudomonal antibiotics when patients have risk factors for antimicrobial resistance, when the prevalence of gram-negative isolates resistant to the proposed monotherapeutic agent exceds 10 %, and when antimicrobial susceptibaility rates are unavailable [2]. However, it is worth noting that synergy of medications has only been demonstrated in vitro and in neutropenic or bacteremic patients and one randomized controlled trial did not demonstrate differences in clinical outcomes between monotherapy and combination therapy groups [6, 23, 24]. An observational cohort study in Lancet suggested combination therapy may be harmful, as the cohort of patients with ATS/IDSA-compliant antimicrobial therapy had a higher risk of death at 28 days than the noncompliant group [25]. This remains controversial, whether these individuals were at higher risk of death from the medications, the infections, or misidentification of them as at higher risk for MDR infection. Further research will be necessary to identify who, if anyone, should be receiving such broad antibiotic coverage from the outset.

While clinical suspicion and identification of ventilator-associated pneumonia should remain high, significant controversy has revolved around establishing a reliable epidemiological surveillance definition. Prior to January 2013 the Centers for Disease Control’s surveillance reporting definition the included several subjective components, including the change in the “character of sputum” and in radiographs [26–30]. As a result, several studies identified little agreement either across infection control experts at a single institution [31] or across multiple institutions [32]. Other definitions that sought to identify episodes of VAP either through greater invasive strategies or through other scoring mechanisms fared equally poorly [33].

In response, an effort of many professional societies and the CDC generated an alternative approach with the creation of the entity Ventilator Associated Event (VAE) [34]. Intended to cast a broader net, this newly-defined condition is intended to identify the majority of iatrogenic harm from mechanical ventilation, including but not limited to pneumonia [35, 36]. Further, it is designed to be reliable as it is solely based on any changes made to the ventilator that would indicate worsening oxygenation after a period of stability and at least three days into the course of mechanical ventilation. Review of radiology has been removed from the definition. There are subsequent sub-categories of harm, including probably or possible pneumonia, which are based on antibiotic changes and evidence of positive qualitative or quantitative cultures. (Table 29.2) [34].