Population Patients with suspected VAP

Intervention Quantitative culture using invasive sampling

Comparator Qualitative culture using tracheal aspirate

Outcome Mortality, length of ICU stays, and ventilator-free days

We suggest against routine quantitative cultures using invasive sampling (e.g., bronchoalveolar lavage [BAL] and protected specimen brush [PSB]) for pathogen identification in patients with suspected VAP (conditional recommendation, moderate-quality evidence).

There were five randomized control trials (RCTs) for this key question (Table 2). In all studies, except one, the mortality rate, length of ICU stay, and duration of ventilator days did not differ between patients with VAP who underwent quantitative culture using invasive sampling and those who underwent qualitative culture using tracheal aspirate for identifying [17-21]. In addition, a meta-analysis including five RCTs and other observational studies showed no difference in terms of length of ICU stay and duration of ventilator days between patients who underwent invasive sampling and those who underwent qualitative culture using tracheal aspirate. However, invasive sampling was more likely to be associated with reduced mortality compared with qualitative culture using tracheal aspirate (odds ratio [OR], 0.91; 95% confidence interval [CI], 0.75 to 1.11) [22]. Therefore, the 2016 ATS/IDSA guidelines for HAP/VAP recommend qualitative culture using tracheal aspirate rather than quantitative culture using invasive sampling for pathogen identification in patients with VAP [3]. However, qualitative culture using tracheal aspirates may increase the proportion of drug-resistant pathogens due to the overuse of antibiotics caused by failure to discriminate between pathogens and colonizers [20]. Therefore, quantitative culture using invasive sampling could reduce the duration of antibiotic treatment, the proportion of drug-resistant pathogens, and the incidence of co-infection due to early antibiotic discontinuation if pathogens are not identified [23-26]. In this regard, the 2017 ERS/ESICM/ESCMID/ALAT guidelines for HAP/VAP recommend obtaining a distal quantitative sample (before any antibiotic treatment) to reduce antibiotic exposure in stable patients with suspected VAP and to improve result accuracy [4]. However, other outcomes (e.g., mortality, length of ICU stay, and duration of ventilator days), except for antibiotic use, were similar between patients who underwent quantitative culture and those who underwent qualitative culture. No other RCTs have been performed since 2006, and patients may develop complications caused by procedures (e.g., bronchoscopy). Therefore, we suggest against routine quantitative cultures using invasive sampling (e.g., BAL and PSB) for pathogen identification in patients with suspected VAP.

There was no difference in terms of mortality, length of ICU stay, and duration of ventilator days between quantitative culture using invasive sampling and qualitative culture using tracheal aspirate (weak recommendation, low-quality evidence). Thus, the 2016 ATS/IDSA guidelines recommend qualitative culture using tracheal aspirate rather than quantitative culture using invasive sampling for pathogen identification in patients with VAP [3]. Regarding the evidence of this recommendation, one RCT showed that quantitative culture is more useful than qualitative culture in distinguishing pathogens from colonizers. However, qualitative cultures can be performed more rapidly, are associated with fewer complications, and require less equipment than quantitative cultures using invasive sampling. Therefore, qualitative culture using tracheal aspirates for pathogen detection is recommended [3].

The 2017 ERS/ESICM/ESCMID/ALAT guidelines recommend obtaining a distal quantitative sample before any antibiotic treatment to reduce antibiotic exposure in stable patients with suspected VAP and to improve result accuracy (weak recommendation, low-quality evidence) [4]. However, the 2017 ERS/ESICM/ESCMID/ALAT guidelines also recommend qualitative culture using tracheal aspirate for pathogen detection in patients with acute respiratory distress syndrome or severe septic shock because of the unclear benefit of invasive procedures and the risk of complications in quantitative culture using invasive sampling (e.g., hypoxia) [4].

Population Patients with suspected HAP/VAP

Intervention Clinical criteria plus procalcitonin

Comparator Clinical criteria only

Outcome Diagnostic accuracy

We suggest against treatment decisions based on procalcitonin plus clinical criteria for patients with suspected HAP/VAP (conditional recommendation, moderate-quality evidence).

The efficacy of procalcitonin in HAP/VAP diagnosis is unclear as the number of observational studies is only limited (Table 3). In addition, procalcitonin can exhibit false positives in patients who underwent surgery and those with trauma, burns, cardiogenic shock, severe pancreatitis, autoimmune disease, severe renal failure, or severe liver failure. Further, it can exhibit false negatives in patients with local infection without signs of systemic infection and early bacterial infection within 6 hours [27]. The accuracy of procalcitonin for diagnosing VAP in a prospective observational study was poor (area under the curve, 0.62; 95% CI, 0.50 to 0.73) [28]. Moreover, the results of other small observational studies were similar to those of previous studies [29-31]. According to a meta-analysis of the 2016 ATS/IDSA guidelines, the sensitivity and specificity of procalcitonin for HAP/VAP diagnosis were 67% and 83%, respectively [3]. As studies on this issue were limited and heterogeneous and the study outcomes were poor, we disagreed that treatment decisions based on procalcitonin and clinical criteria were more effective than those based on clinical criteria alone in patients with HAP/VAP.

To decide whether or not to initiate antibiotic therapy in patients with suspected HAP/VAP, the 2016 ATS/IDSA guidelines recommended using clinical criteria alone rather than procalcitonin plus clinical criteria (strong recommendation, moderate-quality evidence) [3]. The 2017 ERS/ESICM/ESCMID/ALAT guidelines do not include a clinical question about the diagnostic usefulness of procalcitonin for determining whether to initiate antibiotic therapy in patients with HAP/VAP [4].

Population Patients with HAP/VAP

Intervention PCR test for atypical pathogen

Comparator No PCR test for atypical pathogen

Outcome Prevalence of atypical pathogen in patients with HAP/VAP

We suggest against PCR tests for atypical pathogens in patients with HAP/VAP (conditional recommendation, moderate-quality evidence).

Traditionally, atypical pneumonia has been defined as pneumonia caused by pathogens such as Mycoplasma, Chlamydia, and Legionella bacterium. In addition to these bacterial infections, viruses and fungi can also cause atypical pneumonia. Viruses generally cause pneumonia in immunocompromised patients. However, viruses are not infrequently detected as pathogens in patients with HAP/VAP who are immunocompetent [32,33]. In a Korean single-center study, viruses were identified in approximately 22.5% of patients with severe HAP/VAP requiring ICU admission and 11% of immunocompetent patients [32]. In a study of patients with relatively mild HAP who did not require mechanical ventilation, viruses were detected in approximately 22.7% of patients [33]. Regarding bacterial pathogens, Mycoplasma and Chlamydia species are rarely reported in patients with HAP/VAP. Legionella was once a common cause of HAP, accounting for approximately 10% of nosocomial infections in the 1990s [34]. However, since 2010, with proper hospital plumbing and water quality management, only 10 to 15 cases per 100,000 people have been reported worldwide [35]. In a Korean multicenter retrospective study of patients with HAP/VAP published in 2021, approximately 17.5% were tested for atypical pneumonia pathogens. However, all results were negative [6]. Therefore, we agreed not to recommend PCR testing for identifying atypical pathogens in patients with HAP/VAP.

There are no clinical questions or recommendations regarding PCR testing for atypical pathogens in patients with HAP/VAP in the 2016 ATS/IDSA and 2017 ERS/ESICM/ESCMID/ALAT guidelines [3,4].

Population Patients with HAP/VAP

Intervention Empiric piperacillin/tazobactam

Comparator Empiric cefepime

Outcome Mortality

We make no suggestion for using specific antibiotics (piperacillin/tazobactam or cefepime) for the empiric treatment of patients with HAP/VAP (inconclusive, low-quality evidence).

The single agents currently recommended for empirical HAP/VAP treatment include piperacillin/tazobactam and cefepime, which have anti-methicillin-sensitive Staphylococcus aureus and antipseudomonal effects. In a multicenter HAP/VAP study published in 2021, piperacillin/tazobactam (59.3%) and cefepime (6.7%) were the most frequently prescribed empirical antibiotics in Korea [36]. Table 4 shows the results of the two antimicrobial agents evaluated in febrile patients with sepsis and neutropenia. Retrospective studies of septic shock showed that the cefepime group had a higher mortality rate than the piperacillin/tazobactam group. However, the interpretation of results is limited as confounding factors have not been adjusted [36,37]. In a meta-analysis of febrile neutropenic patients, the cefepime group had a high mortality rate [38]. In addition, in a Korean retrospective study of 43 patients with severe community-acquired pneumonia, the mortality rate did not significantly differ between the two agents (both groups used drugs combined with ciprofloxacin) [39].

No study has directly compared the clinical effects of piperacillin/tazobactam and cefepime in patients with HAP/VAP. However, previous studies comparing the pharmacodynamic/pharmacokinetic effects, antibiotic sensitivity, and drug toxicity of the two drugs may be used as a reference for drug selection. In evaluating lung penetration of antimicrobial agents using the lung epithelial fluid/plasma concentration, the lung permeability of piperacillin/tazobactam was 0.568/0.913, and the permeability of cefepime was higher at 0.99 to 1.12 [40]. If the time exceeding the minimum inhibitory concentration for gram-negative bacteria was measured, the probability of achieving the bacteriostatic/bactericidal goal of cefepime against gram-negative bacteria was higher than that of piperacillin/tazobactam (88%/81% vs. 79%/71%). Therefore, the previous study suggested cefepime as a preferred empiric antibiotic for gram-negative pulmonary infections [41].

Piperacillin/tazobactam may have a higher risk of acute kidney injury than cefepime combined with vancomycin [42-47]. However, the causal association and the mechanism of kidney injury are not completely identified. It is mainly related to the inhibition of creatinine secretion, and the clinical significance of the patient’s prognosis is not significant [48]. By contrast, cefepime easily crosses the blood-brain barrier, which causes neurotoxicity that is characterized by symptoms such as decreased consciousness, aphasia, myoclonic myoclonus, seizures, and coma through concentration-dependent γ-aminobutyric acid antagonism [49]. In most cases, symptoms improve if the drug is discontinued. However, caution is required as no improvement is observed in some cases. Thus, clinical data on the difference in terms of mortality rates between cefepime and piperacillin/tazobactam in HAP/VAP is not sufficient to recommend the use of one agent. We agreed to make no recommendation for a preferred agent between piperacillin/tazobactam and cefepime for HAP/VAP.

There are no clinical questions or recommendations regarding a preference between piperacillin/tazobactam and cefepime for HAP/VAP treatment in the 2016 ATS/IDSA and 2017 ERS/ESICM/ESCMID/ALAT guidelines [3,4].

Population Patients with HAP/VAP who are at high risk of multidrug resistance and mortality

Intervention Empiric combination therapy with fluoroquinolone

Comparator Empiric beta-lactam monotherapy

Outcome Mortality

We suggest empiric β-lactam plus fluoroquinolone combination therapy for patients with HAP/VAP who are at high risk of multidrug resistance and mortality (conditional recommendation, low-quality evidence).

In the 2016 ATS/IDSA guidelines, combination therapy with β-lactam and other classes of antipseudomonal antibiotics, which increases the appropriateness and clinical response of empiric treatment against multidrug-resistant gram-negative bacteria, was weakly recommended for patients with HAP/VAP who are at high risk of multidrug resistance and mortality [3]. There was no difference in terms of mortality, clinical response, side effects, or incidence of resistance between monotherapy and combination therapy (Table 5) [50,51]. Nevertheless, the applicability of these results may be limited since many of these studies excluded patients with comorbidities or colonization of resistant strains and allowed additional empiric treatment for Pseudomonas aeruginosa until the actual pathogen was identified. The 2017 ERS/ESICM/ESCMID/ALAT guidelines also strongly recommend combination therapy for high-risk patients with HAP/VAP, including those with septic shock and multidrug resistance. The target strains included methicillin-resistant S. aureus and gram-negative bacteria [4]. However, no subsequent clinical trials have been conducted to support these recommendations. A Korean HAP/VAP multicenter study showed that 47.3% of initial empirical antibiotics were combination therapy. The most commonly used combination antibiotics were piperacillin/tazobactam (59.3%) and respiratory fluoroquinolone (32.1%) [6]. In another Korean study that only analyzed patients from general wards in the same cohort, 70.8% of all combination therapies were β-lactam plus fluoroquinolones. However, combination therapy was not associated with a reduced mortality rate [8]. Therefore, we agreed that empirical combination therapy is unnecessary for patients with HAP/VAP who are at low risk of multidrug resistance and mortality. However, considering the current frequency of combination therapy in Korea with a high multidrug resistance rate and the fact that there are no data on the side effects and costs of combination therapy, we agree that combination therapy with β-lactam plus fluoroquinolone could be an empiric treatment for patients with HAP/VAP who are at high risk of mortality, such as those with septic shock. In addition, after empiric combination therapy, a gradual reduction should be followed. In particular, one or more antibiotics should be discontinued according to microbiological test results, and the antibiotic treatment duration should be decreased if there are improvements in clinical signs [52].

The 2016 ATS/IDSA guidelines made a weak recommendation with low-quality evidence for empiric combination therapy of two antipseudomonal antibiotics, including β-lactams and other classes for patients with VAP who exhibit risk factors for multidrug resistance (history of intravenous antibiotic administration within 90 days, septic shock, acute respiratory distress syndrome, hospitalization for >5 days, and renal replacement therapy during VAP onset), patients admitted to units in which ≥10% of gram-negative bacteria are resistant to a single treatment, or patients admitted to units where local antibiogram data are not available [3]. In addition, empiric combination therapy of two antipseudomonal antibiotics with β-lactams and other classes is weakly recommended for patients with HAP only if there is a risk of mortality or multidrug resistance, such as those with a history of intravenous antibiotic treatment within 90 days (weak recommendation, very low-quality evidence) [3].

In the 2017 ERS/ESICM/ESCMID/ALAT guidelines, combination therapy is strongly recommended with moderate-quality evidence only for patients with high-risk HAP/VAP, including those with septic shock or risk factors for multidrug resistance (hospital environment with a high multidrug resistance rate, previous history of antibiotic use, long-term hospitalization of >5 days, and colonization of previous multidrug-resistant bacteria). In addition, the target strains included methicillin-resistant S. aureus and gram-negative bacteria [4].

Population Patients with HAP/VAP

Intervention Considering anaerobic coverage

Comparator Not considering anaerobic coverage

Outcome Clinical response

We suggest against considering anaerobic coverage when selecting empiric antibiotics in patients with HAP/VAP (conditional recommendation, moderate-quality evidence).

The aging society is associated with an increasing rate of risk factors for aspiration, including chronic neurological disorders and tube feeding, among patients with HAP/VAP [53-55]. This key question was discussed because the limitations in identifying anaerobes may cause an underestimation of the potential role of anaerobes in patients with HAP/VAP [56]. Anaerobes were considered as major pathogens of aspiration pneumonia. In fact, 60% to 90% of pathogens in aspiration pneumonia were anaerobes based on studies published until the late 1990 [56]. However, according to more recent studies published after 2000, community-acquired pneumonia or HAP/VAP caused by aspiration had similar causative pathogens compared with the usual community-acquired pneumonia or HAP/VAP and revealed a low rate of anaerobes (1% to 2%) [55,57]. Moreover, an RCT comparing the efficacy and safety of moxifloxacin and ampicillin/sulbactam in patients with aspiration pneumonia or lung abscess showed no significant intergroup differences in terms of main outcomes between the two agents [58]. Considering the results of a previous study and the notion that a subset of empirical HAP/VAP antibiotics is effective against anaerobes, we suggest against considering anaerobic coverage when selecting empiric antibiotics in patients with HAP/VAP.

The 2016 ATS/IDSA and 2017 ERS/ESICM/ESCMID/ALAT guidelines do not have recommendations for anaerobic coverage in empiric antibiotic selection for patients with HAP/VAP [3,4].

Population Patients with pseudomonas-related HAP/VAP

Intervention Combination therapy

Comparator Monotherapy

Outcome Mortality

We suggest against combination antibiotics for patients with HAP/VAP caused by pseudomonas infection (conditional recommendation, moderate-quality evidence).

If P. aeruginosa was identified as the causative strain in patients with HAP/VAP, combination therapy did not have any benefits [59]. There was also no difference in terms of mortality between monotherapy and combination therapy in pneumonia accompanied by P. aeruginosa bacteremia [60]. Accordingly, the 2016 ATS/IDSA guidelines recommend monotherapy if the risk of septic shock or mortality is not high in HAP/VAP caused by P. aeruginosa [3]. In a Korean retrospective observational study, combination therapy showed a trend toward reduced mortality in P. aeruginosa bacteremia [61]. However, in a recent meta-analysis involving pneumonia and bacteremia caused by P. aeruginosa, there was no evident association between combination therapy and mortality reduction (Table 6) [62]. In multidrug-resistant P. aeruginosa pneumonia, the benefit of colistin-based combination therapy has been reported in severe pneumonia cases [63,64]. However, comparative clinical trials and meta-analyses that have been performed since then have not confirmed the benefits of combination therapy [65-67]. Ceftolozane/tazobactam, a recently approved multidrug-resistant P. aeruginosa therapy, has a higher cure rate and fewer side effects than colistin-based combination therapy. Previous studies have shown that monotherapy is recommended for multidrug-resistant P. aeruginosa [66,68,69]. Based on this notion, we recommend a single susceptible antibiotic treatment for HAP/VAP, in which P. aeruginosa has been identified as the causative strain. We agreed not to recommend combination therapy for HAP/VAP caused by P. aeruginosa.

The 2016 ATS/IDSA guidelines recommend monotherapy for P. aeruginosa-induced HAP/VAP in cases where the risk of septic shock or mortality is not high (strong recommendation, low-quality evidence). However, combination therapy was recommended in cases with a high risk of mortality (weak recommendation, very low-quality evidence) [3]. However, the 2017 ERS/ESICM/ESCMID/ALAT guidelines do not have relevant clinical questions or recommendations [4].

In addition, the 2021 IDSA guidelines on the use of antibiotics related to multidrug-resistant pathogens do not recommend combination therapy for managing difficult-to-treat P. aeruginosa. If ceftolozane/tazobactam or other effective antimicrobials can be used, monotherapy is recommended [68]. The 2022 ESCMID guidelines for treating infections caused by multidrug-resistant gram-negative bacilli also recommend monotherapy for mild infections [69].

Population Patients with VAP caused by CRGNB

Intervention Systemic plus inhaled colistin therapy (adjunctive therapy)

Comparator Systemic colistin therapy alone

Outcome Mortality, clinical resolution, bacterial eradication, and nephrotoxicity

We suggest systemic plus inhaled colistin therapy (adjunctive therapy) for patients with VAP caused by CRGNB (conditional recommendation, low-quality evidence).

Whether colistin is an appropriate VAP treatment was not clear due to its low pulmonary tissue penetration rate when administered intravenously [70]. To overcome this, systemic plus inhaled colistin therapy (adjunctive therapy) has been proposed based on studies showing that inhaled colistin treatment could have a higher concentration in the lung tissue and lung epithelial lining fluid than intravenous colistin [71,72]. Retrospective studies have reported the use of adjunctive therapy in patients with VAP caused by multidrug-resistant Acinetobacter baumannii or P. aeruginosa. However, there was no significant difference in terms of mortality between VAP patients with adjunctive therapy and those without [73-78]. However, adjunctive therapy was associated with a higher clinical cure rate (69.2% vs. 54.8%, p=0.03) and shorter mechanical ventilation time (8 days vs. 12 days, p=0.001) [78]. In addition, in an RCT on this issue, there was no difference in clinical response (51% vs. 53%, p=0.84). However, regarding microbiological response, adjunctive therapy was superior to intravenous colistin monotherapy (60.9% vs. 38.2%, p=0.03) [79]. Meanwhile, there was no significant difference in side effects, such as renal toxicity and bronchoconstriction related to drug inhalation, between the two groups. In a meta-analysis comparing intravenous colistin monotherapy and adjunctive therapy for patients with VAP, no significant difference was observed in mortality between the two groups. However, adjunctive therapy was superior to intravenous monotherapy in terms of clinical response, microbiological eradication, and infection-related mortality. Further, there was no difference in terms of nephrotoxicity between the two groups [80]. Based on this, the 2016 ATS/IDSA guidelines recommend adjunctive therapy rather than intravenous monotherapy if the drug is the only sensitive antibiotic for patients with VAP caused by gram-negative rod bacilli [3]. In a Korean retrospective observational study of VAP caused by CRGNB, adjunctive therapy had a higher microbiological eradication rate and a lower overall mortality rate than intravenous monotherapy [81]. In a Taiwanese multicenter observational study, adjunctive therapy had a lower treatment failure rate than intravenous monotherapy [82]. Based on these results (Table 7), we agreed that inhaled colistin therapy can be added to systemic therapy for treating VAP caused by CRGNB [73-79,81-84].

The 2016 ATS/IDSA guidelines recommend adjunctive therapy if colistin is the only susceptible antibiotic (weak recommendation, very low-quality evidence) [3]. There are no relevant clinical questions or recommendations in the 2017 ERS/ESICM/ESCMID/ALAT guidelines [4].

However, the 2022 ESCMID guidelines for treating infections caused by multidrug-resistant gram-negative bacilli recommended avoiding inhaled therapy because there is no sufficient evidence showing that adjunctive therapy have clear clinical benefits. Further, safety, particularly the prevention of respiratory side effects, is challenging to ensure (weak recommendation, very low-quality evidence) [69]. In addition, the recently published IDSA guidelines for treating carbapenem-resistant A. baumanii infection do not recommend the use of inhaled colistin as adjunctive therapy as it lacks clinical benefit and there are concerns regarding unequal distribution in the infected lungs and respiratory complications such as bronchoconstriction in 10%-20% of patients receiving inhaled antibiotics [85].

Population Patients with HAP/VAP

Intervention Antimicrobial therapy for 7-8 days

Comparator Antimicrobial therapy for 10-15 days

Outcome HAP/VAP relapse

We suggest shortening the duration of antimicrobial therapy to 7 to 8 days in patients with HAP/VAP who exhibit a good clinical response to antimicrobial therapy (conditional recommendation, moderate-quality evidence).

In a previous study, the ATS guidelines recommended that HAP/VAP should be treated for at least 14 to 21 days [86]. However, the recommendations differed based on the severity of diseases, time to clinical response, and the causative organisms. Moreover, short-term treatment (7 to 10 days) had been recommended for HAP/VAP caused by methicillin-sensitive S. aureus or Haemophilus influenzae [86]. Subsequent comparative clinical studies revealed that short-term treatment did not differ with traditional long-term treatment in terms of clinical results [87,88] based on the 2005 revised ATS/IDSA guidelines that exerted efforts to shorten the treatment period from 14-21 to 7 days [1]. However, pneumonia caused by non-glucose fermenting gram-negative bacillus (GNB) was more likely to have a higher recurrence rate in patients receiving short-term treatment88. Thus, short-term treatment was recommended only if the causative organism was not P. aeruginosa and if the patient had a good clinical response [1]. Moreover, the 2016 ATS/IDSA guidelines also confirmed no significant difference in terms of mortality, cure, and recurrence rates between patients receiving short-term treatment (7 to 8 days) and those receiving long-term treatment (10 to 15 days) [3]. Previous meta-analyses of VAP caused by non-glucose-fermenting GNBs, mostly containing Pseudomonas and Acinetobacter, showed that patients receiving short-term were at higher risk of recurrence [89,90]; however, the updated meta-analysis of the 2016 ATS/IDSA guidelines did not show difference in terms of recurrence and mortality [3]. In a recent clinical trial of the non-inferiority for pneumonia recurrence between patients who received short-term antibiotic treatment (8 days) and long-term antibiotic therapy (15 days) in patients with VAP caused by P. aeruginosa, the recurrence of pneumonia differed by 7.9% (9.2% in the 15-day group and 17% in the 8-day group). Moreover, there was an increasing trend in the length of ICU stay and mortality rate in the 8-day group [91]. However, the results should be cautiously interpreted because the study was terminated early due to difficulties in registering the participants. Based on the recent data, we agreed that short-course antibiotic therapy requires attention in patients with VAP caused by non-glucose-fermenting GNB, in light of the recent evidence indicating that the risk of recurrence may increase in patients with VAP caused by non-glucose-fermenting GNB who receive short-course therapies (Table 8) [88,91-93].

The 2016 ATS/IDSA guidelines recommend 7-day antimicrobial therapy rather than a longer-course treatment (strong recommendation, moderate-quality evidence) [3]. Based on existing evidence showing that the risk of recurrence may increase in patients receiving short-course antibiotic therapy, separate recommendations were considered for patients with VAP caused by glucose non-fermenting GNB. However, no other recommendations were made as a slight increase in recurrence rates did not affect mortality and clinical cure rates. The 2017 ERS/ESICM/ESCMID/ALAT guidelines suggest the use of 7- to 8-day antibiotic therapy in patients with HAP/VAP patients without immunodeficiency, cystic fibrosis, empyema, lung abscess, or cavitation or necrotizing pneumonia and with a good clinical response to therapy (weak recommendation, moderate-quality evidence) [4]. In addition, the guidelines recommend that patients who have received inadequate initial empirical treatment may require a longer antibiotic treatment and that the treatment must be individualized according to the patient’s clinical response, specific bacterial findings, and serial biomarker measurements [4].

Population Patients with HAP/VAP

Intervention Antimicrobial de-escalation

Comparator No antimicrobial de-escalation

Outcome Clinical outcomes (mortality, length of stay, and recurrent infection), superinfection, duration of antimicrobial therapy, treatment cost, and the emergence of a resistant pathogen

We suggest antimicrobial de-escalation via one or more of the following in patients with HAP/VAP (conditional recommendation, moderate-quality evidence):

(1) Narrowing the spectrum of an antimicrobial based on the results of microbiology studies

(2) Discontinuation of treatment with one or more antimicrobials based on the results of microbiology studies

(3) Shortening the therapy if the patient shows signs of clinical improvement

In 2001, antimicrobial de-escalation in patients with HAP/VAP was publicized for the first time at the Consensus Conference on the Diagnosis and Treatment of VAP [94]. Subsequently, the ATS, Task Force of Three European Societies, ERS, ESCMID, and ESICM recommended the de-escalation of antibiotics in HAP/VAP treatment [1,2]. Two randomized clinical studies and five observational studies were analyzed in the 2016 ATS/IDSA guidelines [3]. They recommend antimicrobial de-escalation in HAP/VAP treatment based on expert opinions that they are beneficial because of reduced antibiotic side effects, resistance, and low antibiotic costs. According to a meta-analysis of antimicrobial de-escalation in patients with pneumonia who are admitted to the ICU that was published later, antimicrobial de-escalation is advantageous over fixed treatment in terms of 15-day mortality, length of hospital stay, and antibiotic cost. However, most evaluation parameters were not significant [95]. Subsequent observational studies have not identified consistent advantages in evaluation parameters other than cost reduction due to antimicrobial de-escalation [96-98]. However, only a few studies have evaluated antibiotic de-escalation in patients with HAP/VAP patients, and most of them are observational studies. Hence, there is a possibility of selection bias. Further, there are few studies on the development of resistance due to the de-escalation therapy are accepted as limitations. Therefore, the results of related studies cannot be accepted as they are. The recently announced definition of antimicrobial de-escalation is narrowing the spectrum of antibiotics based on microbiological test results, causing the discontinuation of one or more antibiotics based on microbiological test results, and shortening the duration of antibiotic treatment if clinical signs improve [52]. Based on the recently published definition of antimicrobial de-escalation, we agreed on the need for antimicrobial de-escalation, with consideration of the benefits of reducing the length of hospital stay and antibiotic cost and the antibiotic stewardship program (Table 9).

In the 2016 ATS/IDSA guidelines, two RCTs and five observational studies were analyzed [3]. Antimicrobial de-escalation was not associated with a significant difference in mortality rate and length of ICU stay compared with fixed treatment. The results on recurrence of pneumonia, duration of antibiotic use, presence of superinfection, and development of resistant strain were conflicting. Nevertheless, antimicrobial de-escalation was recommended, and this reflects the experts’ opinion that antimicrobial de-escalation has advantages in terms of reducing antibiotic costs and reducing side effects and resistance caused by antibiotic use (weak recommendation, very low-quality recommendation). The 2017 ERS/ESICM/ESCMID/ALAT guidelines had no relevant clinical key questions or recommendations [4].

Population Patients with HAP/VAP requiring antibiotic treatment with individualized durations

Intervention Procalcitonin plus clinical criteria

Comparator Clinical criteria alone

Outcome Antibiotic duration and treatment outcomes (mortality rate, mechanical ventilation duration, and length of stay)

We suggest discontinuing antibiotic therapy according to procalcitonin plus clinical criteria in patients with HAP/VAP whose therapy duration should be individualized (e.g., resistant pathogens, initially inappropriate antibiotics, and immunocompromised hosts) (conditional recommendation, moderate-quality evidence).

In a meta-analysis including 14 studies with 4,221 patients with acute respiratory infections, discontinuing antibiotic therapy according to procalcitonin level plus clinical criteria reduced the therapy duration for approximately 3.5 days compared with discontinuing antibiotic therapy based on clinical criteria alone. In addition, there were no intergroup differences in terms of mortality and treatment failure rates [99]. Previous meta-analysis evaluating acute respiratory infections in patients with HAP/VAP has limitations. Thus, the 2016 ATS/IDSA guidelines performed a meta-analysis, including three RCTs of 308 patients with VAP [3]. Other studies also revealed that the procalcitonin group had a significantly shorter duration of antibiotic therapy than the control group (9.1 days vs. 12.1 days, p<0.001). Further, there were no intergroup differences in terms of mortality rate, days of mechanical ventilation, and length of ICU and hospital stay [3,100-102]. In a succeeding RCT evaluating approximately 1,600 patients with critical illness, the procalcitonin group also had a shorter antibiotic duration by 2.7 days than the control group (95% CI, 1.26 to 4.12 days; p<0.001) [103]. However, previous studies have shown that the duration of antibiotic therapy has decreased in the control groups who discontinued antibiotic treatment without considering procalcitonin levels (from approximately 15 to 9.3 days in a study published in 2009 and 2016) (Table 10). Considering the decreasing tendency in antibiotic duration, the role of procalcitonin in decreasing the duration of antibiotic therapy in patients with HAP/VAP who can be treated with a short (7 to 8 days) course of antibiotics may be extremely limited. Thus, we suggest discontinuing antibiotics according to procalcitonin plus clinical criteria in patients with HAP/VAP whose therapy duration should be individualized, which included those with HAP/VAP caused by non-glucose fermenting gram-negative bacilli (Pseudomonas and Acinetobacter) [88], those with HAP/VAP caused by other resistant pathogens, including carbapenem-resistant Enterobacteriaceae and methicillin-resistant S. aureus, and a subset of patients with HAP/VAP who were excluded from previous RCTs (i.e., inappropriate antibiotics used as initial agents and immunocompromised hosts).

The 2016 ATS/IDSA guidelines recommend considering both procalcitonin level and clinical criteria when discontinuing antibiotic treatment in patients with HAP/VAP (weak recommendation, low-quality evidence) [3]. It was weakly recommended as the benefits of using procalcitonin levels, which are used to determine whether or not to discontinue antibiotic therapy in cases where standard antibiotic therapy for HAP/VAP is already 7 days or less, have not been identified [3]. The 2017 ERS/ESICM/ESCMID/ALAT guidelines do not recommend the routine measurement of serial procalcitonin levels for reducing the duration of antibiotic therapy in patients with HAP/VAP patients if the anticipated duration is 7 to 8 days (strong recommendation, moderate-quality evidence) [4]. However, the guidelines recommend the measurement of serial serum procalcitonin levels along with clinical assessment in specific clinical circumstances (i.e., HAP/VAP caused by non-glucose fermenting gram-negative bacilli or other resistant pathogens or immunocompromised hosts [good practice statement]).