Antibiotic De-Escalation

Antibiotic De-Escalation Robert G. Masterton, FRCPath, FRCP [Edin & Glas] KEYWORDS De-escalation Antimicrobial streamlining Antimicrobial stewardship The present topography of clinical sepsis is a landscape populated by increasing and developing antimicrobial resistance, with a future where ever fewer new antibiotics, particularly innovative classes, 1 are becoming available to meet these challenges. This prospect has resulted in a new focus on making the best use of the antibiotics available to maximize their clinical impact and longevity. Such initiatives have become condensed into 2 main themes that are integrated, with the new treatment paradigm that deals with serious sepsis, of hit it hard and hit early 2 being embedded within the overall encompassing concept of antimicrobial stewardship. 3 De-escalation forms one of the key features of the new treatment paradigm (Box 1). Within this paradigm de-escalation presents probably the most challenging element. Notwithstanding this, the literature shows that de-escalation has received widespread support in various review and recommendation documents 4 6 over the last decade, but in a manner that perhaps does not reflect its true standing against the difficulties attendant to its implementation. Whereas its step-down concept of changing to a more targeted antibiotic is intrinsically logical, in clinical practice it faces the natural instinct of the clinician to continue with a treatment that is proving to be effective in managing the often life-threatening infection affecting a patient. This remains true, notwithstanding the positive conclusion reached within the recently released guidelines on antimicrobial stewardship 3 stating that: Streamlining or de-escalation of empirical antimicrobial therapy on the basis of culture results and elimination of redundant combination therapy can more effectively target the causative pathogen, resulting in decreased antimicrobial exposure and substantial cost savings. Crucially, whereas the strength of this recommendation was assigned the top rating of an A, it was acknowledged that the quality of the clinical evidence underpinning this was only in the middle band. This article therefore reviews the issue of de-escalation to present the current position. Financial disclosures and/or conflicts of interest: The author has nothing to disclose against the subject matter and materials discussed in this article. Funding support: None. Department of Microbiology, Ayrshire & Arran NHS Board, The Ayr Hospital, Dalmellington Road, Ayr KA6 6DX, UK E-mail address: robert.masterton@aaaht.scot.nhs.uk Crit Care Clin 27 (2011) 149 162 doi:10.1016/j.ccc.2010.09.009 criticalcare.theclinics.com 0749-0704/11/$ see front matter Ó 2011 Elsevier Inc. All rights reserved.

150 Masterton Box 1 Key principles of the new treatment paradigm Get effective antibiotic selection right first time Base antimicrobial selection, both empiric and targeted, on knowledge of local susceptibility patterns Use broad-spectrum antibiotics early Optimize the antibiotic dose and route of administration Administer antibiotics for the shortest possible duration AND Adjust or stop antibiotic therapy as early as possible to best target the pathogen(s) and remove pressure for resistance development (ie, de-escalation) DEFINITION OF DE-ESCALATION The definition of antimicrobial de-escalation is that it is a mechanism whereby the provision of effective initial antibiotic treatment, particularly in cases of severe sepsis, is achieved while avoiding unnecessary antibiotic use that would promote the development of resistance. This definition therefore encompasses 2 key features. First, there is the intent to narrow the spectrum of antimicrobial coverage depending on clinical response, culture results, and susceptibilities of the pathogens identified, and second, there is the commitment to stop antimicrobial treatment if no infection is established. 7 Vidaur and colleagues 8 added to this the criterion that where possible a single rather than multiple antibiotics should be used. The problem as it relates to clinical practice is the lack of convincing trial evidence demonstrating that de-escalation does not result in a poorer clinical outcome. Solid study data establishing exactly what criteria should be used, and when, to determine changing and stopping therapy do not exist. ANTICIPATED BENEFITS FROM DE-ESCALATION When considering de-escalation studies, it is important to be aware of the benefits that this approach is intended to produce (Box 2). Perhaps, peculiarly in assessing therapeutic management lines, the key feature for the studies to date in response to the challenges described above has been to show no detriment to individual patients rather than a potential improvement in clinical outcome. The primary focus of de-escalation is actually to demonstrate longer-term benefits through a positive impact on Box 2 Benefits realization in de-escalation therapy Treatment outcomes are unaltered from the conventional therapy approach of maintaining patients on their initially selected antimicrobials There is a beneficial impact observed through surveillance on the antimicrobial resistance profile for the institution at both micro and macro level Decrease in antibiotic related adverse events, for example, the incidence of Clostridium difficile infection and/or of superinfection with resistant bacteria and Candida organisms There is a reduction in overall antimicrobial costs

Antibiotic De-Escalation 151 antibiotic resistance development, with a significant secondary goal being financial savings through improved cost effectiveness. THE EVIDENCE FROM CLINICAL DE-ESCALATION STUDIES The overwhelming majority of the clinical studies assessing patient outcomes performed to date have been focused on nosocomial pneumonia (NP), especially ventilator-associated pneumonia (VAP). Their findings are reviewed in Table 1. The table begins with the seminal short-course antibiotic treatment study of Singh and colleagues. 9 Although not normally found on a literature search for articles on deescalation, it is among the first and strongest examples of de-escalating therapy to a stop when infection is assessed as not being present. There are some further studies reported that are not presented in Table 1, which are very few and diverse. These studies demonstrate centers beginning to expand the boundaries, including the difficulties of evaluating de-escalation outside the narrow field of NP. A prospective audit among 72 hospital inpatients suffering from amoxicillinsusceptible Escherichia coli infection in blood and/or urine 17 found that only 19% of those who could be de-escalated to amoxicillin actually received this agent. Amongst the 54% who had their antibiotic changed 64% of these were actually moved, perversely, to an unnecessarily broad-spectrum agent, and in 10% this was actually to a drug reported to be resistant. The investigators concluded that successful implementation of de-escalation demands more than simple protocol distribution. In another recent study that explored the practical application of de-escalation, 18 data from 113 intensive care unit (ICU) meropenem prescriptions were evaluated. De-escalation was defined as the administration of an antibiotic with a narrower spectrum within 3 days of the start of meropenem. The study found a trend toward a lower mortality rate (7% vs 21%, P 5.12) in patients who had been de-escalated. The majority of the infections were either pneumonia (46%) or intra-abdominal (31%). De-escalation was performed in 42% of patients, with the most common reasons for not doing so being the absence of conclusive microbiology or colonization with a multiresistant gram-negative bacillus. Finally, a retrospective study 19 has explored the issue of de-escalation in 102 cases of health care associated pneumonia where high-quality samples from intubation were not available and the microbiology findings were culture negative. The Pneumonia Severity Index was used for risk adjustment. The findings were that de-escalation occurred in 75% and 77% of the culture-negative and -positive groups, respectively (P 5 1.00). However, in culture-negative cases de-escalation took place approximately 1 day earlier (3.93 vs 5.04 days, P 5.03) and this group also showed a shorter length of hospitalization, lower hospital costs, and lower mortality rates. In the vast majority of the studies presented, the exact time to re-evaluation with a view to de-escalation was not set, which tended to reflect the time taken for the results to become available. In most studies, and against a continuous evaluation of the clinical response of the patient, microbiology results became available at around 48 to 72 hours, and this seems to be a pragmatic and sensible approach. However, the results 20 of a university hospital ICU are illuminating in this regard because they showed that although de-escalation was successfully implemented in 69% of patients for whom the microbiological data supported this, there was a mean interval of around 48 hours from the microbiological data being available to de-escalation action being taken. BENEFITS REALIZATION IN DE-ESCALATION PRACTICE Based on the clinical evidence from the now significant numbers of studies that have been performed, it is reasonable to conclude, given the consistency of these data, that

152 Table 1 De-escalation studies in nosocomial and ventilator-associated pneumonia Masterton Year Reference Principal Features of Study Design 2000 9 Randomized prospective trial involving 81 patients to minimize unnecessary antibiotic by using the clinical pulmonary infection score (CPIS) to assess ventilator-associated pneumonia (VAP) as operational tool for antibiotic therapy decision-making 2004 10 Prospective observational study over 43 months in a medical-surgical ICU involving 115 patients Brief Methodology Overview Quantitative bronchoalveolar lavage (BAL) used for diagnosis. Patients with CPIS 6 (implying low likelihood of pneumonia) received either standard therapy (antibiotic choice and duration physician discretion) or the study group with ciprofloxacin monotherapy with reevaluation at day 3 when ciprofloxacin was discontinued if CPIS remained 6 All episodes received initial broad-spectrum coverage followed by reevaluation according to clinical response and microbiology. Quantitative cultures by bronchoscopic examination or tracheal aspirates were used to modify therapy Principal Findings Relevant to De-Escalation Antibiotics continued beyond day 3 in 90% of those on standard approach compared with 28% in the study group (P 5.0001). Where CPIS remained 6 at day 3 antibiotics were continued in 96% in standard group but 0% in the others (P 5.0001). Mortality and length of intensive care unit (ICU) stay did not differ despite a shorter duration (P 5.0001) and lower cost (P 5.003) in the study arm. Antimicrobial resistance, or superinfections, or both, developed in 15% of study versus 35% of standard therapy patients (P 5.017) Change of therapy in 56.2%, including de-escalation in 31.4% (increasing to 38% if isolates were sensitive). Overall ICU mortality was 32.2%. The deescalation mortality rate was lower than those who continued initial regimen (18% vs 43%; P<.05). De-escalation was lower (P<.05) in the presence of nonfermenting gram-negative bacillus (2.7% vs 49.3%) and in the presence of late-onset pneumonia (12.5% vs 40.7%). When the pathogen remained unknown, 50% died and de-escalation was not performed

2006 11 An observational cohort study involving 104 patients to assess the impact of locally developed antimicrobial guidelines in the initial empiric treatment of ICU patients with severe hospital-acquired pneumonia (HAP) 2006 12 Observational prospective study conducted in 24 Spanish ICUs to assess a carbapenem-based de-escalating strategy in nosocomial pneumonia (NP) in 244 critically ill patients Evaluation pre- and post-guideline introduction where decision to change was based on quantitative BAL. Imipenem the main guideline empiric drug NP treated empirically with imipenem aminoglycoside/ glycopeptide. 91% were late-onset NP where primary outcome was therapeutic success 7 9 days post therapy. Culture-positive rate 54%, based on tracheal aspirates in 82%, protected specimen brush in 33%, and BAL in 4% Guideline followed in 75%. Imipenem continued as directed therapy in 27 cases. De-escalation possible in 34 patients but therapy continued beyond recommendation in 16 of these. The guideline patients showed greater adequate treatment (81% vs 46%; P<.01) with a lower mortality rate at 14 days (8% vs 23%, respectively; P 5.03). There was no increase in imipenem resistance related to its increased used during the study period Initial antibiotics inadequate for 9%. De-escalation in 23% and unchanged for 6%; 16% of patients did not receive de-escalation despite favorable microbiology. 46% therapy not altered, as no pathogen found. De-escalation implemented in only 23% with potentially multiresistant pathogens, compared with 68% in the rest (P<.001). Response rates were 53% and 50%, respectively in those receiving and not receiving de-escalation. No differences in superinfection rates and costs associated with de-escalation mainly dependent on the duration of hospitalization, though the duration of stay and costs in the ICU were higher for patients who were not de-escalated (P<.001 and P 5.001, respectively) (continued on next page) Antibiotic De-Escalation 153

154 Table 1 (continued) Year Reference Principal Features of Study Design 2006 13 Prospective, observational, cohort study in 20 ICUs in United States involving 398 ICU patients with suspected VAP to evaluate clinical characteristics and treatment patterns amongst VAP cases, including the implementation of and outcomes associated with de-escalation therapy 2007 14 Prospective observational study over 36 months to evaluate de-escalation in 115 medical-surgical ICU patients treated according to local pathway Brief Methodology Overview Therapy escalation was a switch to or addition of a drug class or classes with a broader spectrum or additional coverage. De-escalation was a switch to or discontinuation of a drug class resulting in a less broad spectrum of coverage. Diagnosis was by a range of methods All enrollments had positive cultures and were treated with limited-spectrum antibiotics (ie, without activity against Pseudomonas aeruginosa) if they had no prior hospitalization (within 21 days) or prior administration of antibiotics (within 10 days). Quantitative cultures by bronchoscopy or tracheal aspiration used to reassess empiric therapy Principal Findings Relevant to De-Escalation 61.6% of patients had neither escalation nor de-escalation. Overall, escalation occurred in 15.3% and de-escalation in 22.1%. De-escalation was significantly more common where 3 or more antibiotics were prescribed. De-escalation occurred more frequently when a major pathogen was isolated (26.8% vs 6.5%). Most patients with no pathogen had no change in therapy (87.1%), compared with 57.6% when a major pathogen was isolated. In those who initially received adequate therapy de-escalation took place in 27.1%, compared with 16.6% in those whose initial therapy was not adequate (c 2 5 6.15; P 5.013). Deescalation took place in 27.7% of BAL (c 2 5 3.59; P 5.06); 20.5% of tracheal aspirate patients (c 2 5 0.84; P 5.36) and 8.3% of those where neither was performed. The mortality rate was significantly reduced in de-escalated cases (17% de-escalation; 23.7% no change & 42.6% escalation [c 2 5 13.25; P 5.001]) Limited spectrum used in 79 patients (69%). De-escalation in respectively 26% and 72% of patients with early- and late-onset ventilator-associated pneumonia; treatment escalated in 27 (23%). Overall de-escalation was feasible in 42% of patients with no differences in outcome observed Masterton

2007 15 Prospective observational study involving 143 patients with VAP in a multidisciplinary ICU 2009 16 An observational study to evaluate de-escalation in 138 surgical patients, including those with septic shock Diagnosis by positive quantitative cultures of either tracheal aspirate or BAL and assessment by appropriateness of treatment for all significant isolates Surgical ICU patients with quantitative bronchoalveolar lavage with a positive threshold of 10,000 CFU/mL Therapy de-escalated in 40.5% with decreased mortality at day 15 (5.1% vs 31.7%) and day 28 (12% vs 43.5%) and shorter intensive care unit (17.2 1.2 vs 22.7 6.3 days) and hospital (23.7 2.8 vs 29.8 11.1 days) stay (P<.05). In tracheal aspirate patients there was 21% de-escalation with reduced 15-day mortality (5.8% vs 34.3%), reduced 28-day mortality (11.6% vs 45.3%), and shorter intensive care unit (17.2 1.6 vs 22.4 6.4 days) and hospital (23.1 4.4 vs 29.9 11.1 days) stay (P<.05). In BAL patients there was 66.1% de-escalation with decreased 15-day mortality (4.8% vs 23.8%), decreased 28-day mortality (12.1% vs 38%), and shorter ICU (17.2 1.1 vs 23.2 6 days) and hospital (23.8 2.4 vs 29.8 11.4 days) stay (P<.05) The recurrent pneumonia rate was not significantly different at 27.3% and 35.1%, respectively in those receiving and not receiving de-escalation. Mortality did not differ (33.8% vs 42.1%, respectively). De-escalation of therapy occurred in 55% of patients with appropriate initial therapy whereas 8% required escalation Antibiotic De-Escalation 155

156 Masterton de-escalation is appropriate to implement and delivers at least the same clinical outcome as the conventional approach of maintaining the initial therapy started when this was being successful. Although the majority of these findings come from work in NP, there is now evidence for other infections including septicemia and intra-abdominal infections. With particular regard to NP, the specter of de-escalation provoking an increased incidence of relapses or recurrences has been proven not to be real. This position should now promote a scenario where further trials are used to explore how best to de-escalate rather than whether to do it. There are several studies 10,11,13,15,19 suggesting that clinical outcome may actually be improved where de-escalation is practiced. It is difficult to hypothesize why the impact of de-escalation should be to improve clinical outcome, and therefore it remains to be determined whether this effect is genuine or merely reflects the characteristics of the patients in whom de-escalation is both feasible and chosen. Another possibility is that continued potent, broad-spectrum empiric therapy may be intrinsically detrimental in some patients. A recent meta-analysis/meta-regression demonstrated that empiric combination therapy in serious infections can be detrimental in patients at low risk of mortality even while providing significant clinical benefit in high-risk patients. 21 Patients who have already responded to potent, broad-spectrum antimicrobial therapy are similarly at a low risk of death and therefore may derive more harm than benefit from continued broad-spectrum therapy where de-escalation is not implemented, perhaps as a consequence of the modest but measurable toxicity/side effects of such regimens. The data available on cost-effectiveness is much less in both quantity and quality. From the perspective of drug acquisition and administration, it is persuasive that de-escalation should produce savings and there is evidence to support this. As described earlier, a small number of trials that demonstrate clinical benefit appear to point to potential cost reductions through reduced complications and shorter lengths of stay, both in the ICU and in the hospital. Further determination of these aspects will clearly be linked to the future work that explores whether such clinical benefits are truly related to de-escalation as opposed to being an apparent effect. Paradoxically, given that one of the fundamental themes underlying its promotion is the perceived potential for this intervention to positively affect antimicrobial resistance, there is a surprising lack of relevant evidence available. Indeed, this is the feature where there are the fewest data in terms of the impact of de-escalation, with very few studies exploring this aspect at all. The main question to be addressed is whether the initial use of potent broad-spectrum antibiotics in association with de-escalation will successfully protect against the development of resistance to the primary agent being used. Only one study has presented on this question 11 and, although it did demonstrate no increase in resistance to the carbapenem being used, its conclusion must be viewed with caution because of the relatively short reporting period of 6 months related to the trial. Another study 9 evaluated whether the use of de-escalation would have a positive impact on resistance development in the individual patient being treated, and concluded that it did indeed reduce the propensity to resistance development. There is clearly an overwhelming need for well-constructed de-escalation studies to identify whether short- and/or long-term benefits are truly associated with this tactic in terms of modifying the risk of resistance development. CHALLENGES IN IMPLEMENTING DE-ESCALATION The outcome of the adoption of de-escalation is intended to be that, based on microbiology results around the day 3 therapy point, the empiric antibiotic(s) that were

Antibiotic De-Escalation 157 started are stopped or reduced in number and/or narrowed in spectrum. Experience has shown that several factors in practice work against achieving this goal. Adoption of De-Escalation Rates of de-escalation range from about 10% in studies of clinical practice to about 70% in specifically designed trials; this suggests that getting clinicians to actually use de-escalation is a principal barrier. There is a natural propensity, particularly in severe sepsis when the patient who has been very seriously ill is starting to get better, to stick with a treatment regimen that is working rather than change to an alternative agent. The solution to this is to gain clinical confidence in de-escalation. Two principal lines of attack exist to help achieve this result. First, there is the ability for the health care professional to use a robust clinical assessment so that there can be a reasonable degree of certainty about whether an infection is present. Two studies, both in respiratory tract infections, 9,19 have evaluated this type of approach, showing that it is a useful component in enabling decision-making toward de-escalation where a risk-based clinical assessment suggests that infection is very probably not present and that this is supported by negative microbiology. The potential exists to apply this type of approach in other clinical areas, for example, intra-abdominal sepsis. 22 Second, several of the trials presented in Table 1 point to the value of diagnostic certainty where they have shown that de-escalation is most likely to take place against higher quality samples, such as bronchoalveolar lavage, when compared with less invasive tests such as tracheal aspirates. This concept was specifically explored in a study 15 where VAP was diagnosed by positive quantitative cultures of either tracheal aspirate or bronchoalveolar lavage, and where it was conclusively shown that the latter promoted the greater adherence to de-escalation. Leaving aside the issue of what best the clinician should do in the face of uncertainty resulting from negative microbiology cultures, which is addressed later, it has also been shown that where other, more general sepsis concerns exist, physicians choose to cover these potentially pathogenic organisms rather than focus on the proven etiology. This situation is documented particularly for multiresistant gram-negative bacilli 10,12 and results in de-escalation not taking place. There are insufficient data yet available to clarify the optimum course of action in these circumstances. It has also been shown that providing appropriate support to clinicians in the decisionmaking frontline can have a positive impact on action being taken when opportunities to de-escalate are present. In a before-and-after study, prescriptions of 13 selected intravenous antibiotics from surgical and medical wards were evaluated against 3 strategies over 3 consecutive 8-week periods: conventional management by the attending physician (control group); distribution of a questionnaire to the physician (questionnaire group); or distribution of the questionnaire followed by advice from an infectious disease physician (Q-IDP group). 23 The primary outcome was the percentage of modifications of antibiotic therapy at day 4, including withdrawal of therapy, de-escalation, oral switch, or reducing the planned duration of therapy. The greatest effects were seen in the Q-IDP phase where discontinuation was much more likely to happen than within the control group (P<.001). In addition, more prescriptions were modified in the Q-IDP group as compared with the control group (P 5.004), and stopping therapy in the absence of apparent infection also occurred significantly more often in the Q-IDP group than in the control (P<.0001) or questionnaire groups (P 5.002). Those seeking to introduce de-escalation strategies must recognize this fundamental issue of clinician confidence, and respond positively and proactively to it in order to give implementation its greatest chance of success.

158 Masterton Patient on empiric antibiotic and improving Stop antibiotic De-escalate antibiotic Maintain antibiotic Clinical risk score does not support presence of infection Clinical assessment of patient does not support presence of infection Clinical assessment of patient demonstrates infection improving Microbiology cultures positive & show evidence of de-escalation opportunity Clinical assessment of patient demonstrates that infection is improving Microbiology investigations negative Microbiology cultures do not support presence of infection Other investigations do not support presence of infection Other investigations demonstrate that infection is improving Fig. 1. Algorithm for de-escalation decision-making at day 3 in an improving patient.

Antibiotic De-Escalation 159 Patient on empiric antibiotic and not improving Escalate antibiotic and repeat microbiology investigations Maintain/escalate antibiotic and repeat microbiology investigations Clinical assessment of patient demonstrates infection getting worse Microbiology cultures positive & show evidence of escalation opportunity Clinical assessment of patient demonstrates that infection is not improving Microbiology investigations positive and appear to be covered Other investigations demonstrate that infection is getting worse OR Microbiology investigation negative Other investigations demonstrate that patient is not improving OR Microbiology investigation negative Fig. 2. Algorithm for de-escalation decision-making at day 3 in a patient not improving on the empiric antibiotic therapy. Decision-Making in De-Escalation One of the features that promotes clinician uncertainty is a lack of clarity about how to take decisions to de-escalate; this permeates to varying degrees each of the scenarios that the health care professional faces. However, the therapeutic approach to each setting is actually similar, though the respective contributions of the various components depend on the different clinical situations. The part that clinical risk assessments can contribute to the decision-making moment has been described in the preceding section. The other 2 features that are relevant are consideration of the clinical progress of the patient and the investigation information that is available. Box 3 A practical clinical bedside approach to de-escalation 1. Every patient with severe sepsis on antibiotic therapy should have the need for this considered and formally documented every day 2. No later than day 3, a full assessment of investigation results and clinical progress should be performed and a positive decision should be captured to: Stop the treatment (eg, no infection is present) Narrow the spectrum of the therapy Reduce the number of antibiotics being used, for example, there is redundancy in the therapy or such clinical progress that multiple agents active against the same pathogen(s) are not necessary Not to de-escalate, for example, the specific reason for not de-escalating is documented (eg, lack of microbiology results, lack of clinical improvement) 3. Every day thereafter a positive decision to stop, change, or continue the therapy should be made against specific reasons 4. At every assessment the goal is to stop the therapy, or elements of the therapy, unless a positive and persuasive need for their continuation exists

160 Masterton Figs. 1 and 2 present algorithms for de-escalation decision-making at around day 3 of therapy, when the microbiology results normally first become available, in the 2 common scenarios where the patient either is improving or is not. Each of these is set against the assumptions that effective therapy with appropriate empiric antibiotic selection and dosage has been delivered and that source control for the infection has been identified and achieved. The clinical assessment of the patient incorporates not just the standard observations such as pulse, temperature, blood pressure, and oxygen saturation, but also a full physical evaluation. It is essential to consider other noninfectious causes for the patient s condition at each overall evaluation. Other investigations that can shed light on the decision-making process are evaluation of inflammatory markers such as white blood cell count, C-reactive protein, and procalcitonin as well as appropriate use of imaging modalities. Escalation of the antibiotic therapy in the face of a deteriorating patient with negative microbiology is a value judgment set against a risk assessment of the perceived gaps in the spectrum of cover and any likely potential foci of infection. Following the initial day 3 assessment, continued consideration must be given to further opportunities for de-escalation. Key in this context is the situation where the patient is improving but microbiology cultures were negative and so the initial empiric antibiotics were continued at day 3. At a future juncture it will be necessary to decide when the treatment can be stopped at its earliest or converted to a narrower spectrum and/or oral agent. This last will be a judgment against an assessment of the clinical response, the likely source of sepsis, and also the potential pathogens to be covered. Within this envelope is the natural concern of the attending clinician to avoid making a mistake. This factor is more pressing when the patient remains seriously unwell as opposed to when the patient has improved so that organ support and mechanical ventilation are either much reduced or no longer needed. Where ongoing clinical concerns are high, it may be felt necessary to maintain the empiric antibiotic throughout the course of treatment, when it should be ensured that this is for the minimum duration possible. Fig. 2 captures the very problematic, but by no means uncommon, midway position where the patient s condition has not improved at day 3. This scenario has never been explored in any of the clinical trials available, so it is not possible to state the best course of action. In view of this, it is probably unrealistic at present to expect de-escalation to be applied to this group of patients. Although clearly continuing clinical and investigation assessment, including microbiology tests, must be maintained, the decision whether to continue with, or escalate, the antibiotic care will be constructed on an individual patient basis depending on an overall assessment of the risks. In summary, key elements within a successful de-escalation program at the bedside level are captured in Box 3 as a practical aid. The formal and rigorous consideration and documentation suggested is designed to ensure that the approach is applied consistently and in a sustained way to every relevant patient every day. It is only by doing so that the benefits of such schemes can be captured. SUMMARY De-escalation is a critical component that lies at the center of antimicrobial stewardship programs and the hit it hard, hit it early serious sepsis paradigm. The data presented demonstrate that it is a clinically effective concept. However, significant and serious shortfalls in the available evidence are highlighted. These shortfalls include the need to establish the real impact of de-escalation on antimicrobial resistance development; its true cost-effectiveness profile; and, while emphasizing that it is

Antibiotic De-Escalation 161 now well demonstrated that there is no downside for patients, whether it genuinely does improve clinical outcomes. Significant work needs to be done to establish the most effective tools to implement de-escalation, particularly in terms of providing clear guidelines to clinicians to enable them to be confident in applying this maneuver. It is interesting that this concept of de-escalation is now being explored in other types of infection. 24 There is little doubt that de-escalation is now here to stay, and the issue should no longer be whether the clinician should do it but how best it can be delivered. REFERENCES 1. Theuretzbacher U. Future antibiotics scenarios: is the tide starting to turn? Int J Antimicrob Agents 2009;34:15 20. 2. Masterton RG. The new treatment paradigm and the role of carbapenems. Int J Antimicrob Agents 2009;33:105 10. 3. Dellit TH, Owens RC, McGowan JE, et al. Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America guidelines for developing an institutional program to enhance antimicrobial stewardship. Clin Infect Dis 2007;44:159 77. 4. Kollef MH. Optimizing antibiotic therapy in the intensive care unit setting. Crit Care 2001;5:189 95. 5. Lisboa T, Rello J. De-escalation in lower respiratory tract infections. Curr Opin Pulm Med 2006;12:364 8. 6. Petrosillo N, Drapeau CM, Agrafiotis M, et al. Some current issues in the pharmacokinetics/pharmacodynamics of antimicrobials in intensive care. Minerva Anestesiol 2010;76:509 24. 7. Park DR. Antimicrobial treatment of ventilator-associated pneumonia. Respir Care 2005;50:932 52. 8. Vidaur L, Sirgo G, Rodriguez AH, et al. Clinical approach to the patient with suspected ventilator-associated pneumonia. Respir Care 2005;50:965 74. 9. Singh N, Rogers P, Atwood CW, et al. Short-course empiric antibiotic therapy for patients with pulmonary infiltrates in the intensive care unit: a proposed solution for indiscriminate antibiotic prescription. Am J Respir Crit Care Med 2000;162: 505 11. 10. Rello J, Vidaur L, Sandiumenge A, et al. De-escalation therapy in ventilator-associated pneumonia. Crit Care Med 2004;32:2183 90. 11. Soo Hoo GW, Wen YE, Nguyen TV, et al. Impact of clinical guidelines in the management of severe hospital acquired pneumonia. Chest 2005;128:2778 87. 12. Álvarez-Lerma F, Alvarez B, Luque P, et al. Empiric broad-spectrum antibiotic therapy of nosocomial pneumonia in the intensive care unit: a prospective observational study. Critical Care 2006;10:R78. 13. Kollef MH, Morrow LE, Niederman MS, et al. Clinical characteristics and treatment patterns among patients with ventilator-associated pneumonia. Chest 2006;129:1210 8. 14. Leone M, Garcin F, Bouvenot J, et al. Ventilator-associated pneumonia: breaking the vicious circle of antibiotic overuse. Crit Care Med 2007;35:379 85. 15. Giantsou E, Liratzopoulos N, Efraimidou E, et al. De-escalation therapy rates are significantly higher by bronchoalveolar lavage than by tracheal aspirate. Intensive Care Med 2007;33:1533 40. 16. Eachempati SR, Hydo LJ, Shou J, et al. Does de-escalation of antibiotic therapy for ventilator-associated pneumonia affect the likelihood of recurrent pneumonia or mortality in critically ill surgical patients? J Trauma 2009;66:1343 8.

162 Masterton 17. Donaldson AD, Barkham T. De-escalation for amoxicillin-susceptible Escherichia coli: easier said than done. J Hosp Infect 2010;74:304 5. 18. De Waele JJ, Ravyts M, Depuydt P, et al. De-escalation after empirical meropenem treatment in the intensive care unit: fiction or reality? J Crit Care 2010. [Epub ahead of print]. 19. Schlueter M, James C, Dominguez A, et al. Practice patterns for antibiotic deescalation in culture-negative healthcare-associated pneumonia. Infection 2010;38(5):357 62. 20. Fox BC, Fish J, Zheng L, et al. Prospective study of de-escalation of antimicrobial therapy in an ICU [abstract 217]. In: Program and abstracts of the 42nd Annual Meeting of the Infectious Diseases Society of America [Boston]: Infectious Diseases Society of America, Arlington (VA), 2004: 73. 21. Kumar A, Safdar N, Kethireddy S, et al. A survival benefit of combination antibiotic therapy for serious infections associated with sepsis and septic shock is contingent only on the risk of death: a meta-analytic/meta-regression study. Crit Care Med 2010;38:1651 64. 22. Sartelli M. A focus on intra-abdominal infections. World J Emerg Surg 2010;5:9. 23. Lesprit P, Landelle C, Girou E, et al. Reassessment of intravenous antibiotic therapy using a reminder or direct counselling. J Antimicrob Chemother 2010; 65:789 95. 24. Lichtenstern C, Nguyen TH, Schemmer P, et al. Efficacy of caspofungin in invasive candidiasis and candidemia de-escalation strategy. Mycoses 2008; 51(Suppl 1):35 46.