Guy W. Soo Hoo, MD, MPH; Y. Eugenia Wen, MD; Trung V. Nguyen, DO; and Matthew Bidwell Goetz, MD

Impact of Clinical Guidelines in the Management of Severe Hospital- Acquired Pneumonia* Guy W. Soo Hoo, MD, MPH; Y. Eugenia Wen, MD; Trung V. Nguyen, DO; and Matthew Bidwell Goetz, MD Study objectives: To asses the impact of locally developed antimicrobial treatment guidelines in the initial empiric treatment of ICU patients with severe hospital-acquired pneumonia (HAP). Design: Observational cohort study with preguideline and postguideline data collection. Patients: A total of 48 preguideline patients with 56 episodes of severe HAP defined by the National Nosocomial Infections Surveillance (NNIS) compared with 58 guideline-treated (GUIDE) patients with 61 episodes of severe HAP. Results: The two groups were similar in terms of mean ( SD) age (NNIS group, 67.7 9.6 years; GUIDE group, 68.0 11.5 years) and simplified acute physiology score (NNIS group, 12.9 3.9; GUIDE group, 12.6 3.1) at the HAP diagnosis, and the proportion of the most frequent isolates (ie, Pseudomonas and methicillin-resistant Staphylococcus aureus). There was wide variation in initial antibiotic use in NNIS-treated patients, with cefotaxime, ceftazidime, and piperacillin being the most common agents compared with all of the GUIDE patients who received an imipenem-cilastin-based regimen. Vancomycin use was similar in both groups. The GUIDE patients had a higher percentage of adequately treated patients (81% vs 46%, respectively; p < 0.01) with a lower mortality rate at 14 days (8% vs 23%, respectively; p 0.03). A lower mortality rate was also noted at the end of 30 days and the end of hospitalization but was not statistically significant. Appropriate imipenem use (as defined by the guidelines) occurred in 74% of the cases, and there was no increase in the number of imipenem-resistant organisms isolated during the course of the study. Conclusions: These guidelines represent a successful implementation of a deescalation approach, because the recommended empiric therapy with broad-spectrum antibiotics was switched to therapy with narrower spectrum agents after 3 days. Based on our experience, this approach improves the adequacy of antibiotic treatment, with improvement in short-term survival and without increasing the emergence of resistant organisms. (CHEST 2005; 128:2778 2787) Key words: guidelines; nosocomial pneumonia; ventilator-associated pneumonia Abbreviations: ATS American Thoracic Society; GUIDE guideline-treated; HAP hospital-acquired pneumonia; MRSA methicillin-resistant Staphylococcus aureus; NNIS National Nosocomial Infectious Surveillance; VAP ventilator-associated pneumonia. Among nosocomial infections, hospital-acquired pneumonia (HAP) has the greatest impact in patient management. It is estimated to account for almost half of all ICU infections and accounts for approximately 60% of all deaths from nosocomial infections. 1,2 Moreover, it increases hospital stays by *From the Pulmonary and Critical Care Section (Drs. Soo Hoo, Wen, and Nguyen) and the Infectious Diseases Section (Dr. Goetz), West Los Angeles Healthcare Center, VA Greater Los Angeles Healthcare System, Geffen School of Medicine at University of California Los Angeles, Los Angeles, CA. This study was supported in part by a grant from Merck & Co, Inc. Manuscript received September 1, 2004; revision accepted February 7, 2005. 7 to 9 days, crude mortality by 70%, and attributable mortality by 30%. 3,4 Approximately 60% of all deaths from nosocomial infection are due to HAP. Endotracheal intubation and mechanical ventilation are the greatest risk factors for the development of HAP. 5 Intubated patients with ventilator-associated pneumonia (VAP) experience a significantly longer Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Correspondence to: Guy W. Soo Hoo, MD, MPH, Pulmonary and Critical Care Section (111Q), West Los Angeles VAMC, 11301 Wilshire Blvd, Los Angeles, CA 90073; e-mail: Guy.Soohoo@med. va.gov 2778 Clinical Investigations in Critical Care

duration of mechanical ventilation, ICU stay, and hospital stay than matched cohorts. An episode of VAP can increase hospital costs by $40,000. 6,7 For editorial comment see page 1900 With upward of 250,000 cases of VAP annually, the cost of treatment easily approaches $10 billion in the United States alone. The management of HAP in general and VAP in particular has been an area of intense investigation. Nevertheless, there remains considerable uncertainty with the diagnosis of VAP. Up to half of new densities seen on the chest roentgenogram in mechanically ventilated patients may be due to noninfectious causes. 8,9 Furthermore, the microbiology of secretions obtained from endotracheal tubes may not accurately reflect the microbiology of the lower airways but, instead, may represent organisms that have colonized the endotracheal tube. 10,11 Specimens that accurately reflect the microbiology of pneumonia may require invasively obtained specimens (bronchoscopic or nonbronchoscopic BAL). 12 14. The diagnostic challenges result in difficulties in applying the abundant data that link recovery from HAP or VAP with the timely institution of appropriate antibiotic therapy. 15 17 Of necessity, antibiotic choices are made before the return of diagnostic specimens, and, thus, prior to the confirmation of the diagnosis, the identification of the infecting microorganism, and the determination of the antimicrobial susceptibilities. Moreover, the choice of appropriate empiric antibiotic therapy is fraught with difficulty, as there is great variation in the responsible microbiological agents. 18 The relevant factors include the timing of VAP (ie, whether the infection is early [ 3 days] or later [ 7 days] in the course of mechanical ventilation) prior antibiotic use, endemic and epidemic reservoirs of organisms within individual facilities, and variations in the prevalence of antimicrobial resistance. 2,19 One strategy for dealing with these uncertainties entails the early use of aggressive diagnostic maneuvers coupled with very broad-spectrum empiric therapy for patients with suspected VAP. To reduce pressure on the development of antimicrobial resistance, toxicity, and costs, such therapy is then tailored to the final diagnostic findings when they become available several days later. This strategy has often been referred to as culturedirected therapy or deescalation of antibiotic therapy. 20 These considerations led us to the development of HAP antibiotic guidelines based on prior local antimicrobial resistance patterns and the American Thoracic Society (ATS) guidelines for the treatment of HAP for use within our facility. In addition, the emphasis was also placed on invasively obtained microbiological specimens and on the use of these results to narrow the antibiotic spectrum after 3 to 5 days. The following report details our experience with these guidelines. Materials and Methods The West Los Angeles Healthcare Center is a tertiary care facility with 200 medical-surgical beds, of which 25 are ICU beds. No specific local guidelines for the treatment of pneumonia were in place prior to 1999. The management of patients with pneumonia (community-acquired pneumonia or HAP) was at the discretion of the physician staff. Perhaps reflecting the fact that national guidelines or consensus statements for the management of pneumonia provide only general recommendations regarding the choice of specific agents, there was a wide range of practice and antimicrobial prescribing patterns in the institution. Therefore, guidelines for the management of HAP were developed with joint concurrence of the Pulmonary and Infectious Diseases Sections. The recommendations were based on the antibiograms of isolated nosocomial pathogens from ICU patients in the preceding years and guidelines of the ATS. Other elements of the guidelines were an increased emphasis on the invasive collection of respiratory tract secretions for microbiological studies and a focus on culture-directed therapy or deescalation of antibiotic therapy based on these culture results. Definition of HAP and HAP Severity Local criteria for HAP were defined based on a modified Centers for Disease Control and Prevention definition. 21 The criteria include the presence and persistence of a new infiltrate 48 h after admission to the hospital and the following conditions: (1) purulent sputum or changes in sputum quality or isolation of pathogenic bacteria from blood cultures, pleural fluid, sputum, or invasive respiratory sampling; (2) temperature 38.5 C or 35 C; and (3) WBC count of 10 10 6 or 3 10 6 cells/ml. Patients who met the criteria for HAP were further defined as having severe HAP by admission to the ICU and any one of the following conditions: (1) shock defined as systolic BP of 90 mm Hg or diastolic BP of 60 mm Hg; (2) respiratory failure (ie, mechanical ventilation or the need for a fraction of inspired oxygen of 0.35 to maintain an oxygen saturation of 90%; (3) requirement of vasopressor therapy for 4 h; (4) urine output of 20 ml/h or total urine output of 80 ml/h for 4 h, unless oliguria is present due to a condition other than infection/sepsis; (5) acute renal failure requiring dialysis; or (6) rapid radiographic progression, multilobar pneumonia, or cavitation of a lung infiltrate. These criteria were adapted from ATS guidelines. 2 These criteria mirror those used to define organ system failure in those patients with severe sepsis. Patients who did not meet the criteria for severe HAP were considered to have mild-to-moderate HAP. HAP Guidelines: Recommended Antimicrobial Agents A review of antibiograms in the years preceding guideline development identified imipenem-cilastatin (referred to hereafter as imipenem) and amikacin as the most efficacious agents against nosocomial Gram-negative isolates. The combined in vitro efficacy of both agents against Gram-negative bacteria in our facility exceeded 95%. Therefore, an imipenem-based regimen was recommended for patients meeting the criteria for www.chestjournal.org CHEST / 128 / 4/ OCTOBER, 2005 2779

severe HAP. Dual therapy was recommended with amikacin; either ciprofloxacin or levofloxacin could be substituted if there was concern for nephrotoxicity. For patients with mild-to-moderate HAP, piperacillin/tazobactam monotherapy was recommended; ceftriaxone or levofloxacin were recommended as alternatives for those patients with an allergy to penicillin. The addition of vancomycin or azithromycin was recommended for mild or severe HAP if there was consideration of methicillinresistant Staphylococcus aureus (MRSA) or atypical pneumonia. Prior to the development of these guidelines, the aforementioned agents were on the restricted formulary of the hospital. The use of the agents required approval by the Infectious Diseases Service. This consisted of a request to a member of the Infectious Diseases Service for the antimicrobial agent, usually in a brief consultation by telephone before the agent or an alternative agent was approved. This approval was communicated to the Pharmacy Service either by telephone or with a written note in the medical record. The guidelines changed this practice pattern for patients who had been admitted to the ICU. Physicians were given permission to prescribe antimicrobial agents according to the guidelines for patients who met the criteria for HAP (mildto-moderate or severe) without obtaining approval from the Infectious Diseases Service for 3 days. After 3 days, the use of antimicrobial agents was based on the results of microbiological cultures. Patients either continued receiving their original antibiotic prescription or had their regimen adjusted based on the results of sensitivity testing. In the event that no organism was isolated, therapy with antimicrobial agents could be discontinued, or recommendations could be put in place to change therapy to alternative agents (eg, ceftriaxone or levofloxacin). These changes often occurred after a brief consultation with a member of the Infectious Disease Service, as this often required approval for the use of restricted antimicrobial agents. For example, a patient whose condition met the criteria for severe HAP who had started receiving imipenem therapy would continue with that therapy if sensitive Pseudomonas aeruginosa was isolated but would need to be changed if a resistant organism was isolated, or would need to be changed to an agent with a narrower spectrum of activity if the isolated organism was, for example, a fully sensitive Klebsiella sp. This constituted not only culture-directed therapy but also a strategy for the deescalation of the initial therapy. HAP Guidelines: Specimen Collection The other major component of these guidelines included an emphasis on the timely collection of specimens for culture. It is acknowledged that the oral contamination or colonization of endotracheal tube surfaces can obscure the identification of the pathogenic organism in patients with HAP. Endotracheally suctioned specimens were acceptable in patients who had been intubated for no more than 24 h. Otherwise, specimens were obtained either with nonbronchoscopic or bronchoscopic BAL that was performed by previously described methods. 14,22 Briefly, nonbronchoscopic BAL was performed using a commercially available telescoping catheter (BAL Cath; Kimberly-Clark/Ballard Medical Products; Draper, UT). The catheter was inserted into and through an endotracheal tube adapter, and was placed inline between the endotracheal tube and the ventilator circuit. The catheter has a directional tip, and the catheter was directed to the lung (right or left) with the suspected pneumonia. The catheter was advanced about 5 cm beyond the end of the endotracheal tube. The tip of the catheter was then flushed with about 2 ml saline solution instilled by a syringe attached to the three-way stopcock located at the end of the catheter. Holding the outer sheath in place, the inner catheter was then advanced until resistance was felt, signifying a wedge position. A sterile specimen trap was placed in line with the catheter and wide-bore tubing. Then the catheter was locked in place, and 30 ml sterile saline solution was instilled through the three-way stopcock. The wide-bore tubing was connected to a suction device, and, with a twist of the stopcock, suction was applied to system, and the fluid was collected in the inline specimen trap. At least 20 ml fluid was collected. If the sample obtained with the initial BAL was inadequate, the procedure was repeated with another 30 ml sterile saline solution. Bronchoscopic BAL was performed using with a bronchoscope (model EB 153 0T3 or 183-OT3; Pentax Medical; Montvale, NJ). Patients were sedated for the procedure with a combination of midazolam, morphine, or meperidine and were preoxygenated with 100% oxygen, with continuous cardiac monitoring and pulse oximetry. The fiberoptic bronchoscope was advanced down the endotracheal tube and positioned at the orifice of the segment leading to the suspected pneumonia or where the greatest amount of purulent secretions was noted. Lidocaine was use sparingly and primarily for local anesthesia of the upper air or trachea. The tip of the bronchoscope was then wedged into place with the segment lavaged with 30-mL aliquots of sterile saline solution using a total of 120 ml saline solution. The fluid was aspirated through the bronchoscope and BAL fluid obtained after the initial aliquot of saline solution was submitted for semiquanitiative cultures. A threshold of 10 4 cfu/ml was defined as a significant infection for these BAL fluid cultures. 23 HAP Guidelines: Guideline Implementation The guidelines were drafted and completed over several months in early 1999. Prior to the institution of these guidelines, a joint conference was held with members of the Pulmonary and Infectious Diseases Service (ie, faculty, fellows, and residents) as well as the Pharmacy Service. The guidelines were reviewed to acquaint all members with the rationale behind the guidelines, as well as with the changes in hospital policy in the management of HAP patients (including a limited unrestricted use of previously restricted antimicrobial agents based on guideline recommendations, a strategy of invasive sampling of patients with suspected HAP, and a switch or deescalation of antimicrobial agents). The guidelines underwent pilot testing during the spring of 1999 and were posted for full implementation in late June 1999. After posting the guidelines, the guidelines were distributed to each group of house staff rotating through the medical ICU. Guidelines were distributed twice every 4 weeks (intern group and resident group, 26 times per year). The guidelines were reviewed with each group and were also available on the hospital computer intranet. In addition, the guidelines were reviewed regularly by one of the staff physicians during daily rounds or by one of the members of the Pharmacy or Infectious Diseases Service. In the first few months after the introduction of the guidelines, regular conference sessions were held with house staff rotating through the medical ICU to reinforce the guidelines. HAP Guidelines: Data Collection The impact of these HAP guidelines on antibiotic use, disease management, and patient outcomes constitutes the major investigative focus of this report. We specifically made comparisons between guideline-treated (GUIDE) patients who were hospitalized between July 1999 and September 2002, and all of the ICU patients at our institution who had prospectively satisfied the National Nosocomial Infectious Surveillance (NNIS) criteria for VAP during the calendar year 1998. The criteria used to define HAP and HAP severity were the same in both groups of patients and were the same as those outlined earlier in this section. The NNIS database was an existing, prospectively identified, retro- 2780 Clinical Investigations in Critical Care

spectively reviewed database of patients who constituted the comparison or control group. The comparison group consisted of consecutive, prospectively identified, GUIDE patients with severe HAP. A thorough chart review of the management of these NNIS patients and GUIDE patients was conducted by one of the authors. Using a structured case-report form, basic demographic and medical comorbidity data were collected, as were the dates of hospital admission, ICU admission, intubation, diagnosis of HAP, extubation, ICU discharge, and hospital discharge or death. In addition, the results of all of the cultures obtained in the 5 days prior to and 30 days after the date of HAP diagnosis, all of the antimicrobial agents administered, and the potentially adverse effects related to the use of antimicrobial agents (eg, Clostridium difficile colitis, new isolates, and the emergence of drug-resistant organisms) during that 30-day period were tabulated. The review included the identification of coexistent infections (eg, urinary tract, skin, or soft tissue). The review involved the inspection of ICU flow sheets, which included daily Glasgow coma scores and medical records. A severity-of-illness score (ie, the simplified acute physiologic score) was calculated for the date of HAP diagnosis. 24 It should be noted that the medical records at this Veterans Affairs medical center are almost fully electronic, and contain the results of all laboratory studies and virtually all pharmacy data. There was a specific focus on the appropriateness of empiric antimicrobial coverage on the date of HAP diagnosis. The empiric use of prescribed antibiotics was described as adequate if all of the isolated organisms were sensitive to at least one of the prescribed antimicrobial agents. Adequacy was analyzed for HAP and coexistent infections. The duration of the use of antimicrobial agents was designated as being appropriate if continued empiric, broad-spectrum therapy was required based on microbiological results or if antibiotic therapy was narrowed based on these results or the guideline recommendations. Given the logistics of culture processing and reporting at our institution, appropriate use included a switch of antibiotic agents within 5 days of the HAP diagnosis. The continued use of empiric antibiotics beyond 5 days was considered to be inappropriate if the antibiotics were not adjusted despite microbiological results that indicated a resistant organism, no growth, or a pathogen that could be treated with antimicrobial agents in a more focused therapy. Hospital ICU Surveillance Study Antibiograms of isolates that were monitored as part of this study were used to generate a comparison of the percentage of imipenem-resistant Pseudomonas isolates over the time course of this study. These represent data from 100 consecutive microbiological isolates that were collected from any body site in patients who had been hospitalized in one of our critical care units. The survey has been conducted annually since 1995 (except for 1997) and represents data collected over the same time frame in each of those years. 25 Data Analysis All of the case report forms were reviewed for accuracy by one of the authors during a second review of the cases prior to entry into a large database file. In addition, the medical records and case report forms of 10 cases were reviewed in detail to ensure adequate and consistent data abstraction. The data were entered into a spreadsheet (Excel; Microsoft; Redmond, WA) for analysis. Further statistical analysis was conducted using a medical statistics software package (MedCalc; Brussels, Belgium). Continuous variables were compared between the two groups (NNIS and GUIDE patients) using the Student t test, with categoric variables compared using the 2 test or Fisher exact test, where appropriate. The significance level was set at 0.05. A comparison of the impact of treatment on mortality was performed using the Kaplan-Meier method. A univariate analysis of risk factors for mortality was performed with the 2 test, the Fisher exact test, or the unpaired Student t test, where appropriate. Further multivariate analysis of significant risk factors was performed using the Cox proportional hazards regression model, both at 30 days and for the whole duration of hospitalization. This study was reviewed and approved by the Veterans Affairs Greater Los Angeles Healthcare System Institutional Review Board. The committee waived the need for informed consent, because this was an observational study on the impact of guideline implementation, involved the analysis of existing data, and posed minimal patient risk. Results There were 48 patients in the NNIS database with 56 episodes of HAP, and they were designated as the NNIS group. All of the patients had been treated prior to guideline implementation. Fifty-eight patients with 61 episodes of HAP were treated according to the guidelines and comprised the GUIDE group. Basic demographics are summarized in Table 1. In those patients with more than one episode of HAP in the NNIS group, the episodes of HAP occurred a least 1 month apart during the same hospitalization. In the GUIDE group, no patient had more than one episode of HAP during the same hospitalization. Both groups had similar distributions Table 1 Summary of NNIS and GUIDE Patients* Variables NNIS Group (n 48) GUIDE (n 58) Episodes of HAP, No. 56 61 Age, yr 67.7 9.6 68.0 11.5 Comorbidities Alcoholism 10 (21) 7 (12) COPD 16 (33) 18 (30) Congestive heart failure 17 (35) 10 (16) Diabetes mellitus 13 (27) 16 (26) Malignancy 7 (15) 8 (14) CVA 14 (29) 24 (39) Postoperative 20 (42) 10 (17) Renal failure 8 (17) 13 (22) SAPS at HAP diagnosis 12.9 3.9 12.6 3.1 Prior to HAP Hospital days 20.4 16.6 19.7 28.0 ICU days 15.6 14.9 6.2 13.2 Ventilator days 10.4 11.7 5.4 13.2 *Values given as mean SD or No. (%), unless otherwise indicated. CVA cerebrovascular accident; SAPS simplified acute physiology score. Forty-eight men. Fifty-six men and 2 women. Comparison of proportions (p 0.01). p 0.05 (t test). www.chestjournal.org CHEST / 128 / 4/ OCTOBER, 2005 2781

in terms of age, gender, and comorbidities, except that the NNIS group had a greater proportion of congestive heart failure and postoperative patients than the GUIDE group. Both groups had similar severity-of-illness scores at the time of HAP diagnosis. The NNIS patients were hospitalized in the ICU longer and received mechanical ventilation longer prior to HAP diagnosis than the GUIDE patients (p 0.01 [t test]). The results of respiratory tract cultures obtained at the time of HAP diagnosis are listed in Table 2. P aeruginosa and MRSA were the two most frequently isolated pathogens; there was no difference between the two groups in the frequency of their isolated pathogens. The results from nonrespiratory sites at the time of HAP diagnosis showed no significant differences between the two groups. Antimicrobial therapy at the time of the diagnosis of HAP is listed in Table 3. The adequacy of empiric coverage is outlined in Table 4. More than 80% of GUIDE patients received adequate empiric therapy for VAP compared with less than half of the NNIS patients (p 0.01 [two-tailed Fisher exact test]). In the NNIS group, the reasons for inadequate coverage of HAP were equally split between inadequate Grampositive coverage (usually MRSA) and inadequate Gram-negative coverage. In the GUIDE group, the main reason for inadequate coverage was inadequate Gram-positive (MRSA) coverage. In the GUIDE group, there was only one Gram-negative respiratory isolate that was resistant to initial therapy vs 15 Gram-negative respiratory isolates in the NNIS group (p 0.06 [two-tailed Fisher exact test]). The Table 2 Isolates From Respiratory Tract and Other Sites at the Time of HAP Diagnosis* Respiratory Isolates NNIS Group (n 56) GUIDE Group (n 61) P aeruginosa 19 (34) 13 (21) Serratia spp 4 (7) 5 (8) Enterobacter spp 4 5 Acinetobacter spp 4 5 Stenotrophomonas spp 3 (5) 1 (2) Haemophilus influenzae 2 (4) 3 (5) Escherichia coli 2 0 Klebsiella spp 1 (2) 5 Providencia spp 1 3 Proteus spp 0 2 (3) MSSA 1 1 MRSA 27 (48) 19 (31) VSE 4 1 VRE 4 0 Normal oral flora 0 8 (13) *Values given as No. (%). Multiple isolates were reported in some patients. MSSA methicillin-sensitive S aureus; VSE vancomycinsensitive enterococci; VRE vancomycin-resistant enterococci. Table 3 Antimicrobial Therapy at the Time of HAP Diagnosis* Antimicrobial Agent predominant reason for inadequate coverage of the nonrespiratory site infections was insufficient Grampositive coverage (MRSA). The general hospital course of these patients is outlined in Table 5. There was no significant difference in the use of resources or the duration of hospitalization between the two groups after the diagnosis of HAP. There was a trend toward decreased mortality in the GUIDE patients at all points of comparison, which was statistically significant only at 2 weeks after the diagnosis of HAP (Fig 1). For all causes, mortality within 2 weeks of HAP diagnosis was significantly lower in the GUIDE group (5 deaths [8%]) vs that in the NNIS group (13 deaths [23%]; p 0.03 [Kaplan-Meier test]). The mortality rate remained lower at the end of 30 days and at the end of hospitalization for the GUIDE group, but this Table 4 Adequacy of Empiric Antimicrobial Coverage* HAP NNIS Group (n 56) NNIS Group (n 56) GUIDE Group (n 61) Adequate 26 (46) 52 (81) Inadequate 30 9 Resistant Gram-positive 15 8 Resistant Gram-negative 15 1 *Values given as No. (%). p 0.01 (two-tailed Fisher exact test). GUIDE Group (n 61) Imipenem 10 (18) 61 (100) Vancomycin 25 (45) 28 (46) Ciprofloxacin 3 (5) 23 (38) Amikacin 8 (14) 22 (36) Cefotaxime 22 (39) 0 Metronidazole 21 (38) 6 (10) Ceftazidime 11 (20) 0 Piperacillin 11 (20) 0 Gentamicin 11 (20) 2 (3) Fluconazole 8 (14) 3 (5) Clindamycin 7 (13) 1 (2) Cefazolin 2 (4) 0 Oxacillin 2 (4) 0 Ampicillin 2 (4) 2 (3) Piperacillin/tazobactam 0 2 (3) Levofloxacin 0 3 (5) Azithromycin 0 2 (3) Erythromycin 0 1 (2) Trimethoprim-sulfa 1 (2) 0 Ticarcillin/clavulanate 1 (2) 0 Doxycycline 0 1 (2) Antituberculous meds 1 (2) 0 *Values given as No. (%). 2782 Clinical Investigations in Critical Care

Table 5 Hospital and ICU Data of HAP Patients* Variables NNIS Group (n 56) GUIDE Group (n 61) After HAP diagnosis and treatment Ventilator days 16.3 19.1 19.0 22.5 ICU days 24.7 23.2 25.4 24.5 Hospital days 36.3 32.5 36.9 30.9 Total hospital days 51.3 38.0 56.6 44.8 Mortality (all patients) All cause at 14 d 13 (23) 5 (8) All cause at 30 d 21 (38) 14 (23) All cause during hospitalization 25 (45) 18 (30) Adequate initial antibiotic therapy (n 26) (n 52) Mortality All cause at 14 d 7 (27) 4 (8) All cause at 30 d 10 (38) 12 (23) All cause during hospitalization 13 (50) 16 (31) Inadequate initial antibiotic therapy (n 30) (n 9) Mortality All cause at 14 d 6 (20) 1 (11) All cause at 30 d 11 (37) 2 (22) All cause during hospitalization 12 (40) 2 (22) *Values given as mean SD or No. (%), unless otherwise indicated. p 0.03 (Kaplan-Meier survival curve). p 0.036 (two-tailed Fisher exact test). reduction in mortality did not achieve statistical significance (p 0.11 and p 0.15, respectively [Kaplan-Meier test]). These differences between the two groups persist after analysis based on the adequacy or inadequacy of initial antimicrobial therapy for HAP. With each subgroup and at each point in time from the diagnosis of HAP, the observed mortality rate is lower for the GUIDE patients than for the NNIS patients. However, these differences were not statistically Figure 1. Survival curves of the two groups of HAP patients. The solid line traces the survival curve of patients in the GUIDE group, whereas the dashed line represents the survival of patients in the NNIS group. significant except at 2 weeks (p 0.036 [two-tailed Fisher exact test]) for patients receiving adequate antimicrobial therapy. Although the severity-of-illness scores were similar between the NNIS and GUIDE groups, NNIS patients had a greater average duration of ICU stay and endotracheal intubation than GUIDE patients prior to the diagnosis of HAP as identified in a univariate analysis. A multivariate analysis of these influences on survival using the Cox proportional hazards regression model identified guideline treatment as the only significant influence on mortality. This is further summarized in Table 6. Further analysis of imipenem use is presented in Table 7. Overall, patients in the GUIDE group received imipenem for a mean ( SD) duration of 9 6.8 days. In 45 patients (74%), treatment with imipenem was either warranted based on culture results (directed therapy n 27) or was appropriately adjusted within 5 days based on culture results (n 18). The mortality rates are provided for these subgroups, as well as for the subgroup of patients treated with imipenem beyond guideline recommendations. Although the numbers are small and the differences do not achieve statistical significance, those patients treated with imipenem based on the guideline recommendations had the most favorable mortality rates. Imipenem-resistant isolates were identified in seven patients within 30 days of the diagnosis of HAP. Only one identification occurred within 5 days of imipenem use. The mean duration of imipenem use in these patients was 12.6 9.7 days (range, 4 to 31 days). All except one of the imipenem-resistant isolates occurred in the GUIDE group. All except one of the imipenem-resistant isolates occurred in the group managed according to guidelines. Although there was a greater number of imipenemresistant isolates in this group, this was not statistically different from those treated beyond guideline recommendations (p 0.22 [Fisher exact test]). Figure 2 provides further information on the development of imipenem-resistant organisms during the time course of the study. These antibiograms were produced as part of a hospital ICU surveillance study that sought to determine the incidence of multidrug resistance in serial Gram-negative isolates from the ICU at our institution. There was no significant change in the percentage of imipenemresistant isolates of Pseudomonas and other enteric Gram-negative organisms in the time period preceding guideline implementation, the comparison year, and over the course of the 2 years that comprise the course of this report. www.chestjournal.org CHEST / 128 / 4/ OCTOBER, 2005 2783

Table 6 Multivariate Analysis of Potential Influences on Mortality: Cox Proportional Hazards Regression Model* Variables Coefficient SE p Value Hazard Ratio (95% CI) 30 d after HAP Prior hospital days 0.0070 0.0069 0.3136 1.01 ( 0.007 0.021) Prior ICU days 0.0078 0.0207 0.7083 0.99 ( 0.048 0.034) Prior ventilator days 0.0324 0.0286 0.2576 0.97 ( 0.088 0.023) GUIDE 0.7850 0.3836 0.0407 0.46 ( 1.533 0.037) Duration of hospitalization Prior hospital days 0.0089 0.0060 0.1424 1.01 ( 0.003 0.021) Prior ICU days 0.0233 0.0213 0.2736 0.97 ( 0.065 0.018) Prior ventilator days 0.0211 0.0263 0.4229 0.98 ( 0.072 0.030) GUIDE 0.8195 0.3472 0.0183 0.44 ( 1.496 0.143) *CI confidence interval. Discussion Of the many factors that influence the outcome of patients with VAP, the timely administration of effective antimicrobial therapy, well before culture results are known, forms the cornerstone for successful therapy. 26,27 The choice of appropriate therapy has a dramatic effect, with the mortality rate reduced by as much as one half to two thirds of that seen with inadequate antibiotic coverage. Guidelines can help to facilitate appropriate management, and there has been an explosion in the number of guidelines for patient care in the past decade. 28 However, despite the best evidence and good intentions, guidelines are limited in their impact by their acceptance by practitioners and the adherence of those practitioners to them. 29,30 On the other hand, if successfully implemented, guidelines can improve the adequacy of initial therapy. 31 However, guidelines can only provide general recommendations, and the adequacy of initial coverage may be affected by the emergence of resistant pathogens. 32 Guidelines can be further refined after the analysis of local resistance patterns. This report outlines the impact of locally developed clinical guidelines on the management of patients with VAP at our institution. These guidelines were based on ATS recommendations for the management of HAP but were modified based on local antibiotic-resistance patterns. Adherence to guidelines was determined to be about 75% and was crucial to the success of the program. In this observational study, we compared outcomes in patients with VAP whose management was carried out in accordance with our guidelines (ie, the GUIDE group) with a local historical control group of VAP patients (ie, the NNIS group) who were treated prior to the introduction and implementation of the guidelines. Although there are limitations inherent with the retrospective analysis of the NNIS database, as with the comparison group of GUIDE Table 7 Summary of Imipenem Use (Guideline Patients)* Variables Directed Therapy (n 27) Stepdown in Therapy (n 18) Use Beyond Recommendations (n 16) No Imipenem Resistance Imipenem-Resistant Organisms Duration of use, d 13.4 7.9 4.6 2.7 9.3 3.1 Mortality breakdown Mortality within 14 d 0 (0) 3 (17) 2 (13) Mortality within 30 d 4 (15) 5 (28) 5 (32) Mortality during hospitalization 6 (22) 5 (28) 7 (44) Duration of use based on development of imipenem resistance Patients 54 (89%) 7 (11%) Duration of imipenem use, d 9.4 6.3 12.6 9.7 Range, d 3 34 4 31 Resistant organisms by treatment group Appropriate therapy 6 Use beyond recommendations 1 Within 30 d of HAP 6 Within 5dofHAP 1 *Values given as mean SD or No. (%). 2784 Clinical Investigations in Critical Care

Figure 2. Comparison of susceptibility to imipenem of Pseudomonas isolates and all Gram-negative isolates from patients hospitalized in West Los Angeles Veterans Affairs Medical Center critical care units from 1996 to 2001. patients, both groups represented a well-defined group of patients with clearly identified outcome measures. The groups also represented unbiased groups of patients receiving standard care unencumbered by research protocols that may not reflect current or common practice. There are, admittedly, some differences in demographics between the GUIDE group and the NNIS group. The NNIS group had a greater proportion of patients with congestive heart failure and postoperative patients than the GUIDE group. Patients in the NNIS group were hospitalized in the ICU and received mechanical ventilation for a longer time prior to their severe HAP than patients in the GUIDE group. However, despite these disparities, the severity-of-illness scores were comparable between the two groups. Multivariate analysis did not identify these patient characteristics as influences on survival (mortality). Furthermore, the distribution of respiratory and nonrespiratory isolates at the time of HAP diagnosis was also comparable. The similarities in the severity-of-illness scores and pathogens should allow valid comparisons and conclusions in the outcomes of these two groups. As expected, the adherence to guideline recommendations resulted in clear differences in initial antibiotic use for VAP. As can be appreciated in Table 3, in the NNIS group initial Gram-negative coverage was divided among the antibiotic classes of cephalosporins, semisynthetic penicillin, and carbapenem, with or without the addition of an aminoglycoside. There was not a single predominant regimen. After guideline implementation, all of the treatment consisted of an imipenem-based regimen with the addition of either an aminoglycoside or a fluoroquinolone. Vancomycin use was similar in both groups. This uniformity in treatment resulted in a greater number of GUIDE patients receiving adequate therapy. More than 80% of GUIDE patients received adequate empiric therapy for VAP compared with less than half of the NNIS patients, which represented a significant improvement in the adequacy of initial therapy. In the NNIS group, inadequate treatment can be attributed to resistant Gram-negative organisms, which are better covered by the GUIDE group. The adequacy of antibiotic coverage in the GUIDE group would be better if not for resistant Gram-positive organisms. Empiric coverage for resistant Gram-positive organisms (eg, MRSA) was recommended in the HAP guidelines only for those patients with prior resistant isolates or those with an increased risk of acquiring this infection. However, these isolates of Gram-positive organisms are beginning to occur frequently and are the major reason for inadequate initial antibiotic coverage in our study. Others have also noted 33,34 an increase in their incidence. Some institutions have included empiric vancomycin therapy in their recommendations for HAP. 31 The increased adequacy of the treatment of HAP with antimicrobial agents was associated with statistically significant lower mortality rates for GUIDE patients after 14 days of therapy and a trend toward decreased mortality at all of the time points. This finding is similar to that noted by Fagon et al 35 in their comparison of invasive and noninvasive strategies in the management of these patients. They also noted a reduction in mortality at 2 weeks but not at the end of a month or at the end of hospitalization. This is probably more a reflection of the underlying condition of patients who were managed in critical care units with severe HAP or VAP. Patients may recover from their infection, but it is often the underlying process that has the greater influence on overall survival. While the reduction in mortality was most apparent in analyses of the adequately treated NNIS patients, a trend toward decreased mortality was also present in analyses of the inadequately treated NNIS patients. The reason for this latter observation is not readily apparent but may be a testament to the greater anaerobic coverage afforded to GUIDE patients or to other changes in the processes of care. Although anaerobic cultures are not routinely obtained with respiratory cultures, these pathogens are undoubtedly important entities in VAP. 36 Initial therapy with imipenem was not associated with an increased antibiotic resistance. This is outlined in Table 7 and Figure 2. The development of imipenem resistance was associated with a longer average duration of use but was not significantly different from those without resistance. Although www.chestjournal.org CHEST / 128 / 4/ OCTOBER, 2005 2785

such resistance occurred more often in those appropriately treated with imipenem as opposed to patients treated outside of the guideline recommendations, this difference was not statistically significant. This lack of emergence of resistance can also be appreciated in the analysis of isolates from a cohort of critical care unit patients during the time frame of the study. The percentage of imipenem-resistant isolates (Pseudomonas and all other Gram-negative organisms) was unchanged (Fig 2) over the course of the study and was similar to that noted prior to guideline institution when imipenem use was more restricted. This is similar to the experience of another group that limited empiric therapy with imipenem to 72 h and also found no increase in antibiotic resistance. 37 These guidelines further support a strategy that has been labeled as the deescalation of antibiotic therapy. 20 This refers to short-term, very broadspectrum antibiotic coverage followed by changes to more narrowly focused regimens that are driven by culture results. This limited use does not expose the patient to the potential adverse effects of untreated infection or to the complications associated with long-term broad-spectrum antibiotic use, which are primarily the emergence of resistant organisms or new infections (eg, fungus-associated or antibioticassociated colitis). This mirrors a similar concept in which a clinical pneumonia infection score was used to limit antibiotic use, 38 although in this case therapy with antibiotics was not stopped but was changed. Very broad-spectrum initial therapy did not result in the emergence of antibiotic resistance as long as the duration of antibiotic use was limited. In summary, our experience indicates that guidelines can be developed to more effectively treat patients with severe HAP. Patients treated after guideline implementation had more uniform approaches to initial antibiotic therapy, resulting in more appropriate empiric therapy with a trend toward a reduction in mortality and without an increase in antibiotic resistance. Our findings mirror those of Ibrahim et al 31 but are notable for a greater mortality benefit than that found in their experience. The question will arise as to whether this guidelinedirected deescalation therapy may be preferable or can complement other approaches, such as the scheduled rotation of antibiotic classes, in terms of reducing pressure for the selection of antimicrobial resistance. 39,40 There may be additional benefit to a strategy that incorporates both rotation of antibiotics and deescalation of therapy in a guideline-directed approach. As demonstrated by other studies, 26,27 in contrast to a strategy in which very broad-spectrum agents are held in reserve and their use is culturedirected, guidelines in which broad-spectrum agents are utilized in a limited manner as first-line therapy are the preferred approach to the treatment of VAP. The findings of improvement in patient outcome are encouraging but will require a prospective evaluation for confirmation. References 1 Vincent JL, Bihari DJ, Suter PM, et al. The prevalence of nosocomial infection in intensive care units in Europe: results of the European Prevalence of Infection in Intensive Care (EPIC) Study; EPIC International Advisory Committee. JAMA 1995; 274:639 644 2 American Thoracic Society. Hospital-acquired pneumonia in adults: diagnosis, assessment of severity, initial antimicrobial therapy and preventive strategies. Am J Respir Crit Care Med 1995; 153:1711 1725 3 Fagon JY, Chastre J, Hance AJ, et al. Nosocomial pneumonia in ventilated patients: a cohort study evaluating attributable mortality and hospital stay. Am J Med 1993; 94:281 288 4 Leu HS, Kaiser DL, Mori M, et al. Hospital-acquired pneumonia: attributable mortality and morbidity. Am J Epidemiol 1989; 129:1258 1267 5 Chastre J, Fagon JY. Ventilator-associated pneumonia. Am J Respir Crit Care Med 2002; 165:867 903 6 Rello J, Ollendorf DA, Oster G, et al. Epidemiology and outcomes of ventilator-associated pneumonia in a large US database. Chest 2002; 122:2115 2121 7 Warren DK, Shukla SJ, Olsen MA, et al. Outcome and attributable cost of ventilator-associated pneumonia among intensive care unit patients in a suburban medical center. Crit Care Med 2003; 31:1312 1317 8 Meduri GU, Mauldin GL, Wunderink RG, et al. Causes of fever and pulmonary densities in patients with clinical manifestations of ventilator-associated pneumonia. Chest 1994; 106:221 235 9 Winer-Muram HT, Rubin SA, Ellis JV, et al. Pneumonia and ARDS in patients receiving mechanical ventilation: diagnostic accuracy of chest radiography. Radiology 1993; 188:479 485 10 Salata RA, Lederman MM, Shlaes DM, et al. Diagnosis of nosocomial pneumonia in intubated, intensive care unit patients. Am Rev Respir Dis 1987; 135:426 432 11 Cook D, Mandell L. Endotracheal aspiration in the diagnosis of ventilator-associated pneumonia. Chest 2000; 117:195S 197S 12 Torres A, el Ebiary M, Padro L, et al. Validation of different techniques for the diagnosis of ventilator-associated pneumonia: comparison with immediate postmortem pulmonary biopsy. Am J Respir Crit Care Med 1994; 149:324 331 13 Pugin J, Auckenthaler R, Mili N, et al. Diagnosis of ventilatorassociated pneumonia by bacteriologic analysis of bronchoscopic and nonbronchoscopic blind bronchoalveolar lavage fluid. Am Rev Respir Dis 1991; 143:1121 1129 14 Kollef MH, Bock KR, Richards RD, et al. The safety and diagnostic accuracy of minibronchoalveolar lavage in patients with suspected ventilator-associated pneumonia. Ann Intern Med 1995; 122:743 748 15 Chastre J, Fagon JY. Ventilator-associated pneumonia. Am J Respir Crit Care Med 2002; 165:867 903 16 Luna CM, Vujacich P, Niederman MS, et al. Impact of BAL data on the therapy and outcome of ventilator-associated pneumonia. Chest 1997; 111:676 685 17 Kollef MH, Sherman G, Ward S, et al. Inadequate antimicrobial treatment of infections: a risk factor for hospital mortality among critically ill patients. Chest 1999; 115:462 474 2786 Clinical Investigations in Critical Care

18 Rello J, Sa-Borges M, Correa H, et al. Variations in etiology of ventilator-associated pneumonia across four treatment sites: implications for antimicrobial prescribing practices. Am J Respir Crit Care Med 1999; 160:608 613 19 Trouillet JL, Chastre J, Vuagnat A, et al. Ventilator-associated pneumonia caused by potentially drug-resistant bacteria. Am J Respir Crit Care Med 1998; 157:531 539 20 Hoffken G, Niederman MS. Nosocomial pneumonia: the importance of a de-escalating strategy for antibiotic treatment of pneumonia in the ICU. Chest 2002; 122:2183 2196 21 Centers for Disease Control and Prevention. CDC definitions of nosocomial infections, 1988. Am Rev Respir Dis 1989; 139:1058 1059 22 Meduri GU, Chastre J. The standardization of bronchoscopic techniques for ventilator-associated pneumonia. Chest 1992; 102(suppl):557S 564S 23 Baselski VS, el Torky M, Coalson JJ, et al. The standardization of criteria for processing and interpreting laboratory specimens in patients with suspected ventilator-associated pneumonia. Chest 1992; 102(suppl):571S 579S 24 Le Gall JR, Loirat P, Alperovitch A, et al. A simplified acute physiology score for ICU patients. Crit Care Med 1984; 12:975 977 25 Itokazu GS, Quinn JP, Bell-Dixon C, et al. Antimicrobial resistance rates among aerobic Gram-negative bacilli recovered from patients in intensive care units: evaluation of a national postmarketing surveillance program. Clin Infect Dis 1996; 23:779 784 26 Iregui M, Ward S, Sherman G, et al. Clinical importance of delays in the initiation of appropriate antibiotic treatment for ventilator-associated pneumonia. Chest 2002; 122:262 268 27 Kollef MH. Inadequate antimicrobial treatment: an important determinant of outcome for hospitalized patients. Clin Infect Dis 2000; 31(suppl):S131 S138 28 Hackner D, Tu G, Weingarten S, et al. Guidelines in pulmonary medicine: a 25-year profile. Chest 1999; 116: 1046 1062 29 Rello J, Lorente C, Bodi M, et al. Why do physicians not follow evidence-based guidelines for preventing ventilatorassociated pneumonia? A survey based on the opinions of an international panel of intensivists. Chest 2002; 122:656 661 30 Ricart M, Lorente C, Diaz E, et al. Nursing adherence with evidence-based guidelines for preventing ventilator-associated pneumonia. Crit Care Med 2003; 31:2693 2696 31 Ibrahim EH, Ward S, Sherman, G et al. Experience with a clinical guideline for the treatment of ventilator-associated pneumonia. Crit Care Med 2001; 29:1109 1115 32 Ioanas M, Cavalcanti M, Ferrer M, et al. Hospital-acquired pneumonia: coverage and treatment adequacy of current guidelines. Eur Respir J 2003; 22:876 882 33 Rello J, Diaz E. Pneumonia in the intensive care unit. Crit Care Med 2003; 31:2544 2551 34 Ibrahim EH, Ward S, Sherman G, et al. Experience with a clinical guideline for the treatment of ventilator-associated pneumonia. Crit Care Med 2001; 29:1109 1115 35 Fagon JY, Chastre J, Wolff M, et al. Invasive and noninvasive strategies for management of suspected ventilator-associated pneumonia: a randomized trial. Ann Intern Med 2000; 132: 621 630 36 Dore P, Robert R, Grollier G, et al. Incidence of anaerobes in ventilator-associated pneumonia with use of a protected specimen brush. Am J Respir Crit Care Med 1996; 153:1292 1298 37 Namias N, Harvill S, Ball S, et al. Empiric therapy of sepsis in the surgical intensive care unit with broad-spectrum antibiotics for 72 hours does not lead to the emergence of resistant bacteria. J Trauma 1998; 45:887 891 38 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; 162:505 511 39 Kollef M, Vlasnik J, Sharpless L, et al. Scheduled change of antibiotic classes: a strategy to decrease the incidence of ventilator-associated pneumonia. Am J Respir Crit Care Med 1997; 156:1040 1048 40 Gruson D, Hilbert G, Vargas F, et al. Strategy of antibiotic rotation: long-term effect on incidence and susceptibilities of Gram-negative bacilli responsible for ventilator-associated pneumonia. Crit Care Med 2003; 31:1908 1914 www.chestjournal.org CHEST / 128 / 4/ OCTOBER, 2005 2787