Key words: antibiotics; intensive care; mechanical ventilation; outcomes; pneumonia; resistance

Clinical Importance of Delays in the Initiation of Appropriate Antibiotic Treatment for Ventilator-Associated Pneumonia* Manuel Iregui, MD; Suzanne Ward, RN; Glenda Sherman, RN; Victoria J. Fraser, MD; and Marin H. Kollef, MD, FCCP Study objectives: To determine the influence of initially delayed appropriate antibiotic treatment (IDAAT) on the outcomes of patients with ventilator-associated pneumonia (VAP). Setting: Medical ICU of Barnes-Jewish Hospital, St. Louis, a university-affiliated urban teaching hospital. Patients: One hundred seven consecutive patients receiving mechanical ventilation and antibiotic treatment for VAP. Interventions: Prospective patient surveillance and data collection. Measurements and results: All 107 patients eventually received treatment with an antibiotic regimen that was shown in vitro to be active against the bacterial pathogens isolated from their respiratory secretions. Thirty-three patients (30.8%) received antibiotic treatment that was delayed for > 24 h after initially meeting diagnostic criteria for VAP. These patients were classified as receiving IDAAT. The most common reason for the administration of IDAAT was a delay in writing the antibiotic orders (n 25; 75.8%). The mean time ( SD) interval from initially meeting the diagnostic criteria for VAP until the administration of antibiotic treatment was 28.6 5.8 h among patients classified as receiving IDAAT, compared to 12.5 4.2 h for all other patients (p < 0.001). Forty-four patients (41.1%) with VAP died during their hospitalization. Increasing APACHE (acute physiology and chronic health evaluation) II scores (adjusted odds ratio, 1.13; 95% confidence interval, 1.09 to 1.18; p < 0.001), presence of malignancy (adjusted odds ratio, 3.20; 95% confidence interval, 1.79 to 5.71; p 0.044), and the administration of IDAAT (adjusted odds ratio, 7.68; 95% confidence interval, 4.50 to 13.09; p < 0.001) were identified as risk factors independently associated with hospital mortality by logistic regression analysis. Conclusion: These data suggest that patients classified as receiving IDAAT are at greater risk for hospital mortality. Clinicians should avoid delaying the administration of appropriate antibiotic treatment to patients with VAP in order to minimize their risk of mortality. (CHEST 2002; 122:262 268) Key words: antibiotics; intensive care; mechanical ventilation; outcomes; pneumonia; resistance Abbreviations: APACHE acute physiology and chronic health evaluation; IDAAT initially delayed appropriate antibiotic treatment; VAP ventilator-associated pneumonia Several clinical investigations 1 6 have found that initial treatment of ventilator-associated pneumonia (VAP) with an appropriate antimicrobial regimen is associated with lower hospital mortality. Appropriate initial antimicrobial treatment is usually defined as an antibiotic regimen with demonstrated in vitro activity against the identified bacterial species asso- *From the Pulmonary and Critical Care Division (Drs. Iregui and Kollef, and Ms. Ward), and Division of Infectious Disease (Dr. Fraser), Department of Internal Medicine, Washington University School of Medicine; and Department of Nursing (Ms. Sherman), Barnes-Jewish Hospital, St. Louis, MO. For related article see page 140 ciated with infection. 7 The clinical importance of delays in the initial administration of an appropriateantimicrobial regimen to patients with VAP is not This work was supported in part by the Barnes-Jewish Hospital Foundation. Manuscript received October 16, 2001; revision accepted January 17, 2002. Correspondence to: Marin H. Kollef, MD, FCCP, Washington University School of Medicine, Campus Box 8052, 660 South Euclid Ave, St. Louis, MO 63110; e-mail: kollefm@msnotes. wustl.edu 262 Clinical Investigations in Critical Care

well defined. Additionally, the role of the duration of such delays on patient outcomes has not been previously described. Therefore, we performed a prospective clinical investigation with two main goals: to identify the occurrence of initially delayed appropriate antibiotic treatment (IDAAT) for VAP, and to determine the influence of IDAAT on patient outcomes. These study goals were selected to help identify more optimal approaches for the antimicrobial treatment of VAP. Study Location and Patients Materials and Methods This study was conducted at a university-affiliated urban teaching hospital, Barnes-Jewish Hospital (1,000 beds). During a 1-year period (February 2000 through February 2001), all patients receiving mechanical ventilation in the medical ICU were eligible for this investigation. Patients were excluded from participation if they were transferred to the medical ICU from another hospital while receiving mechanical ventilation or were admitted to the medical ICU temporarily due to unavailability of beds in another ICU. The medical ICU is a closed unit with all patient medical care provided by a multidisciplinary team directed by board-certified critical care specialists. This study was approved by the Washington University School of Medicine Human Studies Committee. Only the first episode of VAP was evaluated in this study. Study Design and Data Collection A prospective cohort study design was used. The primary outcome examined was hospital mortality. The secondary outcomes included the duration of mechanical ventilation, hospital and ICU lengths of stay, mortality attributed to VAP, and disposition following hospitalization for survivors. For all study patients, the following characteristics were prospectively recorded by one of the investigators: age; gender; race; indication for mechanical ventilation; health insurance status; premorbid lifestyle scores 8 ; severity of illness based on APACHE (acute physiology and chronic health evaluation) II scores 9 ; the presence of COPD, congestive heart failure, or underlying malignancy; administration of vasopressors; bacteremia; placement of a tracheostomy; and disposition following hospital discharge. One of the investigators made daily rounds in the ICU to identify eligible patients and to determine the onset of VAP based on the diagnostic criteria described below. As this was strictly an observational study, the investigators did not interact with the ICU treating physicians regarding the diagnosis or management of VAP. Patients entered into the study were prospectively followed up until they were successfully weaned from mechanical ventilation, were discharged from the hospital, died, or were transferred to a long-term care facility. Definitions All definitions were selected prospectively as part of the original study design. The premorbid lifestyle score was used as previously defined 8 :1 patient was independent, fully ambulatory, and not employed with restriction; 2 patient had restricted activities, could live alone and get out of the house to do basic necessities, or had severely limited exercise ability; 3 patient was house-bound, could not get out of the house unassisted, could not live alone, or could not do heavy chores; and 4 patient was bed-bound or chair-bound. We calculated APACHE II scores using clinical data available from the first 24-h period of intensive care. 9 Hospital mortality was defined as those patient deaths occurring during the initial hospital admission during which they were studied. The diagnostic criteria for VAP used in this study were modified from those established by the American College of Chest Physicians. 10 VAP was prospectively defined as the occurrence of a new and persistent radiographic infiltrate in conjunction with one of the following: positive pleural/blood culture findings for the same organism as that recovered in the tracheal aspirate or sputum; radiographic cavitation; histopathologic evidence of pneumonia; or two of the following: fever, leukocytosis, and purulent tracheal aspirate or sputum. Persistence of an infiltrate was defined as having the infiltrate present radiographically for at least 72 h. Fever was defined as an increase in the core temperature of 1 C and a core temperature 38.3 C. Leukocytosis was defined as a 25% increase in the circulating leukocytes from the baseline and a value 10 10 9 per liter. Tracheal aspirates were considered purulent if abundant neutrophils were present per high power field using Gram stain (ie, 25 neutrophils per high power field). Tracheal aspirate or BAL cultures were routinely obtained at the time the diagnosis of VAP was being considered. VAP complicating community-acquired pneumonia was considered to be present if a new infiltrate developed at least 48 h after the start of mechanical ventilation and empiric antibiotic treatment for community-acquired pneumonia. The previous infiltrates, attributed to the community-acquired pneumonia, were also required to be stable or improving in their radiographic appearance for at least 48 h prior to the development of the new infiltrate(s). Lastly, the criteria for VAP noted above also had to be met. All patients were also screened prospectively for possible alternative causes for fever and radiographic chest densities as suggested by other investigators. 11 Mortality directly related to VAP was predetermined to be present when a patient died during an episode of VAP and the death could not be directly attributed to any other cause. The administration of IDAAT was arbitrarily defined as a time period of 24 h between the point at which the diagnostic criteria for VAP were first documented, including the first identification of a new radiographic infiltrate, and the subsequent administration of appropriate antibiotic treatment. This was based on several studies 2,3 suggesting that short delays in the administration of effective antibiotic treatment could increase mortality for patients with VAP. The timing of antibiotic administration was determined from the bedside computerized nursing records. Appropriate antibiotic therapy included the administration of at least one antibiotic with in vitro activity against the bacterial pathogens isolated from the patient s respiratory secretions, as well as from blood and pleural fluid when applicable. Antibiotic Treatment for VAP Antibiotic therapy for VAP was based on our prior experience identifying the most common bacterial pathogens associated with VAP in the medical ICU. 1,6,12 All patients with suspected VAP had cultures obtained (blood, tracheal aspirate, or BAL) before antibiotic administration. Antibiotic administration included initial treatment for methicillin-resistant Staphylococcus aureus with vancomycin and Pseudomonas aeruginosa with at least one of the following antibiotics: ciprofloxacin, imipenem, cefepime, or pipercillin-tazobactam. Initial combination antibiotic treatment for P aeruginosa and the total duration of antibiotic www.chestjournal.org CHEST / 122 / 1/ JULY, 2002 263

administration were left to the discretion of the patients treating physicians. The presence of VAP due to antibiotic-resistant bacteria was prospectively defined as Gram-negative bacteria resistant to the prescribed antibiotic(s) for this class of bacteria and VAP due to Gram-positive bacteria resistant to vancomycin. Statistical Analysis All comparisons were unpaired, and all tests of significance were two tailed. Continuous variables were compared using the Student s t test for normally distributed variables and the Wilcoxon rank-sum test for nonnormally distributed variables. The 2 test was used to compare categorical variables. The primary data analysis compared patients receiving IDAAT to all other patients in the cohort. A second data analysis compared patients who died during their hospitalization to those who survived. A logistic regression model was used to control for the effects of confounding variables in order to determine the relationship between hospital mortality (dependent variable) and IDAAT (independent variable). 13,14 A stepwise approach was used to enter new terms into the logistic regression model, where 0.05 was set as the limit for the acceptance or removal of new terms. Variables entered into the logistic regression model were required a priori to have a plausible biological relationship to the dependent outcome variable in order to avoid spurious associations. 15 Model overfitting was examined by evaluating the ratio of outcome events to the total number of independent variables in the final model, and specific testing for interactions between the independent variables was included in our analyses. 14,15 Results of the logistic regression analysis are reported as adjusted odds ratios with 95% confidence intervals. Values are expressed as the mean SD (continuous variables), or as a percentage of the group from which they were derived (categorical variables). All p values were two tailed, and p 0.05 indicated statistical significance. Patients Results A total of 107 consecutive patients receiving mechanical ventilation and antibiotic treatment for VAP were evaluated. The mean age of the patients was 56.6 16.8 years (range, 17 to 89 years), and the mean APACHE II score of the study cohort was 24.6 8.1 (range, 8 to 47). Sixty-four patients (59.8%) were women, and 43 patients (40.2%) were men. The indications for mechanical ventilation included exacerbations of COPD (n 15), congestive heart failure (n 6), community-acquired pneumonia (n 28), drug overdose (n 2), ARDS (n 2), and a category in which multiple factors appeared to contribute to the episode of acute respiratory failure (n 54). The diagnosis of VAP was established using the prospectively determined clinical criteria in 102 patients (95.3%), with histologic confirmation in 4 patients (3.7%), and by the development of cavitation in a new radiographic infiltrate in 1 patient (0.9%). Antibiotic Treatment of VAP All patients were eventually treated with an appropriate antibiotic regimen that was active against the bacteria isolated from their respiratory secretions. Cultures of respiratory secretions identified at least one bacterial pathogen in all patients. Sixty-four patients (59.8%) had tracheal aspirate specimens alone, and 43 patients (40.2%) underwent bronchoscopically guided BAL to obtain cultures. The bacterial pathogens isolated from respiratory secretions included P aeruginosa (n 39), methicillin-resistant S aureus (n 24), methicillin-sensitive S aureus (n 11), Acinetobacter species (n 9), Enterobacter species (n 6), and other Gram-negative bacterial species (n 24). Six patients (5.6%) had two bacterial species isolated from their respiratory secretions. Twelve patients (11.2%) received new antibiotics within 72 h of meeting criteria for the diagnosis of VAP. Thirty-three patients (30.8%) received antibiotic treatment that was delayed for 24 h after first meeting the diagnostic criteria for VAP and were classified as receiving IDAAT. The reasons accounting for the delays in the administration of appropriate antibiotic treatment included the presence of a bacterial species resistant to the initially prescribed antibiotic regimen (n 6), delays in writing antibiotic orders (n 25), and delays in the administration of antibiotics after the initial order was written (n 2). The bacteria resistant to the initially prescribed antibiotic regimen included P aeruginosa (n 3), Acinetobacter species (n 2), and Enterobacter species (n 1). The mean duration of time from when patients initially met the diagnostic criteria for VAP until the administration of antibiotics was 28.6 5.8 h among patients classified as receiving IDAAT, compared to 12.5 4.2 h for all other patients (p 0.001; Fig 1). IDAAT attributed to the isolation of a bacterial species resistant to the initially prescribed antibiotic regimen was associated with a significantly longer duration of delay compared to the other two reasons for classification as IDAAT (37.2 7.7 h vs 26.7 3.2 h; p 0.019). Patients receiving IDAAT had a statistically shorter stay in the hospital prior to the development of VAP compared to patients not receiving IDAAT (Table 1). Secondary Outcomes Forty-four patients (41.1%) with VAP died during their hospitalization. Thirty-seven patients (84.1%) died in the ICU, and 7 patients (15.9%) died on a hospital medical floor. Hospital nonsurvivors were significantly more likely to have an underlying malignancy, require vasopressors, have greater premorbid lifestyle scores, and greater APACHE II scores (Table 1). Patients with VAP receiving IDAAT had a significantly greater hospital mortality compared to the remaining patients with VAP (Table 2). Similarly, 264 Clinical Investigations in Critical Care

patients receiving IDAAT had a significantly greater mortality attributed to VAP. Bacteremia associated with VAP was also significantly more common among patients receiving IDAAT compared to patients without IDAAT. Patients with VAP surviving their hospitalization had significantly longer stays in the ICU and hospital compared to patients with VAP who did not survive their hospitalization. Increased APACHE II scores, the presence of malignancy, and the administration of IDAAT were identified as risk factors independently associated with hospital mortality by logistic regression analysis (Table 3). Discussion Figure 1. Box plots depicting the onset of initial antibiotic administration from the time point when patients first met diagnostic criteria for VAP. The boxes represent the 25th to 75th percentiles, with the 50th percentiles shown within the boxes. The 10th and 90th percentiles are shown as capped bars. We found a statistically significant association between the administration of IDAAT and hospital mortality for patients with VAP. Multiple logistic regression analysis identified IDAAT, increasing severity of illness as measured by APACHE II scores, and the presence of underlying malignancy as important determinants of hospital mortality for this patient cohort. The most common reason identified for the administration of IDAAT was a delay in writing the orders for antibiotic treatment after patients met diagnostic criteria for VAP (75.8%). The presence of a bacterial species resistant to the initially prescribed antibiotic regimen accounted for only 18.2% of the patients classified as receiving IDAAT. Despite all patients eventually receiving Table 1 Patient Characteristics at Study Entry* IDAAT Hospital Mortality Characteristics (n 33) (n 74) (n 44) (n 63) Age, yr 53.7 15.7 57.8 17.3 0.231 59.0 15.8 54.9 17.4 0.208 Gender Male 14 (42.4) 29 (39.2) 0.753 22 (50.0) 21 (33.3) 0.084 Female 19 (57.6) 45 (60.8) 22 (50.0) 42 (66.7) Race White 16 (48.5) 49 (66.2) 0.083 22 (50.0) 43 (68.3) 0.057 nwhite 17 (51.5) 25 (33.8) 22 (50.0) 20 (31.7) Health insurance status Medicare/Medicaid 19 (57.6) 36 (48.6) 0.514 20 (45.5) 35 (55.6) 0.779 Self-pay 3 (9.1) 6 (8.1) 4 (9.1) 5 (7.9) Fee-for-service 11 (33.3) 28 (37.8) 18 (40.9) 21 (33.3) HMO/PPO 0 (0.0) 4 (5.4) 2 (4.5) 2 (3.2) COPD 10 (30.3) 19 (25.7) 0.619 10 (22.7) 19 (30.2) 0.395 Congestive heart failure 5 (15.2) 19 (25.7) 0.228 13 (29.5) 11 (17.5) 0.140 Underlying malignancy 6 (18.2) 18 (24.3) 0.482 15 (34.1) 9 (14.3) 0.016 Vasopressors 23 (67.9) 39 (52.7) 0.100 34 (77.3) 28 (44.4) 0.001 Lifestyle score 1.8 1.1 1.9 1.0 0.858 2.1 1.0 1.7 1.1 0.033 Pao 2 /fraction of 189 117 182 106 0.906 178 109 188 109 0.617 inspired oxygen APACHE II score 26.2 8.3 23.9 8.0 0.228 28.7 8.4 21.8 6.6 0.001 ICU days pre-vap 4.7 2.1 5.8 2.8 0.059 4.7 2.2 6.0 2.8 0.011 Hospital days pre-vap 5.4 2.5 7.1 3.9 0.032 5.3 2.8 7.4 3.9 0.003 *Data are presented as mean SD or. (%). HMO health maintenance organization; PPO preferred provider organization. Represents. of days in ICU and hospital prior to diagnosis of VAP. www.chestjournal.org CHEST / 122 / 1/ JULY, 2002 265

Table 2 Clinical Outcomes* IDAAT Hospital Mortality Outcomes (n 33) (n 74) (n 44) (n 63) Hospital mortality 23 (69.7) 21 (28.4) 0.01 Mortality attributed to VAP 13 (39.4) 8 (10.8) 0.001 Duration of mechanical ventilation, d 7.6 5.7 10.1 8.7 0.346 7.7 6.8 10.5 8.5 0.151 ICU length of stay, d 9.2 6.2 12.7 9.8 0.192 8.6 7.0 13.7 9.7 0.004 Hospital length of stay, d 23.6 22.7 26.5 21.6 0.147 22.3 22.5 27.9 21.3 0.009 Tracheostomy 12 (36.4) 24 (32.4) 0.691 13 (29.5) 23 (36.5) 0.453 Disposition Long-term unit 1 (10.0) 0 (0.0) 0.099 1 (1.6) Home 5 (50.0) 35 (66.0) 40 (63.5) Nursing home/extended care 4 (40.0) 16 (30.2) 20 (31.7) Other 0 (0.0) 2 (3.8) 2 (3.2) Bacteremia 4 (12.1) 1 (1.4) 0.031 4 (9.1) 1 (1.6) 0.157 *Data are presented as. (%) or mean SD. Bacteremia associated with VAP. treatment with an appropriate antibiotic regimen, these data suggest that delaying such treatment for 24 h may be associated with adverse clinical outcomes. This investigation confirms the results of previous VAP studies demonstrating an association between the administration of appropriate antibiotic treatment and clinical outcome. Alvarez-Lerma 3 showed that among 490 episodes of pneumonia acquired in the ICU setting, 214 episodes (43.7%) required modification of the initial antibiotic regimen due to either isolation of a resistant microorganism (62.1%) or lack of clinical response to therapy (36.0%). Attributable mortality from VAP was significantly lower among patients receiving initial appropriate antibiotic treatment compared to patients receiving inappropriate treatment (16.2% vs 24.7%; p 0.034). Similarly, other investigators 2,4 7 have demonstrated increased hospital mortality among patients with VAP not receiving initial appropriate antibiotic treatment. Delays in the administration of appropriate antibiotic treatment for communityacquired pneumonia have also been associated with increased hospital mortality. 16,17 Our investigation is unique in identifying a potentially important temporal threshold for the administration of appropriate Table 3 Independent Predictors of Hospital Mortality Using Logistic Regression Variables Adjusted Odds Ratio 95% Confidence Interval IDAAT 7.68 4.50 13.09 0.001 Underlying malignancy 3.20 1.79 5.71 0.044 APACHE II score (1-point increments) 1.13 1.09 1.18 0.001 antibiotic therapy to patients with VAP. Delays in the administration of appropriate antibiotic treatment beyond 24 h appeared to significantly increase the risk of hospital mortality. The most common pathogens associated with the administration of inappropriate antimicrobial treatment among patients with VAP include potentially antibiotic-resistant Gram-negative bacteria (P aeruginosa, Acinetobacter species, Klebsiella pneumoniae, and Enterobacter species) and S aureus, especially strains with methicillin resistance. 2,3,6,12 This differs from hospital-acquired bloodstream infections in which antibiotic-resistant Gram-positive bacteria (methicillin-resistant S aureus, vancomycin-resistant enterococci, and coagulase-negative staphylococci), Candida species, and less commonly antibioticresistant Gram-negative bacteria account for most cases of inappropriate antibiotic treatment. 18 However, it is important to recognize that the predominant pathogens associated with hospital-acquired infections may vary between hospitals as well as among specialized units within individual hospitals. 19,20 Therefore, updated hospital-specific or unitspecific antibiograms should be very helpful in determining the combination of antibiotics most likely to provide initial appropriate treatment of VAP and other hospital-acquired infections. 21 Our study has several limitations. First, it was performed within a single medical ICU and the results may not be generalizable to other treatment settings. However, these findings are consistent with those observed by other investigators, 2 4 suggesting that they may be applicable to other populations. Second, the study examined a relatively small sample size, limiting our ability to detect all possible differences among the study groups of interest. We only 266 Clinical Investigations in Critical Care

Figure 2. A schema outlining an approach to the antibiotic treatment of suspected VAP. CPIS clinical pulmonary infection score 24 ;F i O 2 fraction of inspired oxygen. identified six patients in whom bacterial isolates were resistant to the initially prescribed antibiotic regimens. Therefore, we cannot determine the impact of antibiotic resistance on clinical outcomes in this study. Third, we employed clinical criteria to establish the diagnosis of VAP. This was purposefully done to study a cohort of patients that more closely reflects the population of patients with VAP in the real world setting. Previous studies 5,22 have demonstrated that clinically diagnosed VAP is a good predictor of clinical outcomes compared to VAP diagnosed using bronchoscopically obtained quantitative cultures. Another important limitation of this study was that we did not attempt to determine the factors influencing physician delays in writing orders for antibiotic therapy once patients met the diagnostic criteria for VAP. Potential explanations for such delays include failure of treating physicians to recognize the presence of VAP, forgetting to write orders for antibiotics, awaiting the results of diagnostic tests such as cultures, and attributing the patient s clinical findings to a noninfectious process. Our results highlight the need for a heightened awareness of VAP among ICU clinicians. This study was an observational study not designed to prove causality between a clinical risk factor (eg, administration of IDAAT) and a specific outcome of interest (eg, hospital mortality). To validate the relationship between IDAAT and increased hospital mortality, a prospective trial randomly assigning patients to IDAAT vs appropriate antibiotic treatment administered without delay would have to be performed. Due to ethical and patient safety concerns, it is unlikely that such a study will ever be carried out. Additional large studies are also required that match patients for severity of illness to determine the attributable mortality of IDAAT, given the conflicting results of another recent smaller study. 23 In summary, our study suggests that patients receiving IDAAT are at greater risk for hospital mortality compared to patients receiving appropriate antibiotic treatment within 24 h of meeting diagnostic criteria for VAP. Given the increasing rates of nosocomial infections due to antibiotic-resistant bacteria, clinicians should consider the following recom- www.chestjournal.org CHEST / 122 / 1/ JULY, 2002 267

mendations for the antibiotic treatment of hospitalacquired infections. Risk stratification should be employed to identify those patients at greater risk for infection due to antibiotic-resistant bacteria. These risk factors include prior antibiotic treatment, prolonged lengths of stay in the hospital, and the presence of invasive devices (eg, central venous catheters, endotracheal tubes, urinary catheters). 21,24 Patients at increased risk for infection with antibioticresistant bacteria should be treated initially with a combination of antibiotics providing coverage for the most likely pathogens to be encountered in that specific clinical setting (Fig 2). Initial antibiotic treatment should also be modified once the agent of infection is identified, or antibiotic therapy should be discontinued altogether if the diagnosis of infection becomes unlikely (ie, de-escalation of antibiotic therapy). 25,26 Such an approach to antibiotic therapy for VAP can be viewed as a strategy to balance the need to provide appropriate initial antibiotic treatment to high risk patients while avoiding unnecessary empiric antibiotic utilization which can further promote antibiotic resistance among potentially pathogenic bacteria. References 1 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 2 Luna CM, Vujachich P, Neiderman MS, et al. Impact of BAL data on the therapy and outcome of ventilator-associated pneumonia. Chest 1997; 111:676 685 3 Alvarez-Lerma F. Modifications of empiric antibiotic treatment in patients with pneumonia acquired in the intensive care unit: ICU-Acquired Pneumonia Study Group. Intensive Care Med 1996; 22:387 394 4 Rello J, Gallego M, Mariscal D, et al. The value of routine microbial investigation in ventilator-associated pneumonia. Am J Respir Crit Care Med 1997; 156:196 200 5 Heyland DK, Cook DJ, Griffith L, et al. The attributable morbidity and mortality of ventilator-associated pneumonia in the critically ill patient. Am J Respir Crit Care Med 1999; 159:1249 1256 6 Kollef MH, Ward S. The influence of mini-bal cultures on patient outcomes: implications for the antibiotic management of ventilator-associated pneumonia. Chest 1998; 113:412 420 7 Kollef MH. Inadequate antimicrobial treatment: an important determinant of outcome for hospitalized patients. Clin Infect Dis 2000; 31:S131 S138 8 Menzies R, Gibbons W, Goldberg P. Determinants of weaning and survival among patients with COPD who require mechanical ventilation for acute respiratory failure. Chest 1989; 95:398 405 9 Knaus WA, Draper EA, Wagner DP, et al. APACHE II: A severity of disease classification system. Crit Care Med 1985; 13:818 829 10 Pingleton SK, Fagon JY, Leeper KV Jr. Patient selection for clinical investigation of ventilator-associated pneumonia: criteria for evaluating diagnostic techniques. Chest 1992; 102: 553S 556S 11 Marquette CH, Georges H, Wallet F, et al. Diagnostic efficiency of endotracheal aspirates with quantitative bacterial cultures in intubated patients with suspected pneumonia: comparison with the protected specimen brush. Am J Respir Crit Care Med 1993; 148:138 144 12 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 13 Meinert CL, Tonascia S. Clinical trials: design, conduct, and analysis. New York, NY: Oxford University Press, 1986; 194 195 14 SAS/STAT User s Guide. Vol 2. Cary, NC: SAS Institute, 1990; 1071 1126 15 Concato J, Feinstein AR, Holford TR. The risk of determining risk with multivariable models. Ann Intern Med 1993; 118:201 210 16 McGarvey RN, Harper JJ. Pneumonia mortality reduction and quality improvement in a community hospital. QRB Qual Rev Bull 1993; 19:124 130 17 Dean NC, Silver MP, Bateman KA, et al. Decreased mortality after implementation of a treatment guideline for community-acquired pneumonia. Am J Med 2001; 110:451 457 18 Ibrahim EH, Sherman G, Ward S, et al. The influence of inadequate antimicrobial treatment of bloodstream infections on patient outcomes in ICU setting. Chest 2000; 118: 146 155 19 Namias N, Samiian L, Nino D, et al. Incidence and susceptibility of pathogenic bacteria vary between intensive care units within a single hospital: implications for empiric antibiotic strategies. J Trauma 2000; 49:638 645 20 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 21 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 22 Timsit JF, Chevret S, Valcke J, et al. Mortality of nosocomial pneumonia in ventilated patients: influence of diagnostic tools. Am J Respir Crit Care Med 1996; 154:116 123 23 Dupont H, Mentec H, Sollet JP, et al. Impact of appropriateness of initial antibiotic therapy on the outcome of ventilator-associated pneumonia. Intensive Care Med 2001; 27: 355 362 24 Richards MJ, Edwards JR, Culver DH, et al. socomial infections in medical intensive care units in the United States: National socomial Infections Surveillance Systems System. Crit Care Med 1999; 27:887 892 25 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 26 Dennesen PJW, van der Ven AJ, Kessels AGH, et al. Resolution of infectious parameters after antimicrobial therapy in patients with ventilator-associated pneumonia. Am J Respir Crit Care Med 2001; 163:1371 1375 268 Clinical Investigations in Critical Care