Evaluation of Outcome for Intubated Patients with Pneumonia Due to Pseudomonas aeruginosa

Transcription

1 973 Evaluation of Outcome for Intubated Patients with Pneumonia Due to Pseudomonas aeruginosa Jordi Rello, Paola Jubert, Jordi Valles, Antonio Artigas, Montse Rue, and Michael S. Niederman From the Department of Intensive Care, Hospital de Sabadell, and the Departments of Molecular Biology and Physiology, Autonomous University, Barcelona, Spain; and the Pulmonary and Critical Care Department, Winthrop-University Hospital, Mineola, New York, USA Thirty consecutively intubated patients with pneumonia due to Pseudomonas aeruginosa (cases) were prospectively observed to establish the attributable mortality rate and the prognostic value of APACHE (Acute Physiological and Chronic Health Evaluation) II scores. Four cases did not receive accurate empirical therapy and were excluded from the study. APACHE II scores were calculated within 24 hours of admission (TO), at the time of the diagnosis of pneumonia (T1), and after 72 hours of therapy (T2). The outcomes for these cases (n = 26) were compared with those for matched controls (n = 52) without pneumonia. Six cases died of causes directly related to pneumonia (group D). Two cases whose conditions clinically improved died of cardiac complications, and 18 cases had clinical resolution (group R); however, only 15 of these cases were alive at discharge. The mean APACHE II score at admission was similar (P >.20) for group R, group D, and controls. In contrast, the mean score at T1 (15.40 ± 6.07 vs ± 4.66; P <.05) and the mean score at T2 (10.40 ± 3.57 vs ± 3.93; P <.01) differed significantly for groups R and D, respectively. The overall observed and predicted mortality rates among cases and controls were 42.3% and 28.1% and 28.8% and 28.7%, respectively, while the attributable mortality rate among cases was estimated to be 13.5% (95% confidence interval, 1.95%-25.04%). We conclude that the attributable mortality rate among intubated patients with pneumonia due to P. aeruginosa is high. The APACHE II score at admission is not useful as a prognostic factor, while progression of organ dysfunction after the onset of pneumonia is an ominous sign. Ventilator-associated pneumonia (VAP) is a very common complication in intubated patients admitted to the intensive care unit (ICU), and some investigators have studied different aspects of its epidemiology [1-4]. For critically ill patients with pneumonia, the major determinants of outcome are the patient's underlying condition (e.g., degree of immunosuppression), the virulence of the pathogen, and the early institution of appropriate antibiotic treatment. Different researchers have described the association of certain pathogens with death directly related to pneumonia itself in patients with VAP [5, 6]. However, some controversy exists about the relative contribution of the underlying disease vs. the microorganism in causing death in patients with nosocomial pneumonia. Indeed, some authors have suggested that VAP should be viewed as a form of organ failure or as a manifestation of systemic inflammatory response rather than as an independent risk factor for death [2]. Received 4 January 1996; revised 8 May This work was presented in part at the 61st Annual International Scientific Assembly of the American College of Chest Physicians held on 29 October to 2 November 1995 in New York. Grant support: This work was supported in part by Fondo de Investigaciones Sanitarias de la Seguridad Social (FISS 94/1456). Reprints or correspondence: Dr. Jordi Rello, Department of Intensive Care, Hospital de Sabadell, Parc Tauli s/n, E08208 Sabadell, Barcelona, Spain. Clinical Infectious Diseases 1996; 23: by The University of Chicago. All rights reserved /96/ $02.00 Despite major advances in caring for critically ill patients and the advent of effective broad-spectrum antibiotics, the morbidity and mortality rates associated with pneumonia caused by Pseudomonas aeruginosa remain unacceptably high [7]. Indeed, it represents the leading etiology of deaths directly related to pneumonia in intubated patients [5]. However, the number of preventable deaths is likely to be much smaller than the total, although the proportion of deaths that might be labeled preventable still needs to be defined. Indices of disease severity [8] provide a method of estimating survival in certain groups of patients admitted to the ICU. Previous experiences have shown the usefulness and accuracy of the APACHE (Acute Physiological and Chronic Health Evaluation) II score in predicting the risk of death for a mixed group of ICU patients [9-12]. However, very few objective data are available [4] that validate the utility of indices of illness severity, such as the APACHE II score, for patients with VAP due to specific pathogens. The principal aims of the present study were to determine the attributable mortality rate associated with VAP caused by P. aeruginosa in patients who received early and appropriate treatment (cases) (each case was matched with two controls) and to determine the usefulness of serial APACHE II scores for estimating mortality rates among these cases. To achieve this objective, we compared the APACHE II score obtained at admission (TO) with scores obtained at the moment that pneumonia was diagnosed (Ti) and again after 72 hours of treatment (T2). Our hypothesis was that serial APACHE II

2 974 Rello et al. CID 1996;23 (November) scores might identify patients at risk of death due to pneumonia as well as serve as a measure of the impact of acquired pneumonia on outcome. Patients and Methods We performed a case-control study to evaluate the attributable mortality due to VAP caused by P. aeruginosa. All patients admitted to the ICU of Hospital de Sabadell (Barcelona, Spain) during an 18-month period (from May 1993 to October 1994) who had an episode of P. aeruginosa VAP were considered to be eligible for the study as cases and were prospectively observed. Controls were selected from ICU patients who received ventilatory support for >48 hours during the study period according to the variables for matching. A control had to have had no evidence of nosocomial respiratory infection at any time during ICU hospitalization. A computer-generated list of potential controls was obtained from a database established during the study period. The list was reviewed for the best possible match, and each individual case was matched with two controls according to the following variables: APACHE II score calculated within the first 24 hours of ICU admission (±4 points), date of birth (± 10 years), gender, and primary diagnostic categories (according to the APACHE II list) for admission. If more than two potential controls on the list were well matched to a case, the two controls whose dates of admission were nearest to that of the case were selected. The diagnosis of VAP was considered when new and persistent pulmonary infiltrates that were otherwise not explained appeared on chest roentgenograms. Moreover, at least two of the following criteria were also required: temperature of _^-38 C; leukocytosis,.^-10,000 leukocytes/mm3; and purulent respiratory secretions. Pneumonia was deemed ventilator-associated when its onset occurred after 48 hours of mechanical ventilation and it was judged not to have been developing before the start of mechanical ventilation [4]. Fiberoptic bronchoscopy with a protected specimen brush (28 cases) or bronchoalveolar lavage (18 cases) was performed within the first 12 hours after the development of a new pulmonary infiltrate. The diagnosis of P. aeruginosa pneumonia was confirmed if the protected specimen brush yielded _^-1,000 cfu of P. aeruginosa/ml or bronchoalveolar lavage fluid yielded. 10,000 cfu of P. aeruginosa/ml. Episodes due to other etiologies were excluded from the analysis. The APACHE II score for each patient was calculated at TO, T1, and T2. Combination therapy was defined as the use of two or more antibiotics to which P. aeruginosa was susceptible in vitro at the time that bronchoscopy was performed. Clinical resolution was defined as complete abatement of all signs and symptoms of pneumonia along with resolution, or lack of progression, of all abnormalities on the chest radiograph [13]. The cases' conditions were deemed clinically improved if fever disappeared and pulmonary infiltrates and physical signs of pneumonia abated [14]. Deaths were considered related to the pulmonary infection if the case died before any objective response to antimicrobial therapy or if the pulmonary infection was thought to be a contributing factor to death in cases with other comorbidities [6]. The attributable mortality due to nosocomial pneumonia is the excess mortality caused by the infection. The rate was determined by subtracting the overall mortality rate among the controls from the overall mortality rate among the cases. The risk ratio was the ratio of the overall mortality rate among the cases to that among the controls [15]. Microbiology Cases in whom VAP was suspected underwent bronchoscopy as previously described [16]. The protected specimen brushes were transected, and the sections were placed into sterile vials containing 1 ml of lactate Ringer's solution; the vials were vigorously agitated for at least 60 seconds to remove all the material from the brushes. Specimens were immediately sent to the laboratory for quantitative cultures. Aliquots of 0.01 ml were taken from the original suspension and inoculated onto plates with blood agar, MacConkey agar, buffered charcoal yeast extract agar, and Sabouraud media. One 0.1-mL aliquot as well as one 0.1 ml of a 10-3 dilution and one 0.1 ml of a 10-6 dilution were inoculated onto plates with chocolate agar. Culture plates were incubated at 37 C under adequate aerobic and anaerobic conditions; all plates except those with Sabouraud media were evaluated for growth at 24 and 48 hours. Bronchoalveolar lavages were performed as previously reported [4]. The specimens of recovered bronchoalveolar lavage fluid underwent two serial 10-fold dilutions, and 0.01-mL aliquots of the original suspension and each dilution were placed onto plates with the same media used for the protected specimen brush samples. All isolates from the protected specimen brushes and bronchoalveolar lavage fluids were identified by standard laboratory techniques [17]. Bacterial counts of 10 3 cfu/ml in a PBS specimen or 104 in a BAL specimen were used as the cutoff point for the diagnosis of pneumonia. Cultures of two blood specimens from all patients were done simultaneously; if pleural fluid was present, cultures of this material were also performed. Statistical Analysis Descriptive statistical analysis was performed. Means were compared with the Student's t test and Mann-Whitney test. Proportions were compared by means of the x2 test with Yates' correction or Fisher's exact test when necessary. Confidence intervals for proportions were obtained on the basis of binomial distribution. The McNemar test with continuity correction was used to determine whether the overall mortality rates among

3 CID 1996; 23 (November) Outcome of Pneumonia Due to Pseudomonas Species 975 Table 1. Diagnostic category at ICU admission of cases with VAP due to Pseudomonas aeruginosa and their matched controls. Diagnostic category Cases (n = 26) No. of patients Controls (n = 52) Postoperative respiratory failure 4 7 Hypovolemic shock 4 7 Ischemic cardiomyopathy 3 7 Chronic obstructive pulmonary disease 3 8 Cardiovascular failure due to infection 2 4 Gastrointestinal* 2 2 Respiratory failure due to infection 2 5 Respiratory failure due to neoplasia 1 1 Multiple trauma 1 1 Gastrointestinal bleeding 1 2 Poisoning 1 2 Respiratory failure due to aspiration 1 3 Respiratory (other) 1 3 NOTE. VAP = ventilator-associated pneumonia. * Pancreatitis in all four patients. cases and controls were significantly different. All P values and confidence intervals were two-sided. Results During the study period, 30 consecutive cases with P. aeruginosa VAP were prospectively observed in our ICU. All cases had received previous antimicrobial therapy. Four cases (two of whom ultimately died) were excluded from the study because the P. aeruginosa isolates recovered from them were resistant to empirical treatment. The remaining 26 patients received combination therapy (amikacin plus piperacillin, ciprofloxacin, or imipenem). The mean age ± SD of these 26 cases was 62.8 ± 20.8 years; 20 cases were men, and six were women. The diagnostic categories of the cases at ICU admission are detailed in table 1. The clinical characteristics of the cases and controls are shown in table 2. The median interval between ICU admission and identification of pneumonia was 10 days (range, 3-43 days). The mean Pao 2/Fio2 ± SD at the time of the diagnosis of pneumonia was ± 59.3 mm Hg. Six cases died of causes directly related to pneumonia (only one of the 12 matched controls died) (group D). Two cases whose conditions clinically improved died of cardiac complications, and 18 cases had complete clinical resolution (group R). However, only 15 of these cases were alive at discharge from the ICU. The overall observed mortality rate among cases was 42.3% (95% CI, 32.6%-51.9%), while the predicted mortality rate (based on APACHE II scores at admission) among cases was 28.1%. The observed mortality rate among controls was 28.8%, while the predicted mortality rate was 28.7% (P >.20). The characteristics of the cases who died in the ICU are shown in table 3. The mean APACHE II scores over time are shown in figure 1. The mean APACHE II score ± SD at TO was similar for group D and group R (15.16 ± 6.91 vs ± 6.17, respectively; P =.41). In contrast, a statistically significant difference (P <.05) was found when the mean APACHE II scores ± SDs for groups D and R at Ti (20.83 ± 4.66 vs ± 6.07, respectively) and at T2 (25.50 ± 3.93 vs ± 3.57, respectively) were compared. Changes in APACHE II scores from Ti to T2 were 4.67 and -5.0 for groups D and R, respectively. The mean APACHE II score ± SD at 48 hours before the diagnosis of pneumonia was 13.3 ± 5.0 for the six cases in group D, and individual scores 48 hours before the diagnosis of pneumonia were lower than scores at Ti for five of these cases. In this study, all cases whose APACHE II scores were _^-20 after 72 hours of treatment for pneumonia died. No significant differences were found in mean scores at TO, T 1, or T2 for the 20 cases in group R independent of the final outcome (survival or death). Nevertheless, the five cases in group R who died in the ICU had a trend of higher APACHE II scores at T2 than did the 15 cases who were alive at discharge (12.4 ± 2.8 vs. 9.7 ± 3.6, respectively; P =.12). Demographic profiles and clinical characteristics of cases and controls were similar (table 2). Controls had a mean APACHE II score ± SD in the first 24 hours of admission of ± No significant differences were found between Table 2. Characteristics of 26 cases with VAP due to Pseudomonas aeruginosa and 52 controls. Characteristic Mean age (y) ± SD No. of males/females No. (%) with Value for indicated characteristic Cases 62.8 ± /6 Controls 60.4 ± /18 P value >.20 >.20 Prior trauma 5 (19.2) 2 (3.8).069 Antecedent surgery 11 (42.3) 18 (34.5) >.20 Steroid therapy 5 (19.2) 7 (13.5) >.20 Digestive bleeding 3 (11.5) 1 (1.9).069 Renal failure 1 (3.8) 3 (5.8) >.20 Antecedent COPD 10 (38.5) 14 (26.9) >.20 Cardiopathy 4 (15.4) 12 (23.1) >.20 Coma 2 (7.7) 10 (19.2).183 Mean period of intubation* (d) (±SD) 13.7 ± ± 24.9 >.20 Mean APACHE II score (± SD) on first day ± ± 6.53 >.20 Estimated mortality rate (%) >.20 NOTE. COPD = chronic obstructive pulmonary disease; VAP = ventilator-associated pneumonia. * For cases, the duration of intubation before pneumonia.

4 976 Rello et al. CID 1996;23 (November) Table 3. Characteristics of cases with VAP due to Pseudomonas aeruginosa who died in the intensive care unit. Case no., age (y)/sex Underlying disease Duration (d) of intubation before pneumonia diagnosis Outcome Day of death after pneumonia diagnosis Cause of death 1, 87/M CAP 4 D 4 MOF 2, 75/M Pancreatitis 9 D 4 MOF 3, 66/M COPD 8 I 4 Cardiac arrest, DNR order 4, 80/F Myocardial infarction 6 I 5 Cardiac arrest, DNR order 5, 82/F FUO 4 D 8 MOF 6, 71/M Digestive bleeding 10 D 8 MOF 7, 27/M CAP 15 D 13 Hypoxia 8, 74/M Biliary sepsis 4 D 18 Hypoxia 9, 22/M Multiple trauma 24 CR 28 Hypovolemia 10, 60/M Nosocomial pneumonia 22 CR 38 MOF 11, 61/M Multiple trauma 32 CR >60 Superinfection NOTE. CAP = community-acquired pneumonia; COPD = chronic obstructive pulmonary disease; CR = clinical resolution; D = death directly related to pneumonia; DNR = do not resuscitate; FUO = fever of unknown origin; I = clinically improved condition; MOF = multiorgan failure; VAP = ventilatorassociated pneumonia. age, gender, and other relevant variables (table 2). The mean duration of ventilatory support ± SD for controls was 14.3 ± 24.9 days compared with 13.7 ± 10.9 days (P >.20) for cases at Ti. Overall, successful matching was achieved for 571 (87.0%) of 656 variables. The attributable mortality rate associated with P. aeruginosa VAP was estimated to be 13.5% (95% CI, 1.95%-25.04%), while the estimated risk ratio for death was 1.46 (95% CI, ). The mean length of ICU stay ± SD for surviving cases was 44.9 ± 31.2 days compared with 15.6 ± 28.2 days for surviving controls. This difference * Time 0 Time 1 Time 2 Figure 1. Mean APACHE II scores ± SD over time. R = cases with ventilator-associated pneumonia (VAP) due to Pseudomonas aeruginosa who had clinical resolution or unrelated deaths; D = cases with P. aeruginosa VAP who died of causes directly related to pneumonia; * = P <.05; TO = time of admission; Ti = time of diagnosis of pneumonia; T2 = after 72 hours of treatment. D R (29.3 days) was statistically significant (95% CI, days). Discussion It remains unclear whether most critically ill patients die as a direct result of infection or simply with infection present. This question is relevant; if the attributable mortality rate is high, the effect of different strategies for prevention or different approaches to treatment may have great implications on outcome. Attributable mortality directly results from nosocomial infection rather than from underlying conditions. The attributable mortality rate associated with pneumonia due to Pseudomonas or Acinetobacter species in a population of intubated patients (most of whom had undergone operations) in France [6] was recently reported to be 27% (95% CI, 8%-46%). The overall observed mortality rate among cases in our study was high (42.3%) and differed significantly from that among matched controls (28.8%). Indeed, the attributable mortality rate among our cases was estimated to be 13.5%, and the estimated risk ratio for death was 1.46 despite early and appropriate combination therapy; these findings highlight the importance of P. aeruginosa as a potentially preventable cause of death. This rate might have been overestimated if patients who received inappropriate therapy were not excluded; however, by including only those patients who received adequate therapy in the study, only the influence of pneumonia on mortality was evaluated. Critical to the validity of this study was successful matching of cases with controls on the basis of relevant confounding variables. Adjustment for different underlying diseases and severity of illness was made, since these are the most important

5 CID 1996; 23 (November) Outcome of Pneumonia Due to Pseudomonas Species 977 variables for accurate analysis of outcome indicators. Of 656 analyzed variables in our cohort, 571 (87.0%) were matched successfully. In addition, no significant differences were found between the duration of total intubation for controls and the duration of intubation before pneumonia for cases, thus ensuring that both cases and controls were exposed to risks for the same duration. Exclusion of patients who did not receive early and appropriate therapy ensures that the excess mortality rate was not due to inappropriate therapy. Although more cases had prior trauma, this difference was not significant. Other potentially confounding variables (including antecedent surgery, presence of coma, and other comorbidities [table 2)) were compared, and no significant differences were found. There was no significant difference in the estimated mortality rates among cases and controls (28.1% vs. 28.7%, respectively), thereby confirming that significant differences in baseline predicted mortality rates were not present in our study. In addition, no significant differences were found between the estimated and the observed mortality rates among controls. Nevertheless, a significantly higher proportion of cases than controls died, thus indicating that the excess mortality was due to infection. In addition, the predicted mortality rates were estimated on the basis of established baseline predicted mortality rates (APACHE II system), and the mortality rate in excess of the predicted mortality rate was noted to be similar to the attributable mortality rate (14.2% vs. 13.5%, respectively; P >.20). There are several different criteria defining organ dysfunction that have been published. Multipurpose scoring systems for ICU patients that are based on physiological abnormalities, age, and chronic health status are validated and could be easily followed serially [4, 18]. The APACHE II score is used by many ICUs to measure disease severity since outcome correlates with the score and models exist that estimate mortality rates among selected groups of patients [19]. This score has also been used to assess ICU efficiency by using criteria for economic and clinical performance [20]. Although APACHE II scoring is a general prognostic system, it has been used successfully for a variety of specified clinical conditions, including pancreatitis [21], myocardial infarction [12], and infection [22, 23]. However, we observed that the mean APACHE II score on the first day of admission was not useful for estimating survival in patients with VAP caused by P. aeruginosa. In fact, once VAP developed, some patients' APACHE II scores rapidly increased. This rapid increase in severity of illness was a direct consequence of the acquisition of pneumonia and predicted the subsequent outcome. Consequently, the APACHE II score on the first day of admission should not be used in epidemiologic studies for purposes of analysis of prognostic factors or for stratification in therapeutic trials of intubated patients with pneumonia caused by P. aeruginosa. In contrast, the score on the day of the diagnosis of pneumonia was statistically higher for cases who died of causes directly related to pneumonia than for the other cases. This finding is in agreement with other studies reporting that the predictive accuracy of the APACHE II score (on the first day) appeared to decrease with the length of time that the patient stayed in the ICU [24]. This observation may also be of interest for studies of nosocomial infections from other sources. However, an important overlap between APACHE II scores at Ti for cases in groups R and D exists (data not shown). This fact is not surprising, since predictive models are useful for describing groups of patients but are very limited when they are applied to specific patients. Nevertheless, we also found that calculation of scores at T2 improved the prediction of outcome for our study population. Indeed, we did not find an overlap between scores at T2 for cases in groups R and D. Our observations suggest that sequential APACHE II scores after the diagnosis of pneumonia may permit an early prediction of outcome. Future studies are required to validate the accuracy of APACHE II scoring for predicting mortality rates in patients with nosocomial pneumonia and to compare it with other models or with the clinical recognition of progressive sepsis. In the present study, all cases with an APACHE II score at T2 of _^-20 died. In addition, most cases who were alive at discharge had scores at T2 that were less than those at T 1. These observations indicate that the progression of organ dysfunction over the first 72 hours after the onset of pneumonia is an ominous sign and a predictor of mortality. Furthermore, these observations stress how critically important the hours immediately after the diagnosis of P. aeruginosa VAP are to resolution, and they support the value of the early institution of appropriate antibiotic treatment. Our study population was not large; therefore, further multicenter studies of data on large groups of patients with pneumonia caused by P. aeruginosa need to be performed. In addition, our findings are valid only for patients with VAP caused by P. aeruginosa susceptible to initially prescribed antibiotics and may not be true for nonintubated patients with P. aeruginosa infection or for intubated patients with pneumonia caused by other organisms. Because risk factors, mortality rates, and antibiotic susceptibility may vary according to the etiologic agent [25], further studies should evaluate the value of sequential scores of the severity of disease due to other pathogens. This evaluation is of interest since a review of the outcome of VAP that is based on scores of disease severity has been recommended for defining groups of patients with similar prognoses [4]. However, since P. aeruginosa is the leading etiology associated with death in patients with VAP and since most cases of pneumonia due to gram-positive pathogens and Haemophilus influenzae are associated with clinical resolution with appropriate antibiotic treatment [5], our analysis may be of great interest to clinical practice. In summary, our results confirm that a high proportion of intubated patients with P. aeruginosa VAP die of infection. In

6 978 Rello et al. CID 1996;23 (November) addition, our findings would be useful to clinicians seeking to identify which patients with P. aeruginosa VAP have a particularly high risk of death. Further studies should investigate the value of sequential scores of disease severity as an early predictor of outcome for patients with pneumonia due to different etiologies. Acknowledgment The authors are indebted to ConcepciOn Montes for technical assistance. References 1. Rello J, Quintana E, Ausina V, et al. Incidence, etiology, and outcome of nosocomial pneumonia in mechanically ventilated patients. Chest 1991; 100: Rodriguez JL, Gibbons KJ, Bitzer LG, Dechert RE, Steinberg SM, Flint LM. Pneumonia: incidence, risk factors, and outcome in injured patients. J Trauma 1991; 31: Torres A, Aznar R, Gatell JM, et al. Incidence, risk, and prognosis factors of nosocomial pneumonia in mechanically ventilated patients. Am Rev Respir Dis 1990; 142: Wunderink RG, Mayhall CG, Gibert C. Methodology for clinical investigation of ventilator-associated pneumonia. Epidemiology and therapeutic investigation. Chest 1992; 102(suppl):580S -8S. 5. Rello J, Ausina V, Ricart M, Castella J, Prats G. Impact of previous antimicrobial therapy on the etiology and outcome of ventilator-associated pneumonia. Chest 1993; 104: Fagon JY, Chastre J, Hance AJ, Montravers P, Novara A, Gibert C. Nosocomial pneumonia in ventilated patients: a cohort study evaluating attributable mortality and hospital stay. Am J Med 1993; 94: Rello J, Ausina V, Ricart M, et al. Risk factors for infection by Pseudomonas aeruginosa in patients with ventilator-associated pneumonia. Intensive Care Med 1994; 20: Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med 1985; 13: Chu DY. Predicting survival in AIDS patients with respiratory failure. Application of the APACHE II scoring system. Crit Care Clin 1993; 9: Kruse JA, Thill-Baharozian MC, Carlson RW. Comparison of clinical assessment with APACHE II for predicting mortality risk in patients admitted to a medical intensive care unit. JAMA 1988; 260: McClish DK, Powell SH. How well can physicians estimate mortality in a medical intensive care unit? Med Decis Making 1989; 9: Moreau R, Soupison T, Vauquelin P, Derrida S, Beaucour H, Sicot C. Comparison of two simplified severity scores (SAPS and APACHE II) for patients with acute myocardial infarction. Crit Care Med 1989; 17: Chow AW, Hall CB, Klein JO, Kammer RB, Meyer RD, Remington JS. Evaluation of new antiinfective drugs for the treatment of respiratory tract infections. Clin Infect Dis 1992; 15(suppl 1): Montravers P, Fagon JY, Chastre J, et al. Follow-up protected specimen brushes to assess treatment in nosocomial pneumonia. Am Rev Respir Dis 1993; 147: Pittet D, Tarara D, Wenzel RP. Nosocomial bloodstream infection in critically ill patients. Excess length of stay, extra costs, and attributable mortality. JAMA 1994; 271: Wimberley N, Falling LS, Bartlett J. A fiberoptic bronchoscopy technique to obtain uncontaminated lower airway secretions for bacterial culture. Am Rev Respir Dis 1979; 119: Lennette EH, Balows A, Hausler WJ Jr, Shadomy HJ, eds. Manual of clinical microbiology. 4th ed. Washington, DC: American Society for Microbiology, Lemeshow S, Klar J, Teres D, et al. Mortality probability models for patients in the intensive care unit for 48 or 72 hours: a prospective, multicenter study. Crit Care Med 1994; 22: Chang RW, Jacobs S, Lee B, Pace N. Predicting deaths among intensive care unit patients. Crit Care Med 1988; 16: Cullen DJ, Chemow B. Predicting outcome in critically ill patients [editorial]. Crit Care Med 1994; 22: Larvin M, McMahon MJ. APACHE II score for assessment and monitoring of acute pancreatitis. Lancet 1989; 2: Dellinger EP, Wertz MJ, Meakins JL, et al. Surgical infection stratification system for intra-abdominal infection: multicenter trial. Arch Surg 1985; 120: Durocher A, Saulnier F, Beuscart A, et al. A comparison of three severity score indexes in an evaluation of serious bacterial pneumonia. Intensive Care Med 1988; 14: Sleigh JW, Brook RI, Miller M. Time-dependent error in the APACHE II scoring system. Anaesth Intensive Care 1992;20: Rello J, Torres A. Microbial causes of ventilator-associated pneumonia. Semin Respir Infect 1996; 11:24-31.