Jean-Yves Fagon,* Jean Chastre, Yves Domart, Jean-Louis Trnuillet, and Claude Gibert

Transcription

1 538 Mortality Due to Ventilator-Associated Pneumonia or Colonization with Pseudomonas or Acinetobacter Species: Assessment by Quantitative Culture of Samples Obtained by a Protected Specimen Brush Jean-Yves Fagon,* Jean Chastre, Yves Domart, Jean-Louis Trnuillet, and Claude Gibert From the Service de Reanimation Medicale, Hopital Bichat, Paris, France Ventilator-associated pneumonia (VAP) due to multiresistant pathogens is associated with a high death rate. We analyzed the relationship betweenvap due to Pseudomonas or Acinetobacter species and death by comparing the outcomes for patients colonizedwith these pathogens (bacterial counts of <10 3 cfu/ml) with those for patients with pneumonia due to these pathogens (bacterial counts of ~103 cfulml). Samples were obtained systematically with a protected specimen brush when pneumonia was suspected. Clinical characteristics at admission to our intensivecare unit and clinical features at the time of suspicion of VAP were not significantlydifferent between colonized patients and those with YAP. Mortality rates were 29% among colonized patients and 73% among patients with VAP (P <.001). These results demonstrate a relationship between a high mortality rate and the development of pneumonia due to multiresistant, nonfermenting, gram-negative bacilli (~103 cfu/ml) in the lower airways of patients receiving ventilatory support. Ventilator-associated pneumonia (VAP) in critically ill patients is associated with a high mortality rate [1]. However, the respective influences ofthe severity ofthe underlying disease(s) and the occurrence of pneumonia on death are difficult to assess in patients treatedwith mechanicalventilationbecause these are the most severely ill patients with the most severe nosocomial infection [2]. In other words, it is difficult to establish whether such patients would have survived if nosocomial pneumonia had not occurred. Nevertheless, nosocomial pneumonias due to nonfennenting, gram-negative bacilli and particularly those due to Pseudomonas or Acinetobacter species have repeatedly been described as being associated with the highest death rates among patients receiving ventilatory support [1, 3]. Analysis ofthe epidemiologic characteristics ofpneumonia developing in patients receiving ventilatory support has been limited by the difficulty encountered in clearly distinguishing between true bacterial pneumonia and lung processes mimicking pneumonia that are or are not associated with colonization ofthe lower airways [4, 5]. The technique ofobtaining uncontaminated lower-airway secretions for quantitative cultures with use ofa protected specimen brush (PSB) during fiberoptic bronchoscopy was introduced by Wimberley et a1. [6]; this technique is considered to be one ofthe more appropriate tools for identifying pneumonia in patients undergoing mechanical ventilation or at least for differentiating between patients with Received 22 November 1995; revised 8 May *Current affiliation: Service de Reanimation Medicale, H6pital Broussais, Paris, France. Reprints or correspondence: Dr. Jean-Yves Fagon, Service de Reanimation Medicale, Hopital Bichat, 46 rue Henri-Huchard, Paris Cedex 18, France. Clinical Infectious Diseases 1996;23: by The University of Chicago. All rights reserved /96/ $02.00 high distal bacterial counts (~10 3 cfu/ml) who are considered to have true pneumonia and those with low distal bacterial counts (< 10 3 cfulml) who are considered to have simple colonization of the lower airways [4-7]. Indeed, the ability to make such a distinction is one of the ways of recognizing the specific epidemiologic characteristics of VAP in a comparison of VAP with ventilator-associated colonization of the lower airways. The aim of this study was to compare the clinical characteristics of and the outcomes for two groups ofpatients receiving ventilatory support who were suspectedofhaving pneumoniaon clinical grounds; pneumonia (bacterial counts of ~ 10 3 cfu/ml) or colonization (bacterial counts of < 10 3 cfulml) due to Pseudomonas or Acinetobacter species was diagnosed for these patients on the basis of the results ofquantitative cultures ofsamples obtained a with PSB. Methods Patient Selection Over a 5-year period, the results of quantitative cultures of PSB samples that demonstrated the presence of Pseudomonas or Acinetobacter species were used to identify two groups of patients receiving ventilatory support: patients with pneumonia due to these pathogens who had bacterial counts of ~ 10 3 cfulml and patients whose lower respiratory tract was merely colonized by these nonfermenting, gram-negative bacilli and who had bacterial counts of < 10 3 cfulml, even if > 10 3 cfu of another potential pathogenlml grew. During the study period, PSB samples were obtained from every patient suspected of having VAP. The clinical suspicion of VAP was based on the presence of a new, persistent lung infiltrate on chest roentgenograms, macroscopically purulent tracheal aspirates, and mechanical ventilation for >48 hours. Fiberoptic broncho-

2 em 1996;23 (September) Mortality Due to VAP with Pseudomonas or Acinetobacter 539 scopic examination and PSB sampling were performed following a protocol previously described in detail [7]. Patients identified as having pneumonia were treated with antimicrobial agents effective against the isolated Pseudomonas or Acinetobacter species. No new therapy with antibiotics effective against these bacilli was given to patients colonized by these microorganisms. For the purpose ofthe present study, patients from whom Pseudomonas or Acinetobacter species were isolated at low concentrations «10 3 cfulml) in association with significant concentrations (~10 3 cfulml) of other pathogens were classified as being colonized with Pseudomonas or Acinetobacter species. They were treated with antimicrobial agents effective against pathogens cultured at counts of ;=:10 3 cfulml. Death in the intensive care unit (lcu) was used as the unique endpoint. Data Collection The following data were recorded at the time of admission to our ICU: age; sex; location prior to admission; severity of underlying medical conditions stratified according to the criteria of McCabe and Jackson [8] as fatal, ultimately fatal, or nonfatal; indication for ventilatory support based on the classification ofzwillich and co-workers [9]; simplified acute physiology score [10]; and presence or absence of infection and/or organ system (respiratory, cardiac, renal, hepatic, neurological, and/or hematologic) failure(s), definitions of which have been previously described [11]. In addition, the following clinical features were noted on the day that VAP was clinically suspected: duration ofmechanical ventilation; use of prior antimicrobial therapy; presence or absence of any of the six organ system failures cited above; temperature; WBC count; arterial oxygen pressure (Pao2)/ inspired oxygen fraction (Fio-) ratio; radiological score determined by a technique previously described [1]; and changes in temperature, WBC count, and Paoj/f'io, ratio during the 3 days preceding bronchoscopy. Statistical Analysis Patient characteristics are expressed as means ± SD or as the number of patients. Comparisons were made with use of the Student's z-test or analysis of variance for continuous variables and the X 2 statistic for categorical variables; P values of <.05 were considered statistically significant. Results Of 941 patients who received ventilatory support for >48 hours, 60 (6.4%) had pneumonia due to Pseudomonas species (45) or Acinetobacter species (15) (bacterial counts of ~ 10 3 cfu/ml), and 48 (5.1%) were colonized with Pseudomonas species (31) or Acinetobacter species (17) (bacterial counts of < 10 3 cfu/ml). Another organism (Streptococcus species, 2; Table 1. Characteristics ofpatients with pneumonia or colonization with Pseudomonas or Acinetobacter species at admission to an rcu. Value Pneumonia Colonization Parameter (n = 60) (n = 48) Mean age (y) ± SD 63 ± II 60 ± 12 No. (%) of males 45 (75) 38 (79) No. (%) admitted from Community 9 (15) 11 (23) Wards 19 (32) 12 (25) Another ICU 32 (53) 25 (52) No. (%) with severity of underlying disease Absence 3 (5) 7 (15) Nonfatal 48 (80) 25 (52) Ultimately or rapidly fatal 9 (15) 16 (33) No. (%) with indication for ventilatory support Chronic airway obstruction 9 (15) 6 (13) Other pulmonary disease 8 (13) 8 (17) Postoperative respiratory failure 31 (52) 22 (46) Drug overdose 0 3 (6) Neurological emergency 8 (13) 3 (6) Miscellaneous 4 (7) 6 (13) SAPS ± SD 12.1 ~ ~ 4.2 No. (%) with organ failure Respiratory 60 (100) 48 (100) Cardiovascular 28 (47) 25 (52) Neurological 18 (30) 18 (38) Renal 15 (25) 7 (15) Hepatic 7 (12) 3 (6) Hematologic 3 (5) 4 (8) Mean total number of organs affected ± SD 2.2 ± ::t: 1.2 NOTE. No differences between groups were significant. ICU = intensive care unit; SAPS = simplified acute physiology score. Branhamella catarrhalis, 2; Haemophilus infiuenzae, 1; and methicillin-sensitive Staphylococcus aureus, I) was simultaneously isolated at a significant concentration from six of 48 colonized patients. Pneumonia due to S. aureus, Legionella pneumophila, and Escherichia coli subsequently developed in three patients; these cases were diagnosed by conventional cultures yielding ~ 10 3 cfu/ml or buffer charcoal yeast extract (BCYE) cultures. According to the results of antibiotype and biotype determinations, no clustering of cases of nosocomial infections due to Pseudomonas or Acinetobacter species was observed in the ICU during the study period. The results ofa comparison ofthe characteristics ofinfected and colonized patients are shown in table 1; no significant differences concerning location before ICU admission, preexisting extensive functional disability, major comorbidities, acute physiological abnormalities, and age were found. Moreover, as indicated in table 2, patients with pneumonia did not differ clinically from patients with colonization at the time that they were suspected of having pneumonia; this difference was evaluated by temperature, WBC count, Pao2IFio2 ratio, radiological score, presence of organ

3 540 Fagon et al. ern 1996;23 (September) Table 2. Clinical features at the time ofclinical suspicion ofpneumonia in patients with pneumonia or colonization with Pseudomonas or Acinetobacter species. Value Parameter Pneumonia(n = 60) Colonization(n = 48) Mean duration (d) of mechanicalventilation ± SD No. (%) with prior antimicrobial therapy Mean temperature (0C) ± SD Mean WBC count (from 3 ) ± SD Mean PaozlFioz ratio (rom Hg) ± SD Mean radiological score* ± SD No. (%) with organ failure Respiratory Cardiovascular Neurological Renal Hepatic Hematologic Mean total number of organs affected ± SD Change" in Mean temperature ec) ± SD Mean PaozlFioz ratio (rom Hg) ± SD Mean WBC count (zmrrr') ± SD 12 ± (78) 38.6 ± ,300 ± 7, ± ± (100) 19 (32) 22 (37) 13 (22) 6 (10) 5 (8) 2.1 ± ± 0.9 ~0.7 ± 80 +2,400 ± 8, ± (69) 38.2 ± ,600 ± 9, ± ± (100) 14 (29) 8 (17) 9 (19) 2 (4) 2 (4) 1.7 ± ± ± ± 6,800 NOTE. No differences between groups were significant. All values were noted at the time of bronchoscopy. PaozlFioz = arterial oxygen pressure/inspired oxygen fraction. * Determinedby a previously described technique [1]. t Changes in temperature, Paoj/Fio, ratio, and WBC count are differencesbetween values at the time ofbronchoscopy and values on the third day before bronchoscopy. failure, and changes in temperature, WBC count, or PaoiFio 2 ratio during the 3 days preceding bronchoscopy as well as by prior antimicrobial therapy (including the number of patients who received therapy with potentially effective antibiotics against Pseudomonas and Acinetobacter species). The mortality rate among patients with pneumonia was 73%, compared with 29% among patients colonized with Pseudomonas or Acinetobacter species (P <.00I); the colonized patients who died included three (50%) of the six patients with pneumonia due to another pathogen. The mortality rate among patients with pseudomonas pneumonia was 73%, and the mortality rate among patients with acinetobacter pneumonia was 73%; the mortality rates among colonized patients were 29% and 29%, respectively (NS). The results of an analysis of mortality as a function of the findings of quantitative cultures of PSB samples are shown in figure 1. Death occurred in 12 of 18, 12 of 14, 12 of 15, and 8 of 13 patients with pneumonia who had bacterial counts in the ranges < 10 4, 10 4 _< 10 5, < 10 6, and ~ 10 6 cfu/ml, respectively (NS). One of 12 colonized patients whose bacterial counts were < 10 2 cfu/ml and 13 of 36 colonized patients with bacterial counts between 10 2 and < 10 3 cfu/ml died (P =.08). Thirty-five ofthe colonized patients were not treated with antibiotics effective against Pseudomonas or Acinetobacter species during the follow-up period. In fact, 13 ofthe 48 colonized patients were receiving therapy with antibiotics potentially active against these nonfermenting, gram-negative bacilli during their ICU stay for treatment of an extrapulmonary infection; no difference in their mortality rate was observed (data not shown). During the study period, 240 (28.8%) ofthe 833 patients who received ventilatory support for >48 hours and did not have pneumonia or colonization with these pathogens died. Discussion This study of patients who received ventilatory support for >48 hours in a medical ICU and were clinically suspected of :! cgi 'OJ III 20 Q. '0 ci z 10 O~-_----J Figure % 66.6% Bacterial counts (cfu/ml) Mortality rates (percentages given above the columns) among patients with pneumonia or colonization with Pseudomonas or Acinetobacter species; the rates are based on the results ofquantitative cultures of samples obtained by a protected specimen brush. Open bars = survivors; solid bars = nonsurvivors.

4 em 1996;23 (September) Mortality Due to VAP with Pseudomonas or Acinetobacter 541 having YAP because ofthe presence ofanew lung infiltrate and purulent tracheal secretions demonstrated that bacteriologically proven VAP due to Pseudomonas or Acinetobacter species was associated with a higher mortality rate than colonization of the lower airways with these pathogens; pneumonia or colonization was determined on the basis of the results of quantitative cultures of PSB samples. These cultures were used to differentiate between true lung infection and distal airway colonization. The technique of obtaining samples with a PSB for quantitative cultures is a diagnostic method that was introduced by Wimberley and associates [6] in 1979 and has been extensively evaluated for 15 years. The results of these evaluations, which have been reported in terms of diagnostic accuracy, vary widely, depending on the reference technique used, the previous use of antimicrobial agents, the experience of the evaluators, and the quality of the bacteriologic processing. However, although some controversy exists concerning the sensitivity of the technique of obtaining PSB samples [12, 13], there is a consensus that considers this technique to be highly specific with low rates of false-positive results. A quantitative culture yielding> 10 3 cfu/ml has been demonstrated in experimental studies and subsequently has been clinically accepted as the most accurate diagnostic threshold for identifying lower respiratory tract infection. Such a bacterial count reflects about 10 6 bacteria/ml of respiratory secretions at the site of infection since the volume of the secretions collected with the brush is ml [14]. Recently, Dreyfuss and colleagues [15] suggested that treating all patients with borderline bacterial counts, ranging from ~ 10 2 to < 10 3 cfujml, would lead to overtreatment in most cases. These authors suggested that colonization of the lower respiratory tract in patients receiving ventilatory support is probably a dynamic process that can lead to either bacterial clearing or proliferation, with the development of pneumonia depending on the ability of the host to eradicate the microorganism. In our study, PSB samples were repeatedly obtained from patients who were persistently clinically suspected of having VAP; these patients were classifiedin the pneumonia group since at least one quantitativeculture of a PSB sample yielded Pseudomonas or Acinetobacter speciesat a significantconcentration. Consequently, the presence of false-negative results (i.e., patients with pneumonia who were not included in the study because of a sterile culture of a PSB sample or who were wrongly includedin the colonization group because of a bacterial count of < 10 3 cfu/ml) was highly improbable even if such results could occur particularlyin patients who received prior antimicrobial therapy. Moreover, because of the overall diagnostic accuracy and especially the high specificityof the technique of obtaining PSB samples, the probability that patients classified as having pneumonia (based on the results of quantitative cultures of PSB samples) effectively did not have pneumonia is very low. Whatever the limitations concerning the use of PSB samples for diagnosing VAP, our study suggests at least that low bacterial counts were associated with low mortality rates. Pneumonia due to Pseudomonas or Acinetobacter species is usually associated with a high mortality rate; this rate is frequently > 70% and is significantly higher than those among patients with pneumonia due to other microorganisms [1, 3, 16]. In our study, the mortality rate among patients with bacteriologically confirmed pneumonia due to Pseudomonas and Acinetobacter species was 73%; this rate was significantly higher than that among patients with the same clinical condition at admission who received ventilatory support for the same duration and were clinically indistinguishable from patients with clinical signs of pneumonia who did not have bacteriologically proven pneumonia. This finding further confirms the prognostic role of nosocomial lung infection by itself in severely ill patients that was suggested in two case-control studies [17, 18]; these studies showed that the mortality directly attributable to nosocomial pneumonia was significant. Similarly, our results demonstrated no significant difference between mortality rates among colonized patients and patients who received ventilatory support for >48 hours and did not have pneumonia or colonization with Pseudomonas or Acinetobacter species (29.1 % vs. 28.8%, respectively; NS); the fact that no difference was demonstrated suggests that such patients were actually free of lung infection or at least that their bacterial counts were so low that they were not associated with mortality other than that due to the underlying conditions. Nevertheless, before such a conclusion can be accepted, two points merit emphasis. First, every time pneumonia was clinically suspected, the patient was closely monitored, and PSB samples were obtained; the patient was classified on the basis of the results of quantitative cultures of PSB samples for only Pseudomonas and Acinetobacter species. Nine patients classified as being colonized actually had pneumonia due to pathogens other than these nonfermenting, gramnegative bacilli; pneumonia was diagnosed simultaneously for six of these patients, and pneumonia subsequently developed in three. These nine patients were treated with antibiotic agents effective against the responsible pathogens. Second, patients with pneumonia were given therapy with antimicrobial agents effective against the responsible pathogens, whereas no antibiotic therapy was prescribed for colonized patients. In addition, the clinical characteristics ofpatients with bacteriologically confirmed pneumonia were not significantly different from those of colonized patients; therefore, this study indirectly confirmed the previously reported difficulty encountered in clinically distinguishing between pneumonia and colonization in patients receiving ventilatory support in the absence of a reliable diagnostic technique [4, 5, 12, 13]. In conclusion, our results suggest that pseudomonas or acinetobacter VAP determined on the basis of stringent bacteriologic criteria is, at least partially, responsible by itself for a mortality rate higher than that among colonized patients receiving mechanical ventilation.

5 542 Fagon et al. ern 1996;23 (September) Acknowledgment The authors thank Mrs. C. Brun for preparing the manuscript. References 1. Fagon J-Y, Chastre J, Domart Y, et al. Nosocomial pneumonia in patients receiving continuousmechanical ventilation: prospective analysis of 52 episodeswith use of a protected specimenbrush and quantitativeculture techniques. Am Rev Respir Dis 1989;139: Craven DE, Kunches LM, KilinskyV, LichtenbergDA, Make BJ, McCabe WR. Risk factors for pneumoniaand fatalityin patientsreceivingcontinuous mechanical ventilation. Am Rev Respir Dis 1986;133: Stevens RM, Teres D, Skillman JJ, Feingold DS. Pneumonia in an intensive care unit: a 30-monthexperience. Arch Intern Med 1974;134: Meduri GU. Ventilator-associatedpneumonia in patients with respiratory failure: a diagnostic approach. Chest 1990;97: Fagon J-Y, Chastre J, Hance AJ, Domart Y, Trouillet J-L, Gilbert C. Evaluation of clinical judgment in the identification and treatment of nosocomial pneumonia in ventilated patients. Chest 1993;103: WimberleyN, Faling LJ, Bartlett JG. A fiberopticbronchoscopytechnique to obtain uncontaminated lower airway secretions for bacterial culture. Am Rev Respir Dis 1979;119: Chastre J, Fagon J-Y, Soler P, et al. Diagnosis of nosocomial bacterial pneumonia in intubated patients undergoing ventilation: comparison of the usefulness of bronchoalveolar lavage and the protected specimen brush. Am J Med 1988;85: McCabe WR, Jackson GG. Gram-negative bacteremia. I. Etiology and ecology. Arch Intern Med 1962;110: Zwillich CW, Pierson DJ, Creagh CE, Sutton FD, Schatz E, Petty TL. Complicationsof assistedventilation:a prospectivestudyof 354 consecutive episodes. Am J Med 1974;57: Le Gall J-R, Loirat P, AlperovitchA, et al. A simplifiedacute physiology score for ICU patients. Crit Care Med 1984;12: II. Fagon JY, Chastre J, Novara A, Medioni P, Gibert C. Characterization of intensive care unit patients using a model based on the presence or absence of organ dysfunctionsand/or infection: the ODIN model. Intensive Care Med 1993;19: Niederman MS, Torres A, Summer W. Invasive diagnostic testing is not needed routinelyto manage suspected ventilator-associatedpneumonia. Am J Respir Crit Care Med 1994;150: Chastre J, Fagon JY. Invasivediagnostic testing should be routinely used to manage ventilated patients with suspected pneumonia. Am J Respir Crit Care Med 1994;150: Bartlett JG, Finegold SM. Bacteriology of expectorated sputum with quantitativeculture and wash technique compared to transtrachealaspirates. Am Rev Respir Dis 1978;117: Dreyfuss D, Mier L, Le Bourdelles G, et al. Clinical significance of borderline quantitative protected brush specimen culture results. Am Rev Respir Dis 1993;147: Bryan CS, ReynoldsKL. Bacteremic nosocomialpneumonia: analysis of 172 episodes from a single metropolitanarea. Am Rev Respir Dis 1984; 129: CraigCP, ConnellyS. Effect of intensivecare unit nosocomialpneumonia on duration of stay and mortality. Am J Infect Control 1984;12: Fagon J-Y, Chastre J, Hance AJ, Montravers P, Novara A, Gibert C. Nosocomialpneumoniain ventilatedpatients: a cohort study evaluating attributable mortality and hospital stay. Am J Med 1993;94:281-8.