Epidemiology, diagnosis and treatment of systemic Candida infection in surgical patients under intensive care.

The incidence of systemic Candida infections in patients requiring intensive care has increased substantially in recent years as a result of a combination of factors. More patients with severe underlying disease or immunosuppression from anti-neoplastic or anti-rejection chemotherapy and at risk from fungal infection are now admitted to the ICU. Improvements in supportive medical and surgical care have led to many patients who would previously have died as a result of trauma or disease surviving to receive intensive care. Moreover, some therapeutic interventions used in the ICU, most notably broad-spectrum antibiotics and intravascular catheters, are also associated with increased risks of candidiasis. Systemic Candida infections are associated with a high morbidity and mortality, but remain difficult to diagnose and ICU staff need to be acutely aware of this often insidious pathogen. A number of studies have identified risk factors for systemic Candida infection which may be used to identify those at highest risk. Such patients may be potential candidates for early, presumptive therapy. Here we review the epidemiology, pathogenesis, morbidity and mortality of systemic Candida infections in the ICU setting, and examine predisposing risk factors. Antifungal treatment, including the use of amphotericin B, flucytosine and fluconazole, and the roles of early presumptive therapy and prophylaxis, is also reviewed.

Invasive diagnostic techniques for pneumonia: protected specimen brush, bronchoalveolar lavage, and lung biopsy methods.

We suggest the following strategy for managing patients with pneumonia. For nonventilated patients with either CAP or HAP, empiric antibiotic treatment should be started according to approved guidelines, and if the clinical evolution of the patient is not adequate, fiberoptic bronchoscopy including PSB and BAL could be considered, with modification of the antibiotic treatment accordingly. In ventilated patients with either CAP or HAP, respiratory secretion sampling using noninvasive techniques should be conducted upon clinical suspicion of VAP and before starting a new antibiotic treatment. Antibiotic therapy according to approved guidelines should be started as soon as possible and maintained during the first 48 hours if the patient's evolution is satisfactory and condition has stabilized. Then, initial antibiotic treatment should be adjusted according to cultures. If there is a clear diagnostic alternative to VAP and cultures are negative, this is the only case in which antibiotic treatment could be withdrawn. If the patient's clinical evolution is inadequate (persistence of fever, leukocytosis, increasing infiltrates, and respiratory failure), fiberoptic bronchoscopy with PSB and BAL and modification of the initial antibiotic regimen should be sought. Open lung biopsy may be indicated in patients with diffuse pulmonary infiltrates in whom a diagnosis has not been achieved by other methods, including bronchoscopy. Transbronchial lung biopsy should not be viewed as a diagnostic technique for pneumonia except in immunosuppressed patients with diffuse alveolar infiltrates.

Respiratory infectious complications in the intensive care unit.

Ventilator-associated pneumonia is the most common infectious respiratory complication in intensive care unit patients, particularly those needing mechanical ventilation. Ventilator-associated pneumonia represents a challenging problem in terms of diagnosis, treatment, and prevention. Nosocomial sinusitis is another respiratory infection, not uncommon in mechanically ventilated patients. This type of infection has to be suspected in nasally intubated patients and may be a hidden focus of fever and sepsis.

Relationship between ventilator-associated pneumonia and intramucosal gastric pHi: a case-control study.

PURPOSE: Prior investigations have suggested a clear relationship between nosocomial pneumonia and intramucosal gastric pH (pHi), a probable marker of bacterial translocation. METHODS: We studied 33 patients (18 with pneumonia and 15 without) admitted to an intensive care unit and hospitalized longer than 72 hours with the aim of assessing the relationship between nosocomial pneumonia, pHi, and outcome. pHi was estimated at the time of inclusion of patients into the study. Arterial pH (pHa) and bicarbonate and stomach pH and tonometer PtCO2 were also recorded. Values of < 7.32 or delta pHa-pHi of > +0.06 were used to differentiate between normal and low pHi. Quantitative cultures of pharyngeal swabs, gastric lumen, and protected specimen brush from lower airways were also done. RESULTS: The mean pHi values were 7.397 +/- 0.105 (range, 7.14 to 7.53) and 7.452 +/- 0.059 (range, 7.37 to 7.56) for patients with and without pneumonia, respectively (P = .073). Five patients, all with pneumonia, had pHi < 7.32. No patients without pneumonia had pHi < 7.32 (P = .04). The mean delta pHa-pHi was 0.04 +/- 0.07 (range, -0.11 to 0.13) and 0.05 +/- 0.09 (range, -0.09 to 0.28; P = .72) for patients with and without pneumonia, respectively. However, there were significant differences when tonometer PtCO2 values of both groups were compared (38.9 +/- 8.3 and 30.6 +/- 4.7 mm Hg, respectively; P = .025). Patients with pneumonia had higher alkaline gastric lumen pH (5.2 +/- 1.0) than those without pneumonia (3.8 +/- 1.4; P = .006). Nonsurvivors (n = 7) had more acidic pHi (7.33 +/- 0.11) than survivors (7.44 +/- 0.06; P = .045). The mean gastric lumen bacterial concentration was 4.14 +/- 1.01 Log10 CFU/mL in patients with pneumonia and 4.28 +/- 1.22 Log10 CFU/mL in patients without pneumonia (P = NS). When patients with and without intramucosal gastric acidosis (pHi < 7.32) were compared, the gastric bacterial burden was 4.42 +/- 0.82 Log10CFU/mL and 4.32 +/- 1.03 Log10 CFU/mL (P = .08), respectively. CONCLUSIONS: Most patients with nosocomial pneumonia had no associated intramucosal gastric acidosis. However, low pHi was associated with increased mortality.

Intrapulmonary neutrophil apoptosis in a pig model of pneumonia.

BACKGROUND: Apoptosis and necrosis are two distinct cell death modalities. Polymorphonuclear leukocytes (PMNL) play an important role in pneumonia and little information is available on whether these cells are cleared by necrosis or apoptosis. We therefore compared the proportion of apoptotic PMNL in lung tissues with and without pneumonia. In addition, the association of PMNL apoptosis and mechanical ventilation (MV) was investigated. METHODS: Samples obtained from a pig model developed to tracheobronchial stenting were analysed. Pneumonia was induced with that model in 10 animals (white-Landrace piglets, Pittman-Moore or Göttingen minipigs). Animals were sacrificed when clinical parameters of pneumonia became evident and 17 lung samples could be analysed for histologic evidence of pneumonia, growth of micro-organisms, and the apoptotic index (proportion of apoptotic PMNL over all PMNL in a 400-power field). In addition, lung samples (n = 7) from two pigs without stent implantation on mechanically ventilated for more than 48 h were studied. RESULTS: A total of 9/17 samples (53%) showed histologic evidence of pneumonia, 11/17 (65%) had significant bacterial growth (> 10(3) cfu/g), and 7/17 (41%) fulfilled both criteria. The apoptotic index was not significantly different in samples with and without histologic evidence of pneumonia (pneumonia: 25.3 +/- 6.1% vs. No pneumonia: 25.2 +/- 11.9%; 95%CI: -9.7 - 9.5, p = 0.989) or in samples with and without significant bacterial growth (significant bacterial growth: 23.5+/-9.3% vs. No significant bacterial growth: 28.5+/-8.2%; 95%CI: -4.6 - 16.7; p = 0.284). However, the apoptotic index was higher in samples of pigs, mechanically ventilated for more than 48 h, compared to the stent group (MV: 36.9+/-10.7% vs. Stent: 25.2 +/- 9.0%; 95% CI: -20.5 - -2.9; p = 0.012). CONCLUSION: In an animal model, the proportion of apoptotic PMNL was not different in lung samples with and without histologic or bacterial pneumonia. However, MV for more than 48 h was associated with a higher proportion of apoptotic PMNL in lung tissue. This study may be helpful for the understanding of the evolution of lung tissue damage induced by bacterial pneumonia and MV.

Markers of ventilator-associated pneumonia.

The diagnosis of ventilator-associated pneumonia (VAP) is difficult for several reasons. Firstly, clinical markers show a large percentage of false-positive and false-negative results. Secondly, microbiological diagnosis based on quantitative cultures of protected specimen brush (PSB), bronchoalveolar lavage (BAL), and endotracheal aspirates also present false-positive and false-negative results. Furthermore, definite results are delayed for 48-72 hours. For all these reasons it would be an advantage to have a biological marker of ventilator-associated pneumonia in clinical practice. Since clinical features of pneumonia in mechanically ventilated patients are neither specific nor sensitive, rapid markers of pneumonia might be of great assistance to the clinician in deciding whether to start an empiric antibiotic regimen. A marker of ventilator-associated pneumonia could be a rapid alternative diagnostic method which permits the definite diagnosis of pneumonia. Accordingly, specific markers of VAP, namely the presence of intracellular microorganisms, the detection of elastin fibres, the antibody-coated bacteria test, the level of endotoxin in bronchoalveolar lavage fluid, the local production of interleukin-8, the levels of lactate dehydrogenase, and decreased surfactant protein A, may be important as they can provide a rapid diagnosis of VAP. Among the markers alluded to above, the search for intracellular bacteria in polymorphonuclear leukocytes or macrophages is the most widely validated technique with an excellent specificity, provided that prior antibiotics are not given. However, this technique has its own limitations; it requires a considerable time effort for the microbiologist, and also compels the performance of BAL, a technique not always harmless to the patient.

Severe pulmonary infections in AIDS patients.

Pulmonary infections are a very common complication in acquired immune deficiency syndrome (AIDS) patients. These infections may be severe enough to initiate the admission of these patients to intensive care units (ICU). Pneumocystis carinii pneumonia (PCP) is the most frequent cause of ICU admission because of acute respiratory failure. Mortality of ICU-admitted patients with this infection has changed with time. Initial reports confirmed a high mortality (80% to 90%). After 1985, the mortality rate decreased (50%). Factors such as the use of corticosteroids, better patient care, and a better knowledge of the disease probably explain this change. In recent years (1990 to 1995), mortality has worsened again, perhaps, because ICU facilities were offered more liberally to patients failing aggressive conventional treatment, including adjuvant therapy with corticosteroids. However, for those patients able to be discharged, the prognosis is not worse than expected according to the stage of their human immunodeficiency virus-1 (HIV-1) infection and immunologic status. Consequently, at least a limited period of ICU care and some respiratory support (either continuous positive airway pressure or mechanical ventilation) should be considered and offered to all HIV-1-infected patients with PCP and respiratory failure. Cytomegalovirus may be another cause of severe pulmonary infection in AIDS patients. This infection is difficult to diagnose; hence, it should be suspected when patients with PCP do not progress appropriately, or when no responsible pulmonary pathogen is found. When associated with PCP, mortality is very high. Disseminated tuberculosis is another potential cause of severe respiratory failure and respiratory secretions should be routinely examined for acid-fast bacilli in AIDS patients with pulmonary infiltrates. Finally, bacterial pneumonia (Streptococcus pneumoniae, Neisseria catarrhalis, Haemophilus influenzae, Staphylococcus aureus, and Pseudomonas aeruginosa) may also be the etiological agents of severe acute respiratory failure. Empiric antibacterial treatment to cover these microorganisms should be given when a bacterial agent is suspected.

Stomach as a source of colonization of the respiratory tract during mechanical ventilation: association with ventilator-associated pneumonia.

The aetiopathogenesis of ventilator-associated pneumonia (VAP) requires abnormal oropharyngeal and gastric colonization and the further aspiration of their contents to the lower airways. VAP develops easily if aspiration or inoculation of microorganisms occur in patients with artificial airways, in whom mechanical, cellular and/or humoral defences are altered. Well-known risk factors for gastric colonization include: alterations in gastric juice secretion; alkalinization of gastric contents; administration of enteral nutrition; and the presence of bilirubin. However, the role of the colonized gastric reservoir in the development of VAP remains debatable. Evidence in favour of the role of the stomach in the development of VAP comes mainly from randomized, controlled trials of selective gut decontamination and stress ulcer prophylaxis in the intensive care unit (ICU), in which reducing the bacterial burden of the stomach decreases the incidence of nosocomial respiratory infections. However, at least three studies of flora have found an absence of stomach origin of pneumonia occurring during mechanical ventilation. Prophylactic measures suggested to prevent VAP in relation to the gastric reservoir include: treatment for stress ulcers with sucralfate; prevention of duodenal reflux with metoclopramide; reduction of gastric burden and bacterial translocation by selective digestive decontamination; acidification of enteral feeding; and jejunal feeding. Gastro-oesophageal reflux can be prevented by using small bore nasogastric tubes and jejunal feeding. The aspiration of gastric contents can be reduced by positioning patients in a semirecumbent position, checking the patency of the tube cuff, and aspiration of subglottic secretions. The role of the stomach as a reservoir for microorganisms causing ventilator-associated pneumonia is still controversial but despite the debate, there is major evidence in the literature in favour of the gastric origin of part of these pulmonary infections.

Evaluation of antimicrobial treatment in mechanically ventilated patients with severe chronic obstructive pulmonary disease exacerbations.

OBJECTIVE: To study microbial and susceptibility patterns and antimicrobial treatment responses in patients with severe, acute exacerbations of chronic obstructive pulmonary disease requiring mechanical ventilation. DESIGN: Microbial investigation using tracheobronchial aspirates, bronchoscopy with a protected specimen brush, and bronchoalveolar lavage, as well as paired serologies. Evaluation of antimicrobial treatment by results of the initial investigation, susceptibility testing, and a repeated microbial investigation (tracheobronchial aspirates, bronchoscopy with a protected specimen brush, and bronchoalveolar lavage) after 72 hrs. SETTING: A respiratory intensive care unit of a 1,000-bed teaching hospital. PATIENTS: Fifty severely exacerbated and mechanically ventilated patients with chronic obstructive pulmonary disease. INTERVENTIONS: Initial empirical antimicrobial treatment according to clinical judgment. MEASUREMENTS AND MAIN RESULTS: Overall, 36 of 50 patients (72%) had evidence of a microbial origin. Community-acquired endogenous pathogens were present in 70% of patients, and Gram-negative enteric bacilli and Pseudomonas/Stenotrophomonas species were present in 30%. All five isolates of Streptococcus pneumoniae were resistant to penicillin (three intermediately and two highly), and three were resistant to multiple antibiotics. Pseudomonas species revealed multiresistance in four of nine isolates (44%), and Stenotrophomonas maltophilia revealed multiresistance in one of two isolates. Antimicrobial treatment was modified according to diagnostic results in 11 of 31 patients (36%) with potentially pathogenic microorganisms. In patients who underwent a repeat investigation after 72 hrs, 24% of the initially present and potentially pathogenic microorganisms persisted. Inappropriate initial antimicrobial therapy was associated significantly with bacterial persistence (p < .002). CONCLUSIONS: Considering the diversity of microbial pathogens and the resistance rates especially to S. pneumoniae in this patient population, antimicrobial treatment should be based on the constant study of local microbial and susceptibility patterns along with routine microbial investigation of the individual patient.

Community-acquired pneumonia in the elderly.

The incidence of community-acquired pneumonia (CAP) in the elderly is higher compared to younger populations. In addition, pneumonia in the elderly is a life-threatening problem. As our demographics have changed, clinicians have developed a heightened interest in managing pneumonia in the elderly. The development of pneumonia in elderly patients differs from that in younger individuals due to a complex array of factors. (1) The organisms involved depend on the setting in which the pneumonia developed: either the nonhospitalized elderly patient with CAP or the institutionalized patient who develops nursing-home-acquired pneumonia. (2) Underlying comorbid conditions commonly exist in the elderly that affect the etiology and outcome of pneumonia. Overall, Streptococcus pneumoniae and Haemophilus influenzae are still the most common etiologies of pneumonia in the elderly. The true role of gram-negative bacilli remains unclear although these micro-organisms may be more common etiologic agents in nursing-home pneumonia. Some recent studies from Mediterranean areas have reported high rates of infection by Chlamydia pneumoniae, but the real role of this micro-organism has to be confirmed. Another important issue is that the presenting symptoms of pneumonia in the elderly can be subtle and sometimes difficult to recognize. Fever is frequently absent, and delirium or alteration of functional physical capacity may be the only manifestations. Mortality in the elderly with CAP is higher when compared to younger populations. However, this may be explained by the concomitant presence of comorbid conditions more than by age per se. This statement has to be kept in mind when considering hospital and, particularly, intensive care unit admissions. Finally, antibiotic pharmacokinetics in the elderly populations with CAP ought to be considered to avoid frequent side-effects and complications. Overall, antibiotic regimens in hospitalized elderly patients with CAP do not differ from other hospitalized CAP populations. An organized approach to assessing elderly patients with suspicion of pneumonia and an awareness of common pitfalls in the management of this pulmonary infection in this population are essential to improving outcomes.

The role of the gastric reservoir in ventilator-associated pneumonia.

Ventilator-associated pneumonia (VAP) is a common complication of mechanical ventilation with an incidence ranging from 9-70% and averaging around 25%. The pathogenesis of VAP requires abnormal oropharyngeal and gastric colonisation and then aspiration of these contents into the lower airways. Another co-existing mechanism could be direct oropharyngeal or lower airways inoculation of microorganisms through contaminated respiratory therapy equipment. Ventilator-associated pneumonia develops easily if aspiration or inoculation of microorganisms occur in patients with artificial airways and in whom mechanical, cellular and/or humoral defences are altered. Both host factors and treatments may alter pulmonary defence mechanisms; these too may contribute to the development of VAP. An alternative mechanism to explain VAP is bacterial translocation, although this mechanism is still under investigation. Figure 1 illustrates a schema of the pathogenesis of VAP. In this paper we review the possible role of the gastric reservoir in the aetiology of VAP, emphasising the following issues: 1. Risk factors for gastric colonisation 2. Clinical evidence of gastric aspiration to the lower airways in mechanically ventilated patients 3. Clinical evidence and controversies surrounding the role of the gastric reservoir in ventilator-associated pneumonia 4. The role of bacterial translocation as a mechanism for the development of VAP 5. A summary of prophylactic measures.

Role of glucocorticoids on inflammatory response in nonimmunosuppressed patients with pneumonia: a pilot study.

The aim of the study was to assess the potential role of glucocorticoids (GC) in modulating systemic and pulmonary inflammatory responses in mechanically ventilated patients with severe pneumonia. Twenty mechanically ventilated patients with pneumonia treated at a respiratory intensive care unit (RICU) of a 1,000-bed teaching hospital were prospectively studied. All patients had received prior antimicrobial treatment. Eleven patients received GC (mean+/-SD dose of i.v. methylprednisolone 677+/-508 mg for 9+/-7 days), mainly for bronchial dilatation. Serum and bronchoalveolar lavage fluid (BALF) tumour necrosis factor (TNF)-alpha, interleukin (IL)-1beta, IL-6 and C-reactive protein levels were measured in all patients. The inflammatory response was attenuated in patients receiving GC, both systemically (IL-6 1,089+/-342 versus 630+/-385 pg x mL(-1), p=0.03; C-reactive protein 34+/-5 versus 19+/-5 mg x L(-1), p=0.04) and locally in BALF (TNF-alpha 118+/-50 versus 24+/-5 pg x mL(-1), p= 0.05; neutrophil count: 2.4+/-1.1 x 10(9) cells x L(-1) (93+/-3%) versus 1.9+/-1.8 x 10(9) cells x L(-1) (57+/-16%), p=0.03). Four of the 11 (36%) patients receiving GC died compared to six (67%) who were not receiving GC (p=0.37). The present pilot study suggests that glucocorticoids decrease systemic and lung inflammatory responses in mechanically ventilated patients with severe pneumonia receiving antimicrobial treatment.

Cytokine expression in severe pneumonia: a bronchoalveolar lavage study.

OBJECTIVE: To assess the cytokine expression (tumor necrosis factor-alpha [TNF-alpha], interleukin [IL]-1beta, and IL-6) in severe pneumonia, both locally (in the lungs) and systemically (in blood). DESIGN: Prospective sequential study with bronchoalveolar lavage (BAL) and blood sampling. SETTING: Six-bed respiratory intensive care unit of a 1,000-bed teaching hospital. PATIENTS: Thirty mechanically ventilated patients (>48 hrs) were allocated to either the pneumonia group (n = 20) or a control group (n = 10). INTERVENTIONS: Protected specimen brush and BAL samples for quantitative cultures, and serum and BAL fluid TNF-alpha, IL-1beta, and IL-6 levels were measured on days 1, 3, and 7. In the control group, the procedure was done on day 1 only. MEASUREMENTS AND MAIN RESULTS: Serum TNF-alpha levels were significantly higher in patients with pneumonia compared with controls (35 +/- 4 vs. 17 +/- 3 pg/mL, respectively, p = .001). IL-6 levels in serum and BAL fluid were higher in pneumonia than in control patients (serum, 837 +/- 260 vs. 94 +/- 35 pg/mL, respectively, p = .017; BAL fluid, 1176 +/- 468 vs. 234 +/- 83 pg/mL, respectively, p = .05). On days 1, 3, and 7 in patients with pneumonia, IL-1beta levels turned out to be higher in BAL fluid than in serum (71 +/- 17 vs. 2 +/-1 pg/mL on day 1; 49 +/- 8 vs. 6 +/- 2 pg/mL on day 3; and 47 +/- 16 vs. 3 +/- 2 pg/mL on day 7 for BAL fluid and serum, respectively, p < .05). No significant correlation between BAL fluid cytokine levels and lung bacterial burden was shown in presence of antibiotic treatment. Although no clear relationship was found between BAL fluid and serum cytokines and mortality, there was a trend toward higher serum IL-6 levels in nonsurvivors (1209 +/- 433 pg/mL) with pneumonia compared with survivors (464 +/- 260 pg/mL). In addition, serum TNF-alpha and IL-6 correlated with multiple organ failure score (r2 = .36, p = .004 for both) and with lung injury score (r2 = .30, p = .01, and r2 = .22, p = .03, for TNF-alpha and IL-6, respectively). CONCLUSIONS: The present study describes the lung and systemic inflammatory response in severe pneumonia. The lung cytokine expression seems to be independent from the lung bacterial burden in the presence of antibiotic treatment. Because of the limited sample size, we did not find a clear relationship between serum and BAL fluid cytokine levels and outcome.

Bacterial colonization patterns in mechanically ventilated patients with traumatic and medical head injury. Incidence, risk factors, and association with ventilator-associated pneumonia.

We prospectively evaluated the relation of upper airway, lower airway, and gastric colonization patterns with the development of pneumonia and its etiology in 48 patients with surgical (n = 25) and medical (n = 23) head injury. Initial colonization was assessed by cultures of nasal and pharyngeal swabs, tracheobronchial aspirates, gastric juice, and bronchoscopically retrieved protected specimen brush. Follow-up colonization was determined until the end points extubation, suspected ventilator-associated pneumonia (VAP), or death. The initial colonization rate at any site at ICU admission was 39/47 (83%). It mainly accounted for Group I pathogens (Streptococcus pneumoniae, Staphylococcus aureus, Hemophilus influenzae) of the upper and lower airways. At follow-up, colonization rates with Group II pathogens (Gram-negative enteric bacilli and Pseudomonas spp.) increased significantly. The high initial bacterial load with Group I pathogens of the upper airways and trachea decreased during Days 2 to 4, whereas that of Group II pathogens increased. Upper airway colonization was an independent predictor of follow-up tracheobronchial colonization (odds ratio [OR], 9.9; 95% confidence interval [CI], 1.8 to 56.3 for initial colonization with Group I pathogens; OR, 23.9; 95% CI, 3.8 to 153.3 for follow-up colonization with Group II pathogens). Previous (short-term) antibiotics had a protective effect against colonization with Group I pathogens of the lower respiratory tract (OR, 0.2; 95% CI, 0.05 to 0.86), but they were a risk factor for colonization with Group II pathogens (OR, 6.1; 95% CI, 1.3 to 29). Initial tracheobronchial colonization with Group I pathogens was associated with a higher probability of early onset pneumonia (OR, 4. 1; 95% CI, 0.7 to 23.3), whereas prolonged antibiotic treatment (> 24 h) independently predicted late-onset pneumonia (OR, 9.2; 95% CI, 1.7 to 51.3). We conclude that patients with head injury are colonized in the airways mainly by Group I pathogens early in the evolution of illness. The upper airways represent the main reservoir for subsequent lower airway colonization with Group I pathogens. Previous (short-term) antibiotic treatment is protective against initial tracheobronchial colonization with Group I pathogens, but it represents a risk factor for subsequent lower airway colonization by Group II pathogens.

Pulmonary complications in patients with haematological malignancies treated at a respiratory ICU.

Patients with haematological malignancies developing severe pulmonary complications have a poor outcome, especially after bone-marrow transplantation (BMT). We studied the aetiology, the yield of different diagnostic tools, as well as the outcome and prognostic factors in the corresponding population admitted to our respiratory intensive care unit (RICU). Overall, 89 patients with haematological malignancies and pulmonary complications treated within a 10 yr period were included. The underlying malignancies were predominantly acute leukaemia and chronic myeloid leukaemia (66/89, 74%). Fifty-two of 89 (58%) patients were bone marrow recipients. An aetiological diagnosis could be obtained in 61/89 (69%) of cases. The aetiology was infectious in 37/89 (42%) and noninfectious in 24/89 (27%). Blood cultures and cytological examinations of bronchoalveolar lavage fluid were the diagnostic tools with the highest yield (13/43 (30%) and 13/45 (29%) positive results, respectively). Necropsy results were coincident with results obtained during the lifetime in 43% of cases with infectious and 60% with noninfectious aetiologies. Overall mortality was 70/89 (79%), and 47/52 (90%) in transplant recipients. The requirement of mechanical ventilation, BMT, and an interval <90 days of BMT prior to ICU admission were independent adverse prognostic factors. The outcome in this patient population was uniformly poor. It was worst in bone marrow recipients developing pulmonary complications <90 days after transplantation and requiring mechanical ventilation. Decisions about intensive care unit admission and mech-anical ventilation should seriously consider the dismal prognosis of these patients.