INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although aspiration pneumonia appears as a frequent condition in patients admitted to an intensive care unit (ICU), numerous features about this disease remain unprecise. Its precise incidence among severe community-acquired pneumonias admitted to an ICU is unknown. Most studies about the bacteriology of this pulmonary infection were performed in patients seen relatively late, when complications such as lung abscess, necrotizing pneumonia, or empyema had occurred (1, 2). Finally, prognosis remains unclear. Mortality rate varied from one study to another in the range 7.5% (3) to 62% (4). In some studies, aspiration appeared as a poor prognosis factor (5, 6), while in others it was the reverse (7). In order to clarify some data about the epidemiology and prognostic factors of aspiration pneumonia, we selected for study, among patients admitted to six ICUs in the north of France for severe community-acquired pneumonia over a 9-yr period, those suffering from aspiration pneumonia.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Selection of Cases

We performed a multicenter, retrospective (1987-1992), and prospective (1993-1995) study of patients exhibiting severe community-acquired pneumonia and requiring admission into an ICU. The study sites were: (1) 1987-1995, Tourcoing General Hospital; (2) 1993-1994, Helfaut General Hospital, Lens General Hospital, and Lille University Hospital B; (3) 1995, Arras General Hospital, Lens General Hospital, and Valenciennes General Hospital.

Community-acquired pneumonia (CAP) was defined by the following criteria observed at initial presentation or occurring within 48 h following hospitalization (6): (1) admission from home or a nursing-home; (2) the presence of a new radiographic pulmonary infiltrate; (3) acute onset of at least one major (cough, sputum production, fever) or two minor (dyspnea, pleuretic chest pain, altered mental status, pulmonary consolidation on physical examination, total leukocyte count greater than 12,000/mm³) clinical or biological findings suggestive of pneumonia. Patients (1) hospitalized within 30 d prior to developing pneumonia, (2) with radiographic abnormalities attributed solely to pulmonary embolus, lung carcinoma or congestive heart failure, or (3) with a clinical diagnosis of AIDS or with a known positive HIV test, were excluded.

The criteria for aspiration pneumonia were those usually used (1, 2, 6): clinical, biological and radiological criteria for community- acquired pneumonia as previously described associated with either witnessed aspiration or risk factors for aspiration (altered mental status, abnormal gag reflex of swallowing mechanism, intestinal obstruction).

Patient Evaluation, Data Collection and Definition of Variables

Patients were categorized into three groups: Group 1 included patients with aspiration pneumonia whose main medical ground for ICU admission was respiratory distress; Group 2 included patients with aspiration pneumonia, but neurological disturbance was the main medical ground of ICU admission; and Group 3 included all the remaining nonaspiration patients.

Within 24 h following CAP diagnosis, all patients underwent clinical, radiological, and biological tests. The following clinical variables were recorded: age; sex; origin (home, nursing home); underlying clinical characteristics (chronic respiratory insufficiency, immunosuppression, alcoholism); initial vital signs; mechanisms of aspiration with categorization into three groups: (1) witnessed large aspiration, (2) altered mental status, and (3) altered reflex of swallowing or gastrointestinal problems. The underlying clinical conditions were classified according to the criteria proposed by McCabe and Jackson (9). Vital sign abnormalities were assessed by Simplified Acute Physiologic Score (SAPS) (10), and Acute Organ System Failure (OSF) scoring system (11). Shock was defined with usual criteria (12). Neurological status and changes in mental status were stratified according to the Glasgow Coma Score (13). The need for mechanical ventilation on admission or within 12 h following admission defined initial acute respiratory failure. The following radiological indices were recorded: number of lobes involved and unilateral or bilateral involvement. Biochemical and hematological tests included measurement of total leukocyte count, total platelet count, serum creatinine level, total serum protein, arterial blood gases, and Pa_O2 /FI_O2. A microbiological diagnosis was assigned according to usual criteria (6, 7, 14, 15).

Therapy and Assessment of Evolution

Information on the following therapeutic topics was recorded for each patient: supportive measures such as mechanical ventilation, use of systemic corticosteroids or inotropic drugs and initial antimicrobial therapy. This latter treatment was started as soon as microbiologic sampling was performed. The choice of antibiotics was based on literature guidelines and local epidemiology. Adequacy and effectiveness of this initial antimicrobial treatment were determined within 72 h after the beginning of the antimicrobial treatment.

Antimicrobial therapy was considered adequate if the causative pathogen, according to antibiotic susceptibility reports, was susceptible to this antimicrobial therapy and/or when no precise microbiologic diagnosis was made, the empirical drugs chosen accomplished the recommendations of the medical literature.

The initial therapy was considered effective if the clinical situation improved and fever decreased within the first 72 h of treatment. In all the other clinical situations, the initial therapy was considered a failure. This evaluation, based only on clinical response, should not be confused with the adequacy of therapy based on dose, route, and sensitivity patterns.

Patients were followed by the medical staff throughout their hospitalization in the ICU, until discharge from the ICU or death. During the patient's stay in the ICU, the occurrence of complications was recorded. We distinguished directly pneumonia-related complications such as (1) sepsis-related complications (secondary septic shock, development of multiple organ failure), (2) bronchopulmonary hospital-acquired superinfections, or (3) occurrence of septic metastasis (i.e., meningitidis); and indirectly or nonspecific pneumonia-related complications such as (4) ICU-related complications (upper gastrointestinal bleeding, catheter-related infection, deep venous thrombosis) and (5) complications attributed only to underlying medical conditions.

Finally, patient outcome during the ICU stay was recorded. Subsequent outcome after discharge from the ICU was not studied. Crude mortality included all deaths of studied patients that occurred during the ICU stay. To determine attributable mortality linked to CAP, we excluded all patients whose the primary or potential cause of death was not directly or indirectly pneumonia-related. So, to study prognostic factors related to the outcome in the ICU (survival or death) in patients with aspiration pneumonia, we excluded all patients whose death was related only to disease leading to aspiration.

Statistical Analysis

Among the patients exhibiting an aspiration pneumonia, Groups 1 and 2 were compared, at baseline, for the epidemiological and clinical data as well as laboratory results collected within 24 h of presentation and, during the ICU stay, for occurrence of complications and outcome. On the same way, such data were compared between the patients exhibiting an aspiration pneumonia (Groups 1 and 2) and the remaining patients (Group 3). The qualitative variables were compared by the Fisher's and the chi-square tests with Yate's correction when necessary. When appropriate, continuous variables were analyzed as categorical variables using clinically meaningful cut-points. Continuous variables were compared using Student's t test. At all instances p levels  0.05 were considered significant.

To study prognosis factors of aspiration pneumonia, all variables collected within 24 h of admission and during the ICU stay were analyzed to compare survivors and nonsurvivors. Univariate analysis was performed using the previously described statistical tests. Moreover, to study the relationship between the outcome and any predictor, taking the significant variables into account, we performed three stepwise logistic regressions: The significant variables collected within 24 h of admission were used for a first analysis, the significant variables collected during the ICU stay were used for a second analysis, and finally, all significant variables were entered into a third stepwise analysis.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Characteristics and Microbiologic Data

During the 9-yr period study, 505 patients were retrospectively (n = 337) and prospectively (n = 168) collected. Among them, 116 patients (23%) were retrospectively (n = 73) and prospectively (n = 43) defined as exhibiting an aspiration pneumonia. For them, main medical grounds of ICU admission were respiratory distress in 54 patients (Group 1) and neurological disturbances in 62 patients (Group 2). For the remaining 389 patients (Group 3) no criteria for aspiration was present.

Mains patients characteristics at presentation are summarized in Table 1. Using monovariate analysis, comparison between Group 1 and Group 2 revealed some significant differences: In Group 1, patients were older (63.8 versus 44.4 yr, p < 0.0001), had a more severe underlying diseases (anticipated death within 5 yr, 50% versus 16%, p = 0.001), suffered more frequently from chronic respiratory insufficiency (30% versus 8%, p = 0.003) and exhibited, at presentation, a lower Pa_O2/ FI_O2 (223 versus 280 mm Hg, p = 0.01). Respectively, in Group 1 and 2, underlying immunosuppression (2 patients versus 1 patient) or alcoholism (15% versus 8%), mean SAPS and OSF score, presence of initial shock, chest X-ray involvement, and biological data such as mean Pa_CO2 (35.9 versus 36.6 mm Hg), mean serum creatinine (17.4 versus 13.7 mg/l), mean total serum protein (60.4 versus 64.7 g/l), mean platelets count (258,000 versus 208,000/mm³) and mean leukocytes count (15,000 versus 12,400/mm³) were not significantly different. Significant differences in comparing characteristics of patients with (Groups 1 and 2) and without (Group 3) aspiration pneumonia were as followed: In patients with aspiration pneumonia mean age was lower (53.5 versus 66.3 years, p = 0.00001), underlying diseases were less severe (anticipated death within 5 yr were, respectively, 32% versus 48%; p = 0.004), chronic respiratory insufficiency was less frequent (18% versus 49%, p = 0.001), alcoholism was more frequent (11% versus 6%, p = 0.04) and, on admission, mean Pa_CO2 (36.3 versus 43.7 mm Hg, p = 0.001) was lower.

The clinical settings for aspiration are listed in Table 2. Witnessed large aspiration was observed in five patients. Risk factors for aspiration were altered mental status in 89 patients and altered reflex of swallowing or gastrointestinal problems in the remaining 22 patients.

A potential causal aerobic organism was isolated from 35 patients in Group 1, 35 patients in Group 2 and 249 patients in Group 3. Main recovered organisms are summarized in Table 3.

Therapy and Outcome

Main therapies required during ICU stay were reported in Table 4. Inotropic support was the unique significant difference between patients with or without aspiration pneumonia.

Antimicrobial therapy (monotherapy versus combination, adequacy and efficacy on reevaluation) was similar in the three groups. More detailed data about patients with aspiration pneumonia were the following: Antimicrobial therapy was empirically initiated in all 116 cases. Initial therapy was monotherapy in 51 (44%) patients and combination in 65 (56%) patients. As a single agent, the administered drugs were amoxicillin ± clavulanic acid (n = 36), ureidopenicillins ± betalactamase inhibitor (n = 7), fluoroquinolones (n = 4), third generation cephalosporins (n = 2) and carbapenems (n = 2). The main drugs administered in combination were amoxicillin ± clavulanic acid + fluoroquinolones or aminoglycosides (n = 25), ureidopenicillins ± betalactamase inhibitor + fluoroquinolones or aminoglycosides (n = 18) and third generation cephalosporins + fluoroquinolones or aminoglycosides (n = 16).

Reevaluation of clinical status 72 h after the beginning of treatment was available in 112 patients; death related to disease leading to aspiration occurred in four cases, before this evaluation (cases 1-4) (Table 5). Initial therapy was considered effective in 92 cases. In the remaining 20 cases, therapy was considered a failure (Table 5).

As far as the adequacy of the initial antimicrobial therapy was concerned, drug regimen, dose and route were always correct. However, among the 70 patients exhibiting a bacteriologically documented pneumonia, the causal organism was resistant to the antibiotic(s) prescribed in six patients (8.6%) (Table 5). Thus, the initial antimicrobial therapy was considered adequate in 110 (95%) patients.

Occurrence during ICU stay of complications was not significantly different in the two groups of patients suffering from aspiration pneumonia: directly pneumonia-related complications were observed in 17 cases: sepsis-related complications n = 5 (septic shock n = 3, multiple organ failure n = 2), lower respiratory tract hospital-acquired superinfections n = 11, septic metastasis n = 1 (Staphylococcal meningitidis). Nonspecific pneumonia-related complications occurred in 20 patients: ICU-related complications n = 6 (gastrointestinal bleeding n = 3, catheter-related infection due to Staphylococcus aureus n = 2, pneumothorax n = 1) and complications only attributed to underlying diseases n = 14 (recurrent cerebral stroke n = 4, recurrent cardiac arrest n = 1, gastrointestinal bleeding due to liver cirrhosis n = 1, ketoacidosis n = 1, peritonitis due to gastric neoplasm n = 1, and miscellaneous n = 6).

Crude mortality rate was, respectively, 28% (15/54), 16% (10/62) and 29% (114/389) in the three studied groups. In patients with aspiration pneumonia, death was, in 14 cases, uniquely related to underlying disease favoring or leading to aspiration (cerebral hemorrhage or infarct n = 4, cardiac arrest n = 3, liver cirrhosis n = 3, gastric neoplasm n = 1, hypernatremia with hypovolemic shock n = 1, failure of mechanical ventilation weaning due to neuro-muscular disorders n = 2). Thus, overall attributable mortality rate of aspiration pneumonia was 11% (11/102). No significant difference in attributable mortality was observed between patients from Group 1 (17%) and those from Group 2 (5%). This lack of significance difference could be a type 2 error due to the small numbers of patients in each group. Patients without aspiration pneumonia exhibited an attributable mortality rate significantly higher than overall patients with aspiration (Groups 1 and 2 combined) (28% versus 11%; p = 0.0003). However, when we compared patients with pure aspiration pneumonia admitted into ICU for respiratory distress (Group 1) and patients without aspiration (Group 3), difference in attributable mortality rates was not statistically significant (17% versus 28% p > 0.05). Subject to a possible type 2 error due to the small number of patients in Group 1, this result could reflect the fact that patients with pure aspiration pneumonia as a primary cause of ICU admission had a relatively high attributable mortality, quite similar to rate observed in patients with severe community-acquired pneumonia without aspiration.

Predictors of Mortality of Aspiration Pneumonia

Because of the small numbers of patients in Groups 1 and 2, results provided by a separate prognosis analysis for each group should appear less relevant than those supplied by a prognosis analysis performed in the overall population without distinction between Groups 1 and 2. So, after exclusion of the 14 patients died without any relation with pneumonia, 102 eligible patients were included into the prognosis study. According to univariate analysis 17 significant variables were isolated (Table 6). Sex, risk factors of aspiration according to usual categorization, immunosuppression, hypoxemia (Pa_O2/ FI_O2 < 300 mm Hg), hypercapnia (Pa_CO2 > 50 mm Hg), low platelet count (< 75,000/mm³), low total serum protein (< 45 g/l), polymicrobial causal flora, use of one or more antibiotics as initial treatment, use of systemic corticosteroids and occurrence of ICU-related complications were not significantly associated with prognosis. Among the 17 factors significantly associated with prognosis, 11 were collected within 24 h of admission, and six were ultimately recorded during the ICU stay. According to a first multivariate analysis, studying the 11 initial factors, only four were independently related to prognosis: (1) positive blood culture (p = 0.0001), (2) chest X-ray involvement > 2 lobes (p = 0.006), (3) initial septic shock (p = 0.036) and (4) underlying chronic respiratory insufficiency (p = 0.031). A second stepwise analysis taking into account the 6 prognostic factors recorded during the ICU stay determined three independent predictors of death: (1) ineffective initial antimicrobial therapy (p = 0.0001); (2) use of inotropic support (p = 0.0042); and (3) hospital-acquired lower respiratory tract superinfections (p = 0.0071). When all the 17 factors were entered into an unique stepwise analysis, only four were independently associated with prognosis: (1) ineffective initial antimicrobial therapy (p = 0.0001); (2) positive initial blood culture (p = 0.0001); (3) hospital-acquired lower respiratory tract superinfections (p = 0.0054); and (4) use of inotropic support (p = 0.0078).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although aspiration is a frequent condition in patients admitted to an ICU, incidence of aspiration pneumonia among overall community-acquired pneumonias remains unprecise: (1) in some papers about aspiration pneumonias (3, 8), only the condition was studied and its prevalence among all pneumonias was not determined; (2) in most series studying community-acquired pneumonia, aspiration pneumonias were either excluded (16) or not clearly identified (19); (3) finally, in some publications (5, 6), diagnosis of aspiration pneumonia was only evoked when etiologic research remained negative. Among the few studies taking aspiration into account among all community-acquired pneumonias admitted to an ICU (7, 15, 20), the incidence of this condition was around 15%. In the present work, incidence reached 24%. This high rate could be explained by the specific features of the study sites, where admission of patients suffering from neurological disturbances was frequent. Indeed, in our series, more than half the patients were initially admitted into ICU for major neurological disturbances and aspiration pneumonia diagnosis was delayed.

Data about microbiological diagnosis of aspiration pneumonia are disparate. Most previous studies have involved patients seen relatively late, when complications such as necroziting pneumonia, lung abscess or empyema had already occurred (1, 2). In a recent publication, Mier and colleagues (8) have studied the bacteriology of early aspiration with the use of protected brushing. Positive results were obtained for only nine of 42 patients. The 14 pathogens recovered were S. pneumoniae (28.5%), other Streptococcus spp. (21.4%), S. aureus (14.3%), and gram-negative bacilli (35.7%). In our study, cultures were positive in 70 (60%) patients. The most common aerobic causative organisms were S. pneumoniae (23%), Staphylococcus spp. (29%), and gram-negative bacilli (40%). Significant isolation of anaerobic bacteria in patients suffering from lower respiratory tract infection requires specific sampling techniques, adequate transport conditions and specific growth media. The design of our study, retrospective and prospective multicenter data collection, did not allow such bacteriological investigations. So, in our series, precise data about anaerobes in etiology of aspiration pneumonia was unavailable.

Prognosis of aspiration pneumonia varies considerably from one study to another. In Cameron and coworker's study (4), for example, the mortality rate was 62% on average and reached 90% when patients had more than one lobe of the lung involved. In numerous series, most deaths appeared to be attributable to the aspiration rather than to the disease leading to aspiration (4, 21, 22). In Fine and associate's publications (5, 6), aspiration pneumonia was identified as a predisposing factor for a complicated course or mortality. Conversely, in a previous work of our group aiming to determine a prognostic index of severe community-acquired pneumonia, aspiration was a predictor of survival (23). In Hickling and colleague's publication (3), the mortality rate appears low (21%), and death was most often attributable to underlying illness rather than to aspiration. This low mortality level does not seem explicable by differences in the severity of studied patients, since in this study (3), mean predicted mortality derived from APACHE II (24) was 43 ± 24%. In our series, crude mortality was 22% and, after excluding patients who died in relation with diseases leading to aspiration, attributable mortality linked to pneumonia was 11%. This mortality rate was significantly lower than observed in nonaspiration pneumonia (28%). In an attempt to explain this low attributable mortality, we have separated patients exhibiting aspiration pneumonia according to medical ground leading to ICU admission. In patients admitted for major neurological disturbances attributable mortality was low (5%). In patients with pure aspiration, admitted for respiratory distress, mortality rate was quite higher (17%), but lower than in nonaspiration pneumonia (28%). However, on account of the small numbers of patients in each group of aspiration pneumonia and the usual criteria of statistical analysis, significance of differences in attributable mortality rates between patients according to the mechanism of pneumonia (Group 3 versus Group 1) or the main medical ground underlying to aspiration (Group 1 versus Group 2) is difficult to assess. So, in our series, even if aspiration pneumonia was associated with a significant lower mortality rate than nonaspiration pneumonia, inclusion of patients suffering from severe neurological disturbances with, perhaps, mild pneumonia could explain this feature. Consequently, further studies including a greater number of patients seem needed to definitely state about mortality of pure aspiration pneumonia in comparison with severe nonaspiration pneumonia.

We identified, using monovariate analysis, 17 prognostic factors. Most of them have been previously described in prognostic studies about severe community-acquired pneumonia (7, 15, 20). Among all variables collected within 24 h of admission, only four were independent predictors of mortality. Three witnessed the initial severity of lung infectionpositive blood culture, large chest X-ray involvement and initial shockand, only one reflected comorbid illnesschronic respiratory insufficiency. When all variables collected within ICU stay were entered into a unique stepwise analysis, only one of these early predictors was maintained as an independent predictor of mortalitypositive blood culturein association with three factors collected during the ICU stay: use of inotropic support, ineffective initial antimicrobial therapy, and occurrence of hospital-acquired lower respiratory tract superinfections. These results suggest some comments: First, whatever was the initial presentation of patient with aspiration pneumonia, the physician's ability to identify low- versus high-risk patients appears questionable. Second, efficacy of initial antimicrobial therapy appears as a primordial prognosis factor. However, as recently underlined by Torres and El-Ebiary (25), such criteria, based on clinical evaluation 48-72 h after the beginning of the treatment, is difficult to define and different assessments between physicians could likely occur, leading to misevaluation and potential mistakes about clinical status impairment. Finally, the detrimental role played by hospital-acquired lower respiratory tract superinfections must be underlined. Whatever almost all uncontrolled studies showed a higher mortality rate in patients who develop ventilator-associated pneumonia than in patients who do not, some recent studies (26) suggest that nosocomial pneumonia was only associated with a significant increased mortality when infection was due to high-risk microorganisms such as P. aeruginosa, Acinetobacter spp., or X. maltophila. In our study, results appeared quite the opposite, but only multivariate analysis was performed. A paired-case study could be required to definitely assess the prognostic value of such superinfections occurring during the course of aspiration pneumonia.

In summary, our study underlines that prompt administration of an effective antimicrobial therapy, as soon as diagnosis is suspected, allows one to achieve a low mortality in severe aspiration pneumonia. According to our bacteriological data and to prior literature results, which emphasize that main causal anaerobes in pulmonary infections (Peptostreptococcus, Bacteroides, Prevotella, and Fusobacterium species) are susceptible to most -lactam agents (29, 30), it could be suggested that most effective therapy in aspiration patients does not differ from that in other patients with severe community-acquired pneumonia, inasmuch as -lactam antibiotics are the cornerstone of the initial antimicrobial therapy.