Survival in Patients with Nosocomial Pneumonia: Impact of the Severity of Illness And

Survival in patients with nosocomial pneumonia: Impact of the severity of illness and the etiologic agent

Critical Care Medicine

(C) Williams & Wilkins 1997. All Rights Reserved.

Volume 25(11), November 1997, pp 1862-1867

[Clinical Investigations]

Rello, Jordi MD, PhD; Rue, Montse PhD; Jubert, Paola MD; Muses, Graciela MD;

Sonora, Rosario MD; Valles, Jordi MD, PhD; Niederman, Michael S. MD, FCCP

From the Intensive Care (Drs. Rello, Jubert, Muses, Sonora, and Valles) and

Epidemiology (Dr. Rue) Departments, Consorci Hospitalari Parc Tauli in Sabadell,

Barcelona, Spain; and the Pulmonary and Critical Care Department (Dr. Niederman),

Winthrop-University Hospital, Mineola, NY.

Supported, in part, by grant 94/1456 from the Fondo de Investigaciones

Sanitarias de la Seguridad Social.

Abstract

Objective: To assess the impact of severity of illness at different times,

using the Mortality Probability Models (MPM II), and the impact of etiologic

agent on survival in patients with nosocomial pneumonia.

Design: Retrospective, observational study.

Setting: Fourteen-bed medical-surgical intensive care unit (ICU) in a teaching

hospital.

Patients: Sixty-two patients with nosocomial pneumonia who were receiving early

appropriate antibiotic treatment.

Interventions: None.

Measurements and Main Results: Severity of illness at the time of admission to

the ICU (M0), 24 hrs after admission (M24), and at the time of pneumonia

diagnosis (M1) was determined using MPM II. Bacteriology was established by

quantitative cultures from bronchoscopic samples. The outcome measure was the

crude mortality rate.

The crude mortality rate in the ICU was 59.7%, compared with average predicted

mortality rates of 43.5% (M0), 36.4% (M24), and 52.2% (M1). We observed

significant differences in mean MPM II determinations between survivors and

nonsurvivors at M1 (39.3% vs. 60.9%, p = .001) but not at M0 and M24. In the

univariate analysis, the variables most predictive of mortality were the

presence of coma (p = .02), inotropic medication use (p = .001), and an MPM II

determination of >50% (p = .001) when pneumonia was diagnosed (M1). Multivariate

analysis showed that, in the absence of Pseudomonas aeruglnosa, an MPM II

determination of >50% at M1 was associated with a relative risk of death of 4.8.

The presence of P. aeruginosa was associated with an increase in the risk of

death of 2.6 and 6.36 in both populations with MPM II determinations at M1 of

50%, respectively.

Conclusions: Severity of illness when pneumonia is diagnosed is the most

important predictor of survival, and this determination should be used for

therapeutic and prognostic stratification. In addition, the presence of P.

aeruginosa contributed to an excess of mortality that could not be measured by

MPM II alone, suggesting the importance of the pathogen in prognosis. (Crit Care

Med 1997; 25:1862-1867)

------

Key Words: respiratory infections; ventilator-associated pneumonia; pneumonia;

outcome; severity of illness; Pseudomonas aeruginosa; MPM II; mortality;

nosocomial infection; survival

The association between severity of illness and the impact of nosocomial

infection by multidrug-resistant organisms on outcome is complex. Given the

overlap of risk factors for pneumonia and risk factors for mortality, the

question is raised concerning how much pneumonia itself, rather than underlying

comorbidity, contributes to mortality in critically ill patients.

Some authors [1-3] have reported that infections with multidrug-resistant

bacteria are associated with an increased attributable mortality compared with

infections with nonresistant organisms. In particular, infection with Pseudomonas

aeruginosa or Acinetobacter baumanii is associated with the highest mortality

rates. However, the impact of these organisms on mortality is uncertain because

they are usually found in patients who are near death. In addition, the

contribution of nosocomial infection to outcome may vary with underlying

severity of illness, the effect of infection being most evident in those who

would not ordinarily have died (i.e., those with less severe disease). In our

opinion, severity systems may be useful as tools for investigating the relative

importance of organism identity and severity of illness with regard to outcome.

In addition, there is the possibility of using severity systems to predict

outcome, although limited information is available about the utility of these

models in specific groups such as intubated patients who develop pneumonia.

The objectives of the present study were a) to analyze the ability of the

Mortality Probability Models (MPM II) system [4] to predict outcome in patients

with nosocomial pneumonia; and b) to evaluate the degree of association between

severity of illness and the nature of the infecting pathogen, with regard to

outcome in patients with nosocomial pneumonia.

Our hypothesis was that the MPM II determinations calculated at admission would

underestimate the actual observed mortality in the population with nosocomial

pneumonia, such as has been the case in intensive care unit (ICU) patients with

sepsis [5]. However, this system might identify patients at risk of dying from

pneumonia if it was employed at the time of pneumonia diagnosis. In addition, we

also hypothesized that the presence of certain organisms, such as P. aeruginosa,

might be associated with excess mortality.

MATERIALS AND METHODS

Patient Data and Definitions.

We performed a retrospective chart review study to identify intubated patients

with nosocomial pneumonia in a 14-bed medical-surgical ICU in a teaching

hospital. Eligible study participants were patients with nosocomial pneumonia

etiologically confirmed by quantitative bronchoscopic cultures. This study was

approved by the Institutional Review Board, as part of the Fondo de Investigaciones

Sanitarias de la Seguridad Social 94/1456 study. We identified 65 episodes in 62

patients, all of whom received empiric antimicrobial therapy after the diagnosis

was suspected. The outcomes of 26 episodes with P. aeruginosa isolates have been

reported elsewhere [6]. Patients with isolates resistant to initial empiric

treatment were excluded from the study so that the impact of incorrect therapy

on outcome could be eliminated as a variable. In patients with polymicrobial

infection, the antimicrobial regimen should be active against all isolated

strains. Appropriate antibiotic treatment for P. aeruginosa infection included

combination therapy from the time that bronchoscopy was performed. In the case

of patients with multiple episodes during their ICU stay, values used in

analysis were taken from the period when the last infection was diagnosed;

previous episodes were not considered.

A diagnosis of nosocomial pneumonia was considered when new and persistent

pulmonary infiltrates, not otherwise explained, appeared on a chest radiograph.

In addition, patients had to have at least two of the following: temperature of

>or=to38[degree sign]C; leukocytosis >or=to10,000/mm3; and purulent respiratory

secretions. Pneumonia was considered nosocomial if its onset occurred after 48

hrs of hospital stay and if it was judged not to have been incubating before

admission [2]. Fiberoptic bronchoscopic examination, using a protected specimen

brush or bronchoalveolar lavage, was performed on each patient within 12 hrs

after the development of a new pulmonary infiltrate. Etiology was considered to

have been determined if the protected specimen brush yielded >or=to1000

colony-forming units (cfu)/mL or the bronchoalveolar lavage yielded >or=to10,000

cfu/mL of a specific microorganism. Patients with a diagnosis involving

uncertain etiology were not included in the study.

Combination therapy for P. aeruginosa infections was defined as the use of two

or more antibiotics to which the organism was susceptible in vitro. Therapy was

initiated at the time that the bronchoscopy was performed. Clinical resolution

was defined as complete resolution of all signs and symptoms of pneumonia, along

with improvement, or lack of progression, of all abnormalities on the chest

radiograph [2]. Patients were deemed clinically improved if fever disappeared

and pulmonary infiltrates and physical signs of pneumonia abated [7]. Mortality

was considered to be related to the pulmonary infection if death ensued before

any objective response to antimicrobial therapy occurred or if the pulmonary

infection was considered to be a contributing factor to death, in patients with

other comorbidity [1].

Severity of Illness.

The severity of illness was evaluated at the time of admission to the ICU, 24

hrs after admission, and on the day the diagnosis of pneumonia was made. MPM II

[5] was used to assess severity. This system was developed to estimate the

probability of dying in the hospital at the time of admission to the ICU and 24,

48, and 72 hrs [4] after admission (indices MPM II0, MPM II24, MPM II48, and MPM

II72, respectively). When the pneumonia was diagnosed >or=to4 days after

admission to the ICU, the severity was measured at the time of pneumonia

diagnosis, using the index MPM II72, and this determination was defined as M sub

1. The severity was also calculated after 72 hrs of therapy, using the index MPM

II72, and this determination was defined as M2.

Microbiology.

In patients with suspected ventilator-associated pneumonia, bronchoscopy was

performed [8]. After the protected specimen brush was transected into a sterile

vial containing 1 mL of sterile lactated Ringer's solution, the vial was

agitated vigorously for >or=to60 secs to suspend all the material from the

brush. Specimens were immediately sent to the laboratory for quantitative

cultures. Aliquot portions of 0.01 mL were taken from the original suspension

and inoculated into blood agar, MacConkey agar, buffered charcoal yeast extract

agar, and Sabouraud's agar media. One 0.001-mL aliquot was also inoculated into

chocolate agar media. Culture plates were incubated at 37[degree sign]C under

adequate aerobic and anaerobic conditions; all plates except plates containing

Sabouraud's agar media were evaluated for growth at 24 and 48 hrs. In the case

of the protected specimen brush, bacterial counts of >or=to103 cfu/mL were

considered to indicate pneumonia. Two serial ten-fold dilutions were then done

on the recovered bronchoalveolar lavage fluid, and 0.01-mL aliquot portions of

the original suspension and each dilution were placed onto plates in the same

way as aliquot portions from the protected specimen brush sample. All protected

specimen brush and bronchoalveolar lavage isolates were identified by standard

laboratory techniques [9]. Two blood cultures were done simultaneously in most

of these patients. Cultures of pleural fluid, if present, were done as well.

Statistical Analysis.

Crude mortality in the ICU was considered the outcome variable. Among the

independent variables included in the analysis were severity of illness at the

time of admission (M0), 24 hrs after admission (M24), and at the day of the

diagnosis of pneumonia (M1) (all severity variables were analyzed as continuous

and were also categorized as 50%); age; gender; number of days on mechanical

ventilation; and etiologic microorganism. In addition, there were several

specific variables, included in the MPM II system, at the day of diagnosis of

pneumonia: presence of coma, urine 2.0 mg/dL, prothrombin time >3 secs above

normal use of continuous vasoactive drugs, and Pao2 10% of the patients

exhibited that characteristic. We built a model in which the variables of

severity of illness on the day of pneumonia diagnosis (M1) and type of etiologic

microorganism were analyzed separately. Another model had three dummy variables

that covered the following categories: a) 50% probability of dying and absence

of P. aeruginosa (group B); c) 50% probability of dying and presence of P.

aeruginosa (group C).

RESULTS

Of the 72 identified patients with nosocomial pneumonia, ten patients were

excluded because isolates (P. aeruginosa in six patients, Enterobacteriaceae in

three patients, and Staphylococcus aureus in one patient) were resistant to

empiric treatment. The study population was composed of 62 patients: 39 patients

with episodes of P. aeruginosa pneumonia (group P) and 23 patients with episodes

of pneumonia caused by other microorganisms (group N). All patients received

early and appropriate therapy. Underlying diseases of the patients are shown in

Table 1. More detailed information on etiology is given in Table 2. Three

patients presented with two episodes (Table 3); only the last episode was taken

into account for the study. The mean age of the 62 patients was 60.9 +/- 17.5

(SD) yrs; 38 patients were men and 24 patients were women. The median interval

between admission to the ICU and identification of pneumonia was 8 days (mean

13.7 +/- 21.4). Thirty-nine (62.9%) patients received antibiotics before the

development of nosocomial pneumonia because of concomitant infection.

Thirty-seven patients (22 patients in group P; 15 patients in group N) died and

25 patients were discharged from the ICU, leading to a crude mortality rate of

59.7%, compared with an average predicted mortality of 43.5% at admission (M0),

of 36.4% at 24 hrs (M24), and of 52.2% at the day of diagnosis (M1). Twelve

deaths were considered to be directly related to nosocomial pneumonia (11 deaths

in group P and one death in group N). One patient died before the third day

after diagnosis. The severity of illness was evaluated again the third day after

diagnosis (M2) except for one patient who died before this time. The M2

determinations in the remaining 11 patients with mortality related to pulmonary

infection were greater than or equal to the M1 determinations. On the other

hand, we found M2 determinations to be 1 determinations in five of 25 patients

with clinical resolution and death not related to pulmonary infection (p

(Table 4) shows patient age; length of stay before onset of pneumonia; and

severity of illness at the time of admission, 24 hrs after admission, and on the

day of pneumonia diagnosis, according to survival status and type of microorganism.

The mean age was lower in survivors, but this difference was not statistically

significant. Correlation between M0 and M1 values, using Pearson's correlation

coefficients, was 0.42 (p = .001). In the univariate analysis, the variables

most predictive of mortality were the presence of coma (p = .02), inotropic

medication use (p = .001), and an MPM II determination >50% (p = .001) when

pneumonia was diagnosed (M1). The remaining variables analyzed, including

bacteriology and M0 determination >50%, were not significant. More detailed

information is given in Table 5.

------

Table 4. Description of patients according to survival status and type of

microorganism (mean +/- SD)

Results of the multivariate analysis with ICU mortality as the dependent

variable are shown in Table 6. Model 1, which considers the same increase in the

risk of death when a factor is present independently of the values of the other

variables, shows that inotropic medication use and coma were strongly related to