Ventilator-Associated Pneumonia in Pediatric Intensive Care Unit Patients: Risk Factors and Outcomes Alexis M. Elward, David K.

Warren and Victoria J. Fraser Pediatrics 2002;109;758-764 DOI: 10.1542/peds.109.5.758

This information is current as of March 15, 2006

The online version of this article, along with updated information and services, is located on the World Wide Web at: http://www.pediatrics.org/cgi/content/full/109/5/758

PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly publication, it has been published continuously since 1948. PEDIATRICS is owned, published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois, 60007. Copyright 2002 by the American Academy of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.

Downloaded from www.pediatrics.org by on March 15, 2006

Ventilator-Associated Pneumonia in Pediatric Intensive Care Unit Patients: Risk Factors and Outcomes
Alexis M. Elward, MD; David K. Warren, MD; and Victoria J. Fraser, MD
ABSTRACT. Objectives. To determine the rates, risk factors, and outcomes of ventilator-associated pneumonia in pediatric intensive care unit (PICU) patients. Methods. A prospective cohort study was conducted at the St Louis Childrens Hospital PICU on all patients who were admitted to the PICU from September 1, 1999, to May 31, 2000, except those who died within 24 hours, were >18 years of age, or were neonatal intensive care unit patients on extracorporeal membrane oxygenation. The primary outcome measured was the development of ventilator-associated pneumonia. Secondary outcomes were death and hospital and PICU length of stay. Multiple logistic regression analysis was performed to determine independent predictors for ventilator-associated pneumonia. Results. There were 34 episodes of ventilator-associated pneumonia in 30 patients of 911 admissions (3.3%) and 595 (5.1%) mechanically ventilated patients. The mean ventilator-associated pneumonia rate was 11.6/1000 ventilator days. By logistic regression analysis, genetic syndrome (odds ratio [OR]: 2.37; 95% confidence interval [CI]: 1.015.46), reintubation (OR: 2.71; 95% CI: 1.18 6.21), and transport out of the PICU (OR: 8.90; 95% CI: 3.8220.74) independently predicted ventilator-associated pneumonia. Conclusions. Ventilator-associated pneumonia occurs at significant rates among mechanically ventilated PICU patients and is associated with processes of care. Additional studies are necessary to develop interventions to prevent ventilator-associated pneumonia. Pediatrics 2002;109:758 764; ventilator-associated pneumonia, pediatric intensive care unit, nosocomial infection.
ABBREVIATIONS. ICU, intensive care unit; PICU, pediatric intensive care unit; NNIS, National Nosocomial Infection Surveillance; SLCH, St Louis Childrens Hospital; PRISM, Pediatric Risk of Mortality Score; TPN, total perenteral nutrition.

he epidemiology and outcomes of ventilatorassociated pneumonia are well described in adults, but few data exist for pediatric patients, particularly with respect to risk factors and outcomes such as attributable morbidity, mortality, and cost. Ventilator-associated pneumonia is defined as nosocomial pneumonia in mechanically ventilated patients that was not present at the time of intubation.
From the Division of Infectious Diseases, Departments of Pediatrics and Internal Medicine, Washington University School of Medicine, St Louis, Missouri. Received for publication Jan 3, 2001; accepted Dec 12, 2001. Reprint requests to (V.J.F.) Campus Box 8051, 660 S Euclid Ave, St Louis, MO 63110. E-mail: elwarda@kids.wustl.edu PEDIATRICS (ISSN 0031 4005). Copyright 2002 by the American Academy of Pediatrics.

The estimated incidence of ventilator-associated pneumonia in adult intensive care unit (ICU) patients is 10% to 65%.1 The lower respiratory tract is the most common site of nosocomial infection in adult ICU patients, accounting for up to 30% of nosocomial infections in this population.1 Risk factors for ventilator-associated pneumonia in adults include duration of mechanical ventilation, exposure to antibiotics, prolonged ICU stay, the presence of invasive devices, treatment with antacids or histamine type 2 receptor blockers, and advanced age.2 Increased mortality has been associated with infection attributable to Pseudomonas aeruginosa and Acinetobacter species, more severe underlying illness, and inappropriate antibiotic therapy.35 In adult ICU patients, the median excess length of hospital stay as a result of ventilator-associated pneumonia is estimated at 7.7 days,1 and the attributable mortality of ventilator-associated pneumonia is estimated at 10%.6 Ventilator-associated pneumonia is estimated to cost in excess of $5000 to $8000 per episode7 with an estimated total US cost of $1.1 billion per year in 1985 dollars.8 (Also, see the following: Kappstein I, Schulgen G, Beyer U, Geiger K, Schumacher M, Daschner FD. Prolongation of hospital stay and extra costs due to ventilator-associated pneumonia in an intensive care unit. Eur J Clin Microbiol Infect Dis. 1992;11:504 508.) Three recent studies have examined the incidence and prevalence of ventilator-associated pneumonia in pediatric ICU (PICU) patients. The National Nosocomial Infection Surveillance (NNIS) program sponsored by the Centers for Disease Control and Prevention collects information on nosocomial infections in several hundred hospitals of different sizes. In NNIS hospitals, the pooled mean ventilator-associated pneumonia rate was 6/1000 ventilator days for PICU patients.9 Ventilator-associated pneumonia was the second most common cause of nosocomial infection, representing 20% of nosocomial infections in this population. The highest age-specific rates of ventilator-associated pneumonia occurred in the 2to 12-month age group, and the most common causative organism was P aeruginosa, which accounted for 22% of cases.9 NNIS data does not contain information on risk factors for or outcomes of ventilatorassociated pneumonia. A second study of 20 PICUs in 8 countries performed by the European Multicenter Study Group found that the incidence of nosocomial infection was 23.6% and the most frequent nosocomial infection was pneumonia (53%). P aerugi-

758

PEDIATRICS Vol. 109 No. 5 May 2002 Downloaded from www.pediatrics.org by on March 15, 2006

nosa caused 44% of ventilator-associated pneumonia.10 In this study, patients with nosocomial infection had a longer mean length of stay in the PICU (26.1 17.3 vs 10.6 6 days; P .001). The Pediatric Prevention Network of the National Association of Childrens Hospitals and Related Institutions performed a cross-sectional observational study to determine the point prevalence of nosocomial infection, including bloodstream infections and ventilator-associated pneumonia on a single day in 35 PICUs. In this study, the overall prevalence of nosocomial infection was 12%. Bloodstream infection was the most common nosocomial infection (41.3%), and ventilator-associated pneumonia was the second most common (22.7%). Infections acquired in the PICU were associated with a significant increased risk of death (relative risk: 3.4; 95% confidence interval: 1.57.6). Risk factors and outcomes specific to ventilator-associated pneumonia from this study have not yet been reported.11 Ventilator-associated pneumonia seems to be a common cause of nosocomial infection in PICU patients. Additional studies are needed to determine risk factors for ventilator-associated pneumonia and the impact of ventilator-associated pneumonia on PICU and hospital length of stay, mortality, and cost. Children have different anatomy and physiology, have different underlying illnesses, and undergo different surgical procedures from adults. Because children are so different from adults, specific studies of the risk factors and outcomes of ventilator-associated pneumonia must be performed in PICU patients. We performed a prospective study in PICU patients in a large childrens hospital to determine 1) the rate of ventilator-associated pneumonia, 2) risk factors for ventilator-associated pneumonia, and 3) PICU and hospital length of stay and mortality associated with ventilator-associated pneumonia.
METHODS Setting
St Louis Childrens Hospital (SLCH) is a 235-bed academic tertiary care center affiliated with Washington University School of Medicine. SLCH has a 300-mile radius referral base in southeastern Missouri and southwestern Illinois. The PICU is a combined medical surgical ICU that had 22 beds from September 1, 1999, to November 15, 1999, which expanded to 26 beds on November 15, 1999. The SLCH PICU admits approximately 1400 patients per year. The PICU is staffed by 2 critical care attendings, 2 critical care fellows, a cardiology attending, a cardiology fellow, and 4 pediatric house staff (second- and third-year residents). Critical care attendings rotate every 2 weeks; cardiology attendings and fellows, critical care fellows, and pediatric house staff rotate monthly. The nurse-to-patient staffing ratio is at most 2 to 1.

underlying illnesses, surgeries and procedures, medications, central and arterial catheters, PICU and hospital length of stay, PICU discharge disposition, and nosocomial infections were recorded. Admission Pediatric Risk of Mortality (PRISM) III scores were calculated according to the method of Pollack et al.12 Institutional Review Board approval was obtained from the Washington University School of Medicine Institutional Review Board; a waiver for written informed consent was obtained because of the observational nature of this study. After the initial multivariate analysis was performed, additional information on transport out of the PICU was obtained retrospectively by reviewing the electronic medical records for radiographic reports of computed tomography scans and magnetic resonance imaging, and for operative reports, to determine the date and destination of the first transport out of the PICU. Blood bank records of PICU patients who received a transfusion were reviewed retrospectively for date and amount and type of blood product transfused. New multivariate models, replacing transfusion with transfusion before the development of ventilatorassociated pneumonia, and using both dichotomous and continuous variables for each type of blood product, were created.

Definitions
NNIS definitions were used for nosocomial infections, bloodstream infections, and ventilator-associated pneumonia. A nosocomial infection was defined as an infection not present or incubating at the time of PICU admission, with onset after 48 hours of ICU stay. For diagnosing ventilator-associated pneumonia, the patient was required to have received at least 48 hours of mechanical ventilation and develop new and persistent radiographic evidence of focal infiltrates 48 hours or more after the initiation of mechanical ventilation. In addition, patients had to have 2 of the following: fever 38C, leukocytosis (white blood cell 12 000/ mm3), and purulent sputum (25 white blood cells/high powered field on tracheal aspirate Gram stain).13 Associated organisms were designated as those organisms recovered from tracheal aspirates or bronchoalveolar lavage from patients with ventilatorassociated pneumonia. Lung disease was defined as having a history of restrictive or obstructive pulmonary disease, including asthma, bronchopulmonary dysplasia, and cystic fibrosis, or chronic hypoxemia or hypercapnia requiring home oxygen or home mechanical ventilation, as documented in the admission history and physical. Congenital heart disease was defined as the possession of abnormal cardiac anatomy at birth, including an abnormal number of chambers, intracardiac shunt, or abnormally structured circulation valves or coronary arteries. Developmental delay was defined as a significant lag in or complete failure to attain normal developmental milestones as documented by a physician in the medical record. Genetic syndrome was defined as a known chromosomal abnormality or cluster of physical findings consistent with a recognized genetic syndrome, as documented in the admission history and physical. Transfusion was defined as the receipt of packed red blood cells, platelets, fresh-frozen plasma, or cryoprecipitate at any time during the PICU stay. Transport out of the PICU was defined as the physical movement of the patient out of the PICU to locations such as the operating room or radiology.

Data Analysis
Data were entered into an Access database. Double data entry was performed. Data analysis was performed using SPSS Version 10 (SPSS, Chicago, IL). Categorical variables were compared using 2 analysis. Continuous variables were compared using the Wilcoxon rank sum test for nonnormally distributed variables. P .05 was considered significant. All tests of significance were 2 tailed. Variables that were significantly associated with ventilator-associated pneumonia in univariate analysis, that were present in 10% of patients with ventilator-associated pneumonia, and that had a priori clinical significance, giving consideration to biological plausibility, and variables that had been identified by previous investigators as risk factors for ventilator-associated pneumonia in adults, such as total parenteral nutrition (TPN) and steroids, were entered into multiple forward stepwise logistic regression models. Significant variables that were thought to co-vary were grouped, and only 1 variable from each group was chosen for entry into the model. The final model was chosen on the basis of biological plausibility and by selecting the logistic regression model with the

Patients
All patients who were admitted to the PICU from September 1, 1999, to May 31, 2000, were enrolled in the study unless they met any of the following exclusion criteria: death within 24 hours of PICU admission, age greater than 18 years, or a neonatal ICU patient on extracorporeal membrane oxygenation occupying PICU bed space but cared for by the neonatal ICU physician team.

Data Collection
Medical records, including charts, daily flow sheets, and laboratory and radiographic reports, were reviewed prospectively by 1 of the investigators (A.M.E.). Data on patient demographics,

Downloaded from www.pediatrics.org by on March 15, 2006

ARTICLES

759

highest Hosmer and Lemeshow test of significance and the lowest 2 log likelihood function.14 First-order interactions among significant variables in the final model were tested, and no first-order interactions were found. After the variable transfusion was noted to be a statistically significant predictor of ventilator-associated pneumonia, new variables including transfusion before ventilator-associated pneumonia, packed red blood cells before ventilator-associated pneumonia, fresh-frozen plasma before ventilator-associated pneumonia, platelets before ventilator-associated pneumonia, and cryoprecipitate before ventilator-associated pneumonia were created and entered into the model in place of the variable transfusion. For each type of blood product, amounts and types received after the development of ventilatorassociated pneumonia were censored; thus, the variables entered into the multivariate model reflect only amounts and types of products received before infection. Both categorical and continuous variables for each type of blood product were tested in the multivariate model.

RESULTS

A total of 911 patients were enrolled from September 1, 1999, to May 31, 2000. Patient demographics, underlying disease, procedures, and medications are summarized in Table 1. There was a slight predominance of male patients and a marked predominance of white patients. The mean age was 5.47 5.90 years (median age: 2.94 years). The mean and median PRISM scores were 8.48 6.13 and 7.00, respectively. Congenital heart disease (29%) and lung disease (25%) were common, whereas burns and congenital immunodeficiency were uncommon. Seventy percent of patients were intubated. Transfusion was common (32%) as were steroid (30%) and histamine type 2 receptor blocker (22.7%) use. The types

TABLE 1. SLCH PICU Patient Characteristics: Demographics and Underlying Disease Total (N 911; %) Demographics Male Age (mean, y) SD White PRISM score (mean) SD Underlying illness Lung disease Congenital heart disease Seizure disorder Transplants (solid organ and BMT) Trauma Burn Genetic syndrome Cancer Congenital immunodeficiency Preexisting renal failure Procedures Intubation Reintubation Tracheostomy Cardiopulmonary resuscitation Dialysis Transfusion Medications Inotropes Steroids Histamine type 2 receptor blockers Outcomes PICU LOS (mean, d) SD Hospital LOS (mean, d) SD Death 526 (58) 5.47 5.90 674 (74) 8.88 8.80 226 (25) 264 (29) 106 (11.6) 49 (5.4) 85 (9.3) 5 (0.5) 161 (17.7) 48 (5.3) 15 (1.6) 25 (2.7) 646 (70.9) 125 (13.7) 51 (5.6) 55 (6.0) 24 (2.6) 294 (32.3) 163 (17.9) 273 (30) 207 (22.7) 5.34 13.82 14.38 29.59 53 (5.8)

SD indicates standard deviation; BMT, bone marrow transplant; LOS, length of stay.

of blood products received and mean and median amounts are summarized in Table 2. Of the patients who received transfusions, most (94.6%) received packed red blood cells; nearly 40% received platelets. Cryoprecipitate (36.7%) and fresh-frozen plasma (18.4%) were less commonly transfused (Table 2). There were 34 episodes of ventilator-associated pneumonia in 30 patients. The pooled mean ventilator-associated pneumonia rate was 11.6/1000 ventilator days. Patients with ventilator-associated pneumonia had been in the PICU for a mean of 8.9 days before the development of ventilator-associated pneumonia. The microbiology of ventilator-associated pneumonia in the PICU is summarized in Table 3. There was a preponderance of Gram-negative organisms, particularly P aeruginosa and Klebsiella pneumoniae. Univariate analysis comparing mechanically ventilated patients with and without ventilator-associated pneumonia is summarized in Table 4. Comparison of intubated patients with and without ventilator-associated pneumonia demonstrated an association between ventilator-associated pneumonia and underlying illness, specifically burns and genetic syndrome. Ventilator-associated pneumonia was associated with the following procedures: reintubation, tracheostomy, transfusion, transport out of the PICU, the presence of a central line, multiple central venous catheters, bronchoscopy, thoracentesis, and burn debridement. Eight of the 30 patients with ventilator-associated pneumonia had bronchoscopy, 5 of which occurred before the onset of infection (range: 217 days). Three of the 30 patients with ventilator-associated pneumonia had thoracentesis; 2 before the onset of infection at 4 and 7 days, respectively. The following medications were associated with ventilator-associated pneumonia: TPN, steroids, and histamine type 2 receptor blockers. Data on the use of neuromuscular blockade were not obtained. The locations where procedures were performed, specifically the PICU and radiology, were significantly associated with ventilator-associated pneumonia in univariate analysis. Ventilator-associated pneumonia was also associated with bloodstream infections and number of central venous catheters. Of the 56 patients with bloodstream infections, 49 had a primary bloodstream infection. Primary bloodstream infection remained significantly associated with ventilator-associated pneumonia (9/30 [30%] vs 40/595 [6.7%]; P .0001). Five of the 9 patients with primary bloodstream infections and ventilator-associated pneumonia developed the primary bloodstream infections before the development of ventilator-associated pneumonia. Among mechanically ventilated patients, there was a statistically significant association between having an admission PRISM score 10 and the presence of ventilator-associated pneumonia (14/16 [87.5%] vs 201/881 [22.8%]; P .002). Patients with ventilator-associated pneumonia had higher mean admission PRISM scores and longer PICU and hospital lengths of stay (Table 4). Patients with ventilator-associated pneumonia had a higher mortality

760

PICU VENTILATOR-ASSOCIATED PNEUMONIA Downloaded from www.pediatrics.org by on March 15, 2006

TABLE 2.

Types and Amounts of Blood Products Transfused n (%) (N 294) 278 (94.6) 108 (36.7) 117 (39.8) 57 (18.4) Mean Number of Units (Median) SD 5.2 (2) 8.8 4.9 (2) 7.2 6.5 (3) 8.5 3.1 (2) 3.1 Range 171 139 141 119

Blood Product Packed red blood cells Fresh-frozen plasma Platelets Cryoprecipitate

SD indicates standard deviation. TABLE 3. Microbiology of Ventilator-Associated Pneumonia: SLCH PICU Patients Organism Pseudomonas aeruginosa Klebsiella pneumoniae Staphylococcus aureus Yeast Haemophilus influenzae Streptococcus pneumoniae Enterobacter species Enterococcus Serratia marcescens Stenotrophomonas maltophilia Escherichia coli n (%) (N 34) 10 (29.4) 5 (14.7) 4 (11.8) 3 (8.8) 3 (8.8) 2 (5.9) 2 (5.9) 2 (5.9) 1 (2.9) 1 (2.9) 1 (2.9)

rate, which approached statistical significance (20% vs 7%; P .065). When comparison was made among patients who were mechanically ventilated for 8 or more days, with (n 30) and without (n 62) ventilator-associated pneumonia, PICU length of stay remained significantly longer for patients with ventilator-associated pneumonia (27.53 20.09 vs 18.72 35 days), as did hospital length of stay (52.63 37.43 vs 33.77 49.51 days); there was no difference in mortality (20% and 21%, respectively). For multivariate analysis, only the cohort of mechanically ventilated patients was used. PRISM score 10, genetic syndrome, transport out of the PICU, reintubation, TPN, transfusion, steroids, and bloodstream infection were entered into 1 serial model. Only transfusion, reintubation, and transport out of the PICU remained significant in the forward stepwise logistic regression model; genetic syndrome approached statistical significance (P .058; Table 5). Of the 20 patients who had ventilator-associated pneumonia and received transfusions, 18 received blood products before infection. When transfusion before infection was entered into the multivariate model in place of transfusion at any time, transfusion before infection did not remain a statistically significant predictor of ventilator-associated pneumonia (Table 6); genetic syndrome, reintubation, and transport out of the PICU remained statistically significant predictors. Receipt of packed red blood cells, freshfrozen plasma, platelets, and cryoprecipitate before the development of ventilator-associated pneumonia were not statistically significant predictors of ventilator-associated pneumonia in multivariate analysis when tested as either categorical or continuous variables (data not shown). Of the 30 patients with ventilator-associated pneumonia, 12 had received diagnoses of genetic syndromes. Two of these cases were patients who were readmitted to the PICU and had a second episode of ventilator-associated pneumonia; thus, 10 different

patients had genetic syndromes. Five of the 10 received diagnoses of syndromes involving neuromuscular abnormalities, and 2 had syndromes involving craniofacial abnormalities. Four of the 5 patients with neuromuscular disorders had scoliosis, 3 had developmental delay, and 2 had seizure disorders. Of the 20 patients who had ventilator-associated pneumonia and were transported out of the PICU, 11 were transported before the onset of ventilator-associated pneumonia. Six of the 11 were transported 72 hours before the onset of ventilator-associated pneumonia (overall range: 124 days). Ten patients traveled to the operating room, 3 patients traveled to radiology, and 2 patients traveled to both radiology and the operating room. Of the 9 patients who had ventilator-associated pneumonia and were transported out of the PICU subsequent to infection, 4 underwent laryngoscopy/bronchoscopy and/or tracheostomy.
DISCUSSION

We performed a prospective cohort study of ventilator-associated pneumonia in PICU patients, controlling for underlying illness, severity of illness at the time of admission, and device days to measure risk factors for and outcomes of ventilator-associated pneumonia. To our knowledge, only 1 previous study of this size describing risk factors and outcomes of ventilator-associated pneumonia in PICU patients has been published, despite the high incidence and prevalence of nosocomial infections in PICUs recently described in the literature.15 Fayon et al15 found that immunodepressant drugs, immunodeficiency, and neuromuscular blockade were independent predictors of ventilator-associated pneumonia. We found the ventilator-associated pneumonia rate in the SLCH PICU (11.6/1000 ventilator days) to be higher than the pooled mean ventilator-associated pneumonia rate reported by the NNIS PICU study (mean 6/1000 ventilator days). There are several differences between our data and NNIS. In NNIS, hospitals are required to submit only 1 month of data, and case-finding methods are not specified, but many NNIS hospitals use microbiology reports to identify patients with ventilator-associated pneumonia. Culture-based patient identification may underestimate the incidence of ventilator-associated pneumonia. Prospective review of all patient data is more likely to determine accurately the ventilator-associated pneumonia rate. The NNIS database also consists of a heterogeneous group of PICUs. The SLCH PICU, with a large volume of admissions and patients with relatively high admission PRISM scores,
ARTICLES 761

Downloaded from www.pediatrics.org by on March 15, 2006

TABLE 4. Pneumonia

Univariate Analysis of Intubated Patients With and Without Ventilator-Associated VAP (N 30) n (%) No VAP (N 595) n (%) 337 (56.5) 469 (78.7) 122 (20.5) 217 (36.4) 71 (11.9) 42 (7) 2 (.34) 11 (1.8) 121 (20) 90 (15.1) 50 (8.4) 107 (17.9) 35 (5.9) 235 (39.4) 81 (13.6) 102 (17.1) 163 (27.3) 144 (24.2) 88 (14.7) 168 (28.2) 15 (2.5) 48 (8.1) 60 (10.1) 9 (1.5) 67 (11.2) 46 (7.7) 42 (7) 4.68 5.88 9.05 6.24 5.93 15.38 14.8 30.87 P Value .346 .161 .108 .332 .562 .267 .0001 .071 .010 .470 .762 .0001 .0001 .0001 .0001 .007 .008 .006 .0001 .164 .175 .001 .222 .001 .012 .0001 .065 .632 .036 .001 .002

Male White Lung disease CHD Seizures Transplant Burn Immunodeficiency Genetic syndrome* Respiratory arrest Cardiopulmonary resuscitation Reintubation* Tracheostomy* Transfusion* Transport out* TPN* Steroids* H2 Blockers* Multiple CVC* Cardiac repair Lung transplant Bronchoscopy* Cardiac catheterization Thoracentesis* Central line* Bloodstream infection* Death Age (mean, y) PRISM score (mean)* PICU LOS (mean, d)* Hospital LOS (mean, d)*

14 (46.6) 22 (73.3) 10 (33.3) 8 (26.7) 2 (6.7) 4 (13.3) 2 (6.7) 2 (6.7) 12 (40) 6 (20) 3 (10) 17 (56.7) 7 (23.3) 22 (73.3) 20 (66.7) 11 (36.7) 15 (50) 14 (46.7) 13 (43.3) 5 (16.7) 2 (6.7) 8 (26.7) 1 (3.3) 3 (10) 8 (26.7) 10 (33) 6 (20) 3.53 6.34 11.64 8.17 27.53 20.09 52.63 37.43

CHD indicates congenital heart disease; CVC, central venous catheter; LOS, length of stay. * Statistically significant. TABLE 5. Logistic Regression Analysis of Factors Associated With Ventilator-Associated Pneumonia in PICU Patients OR Genetic syndrome Transport out of the PICU Reintubation Transfusion 2.27 7.61 2.45 2.65 95% CI 0.9725.29 3.2217.96 1.055.75 1.036.80 P .058 .0001 .038 .043

OR indicates odds ratio; CI, confidence interval. The final model had a Hosmer and Lemeshow test significance of 0.801 and a 2 log likelihood of 193. Variables entered into the model were PRISM score 10, genetic syndrome, transport out of the PICU, reintubation, TPN, transfusion, steroids, and bloodstream infection. TABLE 6. Logistic Regression Analysis of Ventilator-Associated Pneumonia Controlling for Transfusion Before Infection OR Genetic syndrome Transport out of the PICU Reintubation Transfusion 2.37 8.90 2.71 95% CI 1.035.46 3.8220.7 1.186.21 P .043 .0001 .011 .884

The final model had a Hosmer and Lemeshow test significance of 0.542 and a 2 log likelihood of 186. Variables entered into the model were PRISM score 10, genetic syndrome, transport out of the PICU, reintubation, TPN, transfusion before ventilator-associated pneumonia, steroids, and bloodstream infection.

is probably more representative of PICUs in large academic tertiary care centers. We identified an independent predictor of ventilator-associated pneumonia in critically ill children that has not been described in adults, specifically, having a congenital syndrome. We also identified
762

independent predictors of ventilator-associated pneumonia that have been described in adult ICU patients: reintubation and transport out of the ICU.16,17 We found that ventilator-associated pneumonia in PICU patients was associated with a greater than 4-fold increase in PICU length of stay and a

PICU VENTILATOR-ASSOCIATED PNEUMONIA Downloaded from www.pediatrics.org by on March 15, 2006

3-fold increase in hospital length of stay. Patients with ventilator-associated pneumonia also had a higher mortality, which approached statistical significance. Our study population was demographically similar to those of other PICUs in large academic tertiary care centers with regard to gender, underlying disease, proportion of intubated patients, and mortality. However, the SLCH study population differed in several other ways from other PICU populations described in the literature. The mean age in this study was 5 years, which is older than other published studies, and the mean admission PRISM score was 8.48, which is also higher than in other published studies. The effect of these parameters on the ventilator-associated pneumonia rate is uncertain. The higher mean age might be expected to result in lower infection rates, as the highest age-specific infection rates in the NNIS data were in the 2- to 12-month age group. The relatively high mean PRISM score might be expected to result in a higher infection rate; in 1 prospective study of PICU nosocomial infections, PRISM scores 10 were associated with a 3- to 4-fold increase in the likelihood of nosocomial infections.18 In multivariate analysis, genetic syndrome, reintubation, and transport out of the PICU were significantly associated with ventilator-associated pneumonia. Genetic syndrome has not been previously described as a risk factor for ventilator-associated pneumonia. Genetic syndrome may be a marker for comorbid conditions that might make a child more likely to have a higher PRISM score at admission, a greater number of and length of exposure to invasive devices, a greater number of procedures, or a longer PICU stay with increased opportunity for colonization and infection. Many genetic syndromes in children are associated with neuromuscular weakness, which may predispose patients to aspiration, or craniofacial abnormalities (eg, micrognathia), which may make endotracheal intubation difficult and thus necessitate multiple attempts at intubation and increase the risk of aspiration, or tracheostomy may be required. Reintubation has also been described by previous investigators to be a risk factor for ventilator-associated pneumonia.16 The most likely mechanism is aspiration of gastrointestinal contents during the procedure. A unique risk factor that we identified was transport out of the ICU. This variable has been described as a risk factor in adults in 1 study by Kollef et al.17 Patient transport out of the PICU is unlikely to be a confounder for severity of illness because PRISM score itself is not a significant predictor of ventilator-associated pneumonia in multivariate analysis. There may be a true association between ventilator-associated pneumonia and the physical process of transporting patients. Decreased positive end expiratory pressure as a result of manual ventilation could lead to increased alveolar collapse, pooling of secretions in alveoli, bacterial overgrowth, and pneumonia. Aspiration may be more frequent during transport when patients are positioned and turned and moved from bed to operating room or computed tomography table. Breaches of proper infection control technique may also occur

during transport. It is also possible that patients may be more likely to be emergently reintubated because of unplanned extubation during transport; however, data on the circumstances surrounding reintubation, including unplanned extubation and emergent reintubation, were not collected in this study. Finally, transfusion was identified as an independent predictor for ventilator-associated pneumonia in the initial multivariate analysis. This seemed biologically plausible as the possible immunosuppressive effects of transfusion have been noted in other patient populations. In cardiovascular and colorectal surgery patients, transfusion has been identified as an independent predictor of surgical site infection; the risk of surgical site infection has been shown to increase 7% to 14% with each unit of packed red blood cells transfused.19 22 Transfusions have also been shown to improve survival of renal grafts in renal transplant patients when given before transplant; proposed mechanisms for this immunosuppressive effect include the induction of suppressor T cells and anti-idiotypic antibodies.2327 However, when we retrospectively collected data on the dates and specific types and amounts of blood products transfused and censored transfusions in patients with ventilator-associated pneumonia that were received after infection and added these variables to the multivariate analysis, neither the categorical nor the continuous variables for any type of blood product or for transfusion in general were significant predictors for ventilator-associated pneumonia. We conclude that the original transfusion variable is likely to be a marker for another risk factor, such as severity of illness at the time of infection. Our study had some limitations. Although we measured severity of illness on admission to the PICU, we did not collect data to calculate PRISM III scores for subsequent ICU days. As the mean number of days of mechanical ventilation before the development of ventilator-associated pneumonia was 8.8, it may be that admission PRISM scores are not as informative as PRISM scores several days before the onset of ventilator-associated pneumonia. Our data did not include the number of attempts at intubation, dates or indication for reintubation, or the technique of intubation (ie, fiber-optic bronchoscopy vs laryngoscopy), which may be relevant. This study has a limited ability to determine the attributable morbidity and mortality of ventilator-associated pneumonia. A nested case-control study that matches patients on severity of illness would be able to determine the attributable morbidity and mortality of ventilator-associated pneumonia, and we are in the process of performing this analysis.
CONCLUSION

We performed a prospective cohort study to determine rates of, risk factors for, and outcomes associated with ventilator-associated pneumonia in a PICU. We found a relatively high rate of ventilatorassociated pneumonia and identified independent predictors of ventilator-associated pneumonia, including genetic syndrome, reintubation, and transport out of the PICU. Patients with ventilator-assoARTICLES 763

Downloaded from www.pediatrics.org by on March 15, 2006

ciated pneumonia had longer PICU and hospital lengths of stay, and the difference in mortality approached statistical significance. Studies of interventions to decrease ventilator-associated pneumonia are needed in PICU patients.
ACKNOWLEDGMENTS
This work was supported by grants from Zeneca through the Pediatric Infectious Diseases Society (Dr Elward), Aventis through the National Foundation for Infectious Diseases Society Fellowship Award (Dr Warren), and in part by Centers for Disease Control and Prevention Cooperative Agreement #UR8/ CCU715087 (Dr Fraser).

12. 13.

14.

15.

REFERENCES
1. George DL. Epidemiology of nosocomial ventilator-associated pneumonia. Infect Control Hosp Epidemiol. 1993;14:163169 2. Craven DE, Steger KA. Epidemiology of nosocomial pneumonia: new perspectives on an old disease. Chest. 1995;108:1S16S 3. Celis R, Torres A, Gatell J, Almela M, Rodriguez-Roisin R, Agusti-Vidal A. Nosocomial pneumonia: a multivariate analysis of risk and prognosis. Chest. 1988;93:318 324 4. Kollef MH, Ward S. The influence of mini-BAL cultures on patient outcomes: implications for the antibiotic management of ventilatorassociated pneumonia. Chest. 1998;113:412 420 5. Kollef MH, Silver P, Murphy DM, Trovillion E. The effect of late-onset ventilator-associated pneumonia in determining patient mortality. Chest. 1995;108:16551662 6. Fagon J, Chastre J, Hance AJ, Montravers P, Novara A, Gibert C. Nosocomial pneumonia in ventilated patients: a cohort study evaluating attributable mortality and hospital stay. Am J Med. 1993;94:281288 7. Papazian L, Bregeon F, Thirion X, et al. Effect of ventilator-associated pneumonia on mortality and morbidity. Am J Respir Crit Care Med. 1996;154:9197 8. Jaeger A, Litalien C, Lacroix J, Guertin M, Infante-Rivard C. Protected specimen brush or bronchoalveolar lavage to diagnose bacterial nosocomial pneumonia in ventilated adults: a meta-analysis. Crit Care Med. 1999;27:2548 2560 9. Richards MJ, Edwards JR, Culver DH, Gaynes R, and the National Nosocomial Infections Surveillance System. Nosocomial infections in pediatric intensive care units in the United States. Pediatrics. 1999;103(4). Available at: http://www.pediatrics.org/cgi/content/full/103/4/e39 10. Raymond J, Aujard Y, and the European Study Group. Nosocomial infections in pediatric patients: a European, multicenter prospective study. Infect Control Hosp Epidemiol. 2000;21:260 263 11. Garrett DA, McKibben P, Levine G, Jarvis WR, and the Pediatric Prevention Network, the National Association of Childrens Hospitals and Related Institutions, and the Centers for Disease Control and Preven-

16.

17.

18.

19.

20. 21.

22.

23.

24.

25. 26. 27.

tion. Prevalence of Nosocomial Infections in Pediatric Intensive Care Unit Patients at US Childrens Hospitals. Abstract presented at the Fourth Decennial International Conference on Nosocomial and HealthcareAssociated Infections; March 59, 2000; Atlanta, GA Pollack MM, Patel KM, Ruttimann UE. PRISM III: an updated pediatric risk of mortality score. Crit Care Med. 1996;24:743752 Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for nosocomial infections. In: Olmsted RN, ed. APIC Infection Control and Applied Epidemiology: Principles and Practice. St Louis, MO: Mosby; 1996:A1A20 Menard S. Applied Logistic Regression Analysis. Thousand Oaks, CA: Sage; 1995. University Paper Series on Quantitative Applications in the Social Sciences, 07-106 Fayon MJ, Tucci M, Lacroix J, et al. Nosocomial pneumonia and tracheitis in a pediatric intensive care unit: a prospective study. Am J Respir Crit Care Med. 1997;155:162169 Torres A, Gatell JM, Azar E, et al. Re-intubation increases the risk of nosocomial pneumonia in patients needing mechanical ventilation. Am J Respir Crit Care Med. 1995;152:137141 Kollef MH, Von Harz B, Prentice D, et al. Patient transport from intensive care increases the risk of developing ventilator-associated pneumonia. Chest. 1997;112:765773 Sing-Naz N, Sprague BM, Patel KM, Pollack M. Risk factors for nosocomial infection in critically ill children: a prospective cohort study. Crit Care Med. 1996;24:875 878 Tartter PI, Mohandas K, Azar P, Endres J, Kaplan J, Spivack M. Randomized trial comparing packed red blood transfusion with and without leukocyte depletion for gastrointestinal surgery. Am J Surg. 1998; 176:462 466 Vamvakas EC, Carven JH, Hibberd PL. Blood transfusion and infection after colorectal cancer surgery. Transfusion. 1996;36:1000 1008 Rebollo MH, Bernal JM, Llorca J, Rabasa JM, Revuelta JM. Nosocomial infections in patients having cardiovascular operations: a multivariate analysis of risk factors. J Thorac Cardiovasc Surg. 1996;112:908 913 Vamvakas EC, Carven JH. Transfusion of white-cell containing allogenic blood components and postoperative wound infection: effect of confounding factors. Transfus Med. 1998;8:29 36 Niaudet P, Dudley J, Charbit M, Gagnadoux MF, Macleay K, Broyer M. Pretransplant blood transfusions with cyclosporine in pediatric renal transplantation. Pediatr Nephrol. 2000;14:451 456 Opelz G, Sengar DP, Mickey MR, Terasaki PI. Effect of blood transfusions on subsequent kidney transplantations. Transplant Proc. 1973;5: 253259 van Es AA, Balner H. Effect of pretransplant transfusions on kidney allograft survival. Transplant Proc. 1972;14:608 617 Opelz G, Terasaki PI. Improvement of kidney graft survival with increased numbers of blood transfusions. N Engl J Med. 1978;299:799 803 Opelz G, Varenterghem Y, Kirste G, et al. Prospective evaluation of pretransplant blood transfusions in cadaver kidney recipients. Transplantation. 1997;63:964 967

MOTHER-LESS CHILDREN

AIDS reduces the ratio of healthy workers to dependents (children and the elderly), increasing the care-taking burden of the survivors. The number of orphans and children who have lost their mother, for example, is projected to double to more than 26 million by 2010.

Flavin C. State of the World 2002. World Watch Institute, 2002

Submitted by Student

764

PICU VENTILATOR-ASSOCIATED PNEUMONIA Downloaded from www.pediatrics.org by on March 15, 2006

Ventilator-Associated Pneumonia in Pediatric Intensive Care Unit Patients: Risk Factors and Outcomes Alexis M. Elward, David K. Warren and Victoria J. Fraser Pediatrics 2002;109;758-764 DOI: 10.1542/peds.109.5.758 This information is current as of March 15, 2006
Updated Information & Services References including high-resolution figures, can be found at: http://www.pediatrics.org/cgi/content/full/109/5/758 This article cites 23 articles, 10 of which you can access for free at: http://www.pediatrics.org/cgi/content/full/109/5/758#BIBL This article has been cited by 1 HighWire-hosted articles: http://www.pediatrics.org/cgi/content/full/109/5/758#otherarticle s This article, along with others on similar topics, appears in the following collection(s): Emergency Medicine http://www.pediatrics.org/cgi/collection/emergency_medicine Information about reproducing this article in parts (figures, tables) or in its entirety can be found online at: http://www.pediatrics.org/misc/Permissions.shtml Information about ordering reprints can be found online: http://www.pediatrics.org/misc/reprints.shtml

Citations

Subspecialty Collections

Permissions & Licensing

Reprints

Downloaded from www.pediatrics.org by on March 15, 2006