Nosocomial Pneumonia

Health Care–Associated Pneumonia

Pneumonias acquired outside the hospital by patients with significant contact with the health care system (HCAP) have historically been conflated with HAP because some early case series found that their pathogen distribution, resistance profiles, and mortality rates were similar to pneumonias acquired by patients inside hospitals. The equilibration of HCAP with HAP has been called into question, however, owing to the fear that the HCAP category is unnecessarily broad and drives substantial overuse of broad-spectrum antimicrobials. Prospective case series do document higher rates of methicillin-resistant Staphylococcus aureus (MRSA), Pseudomonas, Acinetobacter, and enteric gram-negative pathogens in patients with HCAP, but the overall prevalence of these pathogens is still low. The pooled prevalences of MRSA and Pseudomonas in HCAP patients across eight prospective studies, for example, were only 2.2% and 2.1%, respectively. Investigators in some case series do report much higher prevalences, but this typically occurs when the analysis is restricted to patients with positive cultures, typically a sicker subset of the total pneumonia population. Pathogen prevalences can also vary substantially by geography, hospital type, and population characteristics. The local prevalence of drug-resistant pathogens, recent exposure to broad-spectrum antibiotics, hemodialysis, poor functional status, and severity of illness appear to be better predictors of drug-resistant pathogens and morbidity than patients’ outpatient health care contacts.

Epidemiology

“Official” VAP rates reported by infection control programs have fallen to extremely low levels over the past 15 years. According to the Centers for Disease Control and Prevention (CDC), the mean VAP rate for medical intensive care units (ICUs) fell from 4.9 to 0.9 VAPs per 1000 ventilator-days between 2002 and 2012. During the same period, VAP rates in surgical ICUs fell from 9.3 to 2.2 events per 1000 ventilator days. Remarkably, the median VAP rate for US medical ICUs in 2012 (the last year for which data are available) was zero. These incidence densities correspond to VAP incidence rates of less than 1 VAP per 100 ventilated patients.

The credibility of these rates has been questioned. They were generated by infection preventionists using the CDC's traditional surveillance definitions for nosocomial pneumonia, but the CDC's old definitions included many subjective criteria such as “new or progressive infiltrates,” “change in the character of sputum,” and “worsening oxygenation.” Many observers were therefore concerned that increasing pressures on US hospitals to report low VAP rates might have influenced surveyors to interpret these very subjective clinical signs more strictly over time. An audit of chart diagnoses of VAP in ventilated patients conducted by the US Centers for Medicare and Medicaid Services found that the rate of clinically diagnosed and treated VAP was in fact stable at approximately 10 cases per 100 ventilated patients between 2005 and 2013. Likewise, cross-sectional surveys of ICUs continue to suggest that about 15% of patients are on antibiotics for nosocomial respiratory infections at any given time.

Fewer data are available on the incidence of nosocomial pneumonia outside ICUs. A 1-day point-prevalence survey of health care–associated infections in 183 hospitals in 10 US states identified 67 nosocomial pneumonias among 10,748 nonventilated patients (0.6% incidence). A second study using discharge diagnosis codes from a 20% sample of US hospitals reported that 1.6% of hospitalized patients had secondary diagnosis codes for pneumonia. A third set of investigators reviewed the charts of all patients with discharge diagnosis codes for pneumonia in 21 US hospitals. Nosocomial pneumonia rates in nonventilated patients ranged from 0.12 to 2.28 cases per 1000 patient-days.

Morbidity and Mortality

VAP is estimated to extend the duration of mechanical ventilation and intensive care by 4 to 6 days. Crude mortality rates for VAP range between 16% and 78%. Crude mortality rates for non-ICU nosocomial pneumonia range from 15% to 53%. Identifying the fraction of these deaths attributable to pneumonia itself rather than to patients’ underlying frailties, however, is controversial. Estimates have ranged from 0% to 50%. More recent studies using multistate models to account for the time-dependent nature of VAP (the longer a patient remains on a ventilator, the higher his or her risk of both VAP and death) have found attributable mortality rates of 8% to 10%. Dutch researchers calculated almost identical figures (9%–13%) using an entirely different method. They estimated attributable mortality as the ratio of the pooled mortality risk reduction to the pooled VAP risk reduction in the treatment versus control arms of randomized controlled trials of VAP prevention interventions. Estimates using marginal structural models are even lower: 6% of ICU deaths are due to VAP, corresponding to a net-attributable mortality rate of 1.5%.

Pathophysiology

The histologic hallmark of VAP is heterogeneity. Autopsies of ventilated patients’ lungs are often notable for widely scattered, patchy areas of inflammation. The lesions vary significantly in age and severity, ranging from bronchiolitis to bronchopneumonia to frank abscess, often within the same lung. Dependent areas tend to be affected more often than nondependent areas. Different organisms can be cultured from different lung segments of the same patient in 25% to 37% of cases. Cultures of histologically benign–appearing lung segments are often positive. Taken together, these findings suggest that ventilated patients are prone to repeated microaspirations around the endotracheal tube cuff, some of which progress to clinically manifest pneumonia, whereas others resolve spontaneously.

Microbiologic, structural, humoral, and pharmacologic factors combine to increase the risk of pneumonia in critically ill patients. The flora of the oral tract rapidly shifts on admission from typical community respiratory organisms (streptococci, Haemophilus ) toward “hospital-associated” pathogens such as S. aureus, Enterobacteriaceae, Pseudomonas, and Acinetobacter species. The likelihood of these organisms being drug resistant steadily increases with time in the hospital, exposure to antimicrobials, and severity of illness. Orogastric and nasogastric feeding tubes disrupt the lower esophageal sphincter and increase the risk of aspiration of gastric contents. The presence of an endotracheal tube disrupts normal ciliary clearance of constitutive bronchial secretions and impairs patients’ capacity to cough. Secretions therefore pool above the endotracheal tube cuff and intermittently seep around folds in the cuff, particularly if the cuff is underinflated or if it shifts during patient movement or repositioning. Biofilm begins to form both inside and outside the endotracheal tube within a day of placement and serves as a bacterial reservoir within the trachea and oropharynx. Suctioning or instillation of aerosols through the endotracheal tube can mobilize and embolize bacteria from the biofilm into the lungs. Critical illness, poor nutrition, sedation, and immobilization may increase patients’ susceptibility to infection. These factors interact with and reinforce one another to enhance the risk of microaspiration and the likelihood that pulmonary parenchymal colonization will lead to invasive infection.

Risk Factors

Understanding the pathophysiology of pneumonia helps predict HAP and VAP risk factors. Factors that enhance the risk of aspiration increase the likelihood of infection. These include mechanical factors such as emergency intubation, reintubation, duration of intubation, supine positioning, enteral feeding through use of orogastric or nasogastric tubes, use of paralytic agents, and underinflation of the endotracheal tube cuff ; factors that affect mental status such as central nervous system disease, impaired consciousness, and depth of sedation ; factors that increase bacterial bioburden in the upper respiratory and orogastric tracts, such as duration of hospitalization, nasogastric intubation, prolonged antibiotic exposures, and the use of proton pump inhibitors or other gastric acid suppressants ; factors that increase handling or breaking of the ventilator circuit, such as inhaled β-agonist therapy ; and patient factors such as age, preexisting lung disease, and severity of illness. Intensive care staffing levels and transportation out of the ICU for diagnostic imaging or procedures are additional factors that can increase risk. Surgical patients in general, and burn and trauma patients in particular, have higher VAP rates than medical patients.

Diagnosis

More than 100 years ago, Sir William Osler noted a striking discrepancy in pneumonia rates between the wards and the autopsy suite. Contemporary autopsy series continue to confirm the limitations of bedside diagnosis for nosocomial pneumonia. Tejerina and colleagues, for example, evaluated common clinical definitions for VAP in a population of 253 patients, with histologic findings as the reference standard. Defining VAP as a new radiographic infiltrate and two additional signs—leukocytosis, fever, or purulent respiratory secretions—was 65% sensitive and 36% specific. Defining VAP more strictly by requiring radiographic infiltrates and all three of leukocytosis, fever, and purulent respiratory secretions increased specificity to 91% but dropped sensitivity to 16%. Further requiring positive cultures (tracheal aspirate culture with pathogenic organisms, no minimum growth threshold) raised the specificity of both the loose and the strict definitions to 93% and 99%, respectively, but with further cost in sensitivity.

HAP and VAP diagnosis are challenging because the cardinal clinical signs of pneumonia are neither sensitive nor specific in hospitalized patients. Hospitalized patients in general, and critically ill ventilated patients in particular, are at risk for a host of pulmonary complications including pulmonary edema, atelectasis, thromboembolic disease, acute respiratory distress syndrome (ARDS), hypersensitivity reactions, and hemorrhage, in addition to pneumonia. All of these complications tend to present with some combination of a common core set of clinical signs: fever, leukocytosis, impaired oxygenation, changes in the character and quantity of sputum production, and radiographic infiltrates. The problem is compounded by the fact that many hospitalized patients have abnormal lung conditions at admission that further complicate the interpretation of radiographs and result in impaired oxygenation and abnormal sputum production. These include cancers, scars from prior surgeries, bronchiectasis, obstructive lung disease, pulmonary fibrosis, and, in the case of trauma patients, contusions, lacerations, pulmonary hemorrhage, inhalation injuries, or combinations of these insults. Most patients are found to have two or more conditions contributing to their “pneumonia-like” syndrome. Sometimes fever, leukocytosis, and inflammatory syndromes caused by extrapulmonary disease (e.g., bloodstream infections, surgical site infections, pancreatitis) can occur concurrently with noninfectious pulmonary conditions (e.g., atelectasis, pulmonary edema, ARDS) to render a net clinical picture that looks like pneumonia even though the contributing conditions in isolation would rarely be confused with pneumonia. All told, VAP is confirmed at autopsy in only two-thirds of patients with clinically suspected pneumonia.

Alveolar infiltrates and air bronchograms are the most sensitive radiographic signs for autopsy-proven VAP, but neither is specific (sensitivity 83% and 88%, specificity 58% and 26%, respectively). Focal, unilateral infiltrates are more specific than bilateral infiltrates (specificities of 80% and 47%, respectively) but are present in only 21% of cases. Radiographic progression alone is a poor predictor of the presence or absence of histologic pneumonia. Accurate radiographic diagnosis is even more challenging in patients with ARDS. In general, there is no correlation between the presence or absence of radiographic infiltrates and positive cultures on quantitative bronchoalveolar lavage (BAL) or autopsy. Computed tomography (CT) is more accurate than portable radiographs to diagnose pneumonia, but few studies have been published on the use of CT specifically for nosocomial pneumonias. On blinded comparison of immediate postmortem CT versus autopsy in 182 patients, radiologists missed 32% of pneumonias and overcalled 14%.

Quantitative BAL cultures have been advocated to increase the accuracy of VAP diagnosis, but studies comparing quantitative bronchoscopic cultures and histologic findings only partially confirm this. Specimens are prone to contamination with organisms residing in the mouth and on the endotracheal tube; sampling the incorrect lung segment can generate both false positives (subclinical bronchiolitis rather than frank pneumonia) and false negatives (an uninfected lung segment instead of the infected segment); and recent antibiotic exposure can misleadingly suppress growth. Bronchoscopy studies bear witness to the microbial complexity and diversity of intubated patients. There is little or no correlation between positive quantitative cultures and the presence of infiltrates or leukocytosis. Cultures taken from different lung segments are discrepant in more than one-third of patients with suspected VAP. There is little correlation between Gram stain characteristics (gram positive vs. gram negative) and final culture results. In addition, VAP exists on a microbiologic and histologic spectrum that may not cross arbitrary quantitative thresholds for positivity. Studies evaluating the accuracy of quantitative BAL for VAP relative to histologic findings have found sensitivities of 14% to 90% (weighted average, 57%), specificities of 42% to 100% (weighted average, 80%), and positive predictive values of 29% to 100% (weighted average, 77%). Of note, negative cultures are not protective: Patients with culture-negative VAP have outcomes similar to or worse than those of patients with culture-positive VAP.