Nosocomial Pneumonia

Diagnosis is difficult, as 40 – 50% of ICU patients with fever and new onset pulmonary infiltrates do not have pneumonia [Chest 101: 1005, 1992; Chest 106: 221, 1994]. Note that nosocomial pneumonias are often polymicrobial, thus the values in the table below exceed 100%.

Microbial Origin of Pneumonia

Microbial Origin of Pneumonia

Ventilator+ Ward Patients++ Pseudomonas 18.9 9 S. Aureus 18.9 26 S. Pneumonia 13.2 31 E. Coli 9.2 14 Hemophilus 7.1 17 Enterobacter 3.8 (not recorded) Proteus 3.8 11 Klebsiella 3.2 14 Enterococcus 1.4 (not recorded) Others 19.1

+JAMA 290: 2558, 2003 ++Intensive Care Med 21: 365, 1995
Aspiration of oral secretions into the mouth is the likely cause of most cases of pneumonia. The CDC criteria for pneumonia (which require new or progressive infiltrate, cavitation, or pleural effusion on CXR and do not include sputum gram stain or culture) are not ideal – CXR does not have a sensitivity of 100% and its specificity is ~ 50% [Chest 101: 1005, 1992; Chest 106: 221, 1994]. Also, traditional methods of diagnosing pneumonia fail when compared to autopsy [Chest 117: 191S, 2000] – it is NOT possible to diagnose pneumonia on clinical criteria alone. Also, note that only 1/3 of infiltrates in the ICU are pneumonia [Chest 114: 1129, 1998] – ARDS is the most common non-infectious cause, also edema and/or atelectasis

Predictive Value of Clinical Criteria for Pneumonia

Criteria Study Likelihood Ratio+ Infiltrate + purulent sputum + fever or leukocytosis Fagon (1988) 1.03 Infiltrate + 2 of: purulent sputum, fever, or leukocytosis Timsit (1995) 0.96

Chest 117: 191S, 2000 (+when compared to autopsy)

Tracheal aspirates have a sensitivity of > 90% but a specificity of 15 – 40% because they are often contaminated [Chest 117: 195S, 2000]. Sputum cultures can be analyzed only after screening them for squamous cells (should be < 25/LPF) and lung macrophages (only need one to confirm lower respiratory tract), and also looking for elastin fibers (indicate lung necrosis if present) and PMNs (> 25/LPF is evidence of infection [J Clin Micro 16: 627, 1982]) – by doing quantitative cultures and requiring 106 CFU/mL (105 for patients on antibiotics), sensitivity and specificity are 76% and 75%, respectively [Am J Resp CCM 171: 388, 2005; Chest 117: 195S, 2000]

Protected specimen brushings allow for quantitative cultures (> 103/mL is a positive culture). Sensitivity is only 66%, but specificity is 90%. Note that the diagnostic accuracy of protected specimen brushings is dependent on whether or not the patient is taking Abx – while patients off Abx only show 24% false negatives, patients currently taking Abx will show almost 46% false negatives [Chest 102S: 557, 1993], thus this technique should only be used in patients off antibiotics A protected broncheoalveolar lavage uses at least 120 mL of saline, although a mini-BAL can sometimes use as little as 10 mL. For BAL, the threshold for positive cultures is 105 colonies/mL (104 on abx). The sensitivity and specificity of the BAL is 70 – 100% [Intensive Care Med 21: 365, 1995] when not on Abx, but markedly reduced when the patient is already taking antimicrobial therapy [Am J Med 95: 601, 1993]. Initial studies suggest that non-bronch BAL are equivalent to standard BAL [Chest 117: 207S, 2000; NEJM 355: 2619, 2006]. Many people still use tracheal aspirates because no studies show a survival benefit to using more invasive techniques, however, tracheal aspirates do increase antibiotic usage

Quantitative Cultures for Diagnosis of VAP

Technique Tracheal Aspirate Protected Brush BAL Diagnostic Threshold 105 – 106 103 104 – 105 Sensitivity 76% 66% 73% Specificity 75% 90% 82% Relative performance most sensitive most specific most accurate Am J Resp CCM 171: 388, 2005; Chest 117: 195S, 2000; Crit Care Med 28: 962, 2000

Parapneumonic effusions occur in 50% of bacterial pneumonia and should be analyzed if possible – Gram stain and culture, glucose, pH. Transudate vs. exudate does not reliably differentiate between infected and sterile effusions [Marino]. Indications for Drainage include: air-fluid level, hydropneumothorax, purulent fluid, pH < 7.0, and/or glucose < 40 mg/dL unless they are already improving on antibiotics

Prevention of VAP

Clearly Efficacious

Selective Oral / Digestive Decontamination

Largest study comes from the Netherlands, compared SDD and SOD to control. Both SDD and SOD lowered the risk of death (OR 0.83 [0.72-0.97] and 0.86 [0.74-0.99], respectively) and gram negative infection (OR 0.43 [0.24-0.77] and 0.49 [0.27-0.87], respectively), without increasing the incidence of antimicrobial resistance. Importantly, antibiotic usage patterns were significantly different between groups (“…During treatment with SDD as compared with standard care, the use of antimicrobial agents with antianaerobic activity was reduced by 27.8% for broad-spectrum penicillins, 45.7% for carbapenems, and 11.6% for lincosamides. Furthermore, quinolone use (mainly ciprofloxacin) was reduced by 31.4%. In contrast, systemic use of cephalosporins increased by 86.6%”). Notably, the cost per-day of SOD and SDD were $1 and $12, respectively [de Smet AM et al. NEJM 360: 20, 2009]

A recent Cochrane meta-analysis of 36 RCTs, which included 6914 patients, demonstrated that SDD reduces both mortality and RTI in adult patients receiving intensive care treatment, and that SOD reduces RTI (but not mortality). Further, the authors state that “The risk of resistance occurring as a negative consequence of antibiotic use was appropriately explored only in one trial which did not show any such effect” [Liberati A et al. Cochrane Database Syst Rev CD000022: 2009]. Importantly, this analysis does not include data from [de Smet AM et al. NEJM 360: 20, 2009; FREE Full-text at NEJM]. Although most studies show a trend towards improved survival in SDD-treated patients, the majority are too small to show a significant survival benefit.

For more on SDD, please see Selective Digestive Decontamination

Likely Efficacious

Elevating The Head of Bed

In a retrospective review of 109 mechanically-ventilated patients, Kollef found an increased risk of pneumonia in supine patients (adjusted odds ratio 2.9) [Kollef MH JAMA 270: 1965, 1993]. A randomized trial of 50 patients suggested that gastroesophageal reflux (detected with scintigraphy) was more likely in supine patients as compared to semi-recumbent [Ibáñez J et al. JPEN J Parenter Enteral Nutr 16: 419, 1992]. Another randomized controlled trial of semi-recumbent versus supine positioning in mechanically ventilated patients was stopped early because of an increased risk of both clinically suspected (34% vs. 8%, p = 0.003) and microbiologically-confirmed (23% vs. 5%, p = 0.016) pneumonia in the supine patients [Drakulovic MB et al. Lancet 354: 1851, 1999]

Refuted by van Nieuwenhoven CA et al. Feasibility and effects of the semirecumbent position to prevent ventilator-associated pneumonia: a randomized study. Crit Care Med 34: 396, 2006

Lower Transfusion Trigger

In the Hebert study (NEJM 2001), there were trends towards lower overall rates of infection (10 vs 11.9%), bacteriamia (7.2 vs 9.5%), and CRBSI (5 vs 4%), although a trend towards more septic shock in the restrictive group (9.8 vs 6.9%)

Leal-Noval et. al. studied 738 patients in a post-operative ICU, and examined the influence of 36 variables on the development of severe postoperative infections (SPIs) in general and individually for pneumonia, mediastinitis, and/or septicemia. After multivariate analysis, the variables associated with SPI (incidence, 9.4%) were reintubation, sternal dehiscence, mechanical ventilation (MV) for > or = 48 h, reintervention, neurologic dysfunction, transfusion of >= 4 U RBCs, and systemic arterial hypotension. Keep in mind that this was not a prospective study, thus the sicker patients may have been transfused [Leal-Noval SR et. al. Chest 119: 1461, 2001]

Bochicchio GV et. al. conducted a prospective observational cohort study of 766 trauma patients admitted to the intensive care unit (ICU), who received mechanical ventilation (MV) for >= 48h, and who did not have pneumonia on admission. Logistic regression analyses controlled for all variables related significantly to VAP by univariate analysis (sex, Injury Severity Score, and ventilator days and ICU length of stay prior to VAP) and found that transfusion of blood products was an independent risk factor for VAP [Bochicchio GV et. al. Surg Infect (Larchmt) 9: 415, 2008]

Based on cardiac surgery data from 2872 patients given 2872 units of RBCs, patients who were given older units had higher rates of in-hospital mortality (2.8% vs. 1.7%, P=0.004), intubation beyond 72 hours (9.7% vs. 5.6%, P<0.001), renal failure (2.7% vs. 1.6%, P=0.003), and sepsis or septicemia (4.0% vs. 2.8%, P=0.01) [Koch CG et al. NEJM 358: 1229, 2008]

Nursing / Staff Ratio

Over a one year period in a pediatric cardiac ICU, the strongest linear correlation was observed between the monthly new infection rate (NIR) and patient days (r = 0.89, P = 0.0001). There was an inverse correlation between the monthly NIR and nursing hours:patient day ratio (r = -0.77, P = 0.003) [Archibald LK et al. Pediatric Infectious disease Journal 16: 1045, 1997]

Possibly Efficacious

Avoid Intubation Altogether

Use of CPAP associated with a lower (0.67 to 0.87) relative risk of ventilator-associated pneumonia [Civetta 4th ed. p 1665, 2009]

Sedation Holiday

While clearly reducing the risk of mechanical ventilation [NEJM 342: 1471, 2000; Schweickert WD. Crit Care Med. 2004], does not appear to reduce the incidence of VAP [Strøm T. Lancet. 2010 Feb 6;375(9713):475-80] (probably because the incidence of VAP is higher early in the course of mechanical ventilation)

Supraglottic Suctioning

Associated with a lower (0.45) relative risk of ventilator-associated pneumonia [Civetta 4th ed. p 1665]. Note that this is one of only four recommendations by the CDC to reduce VAP. That said, not all randomized controlled trials show a statistically-significant difference

Silver-impregnated ETTs

Probably reduce VAP but may not affect outcomes. A phase III, randomized, multicenter trial of 1500 patients intubated for 24h or longer tested the effect of a commercially available ETT coated with silver ions dispersed in a proprietary polymer [17