Comparing Patient Characteristics and Mortality of Gram-Negative versus Gram-Positive Ventilator-Associated Pneumonia.

Author(s):
William Wames; Saad Shebrain; Robert Sawyer

Background:

Ventilator Associated Pneumonia (VAP) is a common and highly morbid occurrence within critical care units. VAP leads to increased morbidity including increased ventilator days, ICU days, hospital stay, cost, and mortality. Starting effective empiric treatment as early as possible with an appropriate antibiotic regimen is paramount in reducing the impact on patients.

Hypothesis:

Given that VAPs can be caused by both gram-negative (GN) or gram-positive (GP) bacteria, we hypothesized that clinical features of patients could be used to differentiate GN from GP VAP and suggest different empiric antimicrobial choices. Furthermore, we hypothesized that patients with GN VAP have a higher mortality overall compared to those with GP VAPs.

Methods:

Data were compiled from 1996-2022 from surgical critical care patients treated for CDC-defined, culture positive, monomicrobial VAP.  GN and GP VAPs were compared using univariate analysis, and logistic regression including patient age, sex, race, trauma status, transfusion status, APACHE II score, maximum white blood cell count (WBC), maximum temperature (Tmax), days from admission to diagnosis, and era (‘96-‘04, ‘06-‘14, ‘15-‘22) were used to determine predictors of GN versus GP VAP. A second model including similar variables plus GN or GP status was used to define predictors of hospital mortality.

Results:

Among 847 VAPs, 602 were GN and 245 GP. The three most common pathogens were Staphylococcus aureus (n=196), Pseudomonas aeruginosa (n=145), and Enterobacter cloacae (n=69). Independent predictors of GN VAP included prior transfusion (OR 1.7 [95% CI, 1.1-2.7], p=0.013) and days from admission to diagnosis (OR 1.02 [95% CI, 1.01-1.03], p <0.001), with a C-statistic of 0.65. No difference in crude mortality was found when comparing GN to GP VAP (19.6% v 18.8%, p=0.78). Independent predictors of mortality included age, APACHE II score, non-trauma diagnosis, high WBC, lower temperature, days from admission to diagnosis, shorter duration of therapy, and GP VAP (OR 1.7 [95% CI, 1.0-2.7], p=0.013), with a C-statistic of 0.86. GN VAPs were treated longer than GP VAPs (13.1±9.8 v 11.4±7.8 days, p=0.016).

Conclusions:

Clinical characteristics poorly predict GN versus GP VAP and cannot be used to tailor empiric antimicrobial therapy. Although GN VAP is associated with prior transfusion and a longer hospital stay before diagnosis, after controlling for other factors, GP VAP is more deadly.