Locally derived versus guideline-based approach to treatment of hospital-acquired pneumonia in the trauma intensive care unit.

BACKGROUND: Appropriate initial antibiotic therapy for presumed pneumonia in critically ill patients decreases the mortality rate. To achieve this goal, treatment guidelines developed by groups such as the American Thoracic Society (ATS) have been stressed. However, often overlooked is the importance of incorporating local microbiologic data into an empiric algorithm. Our hypothesis was that an empiric algorithm supported by our locally-driven analysis would predict more accurate coverage than one defined strictly by an unmodified guideline-driven approach.
METHODS: Retrospective review of all first hospital-acquired (HAP) and ventilator-associated pneumonia (VAP) pathogens in consecutive trauma intensive care unit (TICU) patients over 18 months. Microbiologic data were analyzed to update our TICU-specific empiric algorithm. The ATS guidelines define patients at risk for multidrug-resistant (MDR) organisms on the basis of standardized criteria and time since admission (early <5 days; late ≥5 days).
RESULTS: A total of 164 pathogens caused 117 pneumonias. For early coverage, ATS guidelines stress identification of MDR risks; these criteria failed to identify 8 of 13 (62%) early MDR pneumonias. For early HAP/VAP with no MDR risks, the ATS guidelines recommend monotherapy; susceptibility differed (49% to ciprofloxacin, 68% to ampicillin-sulbactam, 83% to ceftriaxone). A total of 15% of early pathogens were MDR gram-positive, so addition of vancomycin resulted in adequate predicted coverage of 100%, 79%, and 95% for ciprofloxacin, ampicillin-sulbactam, and ceftriaxone, respectively. For late HAP/VAP, ATS recommends regimens based on broad-spectrum drugs. Vancomycin with ciprofloxacin, cefepime, or piperacillin-tazobactam had predicted coverage of 95%, 94%, and 93%, respectively.
CONCLUSIONS: The empiric algorithm derived from analysis of local microbiologic data predicted significantly better coverage than one defined by an unmodified guideline-driven approach for early HAP/VAP. Our locally-derived TICU algorithm of ceftriaxone+vancomycin for early pneumonia and piperacillin-tazobactam+vancomycin for late pneumonia optimizes the adequacy of initial therapy. Understanding local patterns of pneumonia ensures the creation and maintenance of empiric algorithms that achieve the best clinical outcomes.