To identify predictors of death among patients with clinically diagnosed ventilator-associated pneumonia.
ALARM was a prospective, observational cohort study from 5/1/03 to 12/31/04. Investigators from 20 U.S. medical centers identified 398 adults meeting the ACCP criteria for ventilator-associated pneumonia. Clinical, diagnostic, treatment, and outcomes data were recorded for the duration of ICU stay. Escalation and de-escalation of antibiotic therapy were defined using a novel classification system that incorporated spectrum and number of drugs prescribed (Figure 1). Patients were classified based on the presence of risk factors for multidrug-resistant pathogens using ATS criteria. Multivariate logistic regression was used to identify variables associated with death.
The study population was 60.8% male and had a mean age of 58.3 years. The final model (Chi2=172, p<0.001) included five predictors of death: increased APACHE II score at ICU admission (1 unit increments, RR 1.05, 95% CI 1.40-1.06); inability to de-escalate antibiotic therapy (no change in therapy vs. de-escalation, RR 2.85, 95% CI 2.08-3.89; escalation vs. de-escalation RR 10.75, 95% CI 6.65-17.15); lack of a response to therapy as demonstrated by persistence of or increase in the CPIS score from baseline to 72 hours (RR 1.40, 95% CI 1.33-1.48); failure to use a BAL to diagnose VAP (RR 2.32, 95% CI 1.75-3.03); presence of risk factors for multidrug-resistant pathogens (RR 2.19, 95% CI 1.49-3.23).
This multicenter study confirms several single-center reports that increased severity of illness, failure to use an invasive diagnostic strategy employed, and poor response to therapy are significant predictors of death. Our analysis also validates the recent ATS recommendation that risk factors for multidrug-resistant pathogens should be considered when selecting empiric VAP therapy.
Several predictors of death are modifiable and provide opportunities for improving patient care. Diagnosis of VAP should be aggressively pursued using invasive methods if feasible. Empiric therapy should consider risk factors for multidrug-resistant pathogens and should be de-escalated when possible. Serial CPIS monitoring identifies patients who are failing therapy and have a poor prognosis.
Classification of Antibiotic Therapy Escalation/De-Escalation.Chart notation indicates number of drugs/spectrum of activity. Spectrum of activity was defined using the following hierarchy of drug categories: carbapenems > cefepime > ureidopenicillin/monobactam > fluoroquinolone > other/none.
Lee Morrow, Grant monies (from industry related sources) Elan Pharmaceuticals.