Ventilator-associated pneumonia. A multivariate analysis.

OBJECTIVES

To identify factors associated with the development of ventilator-associated pneumonia (VAP) and to examine the incidence of VAP in different intensive care unit (ICU) populations.

DESIGN

An inception cohort study.

SETTING

Barnes Hospital, St Louis, Mo, an academic tertiary care center.

PATIENTS OR OTHER PARTICIPANTS

A total of 277 consecutive patients required mechanical ventilation for longer than 24 hours from a medical ICU (75 patients), surgical ICU (100 patients), or cardiothoracic ICU (102 patients).

INTERVENTIONS

Prospective patient surveillance and data collection.

MAIN OUTCOME MEASURES

Ventilator-associated pneumonia and ICU mortality.

RESULTS

Ventilator-associated pneumonia occurred in 43 patients (15.5%). Stepwise logistic regression analysis identified four factors to be independently associated with VAP (P < .05): an organ system failure index of 3 or greater (adjusted odds ratio [AOR] = 10.2; 95% confidence interval [CI], 4.5 to 23; P < .001); patient age of 60 years or older (AOR = 5.1; 95% CI, 1.9 to 14.1; P = .002); prior administration of antibiotics (AOR = 3.1; 95% CI, 1.4 to 6.9; P = .004); and supine head positioning during the first 24 hours of mechanical ventilation (AOR = 2.9; 95% CI, 1.3 to 6.8; P = .013). Ventilator-associated pneumonia occurred more often in cardiothoracic patients (21.6%) compared with medical patients (9.3%) (P = .03). Patients with VAP also had a higher mortality (37.2%) than those without VAP (8.5%) (P < .001). An organ system failure index of 3 or greater (AOR = 16.1; 95% CI, 6.1 to 42; P < .001), a premorbid lifestyle score of 2 or greater (AOR = 3.1; 95% CI, 1.3 to 7.3; P = .012), and supine head positioning during the first 24 hours of mechanical ventilation (AOR = 3.1; 95% CI, 1.2 to 7.8; P = .016) were independently associated with mortality.

CONCLUSIONS

These data suggest potential interventions that might affect the incidence of VAP or outcome associated with VAP. Additionally, they indicate that different ICU populations may have different incidences of VAP.