OBJECTIVE: To evaluate the long-term effect of a program of rotating antibiotics on the incidence of ventilator-associated pneumonia and the susceptibilities of Gram-negative bacilli responsible for ventilator-associated pneumonia. DESIGN: Prospective program for the surveillance of antibiotic susceptibilities of microorganisms responsible for ventilator-associated pneumonia. SETTING: Academic, university-based, medical intensive care unit (16 beds). SUBJECTS: 2,856 mechanically ventilated patients. INTERVENTIONS: A new program of antibiotic use was introduced at the end of 1996 that involved the rotation of antibiotics in empirical and therapeutic use of the treatment of ventilator-associated pneumonia. The rotation concerned the beta-lactam and aminoglycoside classes, with a rotation interval of 1 month. The use of antibiotics was monitored monthly. No preference was given to any particular antibiotic. In a previous study, the period before the introduction of this protocol (1995-1996) was compared with the period 2 yrs after (1997-1998): The results indicated a decreased incidence of ventilator-associated pneumonia, a lower incidence of potentially resistant Gram-negative bacilli, and increased sensitivities of Gram-negative bacilli, especially Pseudomonas aeruginosa and Burkholderia cepacia. After 1998, we decided to continue a routine for this rotation. The long-term effect of this program was studied by comparing the incidence of Gram-negative bacilli responsible for ventilator-associated pneumonia and their susceptibilities obtained in a third period: 1999-2001. The long-term effect (5 yrs) of such a strategy-2-yr protocol period (1997-1998) and 3-yr routine period (1999-2001)-could be evaluated. MEASUREMENTS AND MAIN RESULTS: During the 7-yr study period, 2,856 patients were mechanically ventilated for >48 hrs. The incidence of ventilator-associated pneumonia remained significantly lower in period 3 (1999-2001): 23% (period 1, 1995-1996) vs. 15.7% (period 2, 1997-1998) vs. 16.3% (period 3, 1999-2001; p =.002). Late-onset ventilator-associated pneumonia occurred in 86.6% and 94% of cases, respectively, in periods 1 and 3 (p =.02). The decrease of the incidence of early-onset ventilator-associated pneumonia was statistically significant during the 7-yr study period: 13% vs. 9% vs. 5.9% (p =.02). Combined with a higher incidence of late-onset ventilator-associated pneumonia, the incidence of potentially resistant Gram-negative bacilli increased in period 3: 42.2% vs. 34.5% vs. 41.7% (nonsignificant), except for B. cepacia: 11.7% vs. 7.4% vs. 3.7% (p =.005). Nevertheless, the potential antibiotic-resistant Gram-negative bacilli were more sensitive to most of the beta-lactams, especially piperacillin-tazobactam and cefepime. CONCLUSIONS: Rotation of antibiotics could help to avoid ventilator-associated pneumonia. It could greatly improve the susceptibilities of the potentially antibiotic-resistant Gram-negative bacilli responsible for late-onset ventilator-associated pneumonia. This program could be applied in routine with good results 5 yrs after its introduction. Further studies, especially multiple-center trials, are necessary to confirm this result and better define the rotation type and intervals. [ABSTRACT FROM AUTHOR]