We agree with Drs. Asbel and Johnson that the use of uniform definitions for infection rates allows more accurate comparisons among medical reports. The occurrence of ventilator-associated pneumonia in our group receiving circuit changes every 7 days was 17.4 per 1000 ventilator-days. The incidence of ventilator-associated pneumonia in the group receiving no circuit changes was 16.4 infections per 1000 ventilator-days [1].

We selected a 7-day circuit change schedule as the "standard of practice" for our control group on the basis of the available medical literature and our survey of 16 regional medical centers [1]. Our current understanding of the pathophysiology of ventilator-associated pneumonia, along with data provided by Dreyfuss and colleagues [2], makes it unlikely that a greater incidence of pneumonia would have been observed in patients receiving 7-day circuit changes compared with circuit changes done every 48 hours.

The criteria used to establish the diagnosis of nosocomial pneumonia can influence the reported mortality rate associated with this condition. Realizing the importance of this issue, we used a randomized study design so that the same diagnostic criteria would be applied to both arms of our study. In addition, the final determination for the presence or absence of ventilator-associated pneumonia was made by an investigator who was blinded to the patients' treatment group assignments. The success of our randomized procedure is shown by the similar distribution of pathogens for which patients have a high risk for acquiring (for example, Pseudomonas aeruginosa and Xanthomonas maltophilia), associated with their greater mortality rates, between the two treatment arms [3].

The main difference in diagnostic criteria between our study and Dreyfuss and colleagues' study was their inclusion of a diagnostic threshold for ventilator-associated pneumonia using quantitative cultures of lower-airway secretions obtained using a protected specimen brush. Other investigators have found that the accuracy of this technique is limited, primarily because of the influence of antibiotic treatment on the culture results [4]. Therefore, it is incorrect to assume that the use of these invasive diagnostic methods can compensate for relatively small sample sizes in clinical studies. It is also important to note that controversy surrounds the need for invasive lower-airway sampling to establish the diagnosis of ventilator-associated pneumonia. Currently, clinical criteria alone for the diagnosis of ventilator-associated pneumonia can be used in both "high-quality" patient care and clinical investigations [5].

In summary, our study and others have reached similar conclusions on the need for routine ventilator circuit changes. Taken together, these investigations should allow institutions to eliminate the practice of routinely changing ventilator circuits. The rigorous methods used in these investigations should be directed toward examining other standard, but often unproven, respiratory therapy practices.