We read with interest the article by Kollef and colleagues on circuit changes and ventilator-associated pneumonia [1]. This work was intended to confirm our finding that a ventilator circuit may be used throughout the mechanical ventilation of a patient without adverse effects [2]. Unfortunately, their study design precludes this conclusion. The fact that no difference was found between the incidence of pneumonia in patients allocated to a 7-day change and that in patients given no change does not imply that prolonged use of a single ventilator circuit is safe. A group receiving 48-hour circuit changes (current practice in most U.S. hospitals) should have been included so that such a conclusion could be established. A greater incidence of pneumonia associated with 7-day changes than with 48-hour changes may have been overlooked in Kollef and colleagues' study. To our knowledge, only three studies published in full-length form clearly show that prolonged circuit use is safe [2-4].

We question Kollef and colleagues' finding that patients with nosocomial pneumonia had the same mortality rate as patients without nosocomial pneumonia. This finding strongly contradicts the results of all other studies, which have reported that nosocomial pneumonia (whether diagnosed by invasive or other means) is responsible for marked increases in mortality. This discrepancy casts some doubt on the validity of the diagnosis of pneumonia in Kollef and colleagues' study.

The authors indicate that they did this study because of the limitations imposed by small sample size. We agree that our study included only a quarter of the number of patients in the study by Kollef and coworkers. This reasoning, however, raises two questions: 1) What sample size is adequate? and 2) does a larger sample obviate the need for adequate diagnostic technique? Kollef and colleagues based their calculation of power on an unspecified rate of pneumonia and calculated that the required sample was 260 patients. Compared with the number of patients (> 1700) in the recent study by Hess and colleagues [4], one could also say that the study by Kollef and colleagues had a low power. Given the high rate of both false-positive and false-negative diagnoses of nosocomial pneumonia by noninvasive techniques, the sample size used in Kollef and colleagues' study may not have been large enough to compensate for diagnostic inaccuracies. A considerably larger sample, such as that used in the study by Hess and coworkers [4] (which also relied on clinical criteria for diagnosing pneumonia), would probably compensate for the effects of diagnostic imprecision. In contrast, our study [2] used stringent criteria for diagnosing pneumonia, and this factor may have compensated for the relatively small sample.

Despite these criticisms, it is important to note that all carefully conducted studies on prolonged circuit changes [1-4] reach the same conclusion—that a strategy that reduces costs does not put patients at higher risk for infection.