To the Editor:--Hess et al. [1]suggest that ventilator circuit changes could be delayed to 7 days without increasing the incidence of ventilator-associated pneumonia. Indeed, inclusion of a very large number of patients adds tremendous credibility to their paper, which confirms our results. [2]This paper raises a few issues about which we wish to comment. The first concerns the duration of mechanical ventilation and its interaction with nosocomial pneumonia. According to the data reported by Hess et al., the mean duration of ventilation was less than 6 days, during both the 48-h and 7-day circuit changes. These figures are different from those reported in most studies of ventilator-associated pneumonia (VAP), which were conducted with patients whose lungs were ventilated for much longer periods [3-6](10-25 days). Most studies also report that VAP occurs after a mean duration of mechanical ventilation of about 10 days. [2-4]Thus, the data presented by Hess et al. leave some doubt as to whether the VAP cases were essentially early-onset pneumonia, [7]which occurs during the first 4-5 days of ventilation and usually is due to such common pathogens as Hemophilus influenzae or Streptococcus pneumoniae, which do not cause major therapeutic problems because they are usually sensitive to most antibiotics. In contrast, late-onset pneumonia poses very different problems because it may be due to resistant pathogens such as Pseudomonas aeruginosa or Acinetobacter spp. and is responsible for a substantial increase in mortality. [4]It would be interesting to know what pathogens were recovered in the study by Hess et al. and whether patients with pneumonia had been receiving mechanical ventilation for longer durations than patients without pneumonia and had different mortality rates. Second, despite the fact that we studied only 63 patients, we do not agree with Hess et al. that the power of our study was low. Power calculation is relevant to study design but not to the interpretation of data. [8]Confidence intervals are appropriate when interpreting data. [8]In our study, the confidence intervals for the incidence of pneumonia in patients with 48-h circuit changes and those with no change at all were strictly superimposable, and the P value for the pneumonia rates was 0.8. Because we used specific diagnostic criteria, [9]it would be unlikely that increasing the number of patients would yield different results. Furthermore, in another study on a larger population, [10]we confirmed that not changing circuits during ventilation with heated humidifiers does not put patients at greater risk of pneumonia. Our initial study [2]also involved extensive microbiologic surveillance with more than 110 quantitative cultures conducted on four patient and circuit sites in each group (for a total amount of more than 1,000 cultures). There was no difference in patient and circuit colonization, whether or not circuits were changed during the whole duration of mechanical ventilation. Given the pathogenesis of VAP, [11]these findings provide strong additional support for not changing circuits at all during mechanical ventilation. The recommendation to prolong the interval of circuit changes from 24 to 48 h was based on the publication by Craven et al., [12]in which a similar number of cultures was performed. Notwithstanding, we believe that data from a single center always must be questioned by others. We are delighted that the study by Hess et al. adds credibility to the possibility of reducing costs without detriment to the patient.
Didier Dreyfuss, M.D., Service de Reanimation Medicale, Hopital Louis Mourier, 92700 Colombes, INSERM U 82, Faculte Xavier Bichat, Universite Paris VII, France.
Kamel Djedaini, M.D., Service de Reanimation Medicale, Hopital Louis Mourier, 92700 Colombes, France.
(Accepted for publication July 14, 1995.)