The most award winning
healthcare information source.
TRUSTED FOR FOUR DECADES.
Synopsis: Noninvasive ventilation has been shown to be effective treatment for hypercarbic respiratory failure due to exacerbations of COPD. Compared with intubation, this therapy reduces mortality and costs less to provide. Although there are methodological concerns with this paper, it also appears that noninvasive ventilation is associated with a lower risk of nosocomial pneumonia.
Source: Girou E, et al. JAMA 2000;284:2361-2367.
In order to investigate the association of nosocomial pneumonia (NP) with noninvasive ventilation (NIV), Girou and colleagues retrospectively compared 50 patients who had received NIV for treatment of hypercarbia for at least two hours with 50 individually matched patients managed with intubation. Cases were matched one to one for diagnosis, age (± 5 years), simplified acute physiology score II (SAPS II) (± 6 points), organ failure score, and absence of contraindications to NIV. All infections were identified; NP was suspected if the patient had a temperature greater than 38 degrees, lung infiltrates on chest X-ray, and sputum macroscopically indicative of infection, all occurring at least 48 hours following ICU admission. PN was confirmed using quantitative cultures from protected samples in intubated patients and in some patients on NIV. However, most PN in NIV patients were confirmed by clinical measures and response to treatment.
During the study period, of the134 patients receiving NIV, 57 met study criteria. Of these only 50 could be successfully matched. The only variable studied that was different between the groups was the percentage of patients on antibiotics prior to ICU admission, which was 78% in the control patients and 40% in the NIV cases. While the difference did not reach statistical significance, cases and controls were also different in location prior to the ICU admission (70% in community cases vs 53% in controls; P = 0.09), and the percentage of patients who were infected prior to intubation or NIV (30% in cases, 46% in controls; P = 0.10). An equal percentage of patients were treated primarily for cardiac failure in each group (18%).
The total nosocomial infection rate was less in the NIV group (14% vs 38%), NP was also decreased (8% vs 22%). Mortality was greater in the intubated group (26% vs 4%), and both the duration of ventilation and ICU length of stay were also longer. When corrected for mechanical ventilation days, the PN rate remained significantly different.
Previous studies have identified the association of PN with endotracheal intubation and suggested that NIV might reduce the PN rate. This study attempts to further evaluate this presumed association. Unfortunately, the case matching in this study may have failed to remove important differences between the studied groups. Specifically, the groups were different in location prior to receiving ventilation, the NIV patients being primarily a community group while the matched, intubated cases more frequently came from another area of the hospital. Many more patients in the intubated group had already been diagnosed with infection and were receiving antibiotics. These variables may be the most important contributors to determining PN rates.
Another serious concern with this study is the fact that patients treated with NIV received this therapy only 4-6 hours per day while the control patients were continuously intubated. If the incidence of infection was calculated per hour of therapy, rather than per 1000 days of ventilation, the NP risks would have been identical. Another concern is that controls were selected from different years rather than cases. Significant changes over time may have affected the results.
The mortality differences suggest that the groups were not comparable in severity of illness at entry, despite similar SAPS II and organ failure scores. This is an important concern with interpretation of the outcome. Girou et al are to be congratulated on providing adequate information to make the judgment that the conclusions of this study must be interpreted with caution.