High-frequency Percussive Ventilation Improves Oxygenation in Patients With ARDS: Data Collection and Statistical Analysis
Data Collection and Statistical Analysis
Values for the following parameters were collected from a computerized ICU database: Pao2/Flo2, partial Paco2, pH, PIP, MAP, Sao2, mean arterial BP (MBP), oxygen delivery index (Do2I), oxygen consumption index (Vo2I), oxygen extraction ratio (O2ER), and intrapulmonary shunt fraction (Qs/Qt). These data were inserted prospectively into the ICU-computer database and collected retrospectively for this study. Data collection for each patient covered a period of 48 h before and 48 h after HFPV application. All values were abstracted from the ICU database and entered into another database (Excel; Microsoft; Redmond, WA) for analysis.
Each monitored parameter was studied in three ways. The first way was to evaluate the difference between the 48-h CV value and the 48-h HFPV value for each parameter. Patients with three or more values taken at each time period were included in this analysis. The average values during the 48 h before and the 48 h after starting HFPV for each patient were derived and compared using the paired Student’s t test. antibiotics buy
Significance was considered to be the p < 0.05 level. Differences between the average values of each parameter for MICU and SICU patients were examined to determine whether the patterns of ventilatory and hemodynamic changes were similar in the two groups. For the same reason, we evaluated differences in averages between patients who received volume-control ventilation and patients who were given pressure-control ventilation during the 48 h before receiving HFPV. To determine if the differences before and after initiating HFPV were consistent between the two types of ICU patients (SICU or MICU) and between the type of CV (volume- or pressure-control), their interactions were analyzed using the repeated measures analysis of variance model, in which time was the repeated-measure variable and type of ICU patients or type of CV was the main-effect variable. For this purpose, the general linear-model procedure of the computer software (PROC GLM, Statistical Analysis System; SAS Institute; Cary, NC) was used.