Maximal Inspiratory Pressure: Data Analysis
The Pimax was defined as the largest negative pressure generated at the mouth and maintained for at least 1 full second. A minimum of five technically satisfactory trials were conducted for each test and data w^ere discarded if there was an air leak around the mouthpiece or if the pressure was held for less than 1 full second as estimated by the data collector. The initial length of the inspiratory muscles wras controlled by initiating each effort from residual volume (RV). This procedure was adopted because in the clinical situation RV is more reproducible than functional residual capacity (FRC). Patients were instructed to take their time and slow ly empty their lungs to RV, thereby avoiding problems associated with variability in lung volumes caused by dynamic hyperinflation. Patients were not allowed to view the gauge during testing.
Data were collected by two researchers who were experienced in conducting these tests. Tests were conducted in a quiet room with no distractions and the same instructions were given to patients at the beginning of each testing session. Patients were coached in a consistent manner during each maximal inspirator)’ effort and brief rest periods of slightly less than 1 min were taken between repeated Pimax trials. Patients were instructed to take their time and indicate when they were ready to perform each maneuver. Test sessions were conducted at the same time of day and by the same researcher for each patient. Patients were instructed to take their usual medications as scheduled on each day of testing to control for any potential drug effects on respiratory muscle function.
Pulmonary function tests were conducted prior to the first measurement of Pimax using a spirometer that met American Thoracic Society (ATS) standards. Data were reported from the best of three forced vital capacity (FVC) maneuvers. Body mass index was calculated as the ratio of weight (kilogram) to height squared (meter squared).
In a preliminary study, interrater reliability was established for the two data collectors and reliability of the method was established by comparing measurements taken with the aneroid gauge to measurements recorded with a transducer and strip chart recorder. For this purpose, Pimax was measured in a separate group of 15 subjects. Measurements of Pimax were taken simultaneously by both data collectors from the aneroid gauge and at the same time pressures were recorded on a strip chart recorder from a pressure transducer (Viggo Spectromed, Oxnard, Calif). Throughout all reliability tests, both data collectors were kept blind to the results. Data were compared for differences between the two data collectors and for differences between the two methods, aneroid pressure gauge, and transducer with strip chart recorder. Mean net difference was calculated to reflect systematic measurement error such as one data collector or one technique producing higher or lower values than the other. The mean of absolute differences (| Pimax 1 — Pimax2|) was calculated to reflect magnitude of fluctuation in Pimax regardless of the direction. Absolute differences reflect the magnitude of combined systematic and nonsystematic fluctuations in the data. With mean net difference nonsystematic positive and negative differences tend to cancel each other out. When the results from the two data collectors were compared, the mean net difference in Pimax was zero (SD = 4) and mean absolute difference was 3 cm H20 (SD = 2). The interrater reliability coefficient was r = 0.99 (df = 13). When data from the aneroid gauge were compared with data recorded with the transducer, the mean net difference in Pimax was 0 cm H20 (SD = 7) and 0 cm HzO (SD = 8) for each of the data collectors, respectively. The mean absolute difference in Pimax for the two methods was 5 cm H20 (SD = 5) and 6 cm H20 (SD = 5) cm H20 for each of the data collectors. Reliability coefficients for the two methods were r = 0.97 (df= 13) and r = 0.96 (df = 13) for each of the data collectors.
Descriptive statistics were calculated for differences and absolute differences. Analysis of variance for repeated measures was calculated to identify significant differences between test sessions and to determine if there was a linear effect or curvilinear effect over time. Net differences and absolute differences were calculated to describe systematic and nonsystematic fluctuations in repeated measures of Pimax. Test-retest reliability’ was calculated with Pearsons correlations.