Measurement of respiratory mechanics in a mechanically ventilated infant lung simulator: effects of variations in the frequency response of the flow measurement system

Abstract

The frequency responses of systems used to measure flow and pressure in ventilated infants may differ, and hence affect estimates of resistance and compliance. We estimated resistance and compliance in 16 ventilated mechanical lung models using linear regression while varying the frequency response of the flow measurement system. Lung models comprised combinations of four sections of tubing and four bottles filled with steel wool. The cut-off frequencies of a filter in the flow measurement system were chosen to yield time delays of 0, +/-3, +/-6, and +/-9 ms relative to the pressure signal. When the phase lags in the measurement systems were not equal at 10 Hz, a bias in resistance approximately (relative delay) x (elastance) ensued. The bias in the resistance estimate when resistance is 5 Pa ml-1 s and compliance is 2 ml kPa-1 is approximately 28% per ms of delay mismatch. Time-shifting the flow data to eliminate the phase discrepancy reduced the resistance bias by 85%. The residual resistance bias was assumed to be due to inappropriate amplitude response. Compliance measurements were affected by less than 8% and less than 2% after time correction of the flow data. Pressure and flow signals must be synchronized to within 1 ms at 10 Hz and the amplitude responses of the measurement systems must be adequate for reliable resistance measurement.