“Respibands” were taped securely to the chest (nipple level) and abdomen (umbilical level and below the costal margins). Immediately before “lights out” and with the patient in his preferred sleeping posture and prepared to sleep, volume-motion (VM) coefficients were obtained using computerized multiple linear regression (MLR) analysis as described by Stradling et al.s Briefly, RIP outputs from rib cage (RC) and abdomen (AB) together with the integrated spirometer signal (SP) from a pneumotachograph (Gould Godart BV) previously stabilized to eliminate drift and calibrated against a liter syringe were sampled within 20 ms of each other 150 times during a 20-second fixed period of quiet tidal breathing and were analyzed by MLR using a BBC “B” microcomputer programmed in Basic. The 150 raw data values for RC, AB, and SP were used in each calibration to calculate the VM coefficients (a and b), using the equation described by Armitage, te, (ax RC) + (bx AB) + e = SP (“e” representing errors arising from (1) the different voltages of the three variables, and (2) the offset attributable to the intercept of the plot of RIP volume: spirometer volume). more
The calibration procedure was repeated at least twice to ensure consistent results. Where poor calibrations occurred (95 percent confidence intervals exceeding 20 percent), these were usually due to insufficient chest wall or abdominal movement and could be improved by repositioning the abdominal belt and/or having patients deliberately hyperventilate during the calibration period. Application of the VM coefficients (each the mean of three acceptable calibrations) allowed the derivation and minute-by-minute update of ventilation frequency (f), tidal volume (Vt), minute ventilation (Ve), mean inspiratory flow (Vt/Ti), fractional inspiratory time (TV Ttot), and abdominal contribution (AC) to ventilation. Calibration was repeated during the study following any change in the patients’ sleeping posture. Ventilatory data during wakefulness were obtained following a period of sleep. Where possible, data from each sleep stage and from wakefulness were obtained from each patient without a change in sleeping position; if the posture were changed (te, the patient woke and requested to be turned), the instrument was recalibrated before data collection was resumed).