Analysis of error in the determination of respiratory gas exchange at varying FIO2.

On-line gas exchange monitors have been based on the measurement of both inspired and expired flows (method A) or on the measurement of expired flow only (method B). In both methods, the cancellation of systematic errors is appreciable if calibration errors in the inspiration variables are matched with calibration errors in the corresponding expiration variables. The determination of oxygen consumption (VO2) results in a considerable amplification of random measurement error, which becomes larger as the inspired oxygen fraction (FIO2) is made larger. For example, when in method A all gas compositions and flows are in error by +/- 0.01, the computed value of VO2 will be in error by +/- 0.16 when FIO2 = 0.21 and by +/- 0.40 if FIO2 = 0.50. When air is breathed, method B creates smaller errors than method A, but as FIO2 is increased the error in method B increases sharply and eventually overtakes that of method A. Even at high FIO2 values, the random error in VO2 may be reduced to an acceptable value by estimating it from a sequence of 100 measurement samples requiring a minimum of 5 min to acquire. There is no error amplification in the computation of carbon dioxide production.