Relative Agreement between Raman and Mass Spectrometry for Measuring End-Tidal Carbon Dioxide

Abstract

Background

Several technologies can measure carbon dioxide in the exhaled breath and determine the end-tidal partial pressure of carbon dioxide (P_etCO2). Commonly used methods employ an infra-red capnometer, mass spectrometer (MS), or Raman spectrometer (RS). We sought to determine the relative agreement of representative MS and RS devices as a function of respiratory frequency.

Materials & Methods

A bench-top lung system was designed to simulate mechanical respiration. An airway connector with parallel gas sampling ports allowed simultaneous gas sampling. We attached sampling catheters from calibrated MS and RS and simultaneously determined P_etCO2 by MS and RS, collecting data in triplicate at each of 5 frequencies. We analyzed data by paired t test, linear regression, and ANOVA, with p < 0.05 considered significant. Bias and limits of agreement were calculated at each frequency from a total of 60 P_etCO2 pairs.

Results

The RS measurement was consistently higher than MS. The bias (SD) and limits of agreement were 4.98 (2.7) and -0.4 to 10.4 torr for the full range of P_etCO2. At a frequency of 10, the bias (SD), 2.0 (2.9), was significantly lower (p = 0.0007) than at any other frequency.

Conclusions

A statistically significant bias exists between MS and RS measurements. At frequencies greater than 10, this bias is constant. These differences at different frequencies may be due to gas sampling rates, differences in algorithm for determining P_etCO2, or differences in the basic measurement technology. The observed bias should not affect the use or clinical applicability of either of these devices in the clinical setting, although it may be desirable to use the same device for all measurements in a given patient.