A newly designed gas-sampling device using end-tidal CO2 to separate dead s
pace gas from alveolar gas was evaluated in 12 mechanically ventilated pati
ents. For that purpose, CO2-controlled sampling was compared with mixed exp
iratory sampling. Alveolar sampling valves were easily controlled via CO2 c
oncentration. Concentrations of four volatile substances were determined in
the expired and inspired gas. Isoflurane and isoprene, which did not occur
in the inspired air, had ratios of end-tidal to mixed expired concentratio
ns of 1.75 and 1.81, respectively. Acetone and pentane, found in both the i
nspired and expired air, showed ratios of 0.96 and 1.0, respectively. Preci
sion of concentration measurements was between 2.4% (isoprene) and 11.2% (i
soflurane); reproducibility las coefficient of variation) was 5%. Because t
he only possible source of isoflurane and isoprene in this setting was pati
ents' blood, selective enrichment of alveolar gas was demonstrated. By usin
g the new sampling technique, sensitivity of breath analysis was nearly dou
bled.