Fig. 6
Subgroup analyses on PaCO2 level and PaO2 level at 30 min. Transcutaneous data obtained via the chest or forearm sensors were used (n = 272). Bars: SEM, *: P < 0.05, **: P < 0.01, ANOVA with Tukey’s pot-hoc test. a Comparison of biases among severe hypocapnia (PaCO2 < 31 mmHg, n = 7), mild hypocapnia (31 mmHg ≤ PaCO2 < 35 mmHg, n = 24), and normal range (35 mmHg ≤ PaCO2 ≤ 45 mmHg, n = 202). The absolute values of bias observed in the severe hypocapnia group were significantly larger than those observed in the normal range group for both PCO2 and PO2. This effect of hypocapnia seemed to be intensity-dependent. b Comparison of biases among normal range (35 mmHg ≤ PaCO2 ≤ 45 mmHg, n = 202), mild hypercapnia (45 mmHg < PaCO2 ≤ 50 mmHg, n = 26), and severe hypercapnia (50 mmHg < PaCO2, n = 13). PCO2 biases in mild hypercapnia were significantly smaller than in normal range group. c Comparison of biases among hypoxemia (PaO2 < 80 mmHg, n = 158), normal range (80 mmHg ≤ PaO2 ≤ 100 mmHg, n = 102), and hyperoxemia (100 mmHg < PaO2, n = 12). The absolute values of bias observed in the hypoxemia group were significantly smaller than those observed in the normal range group for both PCO2 and PO2. The absolute values of PO2 bias observed in the hyperoxemia group were significantly larger than those observed in the normal range group. These effects on PO2 biases seemed to be PaO2 level-dependent