presents data at the end of oxygen administration (20 min). There were no differences between groups in heart and respiratory rates. Heart rate showed a trend toward a decrease, and respiratory rate was stable in both groups; however, patients treated with 100% oxygen presented a statistically significant decrease in pH (mean difference, 0.03; 95% CI, 0.00 to 0.05) and a significant increase in Paco2 (mean difference, 2.7 mm Hg; 95% CI, 4.7 to 0.7 mm Hg) associated with a significant decrease in PEFR (mean difference, 31.7 L/min; 95% CI, 16.7 to 46.8 L/min) compared with patients treated with 28% oxygen. As anticipated, patients who received 100% oxygen showed a substantial increase in Pao2 compared with patients treated with 28% oxygen (mean difference, 158.4 mm Hg; 95% CI, 185.1 to 131.8 mm Hg). Figure 1
shows the relationship between Paco2 before and at 20 min of oxygen administration. The scatterplot revealed that this relationship is curvilinear, particularly in patients treated with 100% oxygen. We tested different curve estimation regression models, and a polynomial model was the best fit to the data. Both variables correlated significantly (r = 0.53, p = 0.01 in the 28% group, and r = 0.88, p = 0.001 in the 100% group). A close examination of the relationship showed two different responses to oxygen administration. When patients received 28% oxygen, Paco2 had a tendency to fall; on the contrary, patients who received 100% oxygen showed an increase in Paco2, particularly in those patients with Paco2 before oxygen treatment > 40 mm Hg. Post hoc analysis of Paco2 changes from baseline during oxygen administration are shown in Table 3
. Twenty patients (55.6%) in the 28% group and 12 patients (31.6%) in the 100% group had Paco2 decreases (p = 0.05). Eight patients (22.2%) in the 28% group and eight patients (21.0%) in the 100% group (p = 0.9) presented small Paco2 increases (< 2 mm Hg) consistent with the physiologic Haldane effect.21
Finally, only 6 patients (16.6%) in the 28% group presented a Paco2 increase > 2 mm Hg (mean increase, 3.4 ± 1.9 mm Hg; range, 2 to 5.9 mm Hg), compared with 16 patients (42.1%) [mean increase, 5.0 ± 3.8 mm Hg; range, 2.4 to 14.3 mm Hg] in the 100% group (p = 0.02). At the end of the protocol, this subgroup of patients showed a significant difference in PEFR (PEFR deterioration of 11.9 ± 30.0 L/min in the 100% group, compared with an increase of 18.3 ± 25.4 L/min in the 28% group, p = 0.03). Whereas Paco2 did not correlate with PEFR in the 28% group (r = 0.21, p = 0.6), the 100% group showed an inverse curvilinear correlation between PEFR before oxygen administration and Paco2 during oxygen treatment (Fig 2
). Low values of PEFR were associated with high levels of Paco2 (r = 0.53, p = 0.01, polynomial model).