Capillary and arterial blood gases in hemorrhagic shock: a comparative study.

Objectives: To determine whether capillary blood gas samples accurately reflect the information obtained from arterial blood gas samples in a porcine hemorrhagic shock model.

Design: Randomized prospective study.

Setting: Animal research laboratory at a children's hospital.

Subjects: Fifteen adult Yorkshire pigs.

Interventions: Tracheally intubated, ventilated, and anesthetized pigs were randomly assigned to either a control group (n = 10) or a hemorrhagic shock group (n = 5). After instrumentation, the hemorrhagic shock group underwent a 40-mL/kg hemorrhage over 2 hrs, while the control group did not. Animals were then monitored over a 3-hr period.

Measurements And Main Results: Simultaneous arterial blood gas and capillary blood gas samples were obtained from each animal by using standardized technique. Capillary blood gas pH accurately predicted arterial blood gas pH in both control and hemorrhagic shock groups (r = .94, p < .01, and r = .78, p < .01, respectively). Bland-Altman analysis revealed a mean difference between arterial blood gas pH and capillary blood gas pH of 0.016 (95% limit of agreement, +/-0.037) in the control group vs. a mean difference between arterial blood gas pH and capillary blood gas pH of 0.075 (95% limit of agreement, +/-0.142) in the hemorrhagic shock group. Capillary blood gas Pco2 accurately predicted arterial blood gas Pco2 in the control group but not in the hemorrhagic shock group (r = .80, p < .01, and r = .13, p = .41, respectively). By Bland-Altman analysis, the mean difference between arterial blood gas Pco2 and capillary blood gas Pco2 was 3.9, with a 95% limit of agreement of +/-3.61 in the control group. In the hemorrhagic shock group, the mean difference between arterial blood gas Pco2 and capillary blood gas Pco2 was 11.7 with a 95% limit of agreement of +/-38.8. Hemorrhagic shock group capillary blood gas samples significantly overestimated Pco2 (Delta = 11.7, p < .01) and underestimated pH (Delta = 0.075, p < .01) compared with arterial blood gas samples.

Conclusions: Capillary blood gas samples are poor predictors of Pco2 and pH in porcine hemorrhagic shock. These findings should be taken into account when capillary blood gas samples are used in the monitoring and management of subjects diagnosed with shock.