Background: The extent of error, from collection to processing, when measuring PO2, PCO2 and pH in arterial blood samples drawn from critically ill patients with sepsis and leucocytosis, is unknown. Methods: Twenty-nine patients with sepsis and a leucocyte count > 12 000/mm3, who had routine arterial blood analysis were included in the study. Blood was drawn into two 1 ml heparinised glass syringes. One syringe was cooled on ice and tested at 60 minutes. The other syringe was used for analysis at 0, 10, 30 and 60 minutes. Differences in measurements, from the Time-0 results, were described. For PO2, linear mixed models estimated the impact of time to processing, controlling for the potentially confounding and moderating effects of Time-0 leucocyte count and fractional inspired oxygen concentration respectively. Results: PO2 exhibited the most pronounced changes over time at ambient temperature: The mean (SD) relative differences at 10, 30 and 60 minutes were -4.72 (8.82), -13.66 (10.25), and -25.12 (15.55)% respectively; and mean (SD) absolute differences -0.88 (1.49), -2.37 (1.89) and -4.32 (3.06) kPa. For pH, at 60 minutes, the mean (SD) relative and absolute differences were -0.27 (0.45)% and -0.02 (0.03) respectively; for PCO2, 6.16 (7.80)% and 0.25 (0.35) kPa. The median differences for the on-ice 60-minute sample for pH and PCO2 were 0.019 and -0.12 (both P < 0.001), and for PO2 0.100 (P: 0.216). The model estimated that average PO2 decreased by 5% per 10 minute delay in processing (95% CI for effect: 0.94 to 0.96; P < 0.001) at the average leucocyte count, with more rapid declines at higher counts, though with substantial inter-patient variation. Conclusion: Delayed blood gas analysis in samples stored at ambient temperature results in a statistically and clinically significant progressive decrease in arterial PO2, which may alter clinical decision-making in septic patients.

Background: The extent of error, from collection to processing, when measuring PO2, PCO2 and pH in arterial blood samples drawn from critically ill patients with sepsis and leucocytosis, is unknown. Methods: Twenty-nine patients with sepsis and a leucocyte count > 12 000/mm3, who had routine arterial blood analysis were included in the study. Blood was drawn into two 1 ml heparinised glass syringes. One syringe was cooled on ice and tested at 60 minutes. The other syringe was used for analysis at 0, 10, 30 and 60 minutes. Differences in measurements, from the Time-0 results, were described. For PO2, linear mixed models estimated the impact of time to processing, controlling for the potentially confounding and moderating effects of Time-0 leucocyte count and fractional inspired oxygen concentration respectively. Results: PO2 exhibited the most pronounced changes over time at ambient temperature: The mean (SD) relative differences at 10, 30 and 60 minutes were -4.72 (8.82), -13.66 (10.25), and -25.12 (15.55)% respectively; and mean (SD) absolute differences -0.88 (1.49), -2.37 (1.89) and -4.32 (3.06) kPa. For pH, at 60 minutes, the mean (SD) relative and absolute differences were -0.27 (0.45)% and -0.02 (0.03) respectively; for PCO2, 6.16 (7.80)% and 0.25 (0.35) kPa. The median differences for the on-ice 60-minute sample for pH and PCO2 were 0.019 and -0.12 (both P < 0.001), and for PO2 0.100 (P: 0.216). The model estimated that average PO2 decreased by 5% per 10 minute delay in processing (95% CI for effect: 0.94 to 0.96; P < 0.001) at the average leucocyte count, with more rapid declines at higher counts, though with substantial inter-patient variation. Conclusion: Delayed blood gas analysis in samples stored at ambient temperature results in a statistically and clinically significant progressive decrease in arterial PO2, which may alter clinical decision-making in septic patients.