Errors in measuring blood gases in the intensive care unit: effect of delay in estimation

Arterial blood gas measurement is subject to a number of potential sources of error. We investigated some of these in the intensive care unit (ICU). We audited samples for adequate volume and the presence of air and found that all samples were of adequate volume, but 40% contained bubbles or froth. We compared pulse oximeter estimations of oxygen saturation (SpO(2)) with laboratory estimates (SO(2)) from arterial blood samples, and found that there was less than a 5% chance of a difference of 5% or more. We audited the delay between sampling and processing and looked for errors arising as a result. We found that 4% of samples waited longer than 30 minutes to be analyzed in the laboratory, but that there was no correlation between delay and error in partial pressure of oxygen (PO(2)), carbon dioxide (PCO(2)), or SO(2). We performed a bench study to document the changes in PO(2) and PCO(2) over time with samples stored at room temperature and on ice. We found that samples in 1.5-mL PICO 70 syringes (Radiometer Medical A/S, Bronshoj, Denmark) were stable for PO(2) and SO(2) for up to 30 minutes either at room temperature or kept in iced water, and that changes after 60 minutes were small and unlikely to be clinically significant. PCO(2) showed a statistically significant increase after 20 minutes at room temperature, but the changes were not clinically significant.