Analysis of intraosseous samples in endotoxemic shock--an experimental study in the anaesthetised pig

Intraosseous (IO) access is used in emergency situations to allow rapid initiation of treatment. IO access is also sometimes used for blood sampling, although data on accuracy of such sampling in critical illness are limited. There is also a potential risk that bone marrow fragments in IO samples may damage laboratory equipment. It is ethically questionable to perform a simultaneous comparison between IO and arterial/venous sampling in critically ill humans. We have, thus, studied the analytical performance of IO sampling in a porcine septic shock model using a cartridge-based analyser.

Eight pigs with endotoxin-induced septic shock were sampled hourly for 6 h, and analysed for blood gases, acid base status, haemoglobin, glucose and lactate using point of care instruments. Samples were taken from three IO cannulae (tibia bilaterally, one with infusion, and humerus), one arterial and one venous. An interaction test was used to assess changes in agreement between methods over time. Bland–Altman plots were constructed to study bias between methods.

There were, to a varying extent, differences between IO and arterial/venous levels for all studied variables, but agreement did not change significantly during the experiment. A general finding was a large dispersion of differences between methods.

IO sample values should be treated with caution in this setting but may add useful information to the clinical picture. The tibia or humerus may be used for sampling. IO infusion decreases agreement, thus sampling during infusion should be avoided.