Blood sampled from bone for point-of-care testing

Intravenous (iv) catheters are routinely employed for speedy delivery of fluids and drugs to the critically ill; they also provide the means for sampling blood for testing. This iv catheterization occurs in a range of settings including prehospital emergency care, hospital emergency rooms and intensive/critical care units. Insertion of iv catheters can be challenging in emergency situations and may be impossible in the case of the most critically ill with hemodynamic collapse. Cold environments (e.g. at the site of a road traffic accident) may also preclude venous access. 

The technique of intraosseous (IO) access provides an alternative route for rapid delivery of drugs and fluids when iv catheterization is not possible. The technique involves insertion of a needle using a power-driven device into the bone marrow cavity. The most commonly used site is the proximal tibia (just below the knee).

It is normal practice to aspirate a little blood/bone marrow from the site to confirm correct placement of the needle in the bone cavity, before administration of fluids/drugs via the inserted needle. Clearly, venous blood sampling is potentially difficult in such cases and the notion that this intraosseous (IO) blood sample could be used for testing has been the object of several studies including this, the most recently published. 

Thirty-one (31) healthy paramedic volunteers, who were being taught the technique of IO access, were recruited for this study. Two IO blood samples (IO1 and IO2) were taken from each volunteer. The initial sample (IO1) was collected and then after discard of 2 mL blood to waste, a second sample (IO2) was collected. Immediately after this, conventional venous and arterial blood samples were also sampled. 

All four samples from each of the volunteers were collected into dry-heparin blood gas syringes and immediately submitted for analysis using a POCT analyzer. Analytes measured were: sodium, potassium, ionized calcium, glucose, hemoglobin, blood gases (pH, pCO₂, pO₂, base excess and bicarbonate) and lactate. 

Investigators sought to compare results of analysis of IO1 and IO2 samples with those generated from conventional arterial blood samples. As a reference, arterial and venous results were also compared. Finally, comparison of IO1 and IO2 sample results were made to determine if discarding IO blood between sampling had any effect. 

As expected there was very close level of agreement between the two conventional samples (arterial and venous) for all parameters except pH, pCO₂ and pO₂. Differences for these three parameters were as predicted by physiology. 

Comparison of IO1 and IO2 sample results revealed good agreement for all analytes, except potassium and hemoglobin. The authors conclude that it is probably not necessary to discard IO blood, and the initial sample used to confirm position of the needle in the bone cavity is an acceptable sample for POC analysis.

The crucial comparison, IO sample results versus arterial sample results, varied greatly between analytes. Bland-Altman statistical analysis demonstrated good and acceptable level of agreement for only three analytes: pH, glucose and lactate. Significant bias and/or significant imprecision was evident for all other analytes. IO sodium results were clearly lower than arterial sodium (mean IO Na 132 versus mean arterial Na 138).

IO potassium results were clearly higher than arterial potassium (mean IO1 5.8 mmol/L, mean IO2 6.9 mmol/L versus mean arterial 3.7 mmol/L). Blood gases (pCO₂ and pO₂) results of IO samples were between arterial and venous values, which, as the authors argue, is physiologically explicable. Particularly poor level of agreement between IO and arterial blood samples was evident for hemoglobin. 

In discussion of their study results the authors caution that although IO sample analysis is feasible using point-of-care analysis and may be useful to guide treatment in critical emergency care, such analysis should be limited to situations in which no other sample options are available.

Results should be interpreted with caution taking account of analyte bias and imprecision; currently there is not enough scientific evidence regarding agreement between IO and arterial results in the unstable patient. This study, in common with most others, has focused on relationship between IO and arterial blood chemistry in healthy individuals, not critically ill patients.