Is venous blood an acceptable sample for assessment of acid-base status in patients with sepsis/septic shock?

The gold standard sample for blood gas analysis is arterial blood, whereas venous blood is the standard sample for all other biochemical and hematological tests. By comparison with venipuncture, the procedure used to sample venous blood, collection of an arterial blood sample is technically more difficult, as well as more hazardous and painful for the patient.

These considerations, as well as the obvious logistic advantage of using a single venous sample for all biochemical analysis has prompted investigation of the validity of using venous blood, sampled either from a peripheral vein or via a central venous line, for blood gas analysis if only the patient’s acid-base status (pH, pCO₂, bicarbonate and base excess) is of clinical interest.

(There is universal agreement that assessment of blood oxygen status, the other clinical utility of blood gas analysis, depends absolutely on measurement of arterial blood pO₂/sO₂; this consensus reflects the physiologically predictable, very poor correlation and level of agreement between arterial and venous blood for these two oxygenation parameters.)

In excess of 20 studies over the past 15 years have examined correlation and level of agreement between blood gas parameters determined using venous blood with those generated from simultaneously collected arterial blood in a range of conditions that warrant blood gas analysis: diabetic ketoacidosis, COPD, acute respiratory failure, and so on. This most recently published investigation conducted at a US medical intensive care unit focuses for the first time specifically on patients with severe sepsis or septic shock.

Samples of arterial blood (AB), peripheral venous blood (pVB) and central venous blood (cVB) were collected simultaneously (all within 15 minutes of each other) from 67 study patents with severe sepsis/septic shock. All three samples from each study patient were transported on ice to the central laboratory and submitted for blood gas analysis. Results of pH, pCO₂, pO₂, sO₂, bicarbonate and base excess for each sample were recorded for comparative statistical analysis (intraclass correlation coefficient, ICC; and Bland-Altman level of agreement, LOA)

ICC revealed strong correlation between arterial blood and both venous samples (pVB and cVB) for pH, pCO₂, bicarbonate and base excess (ICC all >0.85). As anticipated, there was poor correlation between arterial blood and both venous samples for both pO₂ and sO₂ (ICC <0.3). 

Bland-Altman analysis revealed acceptable LOA between arterial and both venous samples for only one parameter, pH. LOA between arterial and both venous samples was judged inadequate for all other parameters, pCO₂, bicarbonate, base excess, pO₂ and sO₂.

In line with the results of many (though not all) other similarly designed studies, the results of this one examining exclusively sepsis patients suggests to the authors of this study that venous blood should only be used for blood gas analysis if pH is the only parameter of clinical interest. 

Since pH alone is insufficient to fully assess acid-base status, the authors conclude that venous blood should not be used to assess acid-base status in patients with severe sepsis and septic shock. 

The discussion section of this study report includes a useful tabulated summary of the results of previous similar studies. They also report increased utilization of venous blood gases at their institution in recent years. Apparently, blood gas analysis ordered in the emergency department of this hospital is almost exclusively performed on venous rather than arterial blood, whereas in this hospital’s intensive care units approximately 12 % of all blood gas analyses are performed on venous blood.

The authors question the apparent growing acceptance that venous blood gas analysis is a clinically useful tool, since as evidence from most, if not all, studies suggests, it only provides reliable information about the patient’s blood pH.