Are coagulation studies on blood sampled from arterial lines valid?

Coagulation tests were performed on two venous blood samples and two blood samples from arterial lines taken from 79 patients on an intensive care unit. For the first arterial sample the discard volume was 4.5 ml and for the second arterial sample the discard volume was 16 ml. From each pair of venous samples a mean venous coagulation value was calculated. There were statistically significant differences between arterial and venous results and between the two arterial samples for activated partial thromboplastin time and thrombin time assays but not for prothrombin time, reptilase time and fibrinogen assays. However, these differences were sufficiently small to be of little clinical significance. The bias for the difference in activated partial thromboplastin time values between the first arterial sample and the venous sample was +1.24 s (limits of agreement: -4.39 to +6.87 s) and between the second arterial sample and the venous sample the bias was +0.89 s (limits of agreement: -3.25 to +5.03 s). Only 3.8% of first arterial samples and 1.3% of second arterial samples produced activated partial thromboplastin time values that were more than 10% longer than the corresponding venous values. No heparin was detectable in these arterial samples using a heparin assay (< 0.02 iu.ml(-1)). The differences between arterial and venous activated partial thromboplastin times were slightly but not significantly greater in subgroups of patients with moderately or severely deranged coagulation compared with a group with normal coagulation. We conclude that samples from arterial lines provide valid activated partial thromboplastin time results using a discard volume of either 4.5 ml or 16 ml.