We wish to report a misdiagnosis following attempted mixed venous oxygen saturation (Svo2) measurement. A middle-aged female was admitted to our ICU for tracheal intubation and mechanical lung ventilation after developing acute Type II respiratory failure. She had a history of two renal transplants for chronic renal failure, the most recent three months prior to admission. A left internal jugular central line was inserted. The left side was chosen due to the handedness of the operator. Ultrasound guidance was used to ascertain the anatomy and mark the position of the left internal jugular vein before starting the procedure. Under aseptic conditions, the vessel was punctured with a 20-G seeker needle at the first attempt. The blood aspirated was dark and non-pulsatile and at the time of puncture the patient's systolic blood pressure was 140 mmHg and a recent Pao2 was 30 kPa on 50% oxygen. The guidewire passed easily and an 8.5 French quintuple-lumen central line (Arrow International, Erding, Germany) was inserted. The distal lumen was transduced and produced a venous waveform with a central venous pressure of 13 mmHg. We then aspirated blood from the distal lumen to approximate the Svo2. The results were consistent with an arterial sample, with a Pvo2 of 13 kPa, Pvco2 of 5 kPa, and Svo2 of 98%. We then transduced the proximal lumen and still achieved what appeared to be a venous waveform. Samples aspirated from the proximal and a medial lumen gave similar gas tensions and Svo2. Despite the apparently normal central venous pressure, we felt there was some question about the position of the catheter, and it was removed with the patient in the head down position and pressure applied to the site.
It was only in the cold light of day that we realised the patient still had a functioning left forearm arteriovenous fistula and that this, coupled with her hyperoxia following recent intubation and administration of high inspired oxygen concentrations, led to the blood within her superior vena cava having ‘arterial’ values. An arterial sample taken at approximately the same time showed a Pao2 of 20 kPa, Paco2 of 5 kPa and Sao2 of 99%. We did not use ultrasound to re-image the vessels before removal of the catheter. Although this might have allowed us to identify which vessel had been punctured, it would not have excluded a more distal migration of the guidewire or catheter into the artery.
The uncomplicated insertion and normal central venous pressure pointed towards a correctly inserted catheter, but patient safety was paramount, and with some degree of uncertainty we felt the decision to remove the catheter was correct. The operator in this case elected to re-site the patient's central access via the femoral vein, given the assumed carotid artery puncture. With the overlooked significance of her fistula, had access been re-attempted using the right internal jugular or subclavian approach, the same problem would have been encountered after exposing the patient to additional risks of central cannulation.
Problems confirming correct placement of central lines in this group of patients have been reported before. Yee and Despotis reported a significantly higher arteriovenous difference than our case (Pao2 68 kPa, Pvo2 12.5 kPa) , which was the basis on which they excluded intra-arterial cannulation. Angaramo and colleagues encountered an abnormally high central venous pressure (30 mmHg), which triggered further investigation of the catheter's position .
Given that as a result of the Surviving Sepsis Campaign  we are sampling Svo2 in many more patients, and that following renal transplantation patients are more likely to develop sepsis, we felt it important to highlight this misdiagnosis and to emphasise the inaccuracy of Svo2 estimation in patients with an arteriovenous shunt of any cause.