LOWER limb pneumatic compression devices are recommended to prevent venous thromboembolism in many intensive care unit patients. 1These pneumatic stockings increase lower limb venous return, causing acute but transient decrease in pulmonary artery blood temperature. Although they increased the variability between individual measurements, the use of these stockings was reported not to affect the accuracy of thermodilution cardiac output (CO) measurements made using 10 ml room temperature injectate. 2,3
We present a case in which the use of a pneumatic sequential compression device affected CO measurements, thereby delaying decision making, misguiding therapy, and increasing resource use.
Case Report
A 70-yr-old man was transferred emergently to the surgical intensive care unit from an outside hospital for management of a perforated esophagus secondary to dilatation for a peptic stricture. He had a significant medical history for hypertension and for coronary artery disease with coronary artery bypass grafting 21 and 11 yr previously. All saphenous vein grafts and native vessels were occluded. His internal mammary graft was his only patent cardiac vessel. His left ventricular function was estimated to be normal (57%). He had poor exercise tolerance and frequent angina; his medications included atenolol, hydrochlorothiazide, nitroglycerine, and furosemide.
A sequential compression device (SCD) with long sleeves with 45 mmHg (Sequel model 7325; Kendall Co., Mansfield, MA) was applied to each leg as part of routine practice for admission to our surgical intensive care unit.
He was taken to the operating room for a diverting cervical esophagostomy, exploratory laparotomy, and reduction of diaphragmatic hernia as well as placement of a gastrostomy tube. His intraoperative course was uneventful. He arrived back in the intensive care unit intubated, and mechanical ventilation was started.
During the next 6 h, a significant capillary leak developed in the patient, requiring large-volume crystalloid resuscitation (total intake − total output =+6 l), after which we decided to place a pulmonary artery catheter to assist with fluid management. A pulmonary artery thermodilution catheter (PAC; Baxter Swan Ganz ref. 131HF7, thermodilution catheter with antimicrobial coating (AMC) thromboshield-antimicrobial heparin coating 7F; Baxter Healthcare Corp., Edwards Critical Care Division, Irvine, CA) was placed uneventfully by rewiring of the previously placed right internal jugular central venous catheter. The PAC was connected to the CO module (7200 TRAM AR model No. S7200 Tram Module) of the monitor (series 7010 monitors; Marquette Electronics Inc., Milwaukee, WI).
After placement of the PAC, the patient had an arterial blood pressure of 91/46 mmHg, a heart rate of 100 beats/min, a pulmonary artery pressure of 39/20 mmHg, a central venous pressure of 6 mmHg, a temperature of 38.8°C, a CO of 3.7 l/min with an index of 1.8, a pulmonary artery wedge pressure of 6 mmHg, a stroke volume of 37 ml, and a systemic vascular resistance of 1,189 dyn · s−1· cm5.
The effect of further fluid resuscitation and vasoactive support on CO could not be assessed because the nursing staff reported variability of the CO measurements. The individual CO measurements varied between 1.8 and 10.4 l/min. Different physicians and nurses saw erratic waveforms with multiple attempts to measure CO. Without infusing injectate, the bedside cardiac monitor showed cycling output waveforms that varied between 3.5 and 5.0 l/min. Despite changing the monitor, changing the PAC, and having a biomedical engineer investigate the monitor, the problem persisted.
During this struggle with the CO measurements, care for the patient was guided by clinical assessment. After 4 h, the cause of the erratic waveforms was found to be from the cycling of the SCDs. We turned the SCDs off, and CO was measured to be 5.7 l/min with an index of 2.7, stroke volume was 63 ml, systemic vascular resistance was 814 dyn · s−1· cm5when blood pressure was 100/51 mmHg, heart rate was 90 beats/min, pulmonary artery pressure was 39/22 mmHg, and central venous pressure was 11 mmHg. Core temperature measured by PAC was 38.3°C, and injectate temperature was 24.7°C. The skin of the lower extremities was cool, with a toe temperature of 33.1°C, and the patient had 2+ pitting edema of the lower extremities.
We took several steps to establish a causal relation between SCD compression and the autonomous CO. First, we applied the SCDs to one leg and obtained an output from the CO computer of 5.8 l/min, with a stable baseline and good waveforms. Second, we reapplied the SCDs to two legs again; erratic waveforms and cycling autonomous output from CO computer reoccurred again. Third, when the full-leg SCDs were replaced with calf SCDs, erratic waveforms disappeared. Finally, we reapplied the full-leg SCDs, and the autonomous output CO computer reappeared and varied between 3.5 and 5.3 l/min. The rest of the CO measurements were obtained while the SCDs were temporarily turned off. The patient resolved the inflammatory response and was discharged from the intensive care unit 5 days later.
Discussion
Lower limb pneumatic compression devices are commonly used in intensive care units to prevent venous thromboembolism. We present a patient in whom we believe sequential compression devices caused erroneous CO measurements.
When using a PAC, CO is inversely proportional to the area under the thermodilution curve and is determined by injectate volume, blood temperature, density factor, and computation constant. In the current patient, the measure of CO was significantly impacted by the use of SCDs. Given that the thermistor on the PAC measures temperature changes, we hypothesize that the inflation cycle of the compression stockings brought cool blood from the legs into the central circulation, resulting in a decrease in pulmonary artery blood temperature. In this extreme example, because the patient was febrile and edematous with a low stroke volume, the volume of cool blood from the two legs squeezed by the long-sleeve SCD was enough to create a CO waveform and CO calculation by the computer. Neither one-leg SCD nor calf SCD compression produced enough volume to create a CO waveform by itself. We postulate that sepsis and the ensuing volume resuscitation may have resulted in cool extremities and that the patient’s low flow state amplified the impact of cool fluid being delivered to the central circulation.
Horiuchi et al. 2found a larger coefficient of variation on the pulmonary artery pressures when the calf pneumatic compression devices were used. In one patient in whom temperature changes in pulmonary artery blood were compared between the compression device on one or both legs, less temperature change was noted when the device was applied to one leg. They postulated that the larger the amount of cool blood that returns, the greater the variability in temperature. Despite this variation, CO was unaffected by the use of an intermittent compression device. Nonetheless, the variability was greater when they used room temperature injectate instead of iced injectate. With the higher temperature of injectate, the decrease in pulmonary artery temperature would be less than that observed with ice-cold injectate. 2
In the current patient, the impact of the transient decrease in pulmonary artery temperature from the peripheral pooled venous blood was perceived as a new CO curve by the computer. One would also expect that lower stroke volumes would result in less warm blood being ejected from the heart, thus increasing the effect of cool blood being returned from the limbs. 3
The use of SCDs is important for preventing deep venous thromboses in critically ill patients and will likely remain a mainstay of patient care in intensive care units. Physicians need to be aware of the potential for full-leg SCDs to impact thermodilution CO measurements and may need to consider temporarily turning SCDs off during CO measurements when these measurements are highly variable.
