Secondary Logo

Journal Logo

General Article: Case Report

When Drugs Disappear from the Patient: Elimination of Intravenous Medication by Hemodiafiltration

Stricker, Kay H. MD*†; Takala, Jukka MD, PhD*; Hullin, Roger MD, PhD; Ganter, Christoph C. MD*

Author Information
doi: 10.1213/ANE.0b013e3181b9db63


Continuous venovenous hemodiafiltration (CVVHDF) is essential for the treatment of hemodynamically unstable critically ill patients with acute renal failure. However, CVVHDF eliminates both waste products and medications in the venous circulation, requiring dose adjustments. For some drugs, pharmacokinetic data on elimination with CVVHDF are available.1–3 Furthermore, the relative positions of venous infusion and dialysis catheters may influence drug elimination by CVVHDF. We describe a case of acute right heart failure due to elimination of IV epinephrine by CVVHDF.


A 38-yr-old woman with end-stage heart failure complicated by cardiac cachexia (49 kg, body mass index of 18.4 kg/m2) and anuric renal failure requiring CVVHDF underwent orthotopic heart transplantation. Extracorporeal circulation for high-urgency transplant lasted 175 min. Posttransplant transesophageal echocardiography in the operating room revealed normal left and right ventricular systolic function. Her postoperative hemodynamics were stable, with epinephrine at 6.5 μg/min, sodium nitroprusside at 25 μg/min, atrial pacing with 90 bpm, and inhalation of nitric oxide (NO) at 15 ppm. Six hours postoperatively, CVVHDF (HOSPAL-Prisma, Gambro Healthcare, Lakewood, CO) using an AN69 high-flux hemofilter was restarted in the intensive care unit. NO was successively reduced to 7 ppm, whereas dosages of epinephrine and nitroprusside remained unchanged. Twenty-three hours postoperatively, mixed venous saturation decreased by 10 percentage points, accompanied by arterial hypotension. This episode resolved spontaneously after 10 min (Fig. 1, arrow A). On the assumption that there was a technical problem with the pulmonary artery catheter, mixed venous saturation was recalibrated to coincide with the blood gas analysis value, resulting in a lower set point (Fig. 1, arrow B).

Figure 1
Figure 1:
Figure 1.

One hour later, her hemodynamics again became acutely unstable (Fig. 1, arrow C): mean arterial blood pressure decreased from 80 to 43 mm Hg and mixed venous saturation from 63% to 38%. The cardiac output decreased from 4.0 to 2.4 L/min, right ventricular ejection fraction declined from 20% to 11%, central venous pressure increased from 15 to 18 mm Hg, mean pulmonary artery pressure decreased from 23 to 19 mm Hg, and pulmonary artery occlusion pressure decreased from 14 to 11 mm Hg. All values were measured with a pulmonary artery catheter (CCOmbo catheter 777HF8, Edwards Lifesciences, Irvine, CA). Emergency transthoracic echocardiography confirmed acute right heart failure with severe dilation and severely abnormal systolic function of the right ventricle from the apical 4-chamber view. Cardiac tamponade was excluded, and the electrocardiogram remained unremarkable. Initially, right heart failure secondary to overzealous reduction in NO was suspected, and NO was increased to 20 ppm, followed by increasing bolus doses of epinephrine (from 10 to 100 μg) and a bolus of aminophylline (200 mg) through the central venous catheter. However, her hemodynamics did not improve (Fig. 1, arrow D). A check of infusion solutions, syringe pumps, and their connections with the central venous catheter revealed no explanation. Increasing doses of sedative and analgesic drugs were administered to reduce right ventricular afterload caused by the patient’s anxiety and fighting the ventilator (cumulative dose: 18 mg of midazolam and 125 μg of fentanyl), with only moderate effect. For muscle relaxation, 40 mg of atracurium was given without success. The patient was then given 50 mg of rocuronium. Despite these boluses of midazolam, fentanyl, atracurium, and rocuronium, the patient was inadequately sedated and paralyzed. It was suspected that the drugs were not entering the circulation, even though the central venous catheter appeared to be positioned correctly on the chest radiograph (Fig. 2, solid arrow), and blood could be aspirated freely. The port for injection of medications was changed to the additional infusion lumen of the dialysis catheter, with an opening at the tip of the catheter, distal to the dialysis side holes (12F Mahurkar dialysis catheter, Tyco Healthcare, Mansfield, MA) (Fig. 2, hollow arrow). Injection of 100 μg of epinephrine induced a hypertensive crisis, treated with sodium nitroprusside (Fig. 1, arrow E). In addition, anesthesia with 2 mg of midazolam and 25 μg of fentanyl was effective.

Figure 2
Figure 2:
Figure 2.

Review of the postoperative chest radiograph revealed that the tip of the central venous catheter was correctly positioned in the superior vena cava (Fig. 2, solid arrow). However, the tip of that catheter was noted to be adjacent to the “arterial” side port of the dialysis catheter (Fig. 2, dotted arrow), and interference between the catheters was suspected. Through this arterial side port, blood is aspirated to supply the dialyzer, and through the “venous” port at the tip of the dialysis catheter (Fig. 2, hollow arrow), blood that has passed the dialyzer is reinfused into the circulation. The catheters were changed, all infusions and syringe pumps were reconnected, and her hemodynamics improved immediately and remained stable at only half of the dosage of epinephrine compared with before the hemodynamic crisis.


Clinically relevant elimination of drugs injected through a central venous catheter by CVVHDF has rarely been observed. Loss of the thermal indicator for thermodilution cardiac output measurement due to the cooling effect of CVVHDF has been suspected, e.g., blood being aspirated into the dialyzer between the heating coil and the thermistor of the Swan-Ganz catheter. In our patient, the heating coil was distal to the dialysis catheter’s arterial and venous ports and no injectate or “thermal bolus” should have been lost. Furthermore, even if the proximal port of the pulmonary artery catheter is distal to the dialysis catheter, and no thermal indicator is lost within the dialyzer, the lower temperature of the blood returning from the dialyzer may lead to erroneous thermodilution cardiac output measurement. However, Sakka et al.4 did not note an influence of CVVHDF on cardiac output measurements by the transpulmonary thermodilution technique when 15 mL cooled 0.9% saline boluses (<8°C) were injected. Of note, in half of the patients, the dialysis catheter was placed in the inferior vena cava through the femoral vein, whereas injections were given through a central venous catheter in the superior vena cava. Only in 12 patients of Sakka et al. were both catheters placed in the superior vena cava, similar to the case described here.

Extracorporeal drug removal by CVVHDF is dependent on pharmacokinetic parameters (e.g., protein binding, volume of distribution, and molecular weight) and patient-related factors (e.g., concentration of free fatty acids, hyperbilirubinemia, blood pH, and drug charge), the flow and mode of filtration, the type of membrane, the mode of replacement fluid administration (predilution or postdilution), and the amount of blood recirculation within the lumen of the dialysis catheter.5,6 Transmembrane drug elimination during CVVHDF is based for the most part on convection (by hemofiltration) and only partially on diffusion (by dialysis). The main determinant for drug elimination by convection is the protein binding and not the molecular weight, because the molecular weight of almost all drugs is smaller than the membrane cutoff.5 Drugs with expected high protein binding, such as midazolam (95%),7 fentanyl (84%),8 atracurium (82%) ( Accessed June 8, 2009), or aminophylline (60%) ( Accessed June 8, 2009), are less likely to be effectively removed by CVVHDF than drugs with expected high unbound fractions such as epinephrine (24%)9 or rocuronium (30%) ( Accessed June 8, 2009). Nevertheless, it is important to realize that data published on protein binding of drugs were usually determined during steady-state or pseudo-steady-state conditions in pharmacokinetic trials. No protein binding data are available from situations in which a diluted drug enters the extracorporeal circuit but may not even be fully mixed when reaching the membrane. It is conceivable that the protein binding is significantly less in such situations. Moreover, there is a high individual variability in protein binding even in healthy volunteers and it is much more pronounced in critical illness.5 A further factor contributing to drug elimination is the membrane (AN69, acrylonitrile) used for CVVHDF, which has a high adsorptive capacity.5,10 We hypothesize that high plasma protein binding was hardly possible: bolus injections led to high local concentrations of the drugs, and the close proximity of the 2 catheters limited the time that the drugs were in contact with plasma. As a consequence, the drugs were almost completely eliminated by CVVHDF.

In retrospect, it seems that the first hemodynamic decline (Fig. 1, arrow A) was likely due to transient aspiration, and the following crash (Fig. 1, arrow C) was due to constant aspiration of the central venous catheter infusate by the dialysis catheter. As can be observed with fluoroscopy, catheters may migrate within the superior vena cava with heart beat or respiration. When the tip of the infusion catheter lines up with the arterial port of the dialysis catheter, drugs leaving the central venous catheter may directly enter the dialysis catheter, with subsequent elimination by CVVHDF. Thus, we hypothesize that epinephrine was eliminated by CVVHDF when the catheters were adjacent, leading to an acute decrease in cardiac output. As a consequence, the fraction of infused drugs aspirated by the adjacent dialysis catheter increased, leading to a vicious cycle: the more severe the shock, the lower the blood flow in the central vein, and the higher the proportion of elimination of drugs by CVVHDF. In our patient, all 3 central venous catheters were inserted through the left internal jugular vein because of suspected infection of previous catheters and thrombosis on the right side.

In conclusion, when the effects of IV drugs are inadequate in patients receiving CVVHDF, interference of adjacent central venous catheters resulting in elimination of the drug by CVVHDF should be suspected. Although drugs may be injected using a more distal port, temporary interruption of CVVHDF should be considered in an emergency situation. However, to avoid interference, catheters used for infusions and for dialysis should be placed in different veins, such as the superior and inferior caval vein, if feasible.


1. Schetz M, Ferdinande P, van den Berghe G, Verwaest C, Lauwers P. Pharmacokinetics of continuous renal replacement therapy. Intensive Care Med 1995;21:612–20
2. Yagasaki K, Gando S, Matsuda N, Kameue T, Ishitani T, Hirano T, Iseki K. Pharmacokinetics and the most suitable dosing regimen of fluconazole in critically ill patients receiving continuous hemodiafiltration. Intensive Care Med 2003;29:1844–8
3. DelDot M, Lipman J, Tett S. Vancomycin pharmacokinetics in critically ill patients receiving continuous venovenous haemodiafiltration. Br J Clin Pharmacol 2004;58:259–68
4. Sakka SG, Hanusch T, Thuemer O, Wegscheider K. The influence of venovenous renal replacement therapy on measurements by the transpulmonary thermodilution technique. Anesth Analg 2007;105:1079–82
5. Schetz M. Drug dosing in continuous renal replacement therapy: general rules. Curr Opin Crit Care 2007;13:645–51
6. Bugge J. Pharmacokinetics and drug dosing adjustments during continuous venovenous hemofiltration or hemodiafiltration in critically ill patients. Acta Anaesthesiol Scand 2001;45:929–34
7. Swart E, de Jongh J, Zuideveld K, Danhof M, Thijs L, Strack van Schijndel R. Population pharmacokinetics of lorazepam and midazolam and their metabolites in intensive care patients on continuous venovenous hemofiltration. Am J Kidney Dis 2005;45:360–71
8. Meuldermans W, Hurkmans R, Heykants J. Plasma protein binding and distribution of fentanyl, sufentanil, alefentanil and lofentanil in blood. Arch Int Pharmacodyn Ther 1982;257:4–19
9. Sager G, Trovik T, Slordal L, Jaeger R, Prytz P, Brox J, Reikeras O. Catecholamine binding and concentrations in acute phase plasma after surgery. Scand J Clin Lab Invest 1988;48:419–24
10. Clark W, Hamburger R, Lysaght M. Effect of membrane composition and structure on solute removal and biocompatibility in hemodialysis. Kidney Int 1999;56:2005–15
© 2009 International Anesthesia Research Society