Regional citrate anticoagulation in continuous venovenous hemodiafiltration

Regional Citrate Anticoagulation in Continuous Venovenous Hemodiafiltration Demetrios James Kutsogiannis, MD, Irvin Mayers, MD, Wu Dat Nin Chin, MD, and R.T. Noel Gibney, MD ● Over the past several years, continuous venovenous hemodiafiltration (CVVHDF) using pump-driven devices has gained wide acceptance as a form of renal replacement therapy for critically ill patients with acute renal failure. More recently, regional citrate anticoagulation has proven useful as a method of anticoagulating CVVHDF circuits, particularly in those patients at high risk for bleeding. However, an easy and convenient method for guiding the dose of citrate infusion has not previously been described. We describe the use of an algorithm using posthemofilter levels of ionized calcium to guide the dose of administered regional citrate on the survival time and urea and creatinine clearances of 24 Hospal AN69HF hemofilters. Nine patients with acute and chronic renal failure requiring CVVHDF were studied. The median filter survival time when using the postfilter ionized calcium algorithm was 3.4 days, with a survival probability of 46% (95% confidence interval [CI], 17 to 71). Random-effects linear regression analysis did not show a significant decline in blood-side urea clearance (P 5 0.041) or creatinine clearance (P 5 0.308). Moreover, definite bleeding complications occurred with an incidence rate of 0.045/person-day on citrate anticoagulation (95% CI, 0.006 to 0.16), and occult bleeding occurred with an incidence rate of 0.091/person-day on citrate anticoagulation (95% CI, 0.03 to 0.23). Guiding regional citrate anticoagulation through the use of posthemofilter ionized calcium levels is a safe and effective method of prolonging filter life during CVVHDF. r 2000 by the National Kidney Foundation, Inc. INDEX WORDS: Continuous venovenous hemodiafiltration (CVVHDF); continuous renal replacement therapy (CRRT); regional citrate anticoagulation; ionized calcium; hemorrhage; anticoagulation. I N RECENT YEARS, continuous forms of renal replacement therapy have been increasingly used to treat complicated acute renal failure, refractory fluid overload, and life-threatening electrolyte and acid-base disorders. However, the ease of filter clotting and decrease in efficiency compared with intermittent forms of hemodialysis therapy represent major limitations to their use.1 The requirement for prolonged continuous systemic anticoagulation appears to be the major drawback in the use of such continuous forms of renal replacement therapy as continuous venovenous hemodiafiltration (CVVHDF). Other drawbacks include the need for expensive sophisticated continuous dialysis machines and the hemofiltration fluid (up to 70 L/d) required for adequate diffusive and convective solute exchange. From the Division of Critical Care Medicine, the University of Alberta, Edmonton, Canada. The work was performed at the W.C. MacKenzie Health Sciences Center, the University of Alberta, Edmonton, Canada. Received May 5, 1999; accepted in revised form December 10, 1999. Address reprint requests to Demetrios James Kutsogiannis, MD, Rm 4228, Royal Alexandra Hospital, 10240 Kingsway Ave, Edmonton, Alberta, Canada T5H 3V9. E-mail: dkutsogi@telusplanet.net r 2000 by the National Kidney Foundation, Inc. 0272-6386/00/3505-0003$3.00/0 doi:10.1053/kd.2000.6375 802 A recent review of the available methods of anticoagulation in continuous renal replacement therapy emphasized the safety of the patient as the primary determinant guiding the choice of anticoagulant.2 Many critically ill patients cannot tolerate conventional anticoagulation with systemic heparin because of ongoing hemorrhage, severe coagulopathy, or heparin-induced thrombocytopenia and thrombosis. Consequently, regional citrate anticoagulation as a method of anticoagulation in both intermittent and continuous forms of hemodialysis has been advocated as an alternative to heparin.3-8 Citrate is an anticoagulant through its ability to chelate calcium. Intravenously administered citrate is cleared by the tricarboxycyclic acid pathway in the liver, skeletal muscles, and renal cortex, and levels of citrate and ionized calcium return to normal values within 30 minutes of discontinuing a citrate infusion.3,6 Therefore, regional citrate anticoagulation has been advocated as a safe method of preserving filter life. However, a simple and convenient method of dosing and monitoring patients administered regional citrate anticoagulation and its influence on filter survival have yet to be described. The purpose of this study is to determine the survival time of hemodialysis filters, as well as the incidence of bleeding complications, in a American Journal of Kidney Diseases, Vol 35, No 5 (May), 2000: pp 802-811 CITRATE ANTICOAGULATION IN CVVHDF cohort of critically ill patients undergoing CVVHDF and using a predetermined algorithm for regional citrate administration monitored by postfilter ionized calcium levels. METHODS Study Population and Method of CVVHDF Patients entering the general systems intensive care unit (ICU) at a regional tertiary care hospital were screened for the presence of acute or chronic renal failure and entered into the study before the initiation of CVVHDF and after obtaining informed consent. Approval for this study was obtained from the University of Alberta Health Research Ethics Board before enrolling patients. Demographic information, as well as the logistic organ dysfunction (LOD) score,9 was obtained for each patient. The cause of acute renal failure was defined according to the classification of Liano et al.10 CVVHDF was performed using the Prisma CFM machine (Hospal Industrie, Meyzieu, France) with a Hospal M-100 AN69HF predilution acrylonitrile and sodium methallylsulfonate copolymer hemofilter (Hospal Industrie). One of the 24 filters used was a Hospal M-60 AN69HF hemofilter. Venous access was obtained using a 19-cm 11.5 F catheter preferentially inserted into the femoral vein; however, subclavian and internal jugular catheters were placed in three patients at various times during the study. Hemofilters were primed with 5,000 U of heparin in 2 L of 0.9% saline at the discretion of the attending physician, and blood flow rate was maintained at 125 mL/min. Dialysate fluid and prehemofilter replacement (hemofiltration) fluid was manufactured by the hospital pharmacy department and consisted of sodium, 110 mmol/L; chloride, 110 mmol/L; and magnesium, 0.75 mmol/L. Sodium bicarbonate was used as a buffer within the hemofiltration fluid at a concentration of 16.7 to 50 mmol/L. Both the dialysate and prehemofilter replacement fluid were individually run at a rate of 1,000 mL/h, and net ultrafiltration was generally used to maintain a net fluid balance of 0 to –100 mL/h. A solution of 90 mL of 3.9% (wt/vol) trisodium citrate (145 mmol of citrate, 428 mmol of sodium; Tricitrisol; Citra Laboratories, Braintree, MA ) in 1,000 mL of 5% dextrose was used for regional anticoagulation and was initiated at a rate of 190 mL/h (25 mmol/h). This citrate solution was dissipated in the circuit by diffusion into the citrate-, calcium-, and bicarbonate-free dialysate. A separate solution of 80 mL of 0.74% wt/vol calcium chloride (56 mmol of calcium) in 1,000 mL of 0.9% saline was initiated at a rate of 60 mL/h (3.1 mmol/h) in a central venous catheter separate from the CVVHDF circuit. Subsequent adjustments to the rate of the trisodium citrate infusion were guided by the levels of circuit (postdialysis filter) serum ionized calcium drawn every 4 to 8 hours, based on a predefined algorithm and aiming for a level of 1.00 to 1.40 mg/dL (0.25 to 0.35 mmol/L; Table 1). Determination of blood-side urea and creatinine clearances was performed daily on 12 hemofilters and used the methods of Sigler et al.11 Definite bleeding was defined as the observation of gross 803 Table 1. Algorithm for Adjusting the Infusion Rate of Trisodium Citrate Based on Posthemofilter Ionized Calcium Levels Postfilter Ionized Calcium mg/dL mmol/L Change in Trisodium Citrate Infusion Rate .2.00 1.60-1.99 1.41-1.59 1.00-1.40 ,1.00 .0.50 0.40-0.50 0.36-0.39 0.25-0.35 ,0.25 Increase rate by 30 mL/h Increase rate by 20 mL/h Increase rate by 10 mL/h No change in rate Decrease rate by 10 mL/h bleeding at a primary or secondary site and one of the following criteria: (1) a spontaneous loss of 20 mm Hg or greater in systolic or diastolic blood pressure within 24 hours of observing bleeding, (2) transfusion of 2 units of packed red blood cells within 24 hours of observing bleeding, (3) failure of the hemoglobin concentration (in grams per deciliter) to increase after transfusion by at least the number of units transfused minus 2, or (4) decrease in hematocrit of 2% or greater within 24 hours of a witnessed site of bleeding. Occult bleeding was defined in the absence of observing gross bleeding and when one of the following criteria was met: (1) decrease in hematocrit of 2% or greater during CVVHDF confirmed by a repeated determination at 12 to 24 hours, or (2) failure of the hemoglobin concentration (in grams per deciliter) to increase after transfusion by at least the number of units transfused minus 2. Statistical Analysis Survival of the dialysis filters was estimated using the nonparametric method of Kaplan-Meier.12 Noncensored observations were defined as circuit clotting or persistently high filter pressures (.250 mm Hg), prohibiting the continuation of CVVHDF. All other reasons for terminating a filter were treated as censored observations. Bleeding complications were calculated as a bleeding rate per person-time at risk for bleeding while undergoing regional citrate anticoagulation, and 95% confidence intervals (CIs) on this rate were calculated assuming that the incidence rate followed a Poisson distribution.13 The longitudinal change in aggregate blood-side urea and creatinine clearance, posthemofilter ionized calcium, systemic ionized calcium, systemic sodium, and systemic carbon dioxide content (CO2 5 bicarbonate 1 CO2) was described using locally weighted scatterplot smoothing (Lowess) regression techniques.14 Values of posthemofilter ionized calcium less than 1.00 mg/dL were truncated and assigned a value of 1.00 mg/dL, given the limitations of the analyzer. The association between the elapsed time of individual filters and change in blood-side urea and creatinine clearances was determined using random-effects linear regression in which the intercept value for each filter was allowed to vary.15 Coefficients relating the change in blood clearance of urea or creatinine over time were deemed significant if the two-sided P was less than 0.01. All statistical analysis was performed on Stata 5.0 (Stata Press, College Station, TX).16 804 KUTSOGIANNIS ET AL Table 2. Patient Demographic, Illness Severity, Renal, and Renal Hematologic Parameters Before Initiation of CVVHDF Clinical Parameters Mean 6 SD Age (y) Women (%) LOD score* Urea nitrogen† (mg/dL) Creatinine† (mg/dL) Oliguric, ,400 mL/24 h (%) Hematocrit‡ (mg/dL) Platelets‡ (310 ` 9/L) PTT†§ (s) INR† 56 6 15 22 9.3 6 2.3 71.5 6 26.9 5.1 6 2.6 67 25.2 6 4.5 107 6 106 91 6 66 2.0 6 0.6 NOTE. N 5 9. Abbreviations: PTT, partial thromboplastin time; INR, international normalized ratio. *Logistic Organ Dysfunction score during the first 24 hours after admission into the intensive care unit (most severe score, 22). †Represents the greatest value within the 24-hour interval during which CVVHDF was initiated. ‡Represents the lowest value within the 24-hour interval during which CVVHDF was initiated. §One patient was anticoagulated with heparin (PTT 5 200 s) in the 24-hour period preceding the initiation of trisodium citrate anticoagulation for CVVHDF. RESULTS Twenty-four filters in nine patients requiring regional citrate anticoagulation for CVVHDF were studied. Two patients had chronic renal failure and were admitted to the ICU for other reasons. The remaining seven patients underwent CVVHDF for acute renal failure; three patients had septic acute tubular necrosis (ATN), and one patient each had surgical ATN, medical ATN, Wegener’s granulomatosis, and hepatorenal syndrome. The distribution of age, sex, admitting LOD scores, and hematologic and renal parameters are listed in Table 2. Nine of the 24 filters (38%) in five patients clotted or were terminated because of excessively high filter pressures, and the remaining 15 filters (62%) were considered as censored observations. Overall median survival time using the Kaplan-Meier estimator was 3.4 days (82 hours), with a survival probability of 46% (95% CI, 17 to 71). The overall survival probability for all 24 hemofilters is shown in Fig 1. Blood-side urea and creatinine clearances appeared to follow a steady linear course throughout the life of the hemofilters. Mean blood-side urea clearance was 18.93 6 4.63 mL/min (Fig 2), and blood-side creatinine clearance was 13.98 6 4.61 mL/min throughout the life of the filters. Random-effects linear regression analysis did not show a significant decline in blood-side creatinine clearance (P 5 0.308) over time; however, there was a trend toward increasing bloodside urea clearance (P 5 0.041) that was not significant at the level of P less than 0.01. Metabolic acidosis was present in the majority of study patients before the initiation of CVVHDF. After the initiation of CVVHDF and during the duration of time when trisodium citrate anticoagulation was used, the mean systemic carbon dioxide level was 21.8 6 5.3 Fig 1. Kaplan-Meier estimates of 24 hemofilters anticoagulated with regional citrate. CITRATE ANTICOAGULATION IN CVVHDF 805 Fig 2. Lowess smoothed graph of (A) blood-side urea and (B) blood-side creatinine clearances for 12 Hospal M-100 hemofilters anticoagulated with regional citrate. mmol/L (normal, 23 to 31 mmol/L). Three of the nine subjects developed transient metabolic alkalosis, with systemic carbon dioxide concentrations greater than 40 mmol/L. The first subject developed systemic alkalosis 45 hours after initiation of a hemofilter that survived for 73 hours. The alkalosis resolved spontaneously because the patient developed progressive metabolic acidosis from hepatic failure. The second subject developed systemic alkalosis 4 and 8 hours after initiation of a hemofilter that survived for 168 hours. The alkalosis necessitated decreasing the bicarbonate concentration of the replacement fluid from 33.3 to 16.7 mmol/L; however, this concentration had to be increased to 50 mmol/L because the subject subsequently developed a worsening metabolic acidosis. The third subject developed systemic alkalosis 40 and 42 hours after initiation of a hemofilter that survived for 76 hours. The alkalosis resolved spontaneously because the patient developed progressive metabolic acidosis from multisystem organ failure complicating Wegener’s granulomatosis. Moreover, no patients had clinically significant hypernatremic episodes because the mean systemic sodium concentration was 135.1 6 7.2 mmol/L during the duration of CVVHDF (normal, 133 to 146 mmol/L). The temporal changes in systemic carbon dioxide and sodium concentrations are shown in Fig 3A and B. Posthemofilter ionized calcium levels less than 1.40 mg/dL (0.35 mmol/L) were achieved in 83% of 230 measurements on the nine subjects. 806 KUTSOGIANNIS ET AL Fig 3. Lowess smoothed graph of (A) systemic carbon dioxide (CO2), (B) systemic sodium, (C) postfilter ionized calcium, and (D) systemic ionized calcium concentrations on elapsed survival time of hemofilters anticoagulated with regional citrate. The mean posthemofilter ionized calcium level was 1.23 6 0.25 mg/dL; however, the distribution of this parameter was skewed to a greater value given the truncation of values less than 1.00 mg/dL. The corresponding mean value for systemic ionized calcium was 4.34 6 0.83 mg/ dL, and temporal changes in both postfilter and systemic ionized calcium levels are shown in Fig 3C and D. Postfilter ionized calcium levels remained relatively stable throughout the life span of the hemofilters; however, a linear increase in the smoothed regression of systemic calcium on elapsed time was largely attributable to one patient who had received up to 26.6 mmol/h of trisodium citrate and 5.2 mmol/h of calcium for a hemofilter surviving more than 3 days. Assuming a constant blood-flow rate of 125 mL/min, the mean trisodium citrate concentration used to anticoagulate the circuit was approximately 3.1 6 0.78 mmol/L, and the mean calcium concentration infused was 0.49 6 0.18 mmol/L. No systemic cardiac or neuromuscular complications were attributable to hypocalcemia as a result of citrate intoxication. Definite or occult bleeding occurred in four of the nine subjects administered trisodium citrate anticoagulation and was attributed to the severity of their critical illness, rather than to the use of regional citrate anticoagulation. Definite bleeding occurred on two occasions for an incidence CITRATE ANTICOAGULATION IN CVVHDF 807 Fig 3 (cont’d). rate of 0.045/person-day on citrate anticoagulation (95% CI, 0.006 to 0.16). Likewise, four episodes of occult bleeding were identified for an incidence rate of 0.091/person-day on citrate anticoagulation (95% CI, 0.03 to 0.23). One subject diagnosed with myelodysplastic syndrome, febrile neutropenia, and septic ATN developed fresh blood from her nasogastric tube, accompanied by a decrease in hematocrit from 27% to 22%, and required 2 units of blood. The same subject later had a decrease in hematocrit from 22% to 20% with no obvious focus for bleeding and was administered an additional unit of blood. In a second subject diagnosed with hepatic failure, acute respiratory distress syndrome, and hepatorenal syndrome, hematocrit spontaneously decreased from 25% to a low of 16% within a 48-hour period with no initial focus of bleeding (counted as an occult bleed during the first 24 hours). In the latter 24-hour period, the subject developed esophageal variceal bleeding as a terminal event (counted as a definite bleed) that necessitated the transfusion of 8 units of blood, 8 units of cryoprecipitate, 11 units of fresh frozen plasma, and 8 units of platelets. A third subject diagnosed with septic shock from Hemophilus influenzae, acute respiratory distress syndrome, and septic ATN required 2 units of blood for a decrease in hematocrit from 27% to 23%, possibly related to intravascular hemolysis. A fourth subject diagnosed with profound heparin-induced thrombocytopenia causing acute re- 808 KUTSOGIANNIS ET AL nal failure, bilateral superficial femoral artery occlusion, and bilateral above-the-knee amputations required the transfusion of 1 unit of blood for a decrease in hematocrit from 27% to 23% 24 hours after a surgical debridement of a surgical wound. This episode was counted as an occult bleed because no superficial bleeding was noted. Of the nine subjects studied, five subjects died during their hospital stay, for an overall mortality rate of 56% 6 17% (95% CI, 21 to 86). Three of the five subjects who died had life support withdrawn, one patient had intractable bleeding from esophageal varices, and one patient died 48 hours after discharge from the ICU. Of the four surviving patients, two patients had preexisting chronic renal failure and resumed intermittent hemodialysis, and two patients did not require further intermittent hemodialysis at the time of hospital discharge. DISCUSSION Traditionally, systemic anticoagulation with heparin has been the anticoagulant of choice for CVVHDF. More recently, alternatives to conventional heparin have been proposed in an attempt to reduce the incidence of bleeding complications, particularly in patients undergoing major surgery or posttrauma. These alternatives include the following: high–flow rate hemodialysis requiring no anticoagulants17,18; anticoagulation with low-dose heparin by rinsing filters with heparin to bind to filters19,20; limited-dose systemic anticoagulation with heparin20-22; regional heparin anticoagulation by neutralization of heparin with protamine20,23; anticoagulation with prostacycline, which inhibits a platelet activator produced by the endothelium24-26; low-molecular-weight heparin2,27; and regional citrate anticoagulation.3-8 High–flow rate hemodialysis has been limited by the high flow rates (.300 mL/min) required to maintain adequate membrane patency. Regional anticoagulation with heparin is technically complicated because of the continuous requirement to estimate the amount of protamine required to neutralize the postfilter heparin. Compared with heparin, intravenous prostacyclin therapy has been shown to significantly improve filter life and decrease bleeding complications in combined hepatic and renal failure. However, the major limitations to its use are vasodilatation and the accompanying hypotension, as well as its 20% clearance by continuous dialysis techniques.24-28 Both intraindividual and interindividual variability in the metabolism of low-molecular-weight heparin and its partial reversal by protamine currently limit its widespread use in this subgroup of critically ill patients.2 Considering this, regional citrate anticoagulation appears to offer advantages in terms of safety. Its major limitations have been cost and ease of use. The present study shows the ease and feasibility of monitoring and prescribing regional citrate anticoagulation through the use of postfilter ionized calcium levels. The median hemofilter survival time of 3.4 days (82 hours) shown in this study was longer than a median survival time of approximately 48 hours in 36 continuous arteriovenous hemodialysis (CAVHD) hemofilters reported by Mehta et al4 (who used activated clotting time [ACT] measurements to monitor anticoagulation) and 29.5 6 17.9 hours in 85 continuous venovenous hemofiltration (CVVH) hemofilters reported by Palsson and Niles.8 Moreover, the use of postfilter ionized calcium measurements to guide regional citrate anticoagulation in this study and the study of Palsson and Niles8 appears to be more biologically plausible than the use of circuit ACT measurements, given the mechanism by which trisodium citrate acts as an anticoagulant. The present study used a mean citrate infusion rate of 23.2 6 5.9 mmol/h compared with 24 and 18.6 mmol/h administered by Mehta et al4 and Palsson and Niles,8 respectively, and a mean calcium infusion rate of 3.7 6 1.4 mmol/h compared with 2.0 and 2.8 mmol/h described by Mehta et al4 and Palsson and Niles,8 respectively. The mean systemic ionized calcium level in the present study was 4.3 6 0.8 mg/dL compared with ranges of 2.4 to 5.8 and 1.3 to 5.2 mg/dL described by Mehta et al4 and Palsson and Niles,8 respectively. Hence, the longer median survival time described in the present study may be attributable in part to a larger administered dose of trisodium citrate compared with the study by Palsson and Niles8; however, this would not explain the difference compared with the findings of Mehta et al.4 Other factors that may explain the differences in survival times of the hemofilters in these studies are: (1) differences in the patient population and relatively small sample sizes present in CITRATE ANTICOAGULATION IN CVVHDF the three studies that limit statistical comparisons; (2) differences in the site and size of the dialysis catheter used (11.5 F preferentially inserted in the femoral vein in the present study opposed to 14 F or 16 F in the study of Mehta et al4 and 14 F in the study of Palsson and Niles8); (3) mode of continuous renal replacement therapy (CVVHDF in the present study versus CAVHD and CVVH in the studies of Mehta et al4 and Palsson and Niles,8 respectively); and (4) differences in blood flow rates used (125 mL/min in the present study compared with 52 to 125 and 180 mL/min in the studies of Mehta et al4 and Palsson and Niles,8 respectively). In theory, the larger French catheters used by Mehta et al4 and Palsson and Niles8 should provide better blood flow and be less prone to premature clotting; however, this was not reflected in the median hemofilter survival times. Our study attempted to use the femoral site for venous access whenever possible, and this may account for a portion of the improvement in hemofilter survival time because this site recently has been shown to provide the least problems with blood-flow reduction.29 A greater blood-flow rate in the study of Palsson and Niles8 may have also contributed to a lower concentration of hemofilter trisodium citrate per millimole of trisodium citrate administered in the circuit. Finally, the improved hemofilter survival shown in the present study may be attributable to our method of precisely monitoring postfilter ionized calcium levels to guide the dosage of trisodium citrate. Mean blood-side urea clearance was 18.93 6 4.63 mL/min in the present study compared with 24.1 6 0.9 mL/min in the study of Mehta et al.4 However, the rate of change in urea clearance over time was not described by Mehta et al4; therefore, temporal comparisons of urea clearance could not be made. The decreased urea clearance noted in the present study could not be attributed to the hemofilter used (Hospal AN69S hemofilters were used in both studies) or to an increased blood flow rate because a rate of 52 to 125 mL/min was used in the study of Mehta et al.4 Moreover, in contrast to the study of Sigler et al,11 which showed a steady decrease in urea clearance over time when using heparin anticoagulation through a Hospal AN69S hemofilter, 809 the present study failed to show a significant decrease over time. Contrary to the studies of Mehta et al4 and Palsson and Niles,8 which did not observe bleeding complications with regional citrate anticoagulation in CAVHD and CVVH, we encountered a low incidence of bleeding complications and a different proportion of circuit failure because of filter clotting: 38% in the present study versus 49.1% and 24.7% in the studies of Mehta et al4 and Palsson and Niles,8 respectively. The present cohort study defined bleeding episodes in advance of data collection compared with the previous two retrospective studies. These bleeding complications were generally associated with the severity of the illnesses requiring intensive care support rather than as a direct result of regional citrate anticoagulation. With respect to systemic alkalosis, three of the nine subjects in the present study developed transient systemic alkalosis that resolved with conservative measures. Although previous investigators have described the cardiac effects of citric acid intoxication, no systemic cardiac or neuromuscular complications were attributable to hypocalcemia as a result of citrate intoxication, nor was regional citrate anticoagulation terminated because of profound metabolic alkalosis.30,31 This compares with 3 of 11 patients studied by Mehta who required HCl infusions to control their metabolic alkalosis. The overall mortality rate in this study population was 56%, consistent with a 60% mortality rate predicted by the mean LOD scores of the nine patients included on the study and similar to a 53% mortality rate reported by Palsson and Niles8 in 1999 and a 62% mortality in a retrospective study by Jones et al32 from 1991 to 1995. However, it was less than an 89% mortality rate reported by Mehta et al4 in 1990. The most significant limitation of this study is the modest number of filters studied (24 filters), as well as the inclusion of the majority of filters as censored observations (ie, not as failures). Blood-side clearances of urea and creatinine remained linear throughout the duration of filter life with a trend toward increasing urea clearance over time, the latter likely attributable to a smaller number of filters surviving beyond 3 days. This suggests that the most significant component of filter inefficiency and failure occurs rather 810 KUTSOGIANNIS ET AL abruptly rather than through a mechanism of progressive clotting in individual fibers. Moreover, clearance measurements of urea and creatinine were not performed within 30 minutes of hemofilter initiation; therefore, the previously described early permeability decay noted in polyacrylonitrile hemofilters was not present in our study.33 This study has shown that guiding the prescribed dose of regional citrate anticoagulation through the use of postfilter ionized calcium levels is an easy, safe, and biologically plausible way to anticoagulate critically ill patients with renal failure requiring CVVHDF. Following such a regimen results in a respectable filter survival time, stable levels of urea and creatinine clearance, and a minimal risk for developing significant systemic alkalosis. ACKNOWLEDGMENT The authors thank Concetta Carbonne, Dr M. Heule, Dr M. Meier, Margo Miller, Carlos Miranda, Dr D. Muzyka, and Dr M. VanWijngaarden. REFERENCES 1. Ronco C: Continuous renal replacement therapies for the treatment of acute renal failure in intensive care patients. Clin Nephrol 40:187-198, 1993 2. Favre H, Martin PY, Stoermann C: Anticoagulation in continuous extracorporeal renal replacement therapy. Semin Dial 9:112-118, 1996 3. Von Brecht JH, Flanigan MJ, Freeman RM, Lim VS: Regional anticoagulation: Hemodialysis with hypertonic trisodium citrate. Am J Kidney Dis 8:196-201, 1986 4. Mehta RL, McDonald BR, Aguilar MM, Ward DM: Regional citrate anticoagulation for continuous arteriovenous hemodialysis in critically ill patients. Kidney Int 38:976981, 1990 5. Lohr JW, Slusher S, Diederich D: Safety of regional citrate hemodialysis in acute renal failure. Am J Kidney Dis 13:104-107, 1989 6. Pinnick RV, Wiegmann TB, Diederich DA: Regional citrate anticoagulation for hemodialysis in the patient at high risk for bleeding. N Engl J Med 308:258-261, 1983 7. Lohr JW, Slusher S, Diederich DA: Regional citrate anticoagulation for hemodialysis following cardiovascular surgery. Am J Nephrol 8:369-372, 1988 8. Palsson R, Niles JL: Regional citrate anticoagulation in continuous venovenous hemofiltration in critically ill patients with a high risk of bleeding. Kidney Int 55:19911997, 1999 9. Le Gall J, Klar J, Lemshow S, Saulnier F, Alberti C, Artigas A, Teres D: The logistic organ dysfunction system. A new way to assess organ dysfunction in the intensive care unit. JAMA 276:802-810, 1996 10. Liano F, Pascual J, and The Madrid Acute Renal Failure Study Group: Epidemiology of acute renal failure: A prospective, multicenter, community-based study. Kidney Int 50:811-818, 1996 11. Sigler MH, Brendan PT, Van Valkenburgh D: Solute transport in continuous hemodialysis: A new treatment for acute renal failure: Kidney Int 32:562-571, 1987 12. Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1958 13. Selvin S: Measure of risk: Rates and probabilities, in Selvin S (ed): Statistical Analysis of Epidemiologic Data (ed 2). New York, NY, Oxford, 1996, pp 3-40 14. Cleveland WS: Robust locally weighted regression and smoothing scatterplots. J Am Stat Assoc 74:829-836, 1979 15. Diggle PJ, Liang K, Zeger SL: Analysis of Longitudinal Data. New York, NY, Oxford, 1994 16. Stata Press: STATA statistical software, release 5, vol I-III. College Station, TX, Stata Press, 1997 17. Raja R, Kramer M, Rosenbaum JL, Bolisay C, Krug M: Thrombogenicity in hollow-fiber kidney—Use of Cordis Dow-3500 without heparin. Artif Organs 5:S58A, 1981 (abstr, suppl) 18. Naito H, Miyazakai T, Shimizo M, Yoneda K, Yoshikawa K: Non-anticoagulant hemodialysis with synthetic EVAL HF. Artif Organs 5:S50A, 1981 (abstr, suppl) 19. Gretz N, Vienken J: Heparin binding by hemophan membranes and heparin-free dialysis. Semin Dial 8:128129, 1995 20. Swartz RD, Port FK: Preventing hemorrhage in highrisk hemodialysis: Regional versus low-dose heparin. Kidney Int 16:513-518, 1979 21. Swartz RD: Hemorrhage during high-risk hemodialysis using controlled heparinization. Nephron 28:65-69, 1981 22. Gretz N, Quintel M, Ragaller M, Odenwalder W, Bender HJ, Rohmeiss P, Strauch M: Low-dose heparinization for anticoagulation in intensive care patients on continuous hemofiltration, in Sieberth HG, Stummvoll HK, Kierdorf H (eds): Continuous Extracorporeal Treatment in Multiple Organ Dysfunction Syndrome. Contributions to Nephrology. Basel, Switzerland, Karger, 1995, pp 130-135 23. Maher JF, Lapierre L, Schreiner GE, Geiger M, Westervelt FB Jr: Regional heparinization for hemodialysis: Technique and clinical experiences. N Engl J Med 268:451456, 1963 24. Zusman RM, Rubin RH, Cato AE, Cocchetto DM, Crow JW, Tolkoff-Rubin N: Hemodialysis using prostacyclin instead of heparin as the sole antithrombotic agent. N Engl J Med 304:934-939, 1981 25. Turney JH, Williams LC, Fewell MR, Parsons V, Weston MJ: Platelet protection and heparin sparing with prostacyclin during regular dialysis therapy. Lancet 2:219222, 1980 26. Davenport A, Will EJ, Davison AM: Comparison of the use of standard heparin and prostacyclin anticoagulation in spontaneous and pump-driven extracorporeal circuits in patients with combined acute renal and hepatic failure. Nephron 66:431-437, 1994 27. Jeffrey RF, Khan AA, Douglas JT, Will EJ, Davison AM: Anticoagulation with low-molecular-weight heparin (Fragmin) during continuous hemodialysis in the intensive care unit. Artif Organs 17:717-720, 1993 28. Davenport A: The coagulation system in the critically CITRATE ANTICOAGULATION IN CVVHDF ill patient with acute renal failure and the effect of an extracorporeal circuit. Am J Kidney Dis 30:S20-S27, 1997 (suppl 4) 29. Baldwin I, Bellomo R: Flow reductions associated with CVVH: A comparison of frequency, severity and duration between access catheter sites. Blood Purif 17:9A, 1999 (abstr) 30. Bunker JP, Stetson JB, Coe RC, Grillo HC, Murphy AJ: Citric acid intoxication. JAMA 157:1361-1367, 1955 811 31. Bunker JP, Bendixen HH, Murphy AJ: Hemodynamic effects of intravenously administered sodium citrate. N Engl J Med 266:372-377, 1962 32. Jones CH, Richardson D, Goutcher E, Newstead CG, Will EJ, Cohen AT, Davison AM: Continuous venovenous high-flux dialysis in multiorgan failure: A 5-year singlecenter experience. Am J Kidney Dis 31:227-233, 1998 33. Jenkins RD, Kuhn RJ, Funk JE: Permeability decay in CAVH hemofilters. ASAIO Trans 34:590-593, 1988