Continuous renal replacement therapy (CRRT) (Table 1) is well established as a modality for the management of renal failure in the critically ill patient. 1 The benefits of slow, continuous, gentle fluid and urea removal in this population is well supported in the literature. 2–6 As experience with this therapy grows, so do the questions of its efficacy in conditions other than renal failure. In this article, we present an overview of the therapy: indications, management options, issues with vascular access, complications, and technology. Additionally, we discuss the ethical issues that have surfaced as the therapy has developed. Last, we review some of the current research questions.
The development of CRRT began in the 1970s. Both Bergstrom and Henderson noted that conventional intermittent hemodialysis was not the renal replacement method of choice for unstable, critically ill patients. 1 Many of the issues focused on the acute fluid shifts associated with the 4-hour hemodialysis treatment. It was proposed that these hemodynamically and neurologically unstable patients would benefit from a slower, gentler form of therapy.
Peter Kramer is credited with much of the work concerning the development of continuous therapy. Inadvertent cannulation of an artery in the lab resulted in the identification of arteriovenous therapy as a method for effectively removing fluids and solutes. Continuous arteriovenous hemofiltration (CAVH) received U.S. Food and Drug Administration (FDA) approval in 1982. 7 Although therapeutically effective, CAVH has significant potential for complications, many of which are related to arterial cannulation. Although theoretically simple in design, CAVH requires an arterial and venous cannulation and an extracorporeal circuit. Maintenance of this circuit depends on adequate hydrostatic pressure: a mean arterial pressure of at least 60 mm Hg and, ideally, regional anticoagulation of the circuit.
In the mid-1980s the FDA approved continuous venovenous hemofiltration (CVVH). 8 This low-pressure venous system eliminates the risks of arterial cannulation associated with its arteriovenous (AV) counterpart. It does require the addition of a blood pump, which, although eliminating the need for hydrostatic pressure for circuit maintenance, introduces the complexities associated with developing technology.
Indications for Therapy
When CRRT was first developed, the major indications for its initiation were fluid and solute removal associated with renal failure. 9 Initially, candidates for therapy included those patients in whom acute renal failure (ARF) developed while in the intensive care unit and for whom, because of hemodynamic instability, intermittent hemodialysis (IHD) was not an option. Over time it became clear that therapy could also benefit patients with chronic renal failure (CRF) (an IHD patient) who underwent a major surgical procedure (e.g., coronary artery bypass graft) and were too unstable for the first few days to resume hemodialysis.
Although the two broad categories of indications for therapy have remained the same (fluid and solute removal), the conditions under which they are applied have broadened considerably over time. The application of CRRT in the setting of volume overload has become an indication in and of itself. Criteria for initiating therapy include volume overload that is refractory to diuretics (e.g., congestive heart failure). The increased volume requirements associated with rhabdomyolysis after major skeletal muscle injury can also be managed with CRRT. Fluid management of the oliguric patient with ongoing requirements (e.g., parenteral nutrition, antibiotics, blood products) is very effectively accomplished with CRRT. 1
The benefits of meticulous fluid management cannot be overemphasized. Cardiac and pulmonary function improves markedly as fluid is removed in a carefully prescribed manner. Being able to achieve fluid balance on an hourly basis provides “room” to administer antibiotics and parenteral nutrition, which, it is hoped, improve wound healing and ultimately recovery.
The second major indication for the initiation of CRRT is solute removal. 1 As stated, it is a “bridge” between IHD treatments for those CRF patients who are critically ill. Catabolic patients (e.g., thermally injured, necrotizing fasciitis, desquamating skin disorders, trauma) with developing ARF benefit from a controlled method of solute removal. The fluid management component of this population is also addressed with CRRT.
The use of CRRT in the setting of acid-base imbalances must be understood to be a temporizing measure. In the settings of both increased acid production (e.g., lactic acidosis) and bicarbonate depletion (e.g., intestinal fistulas), CRRT will ameliorate the clinical effects of the imbalance for a time while plans for definitive treatment (e.g., surgery) are made. Although there may be an improvement in the hemodynamic and pulmonary status with a resultant decrease in the lactic acidosis, one must not interpret this as resolution of the problem.
In the pediatric setting, CRRT has several unique applications. Patients on extracorporeal membrane oxygenation (ECMO) for the preoperative management of diaphragmatic hernias in whom ARF develops respond well to CRRT connected directly into the ECMO circuit. Those patients with inborn errors of metabolism also demonstrate a positive response. CRRT may be used either prophylactically or as treatment for tumor lysis syndrome (this application applies to adults as well).
The use of CRRT for managing sepsis and cytokine removal potentially presents a new category for application. This issue is further discussed in the Current Questions section.
Temporary vascular access catheters used in CRRT are large-bore two-and three-lumen dialysis catheters, the diameters of the which range from 9.0–13.5F. Catheter length is generally between 10 and 19.5 cm (Mahurkar, The Kendall Company, Mansfield, MA, USA). These catheters are found to be somewhat less flexible than other multilumen catheters.
Appropriate temporary vascular access for CRRT is often quite difficult to achieve and frequently more difficult to maintain. Patients meeting criteria for CRRT are generally bedbound, immobile, and profoundly ill. Selection of appropriate access is dependent on several factors: skill of the accessing clinician, height/weight proportion versus disproportion, mobility of the patient, anticipated duration of therapy, and the ability of the patient to tolerate potential sequelae of line placement.
The internal jugular (IJ) vein is a preferred access site secondary to the limited number of lethal complications associated with catheter placement. It is important to have some idea of reciprocal venous flow in the opposite IJ. Neurological assessment before and after turning the patient's neck may be valuable in predicting those patients who may not tolerate any turning of the neck secondary to compromise of flow in the carotids.
Femoral veins are easily accessible by most clinicians. Complications from femoral access are associated with access of the femoral artery, needling the femoral nerve, and kinking of the catheter after placement. The most frequent complication associated with femoral access is kinking, particularly in the obese patient. Increased angling of the catheter on entrance to the skin and adipose tissue is necessary before cannulation of the vessel. This additional flexion in the catheter leads to a decrease in the intraluminal diameter. It further leads to an increased negative pressure on the “arterial” side of the catheter and may lead to increased positive pressure to drive blood back to the patient on the “venous” side of the catheter. A remedy for this kinking is keeping the head of bed flat while placing the bed in reverse Trendelenberg position to decrease risk of aspiration. An additional unexpected complication from femoral access in the morbidly obese is migration of the catheter or one of the catheter side ports out of the vessel. Although generally the side lumen is considered the “arterial” side, the lumina are used interchangeably to facilitate best “arterial” flow with the least negative pressure. When the lumina are switched (venous for arterial), the venous side may deliver blood back into subcutaneous tissues rather than the intravascular space if the catheter is rendered too short by nature of the depth of the adipose tissue layer. It is essential to have direct visualization of the catheter access site at all times. Additionally, femoral access sites are associated with a higher incidence of catheter-associated bacteremias secondary to the proximity to intestinal flora.
Subclavian access, although associated with greater placement issues (i.e., inability to palpate the vessel, pneumothorax), has fewer issues with maintenance of flow during CRRT. Occasionally, flow issues will occur similar to those found with the IJ and femoral vein sites. Remedies for these include flexing both shoulders back by placing a towel roll behind the patient's back and vertically between the shoulder blades and rotating the patient away from the side with the subclavian access.
CRRT system life is shortened with access difficulty. Increased negative pressure results in decreased blood flow and possibly increases the turbulence within the system. Increased turbulence leads to increased clotting, particularly within the filter and venous chamber.
Chronic renal failure patients who are too ill to undergo intermittent hemodialysis may have their AV grafts and fistulas accessed and connected to the CRRT circuit. These patients will have noticeably higher transmembrane pressures while undergoing CRRT. Protection of the AV fistula or graft and accessing needles is essential to keep the access open. If the CRRT system is shut off for any reason, the needles accessing the AV fistula or graft may be left in but will require a continuous flush to keep the needles from clotting off. Most standard infusion pumps will not be able to overcome the pressure in the access to deliver fluid continuously. A standard transducer setup inflated to 300 mm Hg will infuse a mere 3 mL/hour, keep the AV access open, and avoid fluid overload. Obviously, an access that is not being used for a longer time period should be removed and replaced when needed.
Scrupulous technique is necessary with access of any kind used for CRRT. Aseptic techniques must be used when placing the access. Findings in the literature suggest that prepping with 2% tincture of iodine 10–13 may be associated with significantly less contamination of the skin. This solution is irritating to patient's skin and must be removed with isopropyl alcohol (70%) after catheter placement. Proximal connectors on catheters should be vigorously scrubbed with povodine-iodine solution before connection of the CRRT system. Access should be cared for with occlusive transparent film dressings that are changed every 48 hours. 14 Stopcocks and additional connections of any kind should be discouraged because of increased risk of catheter-associated bacteremia and inadvertent disconnection of the system. If additional connections are necessary, meticulous care must be taken.
Catheter patency needs to be maintained at all times. Catheters may be flushed with heparin, acid citrate dextrose, or normal saline. To avoid administering a “bolus” of heparin to the patient, a “dead space” volume is withdrawn from each catheter and discarded immediately before initiation of therapy.
CRRT uses a blood path at slower flow rates than IHD. Development of clots within the blood path are frequently the cause of decreased or no function of the CRRT system.
Heparin sodium is the most common anticoagulant used for CRRT. Systems are frequently flushed with dilute heparin through the system during the priming procedure (5,000–10,000 U/L normal saline) followed by a constant delivery of heparin for the duration of therapy. Heparin probably can be credited with keeping systems patent for the longest periods. For many years, it was the anticoagulant of choice for all forms of dialysis that used a blood path.
As CRRT was applied to the more profoundly ill patients, heparin was found to be associated with complications caused by coagulation disorders seen in the critically ill. More and more patients every year are identified as having heparin-induced thrombocytopenia (HIT). The increased incidence of HIT is associated with more patients being exposed to intravenous heparin. 15 Frequent monitoring of coagulation studies and platelet counts as well as continual monitoring for bleeding complications is essential for any patient undergoing heparin anticoagulation of the CRRT system. Patients do not require “bolusing” with heparin before initiation of therapy, because the goal is not to anticoagulate patients but rather to provide regional anticoagulation for the system. If the heparin used for priming is not thoroughly flushed from the system, patients will still receive a small heparin bolus from the priming volume. Obviously, patients receiving systemic heparinization for reasons other than CRRT do not require additional heparinization. Most centers aim for a moderate level of anticoagulation and an activated partial thromboplastin time of 50–70 seconds or an activated clotting times of 100–200 seconds. Heparin is detected in only very low levels in the ultrafiltrate, suggesting that, despite the size of the molecule, it does not cross the filter significantly. 16
Trisodium citrate has been used for many years as an anticoagulant for blood products. It was introduced to CRRT as a regional anticoagulant in the early 1990s. It was understood that, if small concentrations of acid citrate dextrose could keep banked blood from clotting, this same concept could be used in CRRT. Arriving at concentrations of sodium citrate that would support regional anticoagulation proved to be more difficult.
Relatively normal hepatic function is needed to metabolize sodium citrate. Citrate toxcity is rare but may be seen on occasion. Citrate toxicity is characterized by a widening anion gap, metabolic acidosis (which is not related to another physiological process), and hypercalcemia. Generally, discontinuation of citrate will ameliorate the symptoms over time.
Currently, two basic methods are used to deliver citrate for regional anticoagulation of the CRRT system. In the concentrated form, 4% trisodium citrate is delivered at a fixed rate (generally 180 mL/hour) every hour with varying amounts of replacement fluid in the form of normal saline. This method was introduced by Mehta and colleagues. 17 Both the citrate and the replacement fluid are delivered prefilter to decrease the viscosity of blood entering the filter and decrease filter clotting. Palsson and Niles 18 established a different method of citrate anticoagulation using a dilute citrate solution and varying the hourly rate as the replacement fluid. The Palsson and Niles protocol combines 40 mEq/L trisodium citrate, 100 mMol/L sodium chloride, 0.75 mMol/L magnesium chloride, and 0.2% dextrose into a single replacement solution in 3,300-mL bag. Niles updated the protocol to refect a total citrate replacement fluid volume of 1.6 L/hr.
This combination of sodium preparations renders this solution essentially physiological in the concentration of sodium. Sodium, therefore, becomes another essential parameter to monitor. Citrate is among the most expensive preparations for replacement fluid. Most hospitals using citrate anticoagulation are having the solution prepared in house. Centers using the original Palsson and Niles method may find it necessary to prepare up to 48 L/day for standard CVVH. This increases the expense of therapy, mostly because of labor costs. In-house manufacturing centers have not had great success in preparing these solutions in advance and storing them for future use. Vendors are beginning to discuss the manufacture of citrate formulations that have more than a 2-week shelf life.
Citrate binds calcium; therefore, frequent intermittent monitoring of ionized calcium and continuous calcium repletion is necessary. Patients undergoing more aggressive forms of therapy (e.g., CVVHD and CVVHDF) may require additional calcium repletion if citrate is used as a replacement fluid and dialysate.
Decreasing the rate of the blood pump for patients with citrate anticoagulation or replacement fluids or a combination is essential to increase the amount of contact time the blood has with the citrate. A suggested pump speed for citrate is 80–120 mL/min. Increasing pump speeds to greater than 150 mL/min may increase clotting of the filter and venous chamber.
Nafamostat mesilate is used in Japan for regional anticoagulation of CRRT circuits. Nafamostat is a protease inhibitor, which prolongs clotting time by inhibiting various factors along the coagulation cascade as well as in the kallikrein and complement series. It has a short half-life of approximately 8 minutes, thus performing as well as a regional anticoagulant. 19
Other methods of anticoagulation are beginning to emerge. New systemic anticoagulants for CRRT include lipirudin and low-molecular-weight heparin. Prostacylin (PGE2) has a short half-life, is used as a regional anticoagulant, and may or may not be given in conjunction with heparin. PGE2 is associated with serious hypotension, which seems to be in opposition to the goal of restoring renal perfusion. 19
Bicarbonate replacement fluid may be given to the patient with reduced hepatic function or metabolic acidosis. Bicarbonate is given in concentrations of 35–40 mEq/L (Table 2). 20 It is important to note that bicarbonate replacement fluid, although capable of ameliorating some of the metabolic acidosis, is not definitive therapy for correction of acidosis from causes other than renal failure. Bicarbonate replacement fluid is generally used in conjunction with heparin as the anticoagulant, but it may be administered without anticoagulation of any sort. CRRT system lives may be significantly shorter without systemic or regional anticoagulation.
Dianeal, (electrolytes and dextrose, Baxter Healthcare, Deerfield, IL), a lactate-based solution, generally comes commercially prepared as peritoneal dialysate. It is available in 1.5%, 2.5%, and 4.5% dextrose solutions. Lactate is suitable for patients who have no acid-base or hepatic function abnormalities. Lactate is delivered as the replacement fluid with heparin anticoagulation. It is the least expensive option for replacement fluid of all of the replacement fluids because it is commercially prepared and has a significantly long shelf life (>1 year). The 1.5% Dianeal concentration is generally used. 21
The patient's nutritional analysis and plan must include the calories that the patient obtains from the dextrose in the Dianeal. Patients who are hyperglycemic because of diabetes mellitus or critical illness may require an increased insulin dose as well as modulation of their carbohydrate intake. Frequent blood glucose monitoring as well as measurements of carbon dioxide production may help prevent excessive carbohydrate administration.
Electrolyte Repletion Solutions
Calcium-free bicarbonate solutions, sodium citrate solutions, and some “home-grown” solutions require calcium replacement (Table 3). Several methods are used to achieve this goal. It is important to note that calcium gluconate infusions should not be infused into the same access as the blood path for the CRRT system, if possible, secondary to the increased risk of access clotting.
Magnesium may also need to be repleted. This is particularly true when solutions do not have additional magnesium. Some centers alternate a solution containing magnesium sulfate and calcium gluconate with solutions containing bicarbonate hourly. Usually these solutions are prepared at bedside and require an enormous amount of preparation time. Serum potassium levels also need to be monitored and replaced as necessary.
Three types of biocompatible synthetic membranes are popular for CRRT: polysulfone, polyacrylonitrile, and AN69. 1,21 These biocompatible membranes are associated with significantly lower mortality rates: more than 20% less compared with cellulose (cuprophane) in most studies. 1,22 Nonbiocompatible membranes are associated with increased catabolism and altered protein metabolism, which is in opposition to the goals of urea removal. 1,21,22
Urea clearance should determine the choice of filters. If the patient is catabolic, a filter with a greater mass transfer coefficient may improve urea clearance. Dialysis of drugs may increase with greater sieving coefficients. 23–28 Ververs and colleagues 28 studied changes required in meropenem dosing in patients on CRRT. They were able to demonstrate a significantly reduced plasma half-life of meropenem under normal dosing schedules in patients receiving CVVH. 27
Filter performance needs to be monitored. Generally, in a system that is flushed, the hollow membrane filter will be visually inspected for the accumulation of clot at the membrane ends. A better analysis of filter performance can be achieved by examining urea nitrogen levels of the ultrafiltrate and blood. Ratios of filtrate urea nitrogen compared with blood urea nitrogen greater than 80% are reflective of excellent filter performance. Depending on the urea clearance desired, filters may be changed electively when performance falls below this level. DeVriese and colleagues 29 speculated that there is greater cytokine adsorption to the filter membrane in the first 24 hours. One can postulate that this may also be secondary to the available membrane surface area in the absence of clot.
Blood pumps have advanced from jury-rigged blood and infusion pumps to entirely integrated systems. It is now possible for the technology to estimate the fluid infused and removed within a <10% rate of error. Systems have improved significantly in air detection, blood leak detection, and reflections of transmembrane pressures. Some systems actually can change modalities “on the fly” from CVVH to CVVHDF. Fluid or blood path warming devices have been included in response to the reports of hypothermia. The three most popular systems available in the United States today are Baxter BM 25 (Baxter Healthcare Corporation, McGaw Park, IL), B. Braun Diapact (Bethlehem, PA), and Gambro Prisma (Lakewood, CO) (Table 4). The European community also has some very advanced technology not available currently in the United States.
Complications associated with CRRT fall into one of two major categories: technical and clinical. The technical complications can be further stratified into those that are directly related to machine performance and those associated with vascular access. Those complications associated with machine performance are no different than for any other piece of high-technology equipment. They must be routinely checked for optimum function and repaired in a timely fashion when problems arise. Vascular access was discussed in detail previously. It cannot be overemphasized, however, that intravascular access that is less than optimal will directly affect machine performance. Poor flows from an access will result in pressure changes in the system, ultimately causing loss of the circuit from clotting. Effectiveness of therapy may be in jeopardy secondary to time lost when circuits must be replaced.
Clinical complications and tend to be related to fluid balance, electrolyte imbalances, and hypothermia. 30 Fluid-balance complications historically occurred because of miscalculation of fluid balances or inaccuracies with home-grown pump-assisted systems, resulting in dehydration or hemodynamic compromise. The newer, integrated CRRT systems have alleviated some of these issues but do not substitute for a knowledgeable clinician.
Patients requiring CRRT often have a very tenuous electrolyte balance. 31 Meticulous monitoring is key in preventing complications. Often the specific electrolyte of concern is related to the replacement or anticoagulant that is used (e.g., ionized calcium with citrate). Most centers have specific protocols for monitoring and replacing electrolytes.
Hypothermia secondary to the presence of an extracorporeal circuit can be a potential hazard for patients on CRRT. It is not uncommon to see a 2°C to 5°C decrease in temperature. To prevent this complication, either the blood or replacement fluid should be warmed. The addition of warming either the blood or the replacement fluid posed a new question to clinicians: are we masking the often first sign of infection, fever? There are several ways to deal with this issue. One such way is to serially culture blood drawn from the circuit. Our practice of daily cultures from the CVVH circuit was discontinued, however, when a review of the culture data found no positive cultures (unpublished data). Another option is to change circuits on a routine basis (i.e., every 24 or 48 hours).
Is CRRT considered to be a life-sustaining therapy? Is there a window of opportunity beyond which CRRT has no effect on outcome and may simply delay or complicate decisions regarding end of life support? Are there certain patients who should not be placed on CRRT? What should we consider a positive outcome of therapy: renal recovery, patient recovery, or some other intermediate criteria such as restoration of fluid balance? Although we have many indications for beginning therapy, what is the end point? When do we stop? The answers to such ethical and moral questions related to CRRT are under debate. 32
Zamperetti and colleagues 33 published the results of a study that examined the ethical approaches of a group of primarily European nephrologists and intensivists to patients undergoing CRRT. This study was conducted via questionnaire. Most respondents (89% of 300) reported that the role of establishing ethical criteria for managing CRRT fell within the realm of medicine rather than within the scope of an ethics committee. These ethical criteria include inclusion and exclusion for therapy, issues related to informed consent, and rationing of therapy (in the situation of equipment unavailability). Most respondents believed that ethics committee consultation could be very helpful when ethical problems interfered with decision making.
The issue of mortality is constant. Across the country, the mortality rate for patients undergoing CRRT is between 60 and 70%. In many ways, this makes sense: only the sickest of the sick have need for it. In and of itself, CRRT is not life saving. Although the therapeutic objectives for instituting CRRT may be achieved (e.g., fluid balance, control of uremia), the myriad other issues facing the critically ill patient may preclude the overall “good” outcome as traditionally defined. For example, although CRRT can temporarily ameliorate a lactic acidosis, it will not be sustained if the root cause of the problem is not addressed (i.e., intestinal ischemia).
It is difficult to draw conclusions about the therapy and its use. For as many clinicians with expertise with the therapy, there at least as many opinions about how to deal with the ethics involved, especially concerning futility of treatment. Some questions require large, randomized, multicenter trials. The ethics of such trials, especially those addressing issues such as when or whether to initiate and when to stop therapy, may preclude them from ever being conducted.
As experience with the therapy grows, so too do the questions relative to indications, techniques, technology, and outcomes. One of the larger debates is over its use for systemic inflammatory response syndrome (SIRS) and sepsis. Theoretically, inflammatory mediators such as tumor necrosis factor and interleukins are of the proper molecular size and structure to be adsorbed to the fibers of the hemofilter or ultrafiltered. If removal of cytokines from the plasma was beneficial, one might expect CRRT to reduce the clinical effects of systemic inflammatory response syndrome. Bellomo and Boyce 2 reported that CVVH using an AN69 membrane decreased plasma cytokine concentrations in 18 patients with sepsis and renal failure. They hypothesized that the extraction of cytokines may lead to the attenuation of multiorgan failure and that the early use of CVVH might actually prevent the development of the sepsis syndrome.
DeVriese and colleagues 29 examined cytokine removal using the AN69 hemofilter in 15 critically ill patients with sepsis and ARF. They measured plasma concentrations of both proinflammatory and antiinflammatory cytokines. Two filters were used with blood flows of 100 and 200 mL/min, respectively. The filters were changed after 12 hours. Cytokine removal was the highest during the first hour after the filter changes. Serum concentration of all cytokines was reduced. An increase in cardiac output and systemic vascular resistance was also noted. The authors concluded that a higher ultrafiltration rate with frequent hemofilter changes significantly reduced plasma cytokine concentrations and improved systemic hemodynamics.
Heering and coworkers, 34 however, measured tumor necrosis factor-α (TNF-α) and interleukin 1 (IL-1) in 33 patients with ARF: 18 with sepsis and 15 with cardiovascular compromise. Unlike the previously mentioned studies, this study used a polysulfone membrane. All hemodynamic parameters returned to normal within 24 hours on CVVH. TNF-α could be detected in the ultrafiltrate, but concentrations did not decrease in blood. There was no difference in plasma concentrations of IL-1 between the septic and nonseptic patients. The authors concluded that, although CVVH removed TNF-α and there was notable improvement in cardiovascular hemodynamics after 24 hours of CVVH, there was no evidence that CVVH decreased blood levels of circulating cytokines.
The limitations of these studies include sample size, type of technology, and type of hemofilter used. Because of the different configurations of hemofilters, one is unable to generalize from one filter to the next. Clearly, the question requires more study.
The technology for CRRT has developed considerably over the last several years. Therapy began with very simple machines—basically a blood pump with an air detector (Hospal BSM22, Gambro AK10)—and progressed to the second generation of blood pump, integrated air detector, and pressure transducers (Baxter bm11), and to the third generation of integrated systems for fluid management and heating as well (e.g., Baxter bm25, Gambro Prisma, B. Braun Diapact). Which technology best meets the needs for CRRT? Each system has its advantages and limitations. Important factors for study include ease of use (best determined by the nurses who manage the patient while on therapy), flexibility (e.g., how easily can a system be converted from CVVH to CVVHD or CVVHDF if necessary? Can different hemofilters be used?), quality of therapy, and cost.
What will the CRRT machine of the future look like? Clearly, none of the current machines have it all. Use of disposable cassettes and computerized screens that guide the setup and priming process (e.g., Gambro Prisma) ranks high by users. The ability to use the postfilter circuit for infusions such as antibiotics and blood products and the flexibility to change filters makes the Baxter bm25 and B. Braun Diapact attractive. With increased flexibility, however, comes increased complexity and cost. It is hoped that the machine of the future will combine all of the desired characteristics: easy to use, flexible, and cost effective. Perhaps filters will be specifically designed for cytokine removal. Perhaps the regulatory issues surrounding the use of citrate will be resolved so it can be commercially prepared, thus lowering the cost for institutions.