Chapter 23. Acute Renal Failure and Renal Replacement Therapy

In keeping with the role of providing organ support, such as managing a ventilator for pulmonary support, or pressors for cardiovascular support, it is often the intensivist's role to provide support for failing kidneys. This chapter will focus on when and how to provide that support.

Normally functioning kidneys are important in several homeostatic mechanisms:

1. The production of hormones such as erythropoietin and renin

2. Partial conjugation required to activate vitamin D that is necessary to absorb enteral calcium

3. The regulation of acid–base status along with the lungs

4. The filtration of the blood and regulation of concentrations of solutes such as sodium and potassium

5. The elimination of fluid and waste products such as urea

It is assisting or replacing these last three functions—solute clearance and volume regulation—that will be the focus of this chapter.

Depending on the population studied, the incidence of renal failure (RF) in ICU patients has been reported to be as high as 25%.1,2 There is disparity, however, in how RF is defined in clinical practice as well as in the literature. This has led to initiation of renal support at different levels of renal function, which makes it difficult to compare studies, construct studies, or extrapolate findings to one's own patient population.

The Acute Dialysis Quality Initiative Group (ADQI), a group formed in 2000 “to provide an objective, dispassionate distillation of the literature description of the current state of practice of dialysis and related therapies,”3 recently proposed a classification scheme for diagnosing acute renal failure (ARF).4 Commonly referred to as the RIFLE criteria, the acronym itself describes the level of renal dysfunction:

• R—Risk of renal dysfunction
• I—Injury to the kidney
• F—Failure of kidney function
• L—Loss of kidney function
• E—End-stage kidney disease

Each level (R-I-F-L-E) of renal dysfunction can be classified or diagnosed by either the glomerular filtration rate (GFR) or urine output (UO).

The GFR is generally considered a better measure of renal function/failure, although it is typically only measured via surrogates such as creatinine clearance. Interpreting a change in GFR requires knowledge of the baseline creatinine that is not always available. For instance, a previously healthy trauma patient who is now acutely ill may have never had his or her baseline serum creatinine measured, while a patient whose primary doctor sends a basic metabolic panel including blood urea nitrogen (BUN) and creatinine every year, or an elective surgery candidate who had preoperative labs drawn, will have a known baseline creatinine.

In the event that a baseline creatinine is not known, or for providers who are more comfortable, UO can also be used. The advantage of UO is that it can be used on all patients except for patients anuric prior to their acute illness (e.g., end-stage renal patients already on dialysis) or in whom UO cannot be measured (e.g., patients with ureteral diversions such as ureterosigmoidostomy). Furthermore, UO is recorded on ICU flow sheets and is a noninvasive or minimally invasive parameter that is easy to measure and follow.

The ADQI group released a graphical representation of the criteria, as shown in Figure 23-1.

The shape of the diagram (wider at the top) is meant to indicate a higher sensitivity (Sn). Therefore, more patients will meet these criteria, while patients who do not meet the criteria are unlikely to have RF (i.e., low risk of false negatives). As might be expected, these same criteria have a lower specificity (Sp), so many patients who meet these criteria will not have ARF (i.e., higher risk of false positives).

Similarly, the criteria at the bottom of the diagram have a lower Sn and may miss patients who do have RF (i.e., higher risk of false negatives). However, these criteria have a high Sp, so as patients progressively meet the criteria toward the bottom of the picture, they are more likely to have true RF (i.e., low risk of false positives).

As more studies use the ADQI Group's RIFLE criteria, it will likely become easier to compare studies with each other, and to identify studies whose populations are similar to one's own population, leading to better evidence-based approaches.

The indications for initiating renal support can be remembered by the mnemonic AEIOU RSI:

• A—Acidosis
• E—Electrolyte abnormality
• I—Intoxication, Ingestion, and Immune modulation (still controversial)
• O—Fluid Overload
• U—Uremia
• R—Rhabdomyolysis
• S—Sepsis (especially in multiorgan failure) (still controversial)
• I—after IV contrast (for patients with renal insufficiency [RI] or ARF)5–7

Renal support should be initiated if the patient has any of the above indications. In the case of patients who will receive IV contrast, this is particularly true if the patient has baseline RI that is resistant to treatment.

Renal replacement therapy (RRT) is a therapy designed to clear the blood of substances and volume that functioning kidneys would normally remove. The most commonly thought of substances that RRT clears are volume, potassium, and urea.

There are two broad categories of RRT: intermittent renal replacement therapy (IRRT) and continuous renal replacement therapy (CRRT). IRRT is the more familiar of these two. The most common modality used in IRRT is hemodialysis (HD), and the abbreviation often seen is intermittent hemodialysis (IHD).

In its most common form, an outpatient with chronic renal failure (CRF) goes to a dialysis center and receives HD for 3 hours, three times per week; usually Monday–Wednesday–Friday or Tuesday–Thursday– Saturday. By following a specific lifestyle, such as a renal diet and limiting fluid intake, these sessions are enough to remove solutes and fluid and thereby approximate what normal human kidneys accomplish with continuous normal blood flow. When such patients are admitted to a hospital (for renal or unrelated issues), they should receive their normal HD sessions, with adjustments made for their acute illness. Either such sessions can be run at the patient's bedside or, if the patient is stable for transport, the patient may be moved to the hospital's dialysis center.

The advantage of IHD is cost, in terms of both time and resources. IHD requires a specialized nurse, but only for 9 hours per week. The general disadvantages of IHD are the requirement of large vascular access (usually initially done with a long-term indwelling dual-lumen vascular catheter, and later via a surgically constructed arteriovenous fistula or graft, aka a “shunt”), and the need to attend a dialysis center three times per week for 3 hours. Another major disadvantage is the buildup of substances (potassium, urea) and body fluid (blood volume, edema) over 2–3 days with rapid resolution over 3 hours. These shifts are not as gentle on a patient as the continuous clearance performed by the patient's own kidneys.

Emergency physicians are acutely aware of this issue since patients with dialysis disequilibrium often present to the emergency department (ED). In addition to lesser symptoms such as headache, vertigo, and weakness, dialysis disequilibrium can cause seizure, intracerebral hemorrhage, cerebral edema, and even death.8

Among outpatients, however, not everyone is a candidate for traditional IHD. For outpatients in whom sufficient vascular access cannot be obtained (either with a long-term indwelling catheter or with construction of an arteriovenous graft), peritoneal dialysis is an option for IRRT. However, patients must be otherwise hemodynamically stable, and still urinate in sufficient quantity.

In peritoneal dialysis, a surgeon implants a port into the patient's peritoneal space. The patient, a visiting nurse, or an automated machine (at night, while the patient sleeps) adds fluid via the access port. The dialysis fluid remains in the abdomen, equilibrating with the patient's interstitial fluid. The fluid along with solutes and electrolytes (such as urea and potassium) is then drained and discarded, and new fluid is added. This must be done manually several times per day or, if at night, automatically several times overnight. The primary advantages of PD are 2-fold. First, it avoids the huge chemical and fluid shifts associated with a 2- to 3-day buildup between traditional HD sessions and removal over 3 hours. Second, it is convenient since a reliable patient can administer this therapy himself or herself, at home. The main disadvantages of PD are complications related to the catheter—specifically local infection (cellulitis) and peritonitis, which can be fatal—and cardiac and pulmonary dysfunction related to negative effects on diaphragmatic excursion secondary to the excess intra-abdominal fluid.9

For outpatients who have poor vascular access and are also not PD candidates, alternative vascular access must be sought. One approach is to place an access port to dialyze using a plexus such as the lumbar plexus.

Outpatient modalities, however, are often not appropriate for use in the ICU. Since, as previously discussed, up to 25% of ICU patients will develop ARF at some point during their illness, intensivists must be able to support the kidneys as they would support any other organ.

Whether the RF is transient and will resolve with the patient's illness or will become permanent, a patient in ARF will require some renal support. If permanent, patients may eventually require outpatient IHD as described above. But while in the acute phase of his or her illness, it is difficult to determine if a patient's renal function will recover. Furthermore, even for patients who were already on IHD, a critical illness may prevent the performance of IHD.

One reason is that during a critical illness, patients become nutritionally deficient. To maintain a patient on every other day renal dialysis, it is necessary to restrict fluid and, more importantly, protein intake. But it is precisely the protein intake that will help the patient recover. When on CRRT, patients can be fed, enterally or parenterally, without concern of urea accumulation over 2–3 days. In other words, there is no need to limit feeding for patients on CRRT while there is such a limitation for patients receiving IHD.

Due to the high blood flow volumes (remember, IHD accomplishes in 9 hours what normal kidneys do in 168 hours), there is a significant blood pressure drop when IHD is initiated. This blood pressure drop is not benign. This hemodynamic instability, however temporary, has been reported to cause loss of consciousness and myocardial infarction, and has even caused patients with partial RF (sometimes called RI) to go into complete RF.10 Furthermore, for patients with hemodynamic instability, IHD may be contraindicated. In the past, some of these patients were treated with PD, but the clearances were generally not sufficient, and in many of those cases (such as abdominal injury or infection) PD was contraindicated.11 For these patients, CRRT is the answer.

Despite the number of critical care patients who have contraindications to IHD, many feel that if a patient can tolerate it, IHD should be the preferred modality, mainly due to cost. This is false. Countless studies have shown a reduction in mortality and morbidity when patients go on CRRT.

As previously mentioned, differences in when renal replacement therapies are initiated, which modalities are used, which settings are used, and when they are halted make reviewing the literature a challenge. However, several key concepts have been proven over and over.

Kellum et al performed a meta-analysis and found that, after pooling studies that enrolled patients with similar APACHE II scores, mortality was lower in patients treated with CRRT.12 Furthermore, when adjusted for study quality and severity of illness, mortality was still noted to be lower in patients treated with CRRT. The only negative conclusion was that it was not possible to determine the optimum timing, modality, and dose of CRRT.

Perhaps the most convincing evidence was an unexpected conclusion from Jacka et al. In their study, they found that among patients who survived their critical illness, those who received continuous venovenous renal replacement therapy (CVVRRT), as opposed to IHD, were more likely to recover their renal function and not require IHD on a long-term or permanent basis.13 This conclusion was also suggested by Waldrop et al,14 though not proven since their study was not powered to detect this outcome. This study drew two conclusions: (1) that patients who suffer ARF secondary to a critical illness have such a high mortality from their inciting event that the massive mortality risk obscures any small benefit CVV might incur and (2) that perhaps studies have been evaluating the wrong endpoint and should instead be powered to study recovery of renal function after CVV versus IHD, rather than mortality benefit.

Finally, one common theme stands out among all the studies. One absolute indication for CVV over IHD is hemodynamic instability (i.e., hypotension) too severe to withstand IHD. Such a degree of hypotension is an indicator of sicker patients. It is difficult to randomize patients to IHD or CVV due to this difference and the fact that these patients are, by necessity, typically placed in the CVV groups. These studies then fail to show a difference between IHD and CVV. One interpretation of this outcome is that CVV may be demonstrating itself to be a superior modality, by achieving survival rates with sicker patients comparable to that achieved by IHD with healthier patients.

At its core, CRRT is the same as IHD. In each modality, blood is removed from the patient, “bad” substances and excess fluid are removed, and the “cleaned” blood is then returned to the patient.

In the most basic of RRT setups Figure 23-2, blood is removed from the patient at a rate called the blood flow rate, or QB. It is put through a hydrostatic pump that contains a filter with holes in it. Any solute dissolved or suspended in the fluid that is smaller than the size of the holes (such as potassium and urea) will go through. Different filters have different size holes, but generally the holes are 500–50,000 d—large enough to allow passage of fluid and solutes, but small enough to prevent loss of plasma proteins such as albumin (80,000 d). This allows fluid and small solutes to be removed without also removing plasma proteins or RBCs. The remainder of the fluid is then returned to the patient.

The fluid extracted is an ultrafiltrate (UF) of plasma. As expected, it has a mild yellow color and resembles urine. The amount of UF generated is expressed in terms of milliliters per minute. Solutes are removed via “solute drag,” that is, the solutes are dragged along with the fluid forced out of the system.

The setup in Figure 23-2 is called slow continuous ultrafiltrate (SCUF), and is the most basic of CRRT modalities. It is not used often because solute clearance is limited to the concentration in the fluid removed. All other CRRT setups, however, build on the foundation of SCUF.

When blood is returned to the patient, it is returned into a central vein. However, the blood draw line can be placed in an artery or a vein.

• If blood is drawn from an artery, this is called continuous arteriovenous renal replacement therapy (CAVRRT, usually shortened to CAVRT or CAVH where the H stands for hemofiltration [HF]).
• If blood is drawn from a vein, this is called CVVRRT (usually shortened to CVVRT, CVVH, or, more commonly, CVV).

When RRT was first developed, blood was taken from an artery. This allowed for a simpler mechanism because no pump was required. The system used the patient's blood pressure (and therefore the patient's heart) as the driving force to keep blood moving through the system. However, CAVRT had many complications, mostly due to the arterial cannulation. Tominaga et al reported arteriovenous fistulas (when an adjacent artery and vein—such as the femoral vessels—were cannulated), pseudoaneurysms, DVTs, ischemia in limbs distal to the arterial cannulation secondary to clot embolization, and persistent bleeding requiring surgical intervention.15 They noted that the rates of vascular complications with CAVRT are similar to those of the arterial access for angiograms—something not reported in CVVRT, at all—suggesting that the complication is related to the arterial cannulation itself, rather than the dwell length of the catheter. In addition to the complications reported by Tominaga et al, Bellomo et al reported shunt failures requiring revisions, infections, bleeding from the shunt itself, recurrent clotting requiring invasive intervention, and hematomas from failed arterial cannulations.16

Even more importantly, patients receiving CAVRT have shown a higher mortality compared with those receiving CVVRT.17 It is postulated that this mortality is a result of the lower clearance rates achievable with CAVRT as opposed to CVVRT. In fact, many patients placed on CAVRT required IHD for additional solute clearance.11,16–18 Finally, CAVRT requires the patient's heart to do all the work. For patients already hypotensive (one of the major reasons to put a patient on CRRT instead of IHD), this places more demand on a heart already strained, results in lower blood flow, and is a contributing factor to the lesser clearance obtainable with CAVRT.11,17

The main hindrance to CVVRT instead of CAVRT was the pump and circuit technology. Specifically, CVV setups require air detectors and bubble traps, to prevent air emboli to the lungs, and closer monitoring.15 Once technology evolved to safely include a pump, CVV became the preferred modality. In this setup, blood is drawn from a central vein, “cleaned” by the CRRT setup, and then returned to a central vein.

For CVV vascular access, a large-bore, dual-lumen catheter placed into a central vessel provides a “take blood” line and a “return blood” line with one stick. If such a catheter is not available, RRT can be administered with two independent large-bore, single-lumen, central lines. However, this doubles the risk of cannulation since two vessels would need to be cannulated instead of one.16

For patients who are dependent on other technology, it may be possible, though generally not preferred, to pull blood from that other technology. For patients on extracorporeal membrane oxygenation (ECMO), cardiac bypass, or cardiopulmonary bypass, it is possible to have the “take blood” line come off the other pump.9 This is, however, not recommended due to infectious risks, and is only to be used as a method of last resort. The blood is still returned directly to the patient, not back to the other pump.

CRRT catheters should be placed as far away as possible from other medication infusion lines. Of course, when hooking up the CRRT circuit to the central venous catheter, the intake line should be upstream from the return line, or the circuit will yield a vastly reduced efficiency.

SCUF as shown in Figure 23-2 limits solute clearance to the concentration in the patient's plasma and the volume of fluid that can be removed. However, consider Figure 23-3. In this setup, blood is removed from the patient at rate QB, just as in SCUF (Figure 23-2). However, the blood is then mixed with “good fluid” called the “substitution fluid” (SF).

This mix is then put through the same hydrostatic pump and the fluid not extracted as UF is then returned to the patient. The advantage here is that by adding the SF, the circuit can filter more volume than if relying on the QB alone.

For example, consider renal support in two identical patients. The first will be supported using SCUF as in Figure 23-2, with the QB set to be 100 mL/min. Using SCUF alone, the amount of solute removed is limited by how much fluid we can remove from the patient. If the patient is hypotensive and cannot suffer to have any plasma volume removed, no solute can be cleared at all.

For the second patient, supported as in Figure 23-3, the same QB of 100 mL/min is set, but also with SF at 2 L/h. If the UF rate is set at 2 L/h, the patient's blood can be filtered, yet the patient left isovolemic. In other words, the renal support is no longer limited by the patient's volume and hemodynamic status.

This setup is called hemofiltration, or HF. Since it is a CVV setup, it may also be referred to as CVVH or CVVHF. This chapter will use the CCVHF abbreviation to contrast it with continuous venovenous hemodialysis (CVVHD), described below.

Remember, in CVVHF, the SF mixes with the blood.

Another way to configure a CVV circuit is to use a membrane instead of a filter (see Figure 23-4).

In this case, blood is still withdrawn from the patient at rate QB. However, instead of using the pump to push the blood against a filter and to obtain a UF, the blood is pumped along a membrane, forcing some plasma across the membrane. On the other side of the membrane is a fluid that may be the same as the SF in the CVVHF setup above (Figure 23-3). However, in this setup, it is called the dialysis fluid, or dialysate, run at rate QD. The dialysate runs along its side of the membrane in a direction opposite, or countercurrent, to the blood. In addition to the small amount of solute cleared in the UF via solute drag (as in SCUF, Figure 23-2), diffusable substances such as potassium and urea flow across the membrane down their concentration gradients into the dialysate, which is then discarded.

Set up in this manner, the CVV circuit is called hemodialysis, or HD and so the full abbreviation is CVVHD.

Many critical care patients on RRT will require volume removal in addition to solute removal in order to avoid volume overload. Taking the example of the CVVHF circuit (Figure 23-3), the QB can be set to be 100 mL/min and the SF is set to normal saline running at 2 L/h. If the UF is set to be 2 L/h, solute will be removed without the removal of volume. If the UF is set to be 2.5 L/h, however, solute will still be removed but now also a net of 0.5 L/h of fluid will be removed, creating a negative fluid balance of 0.5 L/h.

Above, SF was added to the blood before the combined fluid went through the filter (called “prefilter”). The problem with this is that this method dilutes out the “bad substances” and then tries to filter the combined fluid. Consider the setup shown in Figure 23-5 instead.

In this case, QB is still 100 mL/min. But now, as opposed to the setup above, solute is filtered at full concentration. Setting UF rate equal to the SF rate clears solute while maintaining an even fluid balance. Similarly, setting UF higher than the SF clears solute and fluid.

The advantage of this setup is more HF (i.e., higher solute clearance). The disadvantage of this setup is that as more concentrated solute is pushed through the filter, the filter clogs faster. While it seems a straightforward solution to “simply change the filter,” the technical realities entail halting the system (during which time the patient is receiving no RRT), removing the old filter, and recycling and restarting the system. Error—including infection, air emboli, etc.—can be introduced in any one of these steps. An increased number of system halts increases the number of times a complication may develop and thereby increases the complication risk as the access ports and other parts of the system are manipulated.

The setup shown in Figure 23-6 is simply a combination of the previous two modalities (Figures 23-3 and 23-5). In this case, some SF is run prefilter and some is run postfilter. By running some SF prefilter, the solute is diluted, reducing the efficiency of solute clearance but prolonging the life of the filter. As discussed above, this reduces the number of times the filter must be changed and therefore reduces the number of times complications such as DVT or air emboli could occur. Conversely, running some SF postfilter allows for more efficient removal of solutes at the expense of filter life.

This combination of running the SF both prefilter and postfilter may provide the best compromise of solute clearance and filter uptime. However, this has not been well studied and there are no studies demonstrating the optimal combination of fluids (e.g., 30% prefilter and 70% postfilter). In addition, this combined approach adds another layer of complexity to an already complex setup.

As in any extracorporal circuit (cardiac bypass, ECMO), blood will clot in a CVV setup. CVV circuits, therefore, require anticoagulation. The interesting quirk of such systems is that unlike systemic anticoagulation for myocardial infarction or pulmonary embolus, the goal in this case is to anticoagulate only the circuit while not affecting the patient. In practice, the patient is always affected to a degree, but it is the goal to minimize anticoagulation of blood in the patient.

To anticoagulate the circuit, but not the patient, anticoagulant is infused after blood is removed from the patient but before entering the CRRT pump, that is, it is given prefilter. Prefilter measurements must be taken to ensure adequate anticoagulation of the CVV circuit, and postfilter measurements must be taken to ensure minimal effect on the patient or, in the case of some anticoagulants, to counteract their effects once back in the patient.

The most common anticoagulant used is heparin. To measure the effectiveness of heparin and for dosing guidance, a prefilter and postfilter partial thromboplastin time (PTT) should be checked. Figure 23-7 shows a complete CVV circuit with prefilter SF using heparin as the anticoagulant.

In reality, a postfilter PTT can be drawn from the patient rather than from the CVV return line. In fact, it may be more prudent to draw a PTT directly from the patient in order to minimize the infectious risks of the CVV circuit. Protocols should be instituted to allow the nurse running the CRRT to adjust the heparin independently based on the PTT.

There are, however, several contraindications to heparin. If a patient has or develops heparin-induced thrombocytopenia (HIT) or an allergy to heparin (including low-molecular-weight heparin), heparin can no longer be used. Similarly, if a patient is bleeding or has a risk of bleeding, such as trauma or recent surgery, other anticoagulants should be considered.

The next most common anticoagulant is trisodium citrate (TSC). Like the citrate added to stored blood to prevent clotting, the citrate in TSC chelates calcium, preventing activation of platelet adenosine diphosphate (ADP) and thereby inhibiting platelet aggregation and initiation of the clotting cascade.19

As with all anticoagulants in CRRT, the TSC should be added to the blood draw line (i.e., as it comes out of the patient) in order to ensure anticoagulation of the CRRT circuit. But the calcium level must be restored within the patient to prevent systemic anticoagulation. Systemic anticoagulation of the patient will be prevented if the patient's ionized calcium (iCa2+) level is maintained at normal levels.19 Therefore, a calcium drip must also be running.

A postfilter iCa2+ should be checked and the calcium drip adjusted accordingly. Protocols should be instituted to allow the nurse to adjust the calcium drip independently. Note that a calcium drip (usually calcium chloride) requires another central venous access line. Furthermore, calcium drips are compatible with few other medications or drips and therefore require their own, dedicated, central venous port. It is theoretically possible to infuse the calcium in the blood return line; however, since the calcium drip will be changed more often than the CRRT circuit and the indwelling catheter, this puts the patient at additional risk for complications. There are large-bore, dual-lumen catheters that contain a third smaller port. If such a catheter provides adequate blood flow for CRRT, the third port provides a convenient site for the calcium drip. Note that the third port must be sufficiently far from the blood draw port to prevent drawing the calcium drip into the CRRT circuit (which would cause clots in the circuit and fail to restore the patient's coagulation status to normal).

There are two additional disadvantages to TSC. First, for every molecule of citrate, there are three molecules of sodium, which risks hypernatremia. Utilizing a low-sodium SF can compensate to some degree, but many patients will still become hypernatremic requiring a different anticoagulant.

Second, the liver will metabolize the citrate in TSC to bicarbonate. Unchecked, this can lead to a metabolic alkalosis. While in certain situations providing an alkalinizing fluid may be beneficial, most historic indications for administering bicarbonate have been debunked and patients who develop a metabolic alkalosis from the TSC should have the TSC drip altered, the SF changed to compensate, or be switched to another anticoagulant.

There are other less commonly used anticoagulants. These include hirudin (originally isolated from leeches), argatroban, bivalirudin, and others. In fact, if a patient is not able to tolerate any anticoagulant, the circuit can even be run with high blood flows and frequent saline flushes to attempt to minimize clotting.

Special Situations Where Anticoagulation Is Not Required

There are a few instances where patients do not require separate anticoagulation for the CRRT circuit. Note that, in both of these instances, the patients are still receiving anticoagulation of some type. The only difference is they are already systemically anticoagulated and so dedicated anticoagulation of the CRRT circuit is not necessary.

Drotrecogin alfa (recombinant human activated protein C [rhAPC]) is the drug that has shown a measurable decrease in the mortality of patients with severe sepsis. Given as a 96-hour continuous IV infusion, its main risk is that of bleeding that can be significant and even terminal.20 It stands to reason that patients receiving rhAPC who also are receiving CVV do not require additional anticoagulation. Although formal studies are lacking, there is one case series of three patients receiving rhAPC and CVV who were not given any additional anticoagulation.21 Although the filters clotted earlier in patients on rhAPC versus patients on heparin, the difference was not statistically significant. Therefore, given the significant risk of bleeding during rhAPC infusion, it seems reasonable to withhold extra anticoagulation such as heparin or TSC until and unless the filter clots earlier than expected.

Once the rhAPC infusion is ended (either for rhAPC-related complications such as allergy or bleeding or at completion of the 96-hour infusion), patients must be placed back on traditional anticoagulation. If rhAPC is readministered, the anticoagulation should again be held during this second infusion, and restarted when it is complete.

There have been no dedicated studies on the topic of patients who require anticoagulation for reasons other than the CVV circuit (e.g., thromboembolism, atrial fibrillation, mechanical heart valve), but in these patients it is probably not necessary to run separate anticoagulation for the CRRT circuit. Although there have been no studies on this topic, it seems reasonable to only administer anticoagulation prefilter with the goal of achieving postfilter anticoagulation sufficient for the patient's condition. In the case of heparin, administer sufficient heparin prefilter such that the postfilter PTT is at the level required for the patient's condition. If the patient has bleeding complications from this approach and it becomes necessary to stop systemic anticoagulation, one must revert to anticoagulating the circuit in such a way as to not affect the patient.

The selection of an SF starts with some basic principles and then moves into the “art” of medicine. It is also strongly influenced by the fluids available at a particular institution.

Theoretically, as QB and SF infusion rates are raised higher and higher, in either HF or dialysis, the plasma will start to approximate those fluids. Therefore, it might seem that certain fluids should not be used. However, the patient's current condition should be taken into account when choosing an SF. For instance, if a patient is hyperkalemic, it would be prudent to start with a fluid that has no potassium in it, and then change once the patient's potassium level normalizes in order to prevent an equally dangerous hypokalemic state. Furthermore, it is known from other sources that most patients in diabetic ketoacidosis do not benefit from exogenous bicarbonate administration in the absence of cardiac effects. However, a patient with cardiac instability from a low pH secondary to a toxic ingestion might benefit from an SF with significant amounts of bicarbonate. Some commonly available SFs are listed in Table 23-1.

Three other issues are worth mentioning here:

1. A hospital pharmacist should be able to make an SF or dialysate with almost any concentration of ions such as sodium, potassium, bicarbonate, etc. That having been said, it is rather time intensive to do this for a fluid used at a rate of 2–6 L/h. It also introduces another source of possible error since it is something nonstandard that must be mixed ad hoc. It is also significantly more expensive than using fluids that are “off-the-shelf.” Custom fluids are generally used only for the rarest of circumstances where an extra degree of control is required. For instance, it might be useful in pediatrics where the smaller body weights limit the amount of fluid one can administer, regardless of the content of that fluid.

2. Consider fluids that make it easier to run the machine. For instance, Prismasate is often available in 5-L bags, while normal saline is available in only 1-L bags. Therefore, if the patient's condition will allow Prismasate, it is an easier fluid for the nurse to administer as it only requires a bag change every 5 L, rather than every 1 L. This also reduces the number of manipulations to the system's lines and therefore reduces the risk of infection.

3. Finally, fluids can be combined. So, for someone who is dangerously hyperkalemic, it might be prudent to start with Prismasate 0K in order to lower the potassium into a safer range as quickly as possible. As the potassium approaches a normal level, but is still elevated, a bag of Prismasate 0K and a bag of Prismasate 4K can be connected together with a “Y” connector, effectively giving a new fluid, Prismasate 2K. This will still lower the patient's potassium but minimize the risk of dangerous hypokalemia. When the patient's potassium normalizes, a switch to Prismasate 4K alone will maintain the potassium in the normal range. Similarly, a mixture of 0.45% NaCl with 75 mEq of NaHCO3 will yield a fluid with a sodium concentration of 152 mEq/L, chloride concentration of 77 mEq/L, and HCO3 concentration of 75 mEq/L, for a total osmolarity of 304 mOsm/L.

Although there are different filter configurations for CVV, because of cost considerations, generally an institution will have only one or two filters available. In the case that a different filter is desired, or even being considered, a renal consultation can be very helpful with some of the less common configurations. One example might be the use of a charcoal filter for patients with an ingestion or overdose.

Although there are no studies examining minimum required labs for patients while on CVV, it seems prudent to send blood every 6 hours for sodium, potassium, chloride, bicarbonate, urea, creatinine, glucose, calcium, magnesium, and phosphorous. Appropriate anticoagulation labs should also be obtained (prefilter and postfilter PTTs for patients receiving heparin and prefilter and postfilter iCa2+ levels for patients receiving TSC). Arterial blood gases should be sent at regular intervals, or more often as the patient's status or other treatments dictates. Furthermore, bleeding times, activated clotting time, and thromboelastograms may be useful in specific circumstances.

For maximum clearance, CVVHF and CVVHD can be run simultaneously. As a combination of CVVHF and CVVHD, this setup is termed continuous venovenous hemodiafiltration (CVVHDF). This setup combines the advantages of convective clearance with diffusive clearance, giving clearance rates superior to either modality alone. Such a setup, complete with prefilter and postfilter SF, anticoagulation with heparin, and prefilter and postfilter PTTs, is pictured in Figure 23-8.

Note that covering everything above the dotted line leaves a CVVHF circuit as pictured in Figure 23-7. Eliminating the prefilter and postfilter SFs and additional UF leaves the CVVHD circuit pictured in Figure 23-4.

Table 23-2 shows the relative clearances of CVVHF versus CVVHD versus CVVHDF. As you can see, CVVHDF results in higher clearance rates, but not exponentially so.

Just as there is a relative lack of strong evidence-based support for when and how to initiate CRRT, there is a lack of evidence guiding when to withdraw renal support. Uchino et al reported on current practices of 54 ICUs across 23 countries. While not a prospectively derived treatment rule, they found that, by far, UO was the best predictor of successfully weaning renal support, that is, the best indicator of recovery of renal function.22,23 The next best indicator was creatinine clearance, but its predictive power was far inferior to that of UO.

Note that some machines have different terminology for the various settings. Some machines use the terms QB, SF, and UF as described above. But one manufacturer has set the machine such that UF will automatically equal the SF. That means if the machine is set to add SF at 2 L/h, the UF will automatically be 2 L/h. This may seem confusing but it is actually a safety measure. If you do not set the UF (or if you leave it at UF = 0), the machine will automatically keep the patient isovolemic. If you want to make the patient negative, you would then set a UF >0. However, the same machine will not account for any volume from the anticoagulation fluid. Depending on the fluid used and the concentration, especially for TSC, the volume of that fluid may be significant.

The take home point is that familiarity with the machine, or machines, in use is of vital importance. This includes knowing which settings are in milliliters per minute, and which are in liters per hour. Standardized forms greatly aid this effort.

There needs to be a strong working relationship between the physicians and nurses so everyone is using the same vocabulary. All physicians who will be evaluating or writing recommendations for the therapy must be using the same terminology. This should include the intensivist, the renal consulting service, and even consulting cardiologists who will need to know the details of the patient's fluid management.

In-services should be run to educate staff on which machines are in use, how to set them up, and how to troubleshoot them. This should be required for all nurses and doctors directly involved in the running of the therapy (intensivist, nephrologist) and should be made available to all others who may need to understand the therapy (e.g., cardiologist). Online modules greatly aid in this effort and, in fact, may completely suffice for the latter group.

Medication dosing for patients on CVV is beyond the scope of this text. The calculations are complex, and not always completely studied. To calculate appropriate dosing, one must take into consideration the clearance of the substance that is a function of the following items:24

• Volume of distribution of the medication, and whether a loading dose is required
• Whether a certain medicine is cleared more by diffusion (dialysis) or convection (HF)
• What modality the patient is on: CVVHF, CVVHD, and CVVHDF
• If on CVVHF, whether SF is prefilter, postfilter, or both
• Filtration fraction (how much plasma is filtered across the membrane given a certain QB)
• The sieving coefficient (how easily the medicine crosses the membrane)

Given these factors, medication dosing for these patients should involve close consultation with the pharmacology service.

The evaluation of fever in patients requiring RRT is complicated by the fact that all of the CVV modalities involve an extracorporal circuit. The CVV machine itself has a heater to keep the blood warm, but the blood must still pass through tubing exposed to the environment. This means that the blood tubing “bleeds” heat from the patient. This heat loss may make a normothermic patient hypothermic. More importantly, by clearing middle molecules, such as the cytokines that produce fever, CVVHF and CVVHDF (but not CVVHD) may mask a normothermic patient developing a fever. It would seem prudent, then, to lower the threshold for considering a temperature elevation as a true fever, for instance, 100°F (37.8°C) instead of the more traditional 100.4°F (38°C). It would also seem prudent to send routine blood cultures at some predefined interval based on the patient's condition or local practice. There are currently no studies that define the optimal interval for routine blood cultures.

For patients with severe sepsis, CVVHF holds promise as a therapy to modulate the immune response. Specifically, HF, in clearing middle molecules, also clears the cytokines that have been implicated in the harmful immune response of sepsis. In contradistinction to solutes and fluid volume that are cleared by convection, cytokines are cleared by adsorbing to the CVVHF filter. It can be shown that CVVHF clears cytokines because when weighing a filter after being used for CVVHF in a patient with sepsis, the filter weighs more. However, since cytokines exert their influence at the tissue level, and since the filter does not remove all cytokines, it remains to be shown if clearing the cytokines truly modulates the septic response.25,26 Some also argue that in clearing such cytokines, CVVHF also clears the downregulating or beneficial cytokines. While true, the balance in a septic patient seems to be more toward the more harmful cytokines. Until filters can selectively remove the harmful cytokines, nonselective removal of cytokines seems a reasonable therapy. To achieve the clearance rates necessary to clear cytokines in septic shock, CVVHF must be run with high SF rates, and the filter must be changed every 6 hours, which is when the adsorption capability of the filter declines. At that point, the filter will still clear solute, but will not remove additional cytokines. This remains one of the most promising therapies for the future treatment of septic shock.

CRRT, as “the kindler, gentler” RRT, will continue to evolve. Unlike traditional dialysis, CVV does not require a water source, and so can be implemented not only in patients not candidates for IHD but also in locations not set up to handle IHD.

As critical care moves closer to the front lines of medical care, there may be instances where it would be beneficial to start this therapy in the ED. For emergency physicians to run this therapy, they must be familiar with its requirements in order to facilitate its initiation in the ED. This may include placing a large-bore, dual-lumen central line, ordering the CVV machine, ordering the necessary fluids (SF, anticoagulation, etc.), and rearranging staff (or calling in additional staff) to free up a nurse capable of running the CVV machine.

Furthermore, “liver dialysis” is starting to become a technically feasible therapy. While not likely to be an ED therapy, it holds out great promise for hepatic support for any of a number of diseases, including overdoses, “liver help” during a critical illness, and as a bridge to a liver transplant. Its development will naturally draw heavily from the RRT world, and those already equipped to manage RRT will be in the best position to manage hepatic replacement therapy.