Fluid Management during Continuous Renal Replacement Therapy: A Case-Based Approach

Continuous renal replacement therapy (CRRT) is frequently used to provide kidney support for hemodynamically unstable critically ill patients with acute or chronic kidney failure in intensive care units (ICU) [1‒3]. While utilization and technology have favorably evolved, significant practice heterogeneity in the delivery of CRRT has also been observed. One of the areas of CRRT delivery with recognized practice heterogeneity is fluid management [4‒6].

Given that fluid overload has been consistently associated in a dose-response manner with increased mortality and multiorgan dysfunction during acute kidney injury among critical illness [7, 8], net fluid removal (i.e., net ultrafiltration [UF_NET]) represents a critical aspect of fluid management during CRRT. Further, the impact of fluid overload during acute illness could expand toward the post-ICU recovery phase affecting kidney recovery [9, 10]. While observational data have suggested a “U-” or “J”-shaped relationship between UF_NET and mortality [11‒13], the optimal method of fluid management during CRRT is still unclear and may require patient-level individualization. In this mini-review, we discuss current operational definitions and provide two different approaches to fluid management based on recent evidence using a case-based approach.

A 62-year-old female patient with a past medical history of cirrhosis secondary to nonalcoholic steatohepatitis, congestive heart failure, status post prosthetic mitral valve replacement, pulmonary hypertension, and end-stage kidney disease was admitted to the ICU with decompensated liver disease and respiratory failure requiring mechanical ventilation. The patient’s caregiver reported that she has been having a low tolerance to fluid removal during her outpatient hemodialysis sessions for the last 2 weeks despite the use of midodrine before the initiation of hemodialysis. The patient was active on the liver-kidney transplant list before this admission.

Upon arrival to the ICU, her heart rate was 96 beats per min, blood pressure was 90/65 mm Hg requiring norepinephrine 0.2 μg/kg/min and vasopressin 0.03 units/min. The patient was sedated, intubated, and started on assist-control ventilation (tidal volume 350 mL, positive end-expiratory pressure of 8 mm Hg, respiratory rate of 16, and FiO₂ of 0.8). The patient was noted synchronous with the ventilator; her temperature was 36.2 C. Given that her estimated dry weight was 80 kg, and upon her ICU admission, the weight had increased to 100 kg, her fluid overload was estimated to be 25% [14]. The patient was also receiving 70 mL/h of intravenous carrier fluids for delivering various medications, including propofol, fentanyl, norepinephrine, vasopressin, and antibiotics. Findings of subsequent transthoracic echocardiography and right heart catheterization are presented in Table 1.

This female patient with end-stage kidney and liver disease has severe fluid overload due to right and left heart dysfunction and pulmonary hypertension. Her right heart catheterization suggests mitral regurgitation with severe pulmonary hypertension characterized by high pulmonary capillary occlusion pressure (i.e., wedge pressure) and a V wave. Her high cardiac output and low systemic vascular resistance suggest vasodilatory shock, most likely due to liver failure. However, other causes for vasodilatory shock such as sepsis and adrenal failure cannot be ruled out. Her IVC collapsibility of <50% on transthoracic echocardiography suggests elevated right heart filling pressures and fluid overload. Her FiO₂ requirement of 80% suggests pulmonary edema and impaired oxygenation requiring mechanical ventilation. Urgent net fluid removal with CRRT is indicated due to severe fluid overload, hemodynamic instability, and right heart dysfunction.

Fluid management during CRRT requires careful consideration of the patient’s clinical and hemodynamic status to optimize tissue perfusion and allow time for organ recovery. CRRT provides a unique platform for dynamic adjustments according to patient’s volume status to complement concomitant use of vasoactive drugs with fluid repletion or removal to achieve hemodynamic stability [15]. Prescription and delivery of fluids during CRRT is based on a systematic approach to define the goals of therapy over a set period of time, usually 24 h, with frequent evaluations and adjustments. How much fluid is to be given or removed over the set time is established initially considering the underlying pathophysiology and the anticipated adjustments in vasopressors and inotropes. Consequently, the desired patient’s fluid balance (FB) represents the difference in all fluids given and those removed from the patient within and outside the CRRT system and can be expressed as positive, zero (i.e., net even), or negative over the designated time interval. The continuous evaluation of the patient’s FB requires documentation of all the CRRT machine intakes and outputs and those outside the system in flowsheets preferably at hourly intervals to compute the net ultrafiltration rate (UF_NET). UF_NET rate is the net fluid removal rate from the patient, which is the hourly difference between the patient’s intravenous fluid intake and the machine FB. One should note that UF_NET rate is not the same as machine set fluid removal rate. Optimally, a target patient’s FB is selected over a designated time interval and is achieved through dynamic adjustments of the UF_NET. It is important to recognize that insensible losses are usually not objectively captured and may be substantial in burn patients. Additionally, the precision of UF_NET will depend on the completeness and accuracy of documentation at frequent intervals.

UF_NET can be precisely dosed based on the patient’s body weight. For the aforementioned patient, 80 kg was the estimated dry weight that was used in all calculations. It is important to note that ICU admission weight or the most updated weight at the time of CRRT evaluation is sometimes used when the estimated dry weight is not known. Using weight-based dosing of UF_NET rate, one can start fluid removal at 1 mL/kg/h (i.e., 80 mL/h) with close monitoring of hemodynamics. Fluid removal can then be gradually increased at 0.5 mL/kg/h (i.e., 40 mL every h) up to 2 mL/kg/h (i.e., 160 mL/h), as tolerated. When calculating the UF_NET rate, it is important to account for any intravenous fluid infusions the patient receives. This is because the UF_NET rate represents the amount of fluid that will be removed from the circulating intravascular volume in addition to any intravenous fluids being infused simultaneously in the patient. For example, in this patient receiving 70 mL/h of intravenous fluids, the correct UF_NET rate dosing of 1 mL/kg/h would be 150 mL/h (80 mL/h + 70 mL/h). The same principle applies when increasing the UF_NET rate at 0.5 mL/kg/h to a 2.0 mL/kg/h target. In this case, the correct UF_NET rate increase would be 110 mL/h (40 mL/h + 70 mL/h) to a target of 230 mL/h (160 mL/h + 70 mL/h). A conceptual map of UF_NET prescription is presented in Figure 1.

Recent observational studies indicate that UF_NET rates >1.75 mL/kg/h are associated with an increased risk of death, decreased renal recovery, and hemodynamic instability compared with UF_NET rates between 1.01 and 1.75 mL/kg/h during CRRT [10, 12, 13]. Thus, randomized clinical trials are urgently needed to determine the optimal approach to UF_NET dosing that is safe and effective. Moreover, how vasopressors should be managed during net fluid removal is also unclear. In this patient with severe right heart dysfunction, fluid removal may result in improved hemodynamics and less vasopressor requirement. Thus, a cautious approach, including close monitoring of hemodynamics, vasopressor dose, and the UF_NET rate, is warranted until randomized clinical trials (NCT05306964) show optimal dosing of UF_NET. Slow and steady net fluid removal sustained for a longer duration will likely result in better hemodynamic tolerance.

Fluid management during CRRT requires dedicated flowsheets to capture the prescription and metrics of deliverables. When referring to CRRT fluid management, specific parameters should be considered: (1) UF_NET (previously described); (2) patient FB goal or FB_GOAL; (3) patient FB achieved or FB_ACHIEVED; and (4) gap of patient FB achieved versus goal or FB_GAP (%).

A conceptual map of these parameters is presented in Figure 1. One should note that this formula is implausible when fluid removal is not desired (nil).

A recent observational study found an independent association between higher %FB_GAP and increased risk of hospital mortality in critically ill adult patients with acute kidney injury on CRRT in whom clinicians prescribed net negative FB goals of 0.5 L/day on average during CRRT. Interestingly, this study also underpinned that underachievement of patient’s FB goals during CRRT is very frequent in clinical practice. Specifically, the median %FB_GAP in this study was 90.5% with less than 10% of patients reaching the target of %FB_GAP less than 20% proposed by the Acute Disease Quality Initiative [17]. Therefore, identification of what causes %FB_GAP is direly important to develop interventions that narrow this gap. When evaluating potential causes of %FB_GAP, one should consider the broad spectrum of possibilities related to inadequate prescription or ineffective delivery of the prescription (e.g., interrupted CRRT due to access issues, clotting, machine malfunction, etc.), as well as patient intolerance, among other factors [18]. Collectively, the critical care nephrology community should recognize that similar to the % gap between prescribed versus achieved total effluent dose of CRRT [19], %FB_GAP is a relevant CRRT performance indicator that should be monitored during CRRT [20, 21].

The initial course of CRRT fluid removal was inconsistent given patient’s hemodynamic instability and concerns of superimposed sepsis, which was ruled out upon further work-up (daily %FB_GAP ∼50%) (Fig. 1, days 1–3). Then, the clinical team found the “sweet spot” of UF_NET ∼1.0 mL/kg/h that yielded net negative daily (24 h) FBs ∼2 L (Fig. 1, days 4–6). After several days of multiorgan support, including CRRT, the patient was back to her estimated dry weight of 80 kg and was liberated from pressors and mechanical ventilation (Fig. 2). The patient was later transitioned back to intermittent hemodialysis and received intensive physical rehabilitation in preparation for liver and kidney transplantation.

The optimal method of fluid management during CRRT is still unclear. While various methods of net fluid removal exist, such as based on patient FB, FB achieved versus goal (gap), and weight-based dosing of UF_NET, it is important to understand that all of these methods eventually impact the rate of net fluid removal (i.e., UF_NET rate). Therefore, they should be seen as complementary approaches for restoring fluid homeostasis in the critically ill patient. Both slower and faster rates of net fluid removal have benefits and risks. For example, slower net fluid removal may be better tolerated hemodynamically. However, it increases exposure to fluid overload and organ edema, which is associated with an increased risk of death. Whereas faster net fluid removal may aid in the rapid achievement of negative FB and treatment of fluid overload, it also increases the risk of intradialytic hypotension and ischemic organ injury. Thus, until randomized clinical trials are conducted, CRRT fluid management must be individualized to patient needs based on the risk-benefit assessment of treating organ congestion, fluid overload, and hemodynamic tolerance. The evolving utilization of non-invasive hemodynamic monitors and imaging tools to evaluate patients’ fluid status and tolerance to fluid removal could further assist clinicians in individualizing CRRT fluid management.

J.A.N. is supported by grants from NIDDK (R01DK128208, R01DK133539, U01DK12998, and P30 DK079337). R.M. is supported by grants from NIDDK (R01DK128100, R01DK131586, RFA-DK-16-026, R01DK106256, and R01DK121730). R.L.M. is supported by grants from NIDDK (P30 DK079337).

J.A.N. has received consulting fees from Baxter and Outset Medical. R.M. has received consulting fees from Baxter Inc., AM Pharma Inc., Bioporto Inc., and La Jolla Inc.

None.

Javier A. Neyra, Ravindra L. Mehta, and Raghavan Murugan contributed equally to the conceptualization and writing of this article.

This work was conducted at each of the affiliated institutions.