Acute kidney injury (AKI) is commonly seen in critically ill patients, with an estimated incidence ranging from 10% to 30% based on acute kidney injury network (AKIN) or RIFLE (risk, injury, failure, loss, end-stage kidney disease) criteria.1–4 It is well-known that only a minimal increase in serum creatinine will significantly increase mortality.5–7 Although there are a number of strategies existing for the prevention and treatment of AKI, renal replacement therapy remains the mainstay treatment of AKI, as well as its attendant complications such as electrolyte disturbance, fluid accumulation, and acid-base imbalance.8 However, it is still largely unknown whether an intensive or less-intensive dose will benefit patients.9 The past decades have witnessed a rapid development in this controversial area. Ronco et al.10 first demonstrated that intensive continuous renal replacement therapy (CRRT) dose would benefit patients, and a dose >35 ml/kg/hr has been recommended. However, this view was subsequently challenged by the publication of several large trials.11–13 All these later trials consistently show a neutral effect of intensive dose on patient outcomes. A meta-analysis by our group also confirms that increasing effluent volume to >25 ml/kg/hr provides no additional benefits.14
Although the definition of “dose” for CRRT has long been debated,15 most of the previously mentioned trials use effluent flow rate to estimate CRRT dose based on the assumption that small solutes between both sides of the filter membrane are balanced rapidly and the sieving coefficient equals one. This assumption may hold true for a properly functioning filter. In clinical setting however, the filter is soaked in plasma, and many factors such as clotting and concentration polarization will contribute to the impairment of filter function. Sieving coefficient will be reduced to less than one if small solutes cannot be balanced between both sides of the filter membrane. In such circumstance, the prescribed dose represented by effluent flow rate will no longer equal the actually delivered dose.16 Several pilot studies show that the delivered dose is significantly lower than that being prescribed, and this gap can be enlarged over time.17,18 However, because of their small sample sizes and single center nature, the results are waiting to be validated. The current study aimed to examine 1) the gap between prescribed (K) and delivered clearance (Kx), as well as its change over time; 2) the correlation between transmembrane pressure (TMP) and the reduction in delivered dose. We hypothesized that the difference between K and Kx would enlarge progressively over time, and the reduction in delivered clearance positively correlated to TMP.
The study was conducted in a 24-bed intensive care unit (ICU) of a tertiary academic medical center from August 2011 to March 2013. The study was approved by the ethics committee of Jinhua Municipal Central Hospital. Informed consent was obtained from the patient or their next of kin. Patients were deemed eligible if they were admitted to the ICU and received CRRT. Indications of CRRT included 1) fluid overload resulting in pulmonary edema that was refractory to diuretics; 2) severe metabolic acidosis (pH < 7.0) resulting from renal failure; 3) hyperkalemia (>6.5 mmol/L) that was not responsive to medical treatment; 4) intoxication by dialyzable drugs including lithium, theophylline, and myoglobin; 5) uncontrolled electrolyte disturbances such as hyponatremia; 6) uremia (urea >30 mmol/L); 7) sepsis or severe pancreatitis for the removal of cytokines mediating systemic inflammatory response syndrome. Patients were excluded if 1) they were expected to stay in ICU for <48 hours; 2) the informed consent could not be obtained; and 3) the patient or their next-of-kin signed do-not-resuscitation order.
Performance of CRRT
Continuous venovenous hemofiltration (CVVH) was performed in a mixed pre- and postdilution mode. Detailed protocol for performing CRRT has been described elsewhere.19 Systemic heparin anticoagulation was used to maintain circuit patency. Principally, we routinely delivered a dose of 25 ml/kg/hr for the usual patients; but for the clearance of cytokines such as in the situation of sepsis and pancreatitis, we delivered a dose of >30 ml/kg/hr. The performance of CRRT was done by experienced critical care nurses. Renaflo II hemofilter (HF1200; Minntech Corporation, Minneapolis, MN, glycerin-free polysulfone) and Baxter machine were used. Hemofilter was changed routinely every 72 hours according to the instructions provided by the manufacturer.
Blood and effluent samples were collected at 4, 10, 16, 28, 40, 52, and 64 hours after the installation of a new hemofilter. Creatinine concentrations in serum and filtrate were measured simultaneously to determine the actually delivered clearance (Kx). Urea concentrations in filtrate and serum were measured for the determination of filter function. Data on patients’ baseline characteristics including age, sex, acute physiology, and chronic health evaluation (APACHE) II score, causes of AKI, cumulative fluid balance before CRRT, hours in ICU before CRRT, and 24 hour urine output were collected. For each filter, we collected data on life span and the reasons for filter failure.
Equations Used in the Study
Prescribed clearance is adapted from the study by Claure-Del Granado et al.17 and Clark et al.20:
where K is the prescribed clearance, Qr is the replacement fluid rate, Qf is the ultrafiltration rate (fluid removal rate), Hct is the hematocrit, Qb is the blood flow rate, and fpre is the fraction of predilution rate in relation to total replacement fluid rate, and it ranges between 0 (100% postdilution) and 1 (100% predilution). Because creatinine is used as a marker in our study, the diffusion volume flow rate is close to plasma water flow and adjustment for Qb was made.
In postdilution, actually delivered HF clearance is given as:
where Kpost is the postdilution clearance, and σ is the sieving coefficient. In predilution, HF clearance for a solute removed from plasma water is given as:
, where Kpre is predilution clearance. The equation differs from the postdilution clearance by a factor of
which is always less than 1 and greater than 0. However, σ was not used in current analysis, but instead creatinine concentrations in filtrate and serum were measured and their ratio was used to account for the predilution effect:
where Kx is the delivered clearance, Cf is the creatinine concentration in the effluent fluid, and Cs is the creatinine concentration in the serum. When only postdilution is used, the ratio
Clearance reduction is the difference between K and Kx and was calculated using the following equation:
where ΔK is clearance reduction (%).
To investigate the robustness of the result, Cs was adjusted by fluid balance according to the following equation21:
where Ca is adjusted serum creatinine, Cs is measured serum creatinine, Wadmin is body weight measured on admission, and Fcumalative is the cumulative fluid balance.
Filter function was expressed as follows:
where F is filter function, Uf is urea level in filtrate, and Us is the serum urea level.
Continuous variables were expressed as mean and standard deviation or median and interquartile range (IQR) as appropriate. Categorical variables were expressed as the number (n) and frequency (%). The mean differences between prescribed and delivered clearance at each time point were compared using paired Student t-test. The results were presented with box plot. One-way analysis of variance was used to test the differences of the clearance reductions over the treatment period. The correlation between TMP and clearance reduction was tested using Spearman’s test. Two-tailed p < 0.05 was considered statistically significant. All statistical analyses were performed using the software Stata 11.0 (StataCorp, College Station, TX).
Demographic and clinical characteristics of included patients are shown in Table 1. A total of 60 consecutive patients were enrolled in our analysis. The median age was 60 years (IQR: 46–71 years). There were more men than women (38/22). The causes of AKI in descending order were sepsis (58%), acute-on-chronic AKI (20%), trauma (18%), and drug (4%). There were 36 medical and 24 surgical patients in our cohort. The median APACHE II score was 31 (IQR: 25–34). Cumulative fluid balance before the initiation of CRRT was 3,411 ml (IQR: 1,491–5,674 ml). The median length of stay in ICU before the start of CRRT was 23 hours (IQR: 10–47 hours). The median number of filters for each patient was 4 (IQR: 3–6). The median 24 hour urine output was 97 ml (IQR: 52–220 ml). The mean filter life span was 37.7 hours. The reasons for filter failure in descending order were clotting (58%), catheter malfunction (29%), cessation of CRRT due to various reasons (including surgical procedure, death, radiographic examination) (10%), and others (3%). For overall treatment, the prescribed dose was 37.34 ± 2.16 ml/min; dose calculated from the measured effluent rate normalized for effective treatment time was 26.01 ± 8.23 ml/min and actual delivered dose was 23.26 ± 8.02 ml/min.
The differences between K and Kx at different time points are shown in Table 2. Kx was significantly lower than K across treatment period, and the differences enlarged progressively from 1.32 ml/min (n = 248; 95% confidence interval [CI]: 0.98–1.66 ml/min) at 4 hours to 10.47 ml/min (n = 28; 95% CI: 8.94–12.01 ml/min) at 64 hours. On average, K overestimated Kx by 9.3% (95% CI: −4.4% to 32.3%). Ka was significantly higher than unadjusted Kx at most time points (p < 0.05 at time 4, 10, 16, 28, and 40 hours). The filter function declined progressively over time (Figure 1). Figure 2 shows ΔK at different time points over filter life span. The reduction was significantly increased over time.
Figure 3 shows the correlation between TMP and ΔK. A total of 1,003 measurements were obtained during study period. Transmembrane pressure was significantly correlated to ΔK with a Spearman’s rho of 0.44 (p < 0.001).
The current study demonstrated that K significantly overestimated Kx, and the difference enlarged progressively over time. Furthermore, our study for the first time showed that ΔK was positively correlated to TMP, indicating that the reduction in Kx was partly because of filter clotting. Our study confirmed previous findings by Claure-Del Granado et al.17 However, in their study, K overestimated Kx by 26%, which was significantly higher than that of 9.3% in our study. This is probably because filters are not routinely changed within 72 hours, and 68 filters are still working after day 3 in Claure-Del Granado’s study. In contrast, we have routinely changed filters on day 3 even if they were still functioning well (e.g., free of clotting formation, no alarming because of increasing TMP) at that time. We proposed that the sieving coefficient of filter might be further reduced after day 3, and thus, the difference between K and Kx would be more pronounced. Second, Claure-Del Granado’s study used regional citrate anticoagulation to maintain circuit patency, which was more effective in prolonging filter life span when compared with the systemic heparin anticoagulation that have been used in our study. It has been well established that regional citrate anticoagulation is associated with longer filter life span than heparin anticoagulation.22 However, this prolonged life span may not necessarily translate to better clearance efficacy. In this regard, studies comparing different anticoagulation strategies should also take this into consideration. A more recent study by Lyndon et al.23 showed that while there was no difference between the measured total effluent volume (TEV) dose and actual delivered creatinine clearances in standard dose (20 ml/kg/hr) group, measured TEV dose differed significantly from delivered creatinine clearance by 13.9% (p < 0.001) in the high-dose group (35 ml/kg/hr). This is in contrast to our finding that the difference is still statistically significant in standard dose group. The conflicting results cannot be fully explained based on current evidence. However, an important difference between the two studies lies in the difference of CRRT mode (CVVH vs. CVVHDF). It remains to be investigated whether such difference in CRRT mode will lead to discrepancy in clearance reduction.
By using urea as a surrogate marker of small solute, Ricci et al.18 found that the estimated clearance was significantly correlated to delivered clearance, and their difference was not statistically significant within 24 hours. This finding is in contrast with the present result that Kx deviated from K at as early as 10 hours. The possible explanation is that the molecular weight of creatinine is twice lager than urea, and solute with larger molecular weight is more likely to be blocked by clotted filter membrane. This notion has been supported by an in vitro experimental study showing that solute clearance for β2-microglobulin is much lower than that for urea or creatinine.24 Also in another study by Brocklehurst et al.,25 they demonstrated that filter malfunction had significant impact on creatinine clearance but not on urea clearance. Our study showed that K significantly overestimated Kx even after normalizing for effective treatment time (26.01 ± 8.23 vs. 23.26 ± 8.02, p < 0.05). Except for the declining filter function over time, another possible explanation for the significant underestimation is that femoral artery is preferred catheterization site in our institution. One randomized controlled trial shows that the risk of circuit dysfunction is higher in patients using femoral site than those using jugular position.26
Interestingly, the current study shows that some measured Kx is higher than K, which theoretically seems impossible. Such results have also been noticed in the study by Ricci et al.18 As described in equations (1) and (3), if Kx > K, the σ will be >1. One explanation is the laboratory measurement error. On the contrary, the proposed hypothesis of single-pool distribution of solute may not hold true in critically ill patients.27 Creatinine slowly equilibrates across the red blood cell membrane, and predilution will affect the creatinine gradient and re-equilibration of creatinine between plasma and erythrocytes. It is presumed that if the creatinine concentration within blood cells is higher than that in the plasma and this has not been equilibrated when the blood enters the circuit, then the concentration gradient is further enlarged by the use of predilution replacement fluid. Because of the large concentration gradient, the creatinine within blood cells will diffuse into the plasma, and then cross the filter membrane to the ultrafiltrate, making the measured value of Cf in equation (4) higher than expected. However, this is a hypothesis that requires further investigation.
Transmembrane pressure refers to the average applied pressure from the feed to the filtrate side of the membrane. The increase in TMP is thought to be associated with clotting formation, concentration polarization, and membrane fouling.28 However, the impact of high TMP on circuit function (its clearance on small solute) has never been systemically investigated. The current study for the first time investigated the relationship between TMP and ΔK. The result showed that TMP was significantly correlated to ΔK. Because Kx is not routinely monitored in many institutions, TMP can be used to monitor the efficacy of the filter membrane. If TMP remains persistently high, the filter should be exchanged or the effluent volume should be increased to compensate for the reduction in actual delivered dose. However, many other factors may contribute to the increase in TMP. Increased TMP cannot be directly translated to reduction in Kx, as implied by the relatively low correlation coefficient (Spearman’s rho = 0.44). In some cases, the sieving coefficient may well be equal to one even when TMP >300 mm Hg.
In aggregate, our study demonstrates that K overestimated Kx over the entire treatment period, and the difference increases progressively over time. In the current practice, most institutions use the CRRT dose at an effluent flow rate of 25 ml/kg/hr; this rate may need to be increased with the prolonged use of a filter based on our findings. Furthermore, monitoring of TMP can provide useful information on the clearance efficacy of the filter. Based on current evidence, we recommend that filters should be changed at 48–72 hours on a routine basis. Because of the significant difference between K and Kx, further clinical trials on CRRT dose should incorporate data on actually delivered dose.
1. Paudel MS, Wig N, Mahajan S, Pandey RM, Guleria R, Sharma SK. A study of incidence of AKI in critically ill patients. Ren Fail. 2012;34:1217–1222
2. Yunos NM, Bellomo R, Hegarty C, Story D, Ho L, Bailey M. Association between a chloride-liberal vs chloride-restrictive intravenous fluid administration strategy and kidney injury in critically ill adults. JAMA. 2012;308:1566–1572
3. Odutayo A, Adhikari NK, Barton J, et al. Epidemiology of acute kidney injury in Canadian critical care units: A prospective cohort study. Can J Anaesth. 2012;59:934–942
4. Schneider AG, Eastwood GM, Seevanayagam S, Matalanis G, Bellomo R. A risk, injury, failure, loss, and end-stage renal failure score-based trigger for renal replacement therapy and survival after cardiac surgery. J Crit Care. 2012;27:488–495
5. Levy MM, Macias WL, Vincent JL, et al. Early changes in organ function predict eventual survival in severe sepsis. Crit Care Med. 2005;33:2194–2201
6. Chertow GM, Burdick E, Honour M, Bonventre JV, Bates DW. Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. J Am Soc Nephrol. 2005;16:3365–3370
7. Hoste EA, Clermont G, Kersten A, et al. RIFLE criteria for acute kidney injury are associated with hospital mortality in critically ill patients: A cohort analysis. Crit Care. 2006;10:R73
8. Fleming GM. Renal replacement therapy review: past, present and future. Organogenesis. 2011;7:2–12
9. Prowle JR, Schneider A, Bellomo R. Optimal dose of continuous renal replacement therapy
in acute kidney injury. Crit Care. 2011;15:207
10. Ronco C, Bellomo R, Homel P, et al. Effects of different doses in continuous veno-venous haemofiltration on outcomes of acute renal failure: A prospective randomised trial. Lancet. 2000;356:26–30
11. Zhang P, Yang Y, Lv R, Zhang Y, Xie W, Chen J. Effect of the intensity of continuous renal replacement therapy
in patients with sepsis and acute kidney injury: A single-center randomized clinical trial. Nephrol Dial Transplant. 2012;27:967–973
12. Bellomo R, Cass A, Cole L, et al. Intensity of continuous renal-replacement therapy in critically ill patients. N Engl J Med. 2009;361:1627–1638
13. Palevsky PM, Zhang JH, O’Connor TZ, et al. Intensity of renal support in critically ill patients with acute kidney injury. N Engl J Med. 2008;359:7–20
14. Zhang Z, Xu X, Zhu H. Intensive- vs less-intensive-dose continuous renal replacement therapy
for the intensive care unit-related acute kidney injury: A meta-analysis and systematic review. J Crit Care. 2010;25:595–600
15. Bouchard J, Macedo E, Mehta RL. Dosing of renal replacement therapy in acute kidney injury: Lessons learned from clinical trials. Am J Kidney Dis. 2010;55:570–579
16. Marshall MR. Current status of dosing and quantification of acute renal replacement therapy. Part 2: Dosing paradigms and clinical implementation. Nephrology (Carlton). 2006;11:181–191
17. Claure-Del Granado R, Macedo E, Chertow GM, et al. Effluent volume in continuous renal replacement therapy
overestimates the delivered dose
of dialysis. Clin J Am Soc Nephrol. 2011;6:467–475
18. Ricci Z, Salvatori G, Bonello M, et al. In vivo validation of the adequacy calculator for continuous renal replacement therapies. Crit Care. 2005;9:R266–R273
19. Zhang Z, Ni H, Lu B. Variables associated with circuit life span in critically ill patients undergoing continuous renal replacement therapy
: A prospective observational study. ASAIO J. 2012;58:46–50
20. Clark WR, Turk JE, Kraus MA, Gao D. Dose determinants in continuous renal replacement therapy
. Artif Organs. 2003;27:815–820
21. Macedo E, Bouchard J, Soroko SH, et al.Program to Improve Care in Acute Renal Disease Study. Fluid accumulation, recognition and staging of acute kidney injury in critically-ill patients. Crit Care. 2010;14:R82
22. Zhang Z, Hongying N. Efficacy and safety of regional citrate anticoagulation in critically ill patients undergoing continuous renal replacement therapy
. Intensive Care Med. 2012;38:20–28
23. Lyndon WD, Wille KM, Tolwani AJ. Solute clearance in CRRT: Prescribed dose
versus actual delivered dose
. Nephrol Dial Transplant. 2012;27:952–956
24. Troyanov S, Cardinal J, Geadah D, et al. Solute clearances during continuous venovenous haemofiltration at various ultrafiltration flow rates using Multiflow-100 and HF1000 filters. Nephrol Dial Transplant. 2003;18:961–966
25. Brocklehurst IC, Thomas AN, Kishen R, Guy JM. Creatinine and urea clearance during continuous veno-venous haemofiltration in critically ill patients. Anaesthesia. 1996;51:551–553
26. Parienti JJ, Mégarbane B, Fischer MO, et al.Cathedia Study Group. Catheter dysfunction and dialysis performance according to vascular access among 736 critically ill adults requiring renal replacement therapy: A randomized controlled study. Crit Care Med. 2010;38:1118–1125
27. Marshall MR. Current status of dosing and quantification of acute renal replacement therapy. Part 1: Mechanisms and consequences of therapy under-delivery. Nephrology (Carlton). 2006;11:171–180
28. Bouman CS, de Pont AC, Meijers JC, et al. The effects of continuous venovenous hemofiltration on coagulation activation. Crit Care. 2006;10:R150