Clinical factors associated with initiation of renal replacement therapy in critically ill patients with acute kidney injury—A prospective multicenter observational study

Journal of Critical Care (2012) 27, 268–275

Clinical factors associated with initiation of renal replacement therapy in critically ill patients with acute kidney injury—A prospective multicenter observational study☆ Sean M. Bagshaw a,⁎,1 , Ron Wald c,1 , Jim Barton b,1 , Karen E.A. Burns c,1 , Jan O. Friedrich c,1 , Andrew A. House d,1 , Matthew T. James e,1 , Adeera Levin g,1 , Louise Moist d,1 , Neesh Pannu a,1 , Daniel E. Stollery h,1 , Michael W. Walsh e,f,1 a

University of Alberta Hospital, Edmonton, Alberta, Canada T6G 2B7 St Paul's Hospital, Saskatoon, Saskatchewan, Canada S7M 0Z9 c University of Toronto and the Keenan Research Centre in the Li Ka Shing Knowledge Institute, St Michael's Hospital, Toronto, Ontario, Canada M5B 1W8 d University Hospital/Victoria Hospital, London Health Sciences Centre, London, Ontario, Canada N6A SW9 e Foothills Medical Centre, Calgary, Alberta, Canada T2N 2T9 f McMaster University, Hamilton, Ontario, Canada L85 4L8 g St Paul's Hospital, Vancouver, British Columbia, Canada V6Z 1Y6 h Grey Nuns Community Hospital, Edmonton, Alberta, Canada T6L 5X8 b

Keywords: Acute kidney injury; Acute renal failure; Critical illness; Renal replacement therapy; Initiation; Dialysis; Epidemiology; Mortality

Abstract Purpose: Our objective was to describe the current practice for initiation of RRT in this population. There is uncertainty regarding the optimal time to initiate renal replacement therapy (RRT) in critically ill patients with acute kidney injury (AKI). Methods: Prospective study of patients receiving RRT in 6 intensive care units (ICUs) at 3 hospitals from July 2007 to August 2008. We characterized factors associated with start of RRT and evaluated their relationship with mortality. Results: We included 234 patients. RRT was initiated 1 day (0-4) after ICU admission (median [interquartile range]). Median creatinine was 331 μmol/L (225-446 μmol/L), urea 22.9 mmol/L (13.932.9 mmol/L), and RIFLE-Failure in 76.9%. Of traditional indications, PaO2/FiO2 b 200 (54.5%) and oliguria (32.9%) were most common. ICU and hospital mortality were 45.3% and 51.9%, respectively. In adjusted analysis, mortality at RRT initiation was associated with creatinine b332 μmol/L (odds ratio [OR] 2.8; 95% confidence interval [CI] 1.5-5.4), change in urea from admission N8.9 mmol/L (OR 1.8; 95% CI, 1.0-3.4), urine output b82 mL/24 hours (OR 3.0; 95% CI, 1.4-6.5), fluid balance N3.0 L/24 hours (OR 2.3; 95% CI, 1.2-4.5), percentage of fluid overload N5% (OR 2.3; 95% CI, 1.2-4.7), 3 or more failing organs (OR 4.5; 95% CI, 1.2-4.2), Sequential Organ Failure Assessment score N14 (OR

☆ Conflicts: The authors have no conflicts of interest to declare. ⁎ Corresponding author. Tel.: +1 780 407 6755; fax: +1 780 407 1228. E-mail addresses: [email protected], [email protected] (S.M. Bagshaw). 1 The CANadian Acute Kidney Injury (CANAKI) Study Investigators.

0883-9441/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.jcrc.2011.06.003

Renal replacement therapy in critically ill patients with AKI

269

2.3; 95% CI, 1.3-4.3), and start 4 days or more after admission (OR 4.3; 95% CI, 1.9-9.5). Mortality was higher as factors accumulated. Conclusion: In ICU patients requiring RRT, there was marked variation in factors that influence start of RRT. RRT initiation with fewer clinical triggers was associated with lower mortality. Timing of RRT may modify survival but requires appraisal in a randomized trial. © 2012 Elsevier Inc. All rights reserved.

1. Introduction

2. Methods

Acute kidney injury (AKI) is associated with substantial morbidity, mortality, and health resource utilization [1,2]. It is frequently encountered in critically ill patients, and recent data suggest its incidence is increasing [3,4]. An estimated 50% to 70% of those with more severe forms of AKI receive support with RRT, representing approximately 4% to 8% of all critically ill patients [5,6]. These patients have an observed mortality in excess of 50% [1,5,6]. The optimal approach to RRT delivery in critical illness, specifically the timing of initiation, is unknown [7]. The Acute Kidney Injury Network (AKIN) working group has recognized that timing of RRT initiation is an existing knowledge gap and has identified this issue as a high priority for investigation [7]. At present, there is no broad consensus to guide clinicians on when to initiate RRT [7]. However, the decision to initiate RRT is complex and depends on numerous patient-specific, clinician-specific, and logistic factors [7]. Survey data have further shown the practice of RRT initiation to be highly variable [8,9]. Numerous small, mostly retrospective studies and a systematic review have been either inconclusive or suggested earlier RRT initiation may be associated with improved clinical outcomes [1014]. These observations, along with recognition that the timing of RRT initiation is potentially modifiable, argue that there is genuine equipoise for additional clinical trials on whether early initiation of RRT can influence the clinical course and outcomes of critically ill patients with AKI [15]. We performed a prospective multicenter observational study evaluating factors that triggered the initiation of RRT. We hypothesized that initiation with fewer factors considered as relative indications for RRT would represent a surrogate for “earlier” initiation and would be associated with more favorable outcomes. Our objectives were to describe (1) current standard practices for RRT initiation in critically ill patients with AKI; (2) the clinical, physiological, and biochemical characteristics of critically ill patients with AKI at RRT initiation; and (3) the relationship between timing of RRT initiation and hospital mortality.

2.1. Study design, setting, and population This was a prospective observational study performed in 6 intensive care units (ICUs) at 3 Canadian hospitals from July 1, 2007, to August 31, 2008. The study protocol was approved by the Health Research Ethics Boards at each institution before commencement. All adult (age N18 years) patients admitted to participating ICUs (General Systems Intensive Care Unit, University of Alberta Hospital, Intensive Care Unit, Grey Nuns Community Hospital, Edmonton; Medical Surgical Intensive Care Unit, Cardiovascular Intensive Care Unit, Trauma Neurological Intensive Care Unit and Coronary Care Unit at St Michael's Hospital, Toronto) were eligible if they presented with or developed AKI where renal replacement therapy (RRT) was initiated. We excluded patients receiving RRT for non-AKI indications (ie, dialyzable toxin), RRT before ICU admission, and patients receiving maintenance dialysis for end-stage kidney disease.

2.2. Study definitions AKI was defined according to the RIFLE classification scheme, utilizing the serum creatinine (SCr) criteria only [16]. Prehospital baseline SCr values were unavailable for 76 patients (33%). For these patients, we estimated the baseline SCr based on back-calculation of the abbreviated Modification of Diet in Renal Disease (MDRD) as recommended by the Acute Dialysis Quality Initiative (ADQI) [16]. RIFLE category was determined at ICU admission and before initiation of RRT. Percentage of fluid overload (%FO) was estimated for the 24 hours before RRT by the following equation: %FO = [(fluid in − fluid out)/weight] × 100 [17,18]. Prehospital chronic kidney disease (CKD) was defined by an estimated glomerular filtration rate (eGFR) b60 mL/min/1.73 m2, according to the abbreviated MDRD equation, based on the lowest prehospital SCr in the 6 months that preceded the critical illness [19]. We defined RRT as any acute form of extracorporeal blood purification (ie, continuous renal replacement [CRRT], conventional intermittent hemodialysis [IHD], or sustained low-efficiency dialysis [SLED]). Sepsis was defined according to consensus guidelines [20].

270 Based on consensus statements from the ADQI, we defined traditional indications for initiation of RRT as the following: oliguria (b400 mL/24 hours), anuria (b100 mL/24 hours), serum pH b7.15 and/or serum HCO3− b15 mmol/L, PaO2/FiO2 ratio b200 and/or %FO N10%, serum urea N36 mmol/L, serum potassium (K+) N6.0 mmol/L. Shock was defined as a mean arterial pressure b60 mmHg and/or need for vasoactive support. Baseline and changes to organ failure were characterized by the Sequential Organ Failure Assessment (SOFA) score [21].

S.M. Bagshaw et al. was 59.3 (15.7) years, of which 59.8% were men and 32.9% were postoperative. Sepsis was the most common primary admission diagnosis (28.6%). Details of baseline demographics, admission characteristics, and premorbid kidney function are shown in Table 1. Complete prehospital data on baseline kidney function were available for 158 patients (67.5%). CKD was present in 39.2% of those with prehospitalization kidney function data.

3.2. Clinical and physiology characteristics at RRT initiation

2.3. Study protocol Patients were identified by routine screening of all patients admitted to the ICU. Eligible patients underwent a medical record review with documentation of baseline clinical, physiological, and laboratory data. Clinical data extracted included demographics (age, sex, and race) and physiological and laboratory data upon ICU admission and at the time of initiation of RRT. We recorded details of premorbid and enrollment kidney function, diuretic, and/or nephrotoxin exposure; use of mechanical ventilation; need for vasoactive drugs; and organ failure scores (ie, SOFA score). We recorded several parameters related to timing of initiation of RRT and clinical outcomes (ie, lengths stay in ICU and hospital, hospital mortality).

A summary of clinical and physiological parameters at the time of RRT initiation are shown in Table 2. RRT was initiated a median of 1 day after ICU admission (IQR 0-4). Seventy (29.9%) patients were transferred to the study ICUs from other hospitals. We found no significant difference in the median time from ICU admission to RRT initiation in transferred patients (P = .54) versus those not transferred. Patients had a median (IQR) SOFA score of 14 (10-16) and a median (IQR) of 3 (2-4) nonrenal failing organs. The majority of patients were mechanically ventilated (85.0%) Table 1 Baseline demographic, clinical, and physiological characteristics of the cohort at ICU admission Characteristics

2.4. Statistical analysis The primary analysis was predominantly descriptive. Normally or near-normally distributed variables are reported as means with SDs and were compared using a Student t test. Non-normally distributed continuous data are reported as medians and interquartile ranges (IQRs) and compared using a Mann-Whitney U test. Categorical data are reported as proportions and are compared using Fisher Exact Test. We evaluated the association between traditional triggers for RRT and hospital mortality and lengths of stay. In a further exploratory analysis, we also categorized parameters related to timing of RRT initiation into quartiles and assessed their association (as quartiles and medians) with hospital mortality in univariate and multivariable logistic regression analyses. Data are presented as odds ratios (OR) with 95% confidence intervals (CI). We considered P b .05 to be statistically significant for all comparisons unless otherwise stated. All statistical analysis was performed using STATA Release 10.1 (StataCorp, College Station, TX).

3. Results 3.1. Baseline characteristics In total, RRT was initiated on 234 patients with AKI during the study period. The mean (SD) age of our cohort

Age (y), mean (SD) 59.3 (15.7) Sex (male), n (%) 140 (59.8) Weight (kg), mean (SD) 82.8 (21.4) Postoperative admission, n (%) 77 (32.9) Primary diagnostic category, n (%) Sepsis 67 (28.6) Cardiovascular 45 (19.2) Hepatic 27 (11.5) Respiratory 26 (11.1) Gastrointestinal 21 (8.9) Genitourinary 18 (7.7) Metabolic/poisoning 13 (5.6) Trauma 9 (3.9) Other a 8 (3.4) Premorbid (baseline) kidney function SCr (μmol/L) (n = 158), median (IQR) 92 (72-120) Urea (mmol/L) (n = 182), median (IQR) 9.0 (5.5-17.4) eGFR (mL/min/1.73 cm2) (n = 158), median 72 (45-95) (IQR) 62 (39.2) eGFR b60 mL/min/1.73 cm2 (%) (n = 158), mean (SD) ICU admission kidney function SCr (μmol/L), median (IQR) 212 (135-346) Urea (mmol/L), median (IQR) 14.5 (8.9-28.2) RIFLE category at ICU admission, n (%) None 1 (0.43) Risk 25 (10.7) Injury 79 (33.8) Failure 129 (55.1) a

Other includes primary neurological and hematological diagnoses.

Renal replacement therapy in critically ill patients with AKI Table 2 Details of clinical and physiological characteristics at the time of initiation of RRT

271 SLED in 164 (70.1%), 57 (24.4%), and 13 (5.6%) patients, respectively.

Parameters SOFA score, median (IQR) SOFA score (nonrenal), median (IQR) Number of failing organs a, median (IQR) b Vasopressor therapy, n (%) Mechanical ventilation, n (%) PaO2/FiO2 ratio, mean (SD) Glasgow Coma Scale, median (IQR) Bilirubin (μmol/L), median (IQR) Platelets (cells/109), mean (SD) Serum lactate (mmol/L) N4 mmol/L Median (IQR) n (%) Sodium (mmol/L), mean (SD) Potassium (mmol/L), mean (SD) pH, mean (SD) Bicarbonate (mmol/L), mean (SD) SCr (μmol/L), median (IQR) RIFLE category at RRT initiation, n (%) None Risk Injury Failure Worsening RIFLE Category before RRT initiation, n (%) Urea (mmol/L), median (IQR) ΔUrea from premorbid baseline (n = 182): Absolute (mmol/L), median (IQR) Relative (%), median (IQR) ΔUrea from ICU admission Absolute (mmol/L), median (IQR) Relative (%), median (IQR) Urine output mL/24 h, median (IQR) b400 mL/24 h, n (%) b100 mL/24 h, n (%) Fluid balance (mL/24 h) (n = 193), median (IQR) %FO N5%, n (%) %FO N10%, n (%) %FO N15%, n (%) Furosemide (24 h before RRT), n (%) Hospital admission to RRT (d), median (IQR) ICU admission to RRT (d), median (IQR) a b

14 (10-16) 10 (7-13) 3 (2-4) 147 (62.8) 199 (85.0) 206 (167) 9 (3-14) 27 (15-72) 163 (129) 2.2 (1.4-5.8) 74 (34.9) 138 (11) 4.5 (0.9) 7.28 (0.11) 18.4 (5.8) 331 (225-446) 0 (0) 6 (2.6) 48 (20.5) 180 (76.9) 67 (28.6) 22.9 (13.9-32.6) 9.0 (2.3-18.7) 54.4 (14.5-72.3) 1.7 (0-8.9) 8.0 (0-41.9) 265 (82-860) 136 (60.4) 48 (20.5) 3000 (1339-6475) 74 (38.3) 35 (18.1) 14 (7.3) 36 (30) 3 (1-10) 1 (0-4)

Organ failure defined as SOFA score ≥2 points. Excludes kidney failure.

and received vasoactive therapy (62.8%). The median (IQR) SOFA cardiovascular score was 3 (1-4). The median (IQR) SCr and urea were 331 (225-446) μmol/L and 22.9 (13.932.9) mmol/L, respectively. Oliguria (b400 mL/24 hours) and anuria (b100 mL/24 hours) were present 60.4% and 20.5%, respectively. In total, 92.2% were in a positive fluid balance, 38.3% had a %FO ≥5%, and 18.1% has a %FO ≥10%. Initial RRT modality was continuous RRT, IHD, and

3.3. Clinical outcomes The crude ICU and hospital mortality were 45.3% and 51.9%, respectively (Table 3). The median (IQR) duration of RRT for survivors who became RRT independent by hospital discharge was 6 (3-13) days. The median ICU and hospital stays were 13 (6-27) days and 28 (15-52) days, respectively. Patients with premorbid CKD (eGFR b60 mL/ min/1.73 m2) were less likely to recover and become RRT independent by the time of hospital discharge (42.9% vs 12.8%; OR 5.1; 95% CI, 1.6-16.3, P = .009).

3.4. Clinical characteristics at initiation of RRT and clinical outcomes The prevalence and associated outcome for traditional triggers for RRT initiation are shown in Table 4. At RRT initiation, the majority of patients fulfilled RIFLE-Failure (76.9%), whereas fewer were classified as RIFLE-Injury (20.5%) or RIFLE-Risk (2.6%). Additional common factors included hypoxemia, with a PaO2/FiO2 b200 (54.5%), oliguria (32.9%), and a %FO N10% (30.0%). In total, 28.6% of patients developed a worsening RIFLE category before start of RRT. In those with a worsening RIFLE category, RRT occurred later following ICU admission (5 [312] days vs 1 [0-2] days, P b .001) compared with those with no change in RIFLE category. On the contrary, RRT initiation in association with hyperkalemia, acidemia, anuria, or uremia was relatively uncommon. Of these, in crude analysis, anuria was the only factor significantly associated with hospital mortality (OR 3.10; 95% CI, 1.55-6.18, P = Table 3

Summary of crude clinical outcomes

Outcomes Duration of RRT (d), median (IQR) 6 (3-13) ICU survivors a Hospital survivors a 6 (3-13) ICU death, n (%) 106 (45.3) ICU nonsurvivors Duration of RRT (d), median (IQR) 5 (2-13) Time from last RRT to death (d), median (IQR) 1 (0-2) Proportion b2 d (n = 106), n (%) 66.7 ICU length of stay (d), median (IQR) 13 (6-27) RRT post–ICU discharge (n = 127), n (%) 46 (36.2) Hospital death, n (%) 121 (51.9) Hospital length of stay (d), median (IQR) 28 (15-52) Post–RRT hospital length of stay (d), median 19 (7-39) (IQR) RRT posthospital discharge, n (%) (n = 106) 24 (22.6) Creatinine at hospital discharge in survivors 119 (73-183) (μmol/L), median (IQR) a a

In patients not receiving RRT at discharge.

272 Table 4 of RRT

S.M. Bagshaw et al. Summary of traditional indications for initiation

Variable

Prevalence, n (%)

RIFLE category Risk 6 (2.6) Injury 48 (20.5) Failure 180 (76.9) Oliguria 32.9 Anuria 20.5 Urea N36 mmol/L 21.4 K N6 mmol/L 8.1 pH b7.15 10.7 HCO3 b15 mmol/L 27.8 PaO2/FiO2 b200 54.5 %FO N10% 30.0 Number of indications 0 25.6 1 39.3 ≥2 35.0 Any indication 74.36

Crude hospital mortality (%)

Post–RRT hospital length of stay (d), median (IQR)

83.3 52.1 50.6 46.8 72.9 46.0 63.2 60.0 50.8 54.6 63.6

10 30 17 20 13 19 19 19 19 14 14

(2-42) (8-61) (7-35) (8-37) (2-31) (8-56) (1-37) (3-39) (3-40) (7-30) (6-30)

49.2 47.8 58.5 52.9

24 15 22 16

(9-46) (8-34) (3-42) (7-37)

.001). There was no difference in mortality when timing of RRT was stratified by RIFLE class at the time of initiation (Risk 83.0%, Injury 52.1%, Failure 50.8%, P = .31). RRT was generally initiated within 1 day for those with 2 or more traditional indications, whereas for those with none or only 1 indication, RRT was initiated later (P = .0004). In a further exploratory analysis, the association between clinical, physiological, and biochemical parameters at RRT initiation and hospital mortality are shown in Supplementary Tables 1 to 4. By univariate analysis, factors at the time of RRT initiation associated with higher crude hospital mortality included low SCr, low urine output, positive fluid accumulation, greater number of failing organs, greater SOFA score, and longer duration from ICU admission to RRT start. Supplementary Table 3 shows further exploratory analysis of these factors by their observed medians and quartiles. For each factor, these cutoffs values were generally associated with higher adjusted hospital mortality. Patients with multiple factors at the time of RRT initiation had sequentially higher adjusted hospital mortality (Supplementary Table 4).

4. Discussion We performed a multicenter prospective observational study of 234 critically ill patients with AKI who received RRT to describe the current standard of care around RRT initiation and to explore the association of clinical, physiological, and biochemical characteristics at the time of RRT initiation with important clinical outcomes including mortality and length of stay.

We found that critically ill patients with AKI who were initiated on RRT were complex and characterized by a high illness severity and multiorgan dysfunction. The number (median [IQR]) of nonrenal failing organs was 3 (2-4); 62.8% required vasoactive support and 85.0% were mechanically ventilated. Our data also confirmed the high associated morbidity and mortality of critical illness complicated by severe AKI requiring RRT. In our cohort, in-hospital mortality was 52%, median length of stay was 28 days, and 23% of survivors remained dialysis-dependent at hospital discharge. Our study demonstrates that RRT was generally initiated early (median 1 day) following ICU admission. In addition, the majority of patients fulfilled the RIFLE-Failure at the time RRT was initiated; however, the median time to start RRT was longer in patients who had worsening RIFLE category following ICU admission. When we evaluated “traditional” indications for RRT initiation, we found the majority of patients had RRT initiated in the setting of hypoxemia and/or worsening fluid accumulation and oliguria. Interestingly; however, RRT initiation for more conventional “lifethreatening” indications such as hyperkalemia, severe acidemia, and/or uremia was relatively uncommon in this critically ill cohort. These observations indirectly imply the use of RRT may represent an important organ support modality in critically ill patients, not solely to replace selected aspects of kidney function but also to aid in maintaining homeostasis in others. Importantly, we believe these data are the first to provide insight into the current standards of practice for RRT initiation in a Canadian context. In addition to the evaluation of traditional indications for RRT initiation, we also examined whether additional thresholds of clinical, physiological, and laboratory factors had any important associations with mortality. We found that selected factors, in particular urine output, changes in serum pH, and duration to RRT initiation following ICU admission, when stratified by quartiles, showed a U-shaped association for higher mortality. Interestingly, mortality was higher for those in whom RRT was initiated when SCr was b332 μmol/ L; however, we found no differences in mortality for relative changes in creatinine from baseline or ICU admission or by RIFLE class at RRT initiation. Moreover, this finding was independent of baseline CKD status (eGFR b60 mL/min/ 1.73 m2) (38.6% for ≤446 μmol/L vs 41.9% for N446 μmol/ L, P = .84). This observation is consistent with several previous studies showing higher mortality when RRT is initiated at lower creatinine values in critical illness [12,15,22]. Plausible explanations may include the initiation of RRT in patients with low SCr for indications associated with a higher mortality (ie, oligoanuria, fluid accumulation), reduced creatinine generation in critical illness [23], dilution due to fluid accumulation [22,24], or lower baseline patient muscle mass and/or nutritional status [22]. Moreover, our finding may represent “confounding by indication” where RRT is initiated earlier for critically ill patients with higher illness severity and multiorgan dysfunction and hence a

Renal replacement therapy in critically ill patients with AKI higher risk of death, whereas in less severely ill or “healthier” patients, clinicians may adopt a “wait-and-see” approach with creatinine values rising higher before RRT initiation. This hypothesis is supported by the fact that patients whose RIFLE score was lower but worsened had a longer duration before RRT start following ICU admission, and that SOFA scores were higher for patients with lower creatinine values (SOFA scores: creatinine b225 μmol/L 14.6 [4.1] vs creatinine ≥225 μmol/L 12.9 [4.0], P = .003). We observed no association between absolute serum urea values and mortality. This finding is consistent with recent data [12,15]. Plausible explanations include the limitations inherent in retrospective studies that simply dichotomized serum urea by the mean/median value to evaluate outcomes or that static measures of serum urea have limited value as thresholds for RRT initiation [11,25-28]. In addition, changes in serum urea in critical illness have several potential sources beyond reduced renal clearance such as use of corticosteroids, gastrointestinal bleeding, and hypercatabolism. However, we did find that dynamic changes in serum urea from ICU admission to RRT initiation were associated with higher adjusted mortality. There was a significant association between mortality and the volume of urine production before RRT initiation. Mortality was highest for those patients with marked oligoanuria (b82 mL/24 hours). This was also a key indication for RRT initiation in this cohort. Oliguria has repeatedly been associated with higher mortality in AKI [2932]. In a retrospective analysis of 1847 critically ill patients with AKI, Ostermann and Chang [12] found oligoanuria (urine b400 mL/24 hours) before RRT was independently associated with ICU death. Similarly, in 2 small nonrandomized studies, early RRT in response to oliguria following cardiac surgery was associated with lower mortality [33,34]. On the contrary, a nonoliguric state in critical illness has been associated with later RRT initiation and may reflect practice variation and clinician bias in the timing of RRT initiation [35,36]. Liangos et al [35] found increased urine output before RRT was associated with higher mortality (OR 3.8; 95% CI, 1.1-12.8, P = .03). Unfortunately, the assessment of urine output in critically ill patients is known to be confounded by diuretic therapy [37,38]. This draws into question the value of using urine output as a singular parameter for initiation of RRT; however, a nonoliguric state characterized by worsening azotemia, metabolic homeostasis, and/or multiorgan dysfunction may identify a target population wherein early RRT may be beneficial. Positive fluid accumulation in critically ill patients with AKI has recently been identified as an important factor associated with mortality [17,29]. Several observational studies in critically ill children have consistently shown that greater FO at the time of RRT initiation is closely correlated with lower survival [18,39]. Our data also suggest a greater positive fluid accumulation before RRT initiation is associated with higher hospital mortality. Critically ill patients receive considerable resuscitation-related and other

273 obligatory fluid intake [40]. Fluid therapy often greatly exceeds spontaneous urine output, contributing to rapid fluid accumulation, which is further compounded with AKI and/or relative oliguria. Persistent FO may potentiate adverse outcomes through perpetuating respiratory failure with the need for prolonged mechanical ventilation, exacerbating high intra-abdominal pressure, and limiting patient mobilization. Fluid accumulation is an important determinant of RRT initiation in critical illness that may transcend changes in conventional solute/metabolic parameters. In our study, higher organ failure scores (SOFA score N14) and a greater absolute number of (nonrenal) failing organs at the time of RRT initiation were associated with higher mortality. Numerous other studies have reported that a higher burden of organ failure at RRT initiation portends a worse prognosis [12,36,41]. This is not unexpected, as RRT may not be initiated for severe AKI per se but more commonly as part of a strategy of multiorgan and homeostatic support in critical illness. Available data do not provide clear insight into whether early RRT initiation for high burden of nonrenal organ dysfunction can translate into improved survival; however, this merits additional investigation. Finally, longer duration from ICU admission to RRT initiation was associated with higher mortality. The high mortality observed for critically ill patients receiving RRT within the first ICU day could be attributed to the high illness severity and the subsequent need for urgent RRT initiation. For these patients, establishing parameters that constitute “early” RRT may be challenging. On the other hand, a number of previous investigations have shown temporal delays to RRT initiation after ICU admission are independently associated with worse outcome [12,15,29,42]. This may simply be a surrogate for worsening illness severity or development of de novo delayed AKI and may characterize a unique subset of critically ill patients [32]. In the BEST Kidney Study, a prospective multicenter study of 1238 ICU patients with severe AKI receiving RRT, temporal delay from ICU admission to RRT initiation (N5 days) was independently associated with higher hospital mortality (adjusted OR 1.95; 95% CI, 1.3-2.9, P = .001) [15]. This finding was robust in a sensitivity analysis restricted to those patients with severe AKI at the time of admission to ICU (adjusted OR 2.63; 95% CI, 1.6-4.4, P b .0001), implying there may have been an actual “delay” to RRT initiation. Although these patients were more likely to be septic and postoperative, they also had higher serum urea values, lower level of oliguria, and were more likely to have received diuretic therapy. Similarly, Ostermann and Chang [12] recently showed better survival for earlier temporal initiation of RRT after ICU admission. Additional prospective investigation is needed that is focused on this subgroup of critically ill patients who receive RRT in less than 2 days following ICU admission to determine whether pre-ICU support represents an opportunity for early intervention. There are important limitations to our study. First, our study was observational and not a randomized comparison of

274 early versus standard timing of initiation of RRT. Consequently, our study is prone to several forms of bias and we cannot infer a causal relationship between any of these parameters at RRT initiation and clinical outcome. Second, although multicenter, our sample was relatively small and did not permit extensive exploratory analysis of selected subgroups. Third, we did not capture data on critically ill patients not receiving RRT who were “matched” for the parameters described to assess differences in clinical outcomes. Fourth, our study focused on the association between parameters at RRT initiation and mortality; however, we recognize renal recovery and dialysis weaning are also important clinical outcomes. Fifth, there are known intangible clinician-specific and other logistic/operational issues that impact upon when and how RRT is initiated in critical care settings. Our study did not assess these factors. Finally, we did not collect specimens for biomarker analysis that are increasingly proving beneficial in diagnosing and prognosticating in AKI and may therefore guide decisionmaking on RRT initiation. The optimal time to start RRT in critically ill patients with AKI remains uncertain. Our study confirms that RRT initiation in critically ill patients is a complex process and is conditional on numerous patient-specific factors. Although we have identified several clinical parameters at the time of RRT initiation that are associated with worse clinical outcomes, individual patient variation confounds the concept that simple “triggers” can be used across a heterogeneous cohort. Moreover, our data suggest “traditional” indications, aside from oliguria and surrogates for FO, are less commonly encountered in critically ill patients. Randomized trials are needed to establish causality between these parameters and important clinical outcomes, which ideally will allow for an individualized approach to RRT initiation and will establish whether earlier RRT initiation can mitigate the risks associated with RRT itself. Supplementary materials related to this article can be found online atdoi:10.1016/j.jcrc.2011.06.003.

S.M. Bagshaw et al.

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

Acknowledgments This study was funded, in part, by a grant from the University of Alberta Hospital Foundation. Dr Bagshaw is supported by a Clinical Investigator Award from the Alberta Heritage Foundation for Medical Research. Dr Walsh is supported by a Randomized Controlled Trial Mentoring Award from the Canadian Institutes of Health Research and a Clinical Fellowship award from the Alberta Heritage Foundation for Medical Research.

[16]

[17]

[18]

References [1] Bagshaw SM, Laupland KB, Doig CJ, Mortis G, Fick GH, Mucenski M, et al. Prognosis for long-term survival and renal recovery in

[19]

critically ill patients with severe acute renal failure: a population-based study. Crit Care 2005;9(6):R700-9. Wald R, Quinn RR, Luo J, Li P, Scales DC, Mamdani MM, et al. Chronic dialysis and death among survivors of acute kidney injury requiring dialysis. JAMA 2009;302(11):1179-85. Bagshaw SM, George C, Bellomo R. Changes in the incidence and outcome for early acute kidney injury in a cohort of Australian intensive care units. Crit Care 2007;11(3):R68. Xue JL, Daniels F, Star RA, Kimmel PL, Eggers PW, Molitoris BA, et al. Incidence and mortality of acute renal failure in Medicare beneficiaries, 1992 to 2001. J Am Soc Nephrol 2006; 17(4):1135-42. Uchino S, Kellum JA, Bellomo R, Doig GS, Morimatsu H, Morgera S, et al. Acute renal failure in critically ill patients: a multinational, multicenter study. Jama 2005;294(7):813-8. Hoste EA, Clermont G, Kersten A, Venkataraman R, Angus DC, De Bacquer D, et al. RIFLE criteria for acute kidney injury are associated with hospital mortality in critically ill patients: a cohort analysis. Crit Care 2006;10(3):R73. Gibney N, Hoste E, Burdmann EA, Bunchman T, Kher V, Viswanathan R, et al. Timing of initiation and discontinuation of renal replacement therapy in AKI: unanswered key questions. Clin J Am Soc Nephrol 2008;3(3):876-80. Ricci Z, Ronco C, D'Amico G, De Felice R, Rossi S, Bolgan I, et al. Practice patterns in the management of acute renal failure in the critically ill patient: an international survey. Nephrol Dial Transplant 2006;21(3):690-6. Overberger P, Pesacreta M, Palevsky PM. Management of renal replacement therapy in acute kidney injury: a survey of practitioner prescribing practices. Clin J Am Soc Nephrol 2007;2(4):623-30. Bagshaw SM, Uchino S, Bellomo R, Morimatsu H, Morgera S, Schetz M, et al. Timing of renal replacement therapy and clinical outcomes in critically ill patients with severe acute kidney injury. J Crit Care 2009;24(1):129-40. Liu KD, Himmelfarb J, Paganini EP, Ikizler A, Soroko S, Mehta RL, et al. Timing of initiation of dialysis in critically ill patients with acute kidney injury. Clin J Am Soc Nephrol 2006;1:915-9. Ostermann M, Chang RW. Correlation between parameters at initiation of renal replacement therapy and outcome in patients with acute kidney injury. Crit Care 2009;13(6):R175. Seabra VF, Balk EM, Liangos O, Sosa MA, Cendoroglo M, Jaber BL. Timing of renal replacement therapy initiation in acute renal failure: a meta-analysis. Am J Kidney Dis 2008;52(2):272-84. Karvellas CJ, Farhat MR, Sajjad I, Mogensen SS, Leung AA, Wald R, et al. A comparison of early versus late initiation of renal replacement therapy in critically ill patients with acute kidney injury: a systematic review and meta-analysis. Crit Care 2011;15(1):R72. Bagshaw SM, Uchino S, Bellomo R, Morimatsu H, Morgera S, Schetz M, et al. Timing of renal replacement therapy and clinical outcomes in critically ill patients with severe acute kidney failure. J Crit Care 2009;24:129-40. Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P. Acute renal failure—definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care 2004;8(4):R204-12. Bouchard J, Soroko SB, Chertow GM, Himmelfarb J, Ikizler TA, Paganini EP, et al. Fluid accumulation, survival and recovery of kidney function in critically ill patients with acute kidney injury. Kidney Int 2009;76(4):422-7. Sutherland SM, Zappitelli M, Alexander SR, Chua AN, Brophy PD, Bunchman TE, et al. Fluid overload and mortality in children receiving continuous renal replacement therapy: the prospective pediatric continuous renal replacement therapy registry. Am J Kidney Dis 2010;55(2):316-25. Levey AS, Coresh J, Greene T, Stevens LA, Zhang YL, Hendriksen S, et al. Using standardized serum creatinine values in the modification of

Renal replacement therapy in critically ill patients with AKI

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med 2006;145(4):247-54. Levy MM, Fink MP, Marshall JC, Abraham E, Angus D, Cook D, et al. Ramsay G: 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Crit Care Med 2003;31(4):1250-6. Vincent JL, Moreno R, Takala J, Willatts S, De Mendonca A, Bruining H, et al. The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ dysfunction/failure. Intensive Care Med 1996;22: 707-10. Cerda J, Cerda M, Kilcullen P, Prendergast J. In severe acute kidney injury, a higher serum creatinine is paradoxically associated with better patient survival. Nephrol Dial Transplant 2007;22(10):2781-4. Doi K, Yuen PS, Eisner C, Hu X, Leelahavanichkul A, Schnermann J, et al. Reduced production of creatinine limits its use as marker of kidney injury in sepsis. J Am Soc Nephrol 2009;20(6):1217-21. Bouchard J, Macedo E, Soroko S, Chertow GM, Himmelfarb J, Ikizler TA, et al. Comparison of methods for estimating glomerular filtration rate in critically ill patients with acute kidney injury. Nephrol Dial Transplant 2010;25(1):102-7. Gettings LG, Reynolds HN, Scalea T. Outcome in post-traumatic acute renal failure when continuous renal replacement therapy is applied early vs. late. Intensive Care Med 1999;25(8):805-13. Kennedy AC, Luke RG, Linton AL, Eaton JC, Gray MJ. Results of haemodialysis in severe acute tubular necrosis. A report of fifty-seven cases. Scott Med J 1963;8:97-108. Lange HW, Aeppli DM, Brown DC. Survival of patients with acute renal failure requiring dialysis after open heart surgery: early prognostic indicators. Am Heart J 1987;113(5):1138-43. Waikar SS, Bonventre JV. Can we rely on blood urea nitrogen as a biomarker to determine when to initiate dialysis? Clin J Am Soc Nephrol 2006;1:903-4. Payen D, de Pont AC, Sakr Y, Spies C, Reinhart K, Vincent JL. A positive fluid balance is associated with a worse outcome in patients with acute renal failure. Crit Care 2008;12(3):R74. Brivet FG, Kleinknecht DJ, Loirat P, Landais PJ. Acute renal failure in intensive care units—causes, outcome, and prognostic factors of hospital mortality; a prospective, multicenter study. French Study Group on Acute Renal Failure. Crit Care Med 1996;24(2):192-8. Chertow GM, Soroko SH, Paganini EP, Cho KC, Himmelfarb J, Ikizler TA, et al. Mortality after acute renal failure: models for

275

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

prognostic stratification and risk adjustment. Kidney Int 2006;70(6): 1120-6. Guerin C, Girard R, Selli JM, Perdrix JP, Ayzac L. Initial versus delayed acute renal failure in the intensive care unit. A multicenter prospective epidemiological study. Rhone-Alpes Area Study Group on Acute Renal Failure. Am J Respir Crit Care Med 2000;161(3 Pt 1): 872-9. Demirkilic U, Kuralay E, Yenicesu M, Caglar K, Oz BS, Cingoz F, et al. Timing of replacement therapy for acute renal failure after cardiac surgery. J Card Surg 2004;19(1):17-20. Elahi MM, Lim MY, Joseph RN, Dhannapuneni RR, Spyt TJ. Early hemofiltration improves survival in post-cardiotomy patients with acute renal failure. Eur J Cardiothorac Surg 2004;26(5):1027-31. Liangos O, Rao M, Balakrishnan VS, Pereira BJ, Jaber BL. Relationship of urine output to dialysis initiation and mortality in acute renal failure. Nephron Clin Pract 2005;99(2):c56-60. Shiao CC, Wu VC, Li WY, Lin YF, Hu FC, Young GH, et al. Late initiation of renal replacement therapy is associated with worse outcomes in acute kidney injury after major abdominal surgery. Crit Care 2009;13(5):R171. Mehta RL, Pascual MT, Soroko S, Chertow GM. Diuretics, mortality, and nonrecovery of renal function in acute renal failure. JAMA 2002;288(20):2547-53. Uchino S, Doig GS, Bellomo R, Morimatsu H, Morgera S, Schetz M, et al. Diuretics and mortality in acute renal failure. Crit Care Med 2004;32(8):1669-77. Goldstein SL, Currier H, Graf C, Cosio CC, Brewer ED, Sachdeva R. Outcome in children receiving continuous venovenous hemofiltration. Pediatrics 2001;107(6):1309-12. Rivers E, Nguyen B, Havstad S, Ressler J, Muzzin A, Knoblich B, et al. Early goal-directed therapy in the treatment of severe sepsis and septic shock. N Engl J Med 2001;345(19):1368-77. Maccariello E, Rocha E, Valente C, Nogueira L, Rocha PT, Bonomo Jr H, et al. Effects of early changes in organ dysfunctions on the outcomes of critically ill patients in need of renal replacement therapy. Clinics (Sao Paulo) 2008;63(3):343-50. Oudemans-van Straaten HM, Bosman RJ, van der Spoel JI, Zandstra DF. Outcome of critically ill patients treated with intermittent highvolume haemofiltration: a prospective cohort analysis. Intensive Care Med 1999;25(8):814-21.