Initiation Time of Renal Replacement Therapy on. Patients with Acute Kidney Injury: A Systematic Review and. Meta-analysis of 8179 participants

Transcription

1 Initiation Time of Renal Replacement Therapy on Patients with Acute Kidney Injury: A Systematic Review and Meta-analysis of 8179 participants Caixia Wang nephrology research fellow a*, Lin-Sheng Lv research fellow of medicine b*, Hui Huang professor of cardiology c*, Jianqiang Guan vice professor of Anesthesiology d*, Zengchun Ye attending Physician of nephrology a, Shaomin Li nephrology research physician a, Yanni Wang nephrology research fellow a, Tanqi Lou professor of nephrology a#, Xun Liu professor of nephrology a#, a: Department of Nephrology, the Third Affiliated Hospital of Sun Yat-sen University, Guangzhou, China b: Operation Room, The Third Affiliated Hospital of Sun Yat-sen University, Guangzhou, China. c: Department of Cardiology; Guangdong Province Key Laboratory of Arrhythmia and Electrophysiology, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, Guangzhou, China. d: Department of Anesthesiology; the Third Affiliated Hospital of Sun Yat-sen University, Guangzhou, China. * These authors contributed equally to the paper. Correspondence to: # these authors contributed equally to the paper. Dr. Xun Liu. Division of Nephrology, Department of Internal Medicine, The Third Affiliated Hospital of Sun Yat-sen University, Guangzhou , China. Tel: , This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as doi: /nep.12890

2 Prof. Tanqi Lou. Division of Nephrology, Department of Internal Medicine, The Third Affiliated Hospital of Sun Yat-sen University, Guangzhou , China. Tel: , Running title: Initiation Time of RRT on AKI Patients Conflict of interest statement: None Abstract The early initiation of renal replacement therapy has been recommended for patients with acute renal failure by some studies, but its effects on mortality and renal recovery are unknown. We conducted an updated meta-analysis to provide quantitative evaluations of the association between the early initiation of renal replacement therapy and mortality for patients with acute kidney injury. After applying inclusion/exclusion criteria, 51 studies, including 10 randomized controlled trials, with a total of 8179 patients were analyzed. Analysis of the included trials showed that patients receiving early renal replacement therapy had a 25% reduction in all-cause mortality compared to those receiving late renal replacement therapy (risk ratio [RR] 0.75, 95% CI [0.69, 0.82]). We also noted a 30% increase in renal recovery (RR 1.30, 95% CI [1.07, 1.56]), a reduction in hospitalization of 5.84 days (mean difference [MD], 95% CI [-10.27, -1.41]) and a reduction in the duration of mechanical ventilation of 2.33 days (MD, 95% CI [-3.40, -1.26]) in patients assigned to early renal replacement therapy. The early initiation of renal replacement therapy was associated with a decreased risk of all-cause mortality compared with the late initiation of RRT in patients with acute kidney injury. These findings should be interpreted with caution given the heterogeneity between studies. Further studies are needed to identify the causes of mortality and to assess whether mortality differs by dialysis dose.

3 After revision, the manuscript is better organized and suitable for publication. This is an important issue for acute kidney injury and hope this articles will help the readers about the decision of time initiation of RRT. Keywords: Acute kidney injury; Initiation time; Meta-analysis; Mortality; Renal replacement therapy Introduction Acute kidney injury (AKI) is a major health issue that is associated with substantial morbidity and mortality. 1-7 Renal replacement therapy (RRT) has long been used as a supportive treatment for AKI and plays an important role in the recovery of kidney function. In daily practice, there is substantial variation in policies regarding the initiation time of RRT. Data have emerged suggesting that earlier RRT initiation may attenuate kidney-specific and non-kidney organ injury from uremia, acidemia and fluid overload. 8 While an earlier initiation of RRT may reduce AKI symptoms, such as serum ph homeostasis and improve clearancel of toxic solutes, it may also unnecessarily expose patients to potential harms (hemorrhage, thrombosis, bacteremia, intradialytic hypotension, hypersensitivity to the extracorporeal circuit, clearance of trace elements, and antibiotics) along with added resource utilization. There remains no rigorous evidence to guide clinicians on this important issue. 9 The proper time to start RRT is currently one of the top research priorities for AKI. Several reviews have focused on the optimum time to start RRT with few randomized controlled trials (RCTs) included and small sample sizes It is recognized that previous findings were limited by low power and large heterogeneity, meaning this important topic still has not been fully explored. Since then, several RCTs and cohort studies related to this

4 issue have been published and an updated review of the evidence would be of great use to clinicians. Therefore, we conducted a meta-analysis aiming to provide better evidence as to when to start RRT for patients with AKI. Materials and methods Search strategy and selection criteria We conducted a systematic review and meta-analysis in accordance with the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines. 15 We searched for clinical trials published with no start time limitation and the end time was Jun 22, 2016 (the date of the last search). We electronically searched PUBMED, EMBASE, the specialized register of the Cochrane Renal Group and the Cochrane Central Register of Controlled Trials and references of related papers to identify all relevant trials. The keywords included acute kidney injury, renal replacement treatment, early, late, timing, mortality, etc. Restrictions on language were not imposed in our searches. The complete search strategy is available in the appendix (S1). Study selection Studies comparing mortality and other clinical outcomes of early and late RRTs for patients over 18 years old with AKI were included. Editorials, letters to the editor, review articles, case reports, and animal experimental studies were excluded. The primary outcome was all-cause mortality. The secondary outcomes were renal recovery, hospital mortality, duration of hospitalization and mechanical ventilation.

5 Data extraction and quality assessment Data were extracted by two reviewers independently according to the Cochrane guidelines, 15, 16 and a third reviewer checked the extracted data for accuracy. Abstracts and potential manuscripts were reviewed and identified for retrieval according to the inclusion criteria. The following information was extracted: number of patients included, population characteristics (age, weight, sex), presence of diabetes, population setting, RRT modality, dialyzer and membrane and definition of early or late RRT. We also extracted data on trial characteristics, trial intervention and type of study. For randomized controlled trials we recorded blindness, random allocation, adequate allocation concealment, intention to treat, and withdrawal or dropout rate in the quality assessment. The quality of RCTs was assessed by Review Manager 5.3 (Oxford, UK) and that of cohort studies was assessed by Newcastle - Ottawa Quality Assessment Scale_Cohort Studies. A third person was available when there were any disagreements concerning data extraction and/or quality assessment. Synthesis and analysis of data We performed the meta-analysis using Review Manager 5.3 and combined the studies using a random effects model. Additionally, we performed a meta-regression by comprehensive meta-analysis (CMA). We combined dichotomous outcomes (e.g. all-cause mortality) using risk ratios (RR) and continuous outcomes (e.g. duration of hospitalization) using weighted mean difference (MD). Heterogeneity was assessed using the I 2 index, which judges values less than 25% to be minimal, less than 50% to be moderate, and 50% or greater to be substantial. Funnel plots were provided to assess publication bias.

6 We performed subgroup analyses of the primary outcome measure (all-cause mortality) by trial type and sample size, duration of follow-up, dialysis modality, publication year, population setting and disease severity. We used all available data reported in each study. Meta-regression analysis was performed to assess possible sources of heterogeneity according to trial type and sample size, dialysis modality, publication year, population setting and disease severity. We also performed a meta-analysis of baseline patient characteristics according to serum levels of blood urea nitrogen (BUN), creatinine and APACHE II scores. We also performed a sensitivity analysis in all-cause mortality, and finally the influence of each study was examined by omission from the analysis. Publication bias was assessed and visualized with a funnel plot. Results Study Characteristics A total of 9137 potentially relevant citations were identified and screened electronically. We excluded 9002 because they were outside the scope of this review Among the 135 remaining papers, another 81 were excluded: 32 because they did not provide information on the outcomes of interest, 28 because they did not present original data (review articles or commentaries), 21 because they had multiple mortality data, and 3 because they included patients without AKI. Overall, 10 randomized and 41 non-randomized trials fulfilled the inclusion criteria and were included in the meta-analysis (Fig 1)

7 The selected studies included 8179 participants, of whom 4218 received early RRT and 3961 received late RRT. Approximately one-fifth of the patients (1456) were from the 10 RCTs. Three trials only reported data in abstracts. 37, 45, 46 The sample size ranged from 15 to 1238 patients and the follow-up ranged from 14 to 1000 days. Detailed characteristics and population setting of the 52 studies are displayed in Table 1 and appendix S2. Quality of included studies The quality of the included trials was assessed and we regarded all the RCTs as high quality of low risk bias; further quality assessment is detailed in appendix S3-4. Cohort studies with 8 to 10 stars, as assessed by Newcastle - Ottawa Quality Assessment Scale_Cohort Studies, were treated as high quality with low risk bias, those with 5 to 7 stars were treated as moderate quality with moderate risk bias, and those with 0 to 4 stars was treated as low quality with high risk bias. Original assessments of the quality of the cohort studies are displayed in appendix S5. Of all the included studies, 16 are of high quality with low risk, 34 with moderate risk and 1is of low quality with high risk. Primary outcome The primary outcome was based on all the included trials and consisted of 8179 patients with 4375 deaths (Fig 2). Compared with patients assigned to late RRT, patients assigned to early RRT had a significant reduction of 25% in all-cause mortality. However, there was statistically significant heterogeneity (I 2 = 69%). The included studies were quite different in definitions for early RRT, population setting (Medical/ Surgery/ Obstetrical), dialysis modality, disease severity, type and sample size, publication year, dialysis machine and membrane.

8 To investigate factors that may result in the heterogeneity, we performed several subgroup analyses. In our subgroup analysis of all-cause mortality based on population setting (internal medicine, surgery and mix), we identified 12 studies that reported on mortality in patients with internal medicine disease (n = 1938), 18 studies in patients after surgery (n = 2091) and 21 studies in patients with medical or surgical disease (mixed, n = 4150). Compared with those assigned to late RRT, we noted significant reductions of all-cause mortality in patients assigned to early RRT in the surgery group (risk ratio [RR], 0.61, 95% CI [0.52, 0.72]) and mixed group (RR, 0.79, 95% CI [0.71, 0.89]). Patients assigned to early RRT from the internal medicine group showed a 12% reduction in all-cause mortality compared with patients receiving late RRT (RR, 0.88, 95% CI [0.76, 1.03]), but this change was not statistically significant. At present, there are still no uniform criteria for the timing of RRT for patients with AKI; therefore, the studies included in our meta-analysis varied in the definitions of early and late RRT. (detailed in Table 2). In the subgroup analysis of mortality by definitions of early RRT, we could not evaluate all included studies due to the variety of criteria for patients assigned to early and late RRT. After carefully screening, 13 studies were divided into three groups by a relatively uniform criteria (group 1: patients were estimated by the RIFLE criteria; group 2: patients were estimated by clinical symptoms; group 3: patients assigned to dialysis when serum BUN levels < 29 mmol/l were treated as those receiving early RRT) and were evaluated. We noted a statistically significant decrease in mortality in patients assigned to early RRT by no clinical symptoms vs. clinical symptoms (6 trials with 512 patients, RR, 0.60, 95% CI [0.47, 0.76]) and by serum BUN levels <29 mmol/l vs. 29 mmol/l (2 studies with 134 patients, RR, 0.65, 95% CI [0.54, 0.78]), separately, compared with patients assigned to late RRT. A decrease in mortality in patients assigned to early RRT by RIFLE (injury vs. failure) criteria (5 studies with 628 patients, RR, 0.69, 95% CI [0.46, 1.05]) was

9 noted, but the results were not statistically significant (Fig 3). In our analysis of all-cause mortality based on dialysis modality (continuous renal replacement therapy (CRRT), intermittent hemodialysis (IHD) and mixed), we identified 20 studies with 2585 patients receiving CRRT, 16 studies with 1695 patients receiving IHD and 15 studies with 3899 patients receiving CRRT, IHD, sustained low efficiency dialysis (SLED) or early isovolaemic haemofiltration (EIHF). Compared with those assigned to late RRT, we noted a statistically significant decrease in all-cause mortality in patients assigned to early RRT in the CRRT group (RR, 0.75, 95% CI [0.65, 0.86]), IHD group (RR, 0. 68, 95% CI [0.55, 0.83]) and the mixed group (RR, 0.80, 95% CI [0.71, 0.90]). In our subgroup analysis of all-cause mortality based on disease severity, we noted a statistically significant decrease in mortality in patients assigned to early RRT in both critically ill patients (24 trials with 4744 patients, RR, 0.75, 95% CI [0.66, 0.84]) and others (27 trials with 3435 patients, RR, 0.76, 95%CI [0.67, 0.85]). In our subgroup analysis of all-cause mortality based on type of study, we noted a statistically significant decrease in mortality in patients assigned to early RRT in non-rcts (41 studies with 6723 patients, RR, 0.74, 95% CI [0.68, 0.81]); however the decrease in mortality from the RCTs was not statistically significant (10 studies with 1456 patients, RR, 0.82, 95% CI, [0.62, 1.08]). We also analyzed all-cause mortality by sample size (n 100 or n < 100) and by publication year (pre- vs post-2000), and noted that compared with patients assigned to late RRT, patients assigned to early RRT had a significant reduction of all-cause mortality in both subgroups of sample size (RR, 0.61, 95% CI [0.51, 0.73] for n<100; RR, 0.82, 95% CI [0.76, 0.89] for n 100) and publication year (RR, 0.75, 95% CI [0.65, 0.87] for pre-2000 studies; RR, 0.75, 95% CI [0.68, 0.83] for post-2000 studies).

10 Secondary outcomes Renal recovery was reported in 14 studies with 2570 patients. Patients assigned to early RRT showed a statistically significant increase in renal recovery of 30%, compared with those assigned to late RRT (RR, 1.30, 95% CI [1.07, 1.56]). Hospital mortality was reported in 23 trials with 4198 patients, and we recognized a statistically significant reduction of 25% in patients assigned to early RRT, compared with patients assigned to late RRT (RR, 0.75, 95% CI [0.66, 0.86]). According to the analysis of the duration of mechanical ventilation and hospitalization, patients assigned to early RRT had a statistically significant reduction of 2.33 days in duration of mechanical ventilation (mean difference [MD], -2.33, 95% CI [-3.40, -1.26]), as well as a significant reduction of 5.84 days in duration of hospitalization (MD, -5.84, 95% CI [-10.27, -1.41]). Meta-regression analysis and meta-analysis of baseline characteristics. To investigate sources of heterogeneity, we performed a meta-regression analysis according to trial type and sample size, dialysis modality, publication year, population setting and disease severity, and a meta-analysis of baseline characteristics such as serum BUN and creatinine levels, and APACHE II scores. In the meta-regression analysis, we noted that the heterogeneity observed is most likely explained by differences in disease severity, study design and sample size (S6-9). In the meta-analysis of baseline characteristics, the baseline levels of BUN and creatinine, and the baseline APACHE II scores in patients assigned to early RRT were significantly lower than those assigned to late RRT (MD, , 95% CI [-12.59, -7.58] for BUN, MD, , 95% CI [-61.75, ] for creatinine; MD, -1.29, 95% CI [-2.13, -0.45]) for APACHE II scores (S10-12).

11 Sensitivity analysis and publication bias The sensitivity analyses included all 51 studies and were based solely on the primary outcome. The influence of each study was examined by omission, and the omission of any study in our meta-analysis of all-cause mortality did not result in a change of the conclusion (S13). Additionally, the heterogeneity of each analysis is presented in detail in Table 3. Fig 4 shows a funnel plot of the studies included in the meta-analysis that reported on all-cause mortality. Some studies lie outside the 95% CIs with an unbalanced distribution around the vertical axis, indicating an obvious publication bias. The quantitative assessment of publication bias was displayed in appendix S14.. Discussion This systematic review and meta-analysis showed a 25% reduction in all-cause mortality in patients assigned to early RRT compared with those receiving late RRT. We noted a similar reduction in mortality when we considered population setting, dialysis modality, sample size and publication year. However, in the analysis of mortality by study type, we only observed a significant reduction in mortality from non-rcts; while a reduction in mortality was observed in early RRT from RCTs, this result was not statistically significant. Early RRT was also associated with a lower hospital mortality and higher renal recovery. Additionally, patients assigned to early RRT had a lower duration of hospitalization and mechanical ventilation

12 In the application of any new therapy, the efficacy of the procedure is always paramount. The pooled data of dialysis outcomes indicated that early RRT was effective for patients with AKI. One potential explanation for the decrease in mortality associated with early RRT might be avoiding life-threatening complications of uremia, acidosis, hyperkalemia, and volume overload. Interventions that attain solute clearance and fluid balance before the development of more serious disease, attenuate kidney-specific and non-kidney organ injury better than late RRT. This may translate into improved survival and earlier renal recovery rates. Moreover, our systematic review suggests that early RRT can reduce the duration of mechanical ventilation and hospitalization in patients assigned to early RRT, which means patients may recover with less medical expenditures. We firstly performed a subgroup analysis by definitions of early and late RRT. As the definitions of initiation time were quite different in included studies, we only combined data from three groups of patients with relatively uniform criteria. In the subgroup analysis of mortality, we noted a statistically significant decrease in mortality in patients assigned to early RRT but having no clinical symptoms and serum BUN levels <29 mmol/l, separately, which indicated that patients with AKI may benefit more from early RRT when patients were evaluated by clinical symptoms and serum BUN levels. As only part of the data was evaluated in the subgroup analysis due to the variety of criteria for patients assigned to early or late RRT, future studies with uniform criteria are needed. In the subgroup analysis based on population setting, we only noted a statistically significant decrease in mortality in the surgery and mixed patients group assigned to early RRT. Surgery patients were a more homogeneous population than those with internal medicine disease and early RRT might benefit surgery-related multi-organ failure more than other groups.

13 In the subgroup analysis based on dialysis modality, mortality did not differ by dialysis modality, which indicates that CRRT does not offer an advantage, at least with regards to survival, compared with IHD. These results were in accordance with papers performed to 14, 68 evaluate the efficacy of CRRT versus IHD in the management of patients with AKI. In the subgroup analysis based on disease severity, we noted a statistically significant decrease in mortality in both critically ill and other patients assigned to early RRT. Earlier RRT might have a beneficial impact on survival for both critically ill and other patients with AKI. The studies included in our systematic analysis had a great difference in their operational definitions for RRT timing, disease severity, population setting, study design, dialysis dose, solution used, and filtration and membrane type. Sources of heterogeneity were explored through subgroup meta-analyses, meta-regression and meta-analysis of baseline characteristics. We noted that disease severity, study design and sample size and the date of publication were associated with the large heterogeneity, and that subgroup analysis was in accordance with the outcomes. Moreover, in the subgroup analysis, we noted that patients in CRRT group and patients assigned to early RRT according to clinical indications and serum BUN levels showed a lower heterogeneity, indicating that the heterogeneity may be partially explained by dialysis modality and the various definitions of early and late RRT. We also found a significant difference in baseline serum levels of BUN and creatinine, and baseline APACHE II scores in the meta-analysis of baseline characteristics. Such imbalance of baseline characteristics may result in great heterogeneity.

14 This is the largest meta-analysis performed up to date comparing early versus late initiation of RRT in patients with AKI. Previous reports on this issue analyzed no more than 23 trials (compared with our 51 trials), and did not provide data for duration of mechanical ventilation and hospitalization. Moreover, in contrast to previous work, we performed several subgroup analyses based on strict definitions of early and late RRT and noted a significant decrease in mortality in early RRT according to the definitions based on clinical symptoms and serum BUN levels. We also investigated the heterogeneity by three kinds of analyses and found the factors that contributed to the large heterogeneity. However, several limitations must be considered in our systematic review and meta-analysis. Unpublished reports could not be identified, which may have biased our results. Data in this meta-analysis are limited by heterogeneous indicators as we discussed above, and as a result, a multi-centric, suitably-designed randomized trial with a large patient cohort is needed. Evidence from our systematic review and meta-analysis was insufficient to identify specific causes of mortality. We could not assess whether mortality differs from dialysis dose because the dose of dialysis varied between almost every study. As a result, studies designed to identify causes of mortality and to assess whether mortality differs from dialysis dose are needed. Conclusions Patients with AKI are at high risk of mortality. Our systematic review and meta-analysis suggests that early RRT is associated with a decreased risk of all-cause mortality compared with late RRT in patients with AKI. Additionally, we show that early RRT can benefit patients with AKI in renal recovery, duration of mechanical ventilation and hospitalization. However, this meta-analysis was based on heterogeneously-designed studies. Evidence from a multi-centric, suitably-designed randomized trial as well as trials to identify causes of

15 mortality, and to assess whether mortality differs from dialysis dose are still needed. Acknowledgements This study was supported by the National Natural Science Foundation of China (Grant Nos and ), the China Postdoctoral Science Foundation (Grant No ), Guangdong Science and Technology Plan (Grant Nos. 2011B and 2013B ), the Fundamental Research Funds for the Central Universities (Grant No. 11ykpy38), the National Project of Scientific and Technical Supporting Programs Funded by Ministry of Science & Technology of China (Grant No. 2011BAI10B00). This work was also supported in part by NSFC [ , , ], and the Natural Science Foundation of Guangdong Province [2014A ] to Hui Huang. References 1. Ahlstrom A, Tallgren M, Peltonen S, Rasanen P, Pettila V. Survival and quality of life of patients requiring acute renal replacement therapy. Intensive Care Med. 2005;31: Bagshaw SM, Laupland KB, Doig CJ, et al. Prognosis for long-term survival and renal recovery in critically ill patients with severe acute renal failure: a population-based study. Crit Care. 2005;9:R 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:R Korkeila M, Ruokonen E, Takala J. Costs of care, long-term prognosis and quality of life in patients requiring renal replacement therapy during intensive care. Intensive Care

16 Med. 2000;26: Manns B, Doig CJ, Lee H, et al. Cost of acute renal failure requiring dialysis in the intensive care unit: clinical and resource implications of renal recovery. Crit Care Med. 2003;31: Morgera S, Kraft AK, Siebert G, Luft FC, Neumayer HH. Long-term outcomes in acute renal failure patients treated with continuous renal replacement therapies. Am J Kidney Dis. 2002;40: Uchino S, Kellum JA, Bellomo R, et al. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA. 2005;294: Matson J, Zydney A, Honore PM. Blood filtration: new opportunities and the implications of systems biology. Crit Care Resusc. 2004;6: Gibney RT, Bagshaw SM, Kutsogiannis DJ, Johnston C. When should renal replacement therapy for acute kidney injury be initiated and discontinued? Blood Purif. 2008;26: 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: Karvellas CJ, Farhat MR, Sajjad I, 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:R Wang X, Jie Yuan W. Timing of initiation of renal replacement therapy in acute kidney injury: a systematic review and meta-analysis. Ren Fail. 2012;34: Liu Y, Davari-Farid S, Arora P, Porhomayon J, Nader ND. Early versus late initiation of renal replacement therapy in critically ill patients with acute kidney injury after cardiac surgery: a systematic review and meta-analysis. J Cardiothorac Vasc Anesth.

17 2014;28: Pannu N, Klarenbach S, Wiebe N, Manns B, Tonelli M. Renal replacement therapy in patients with acute renal failure: a systematic review. JAMA. 2008;299: Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med. 2009;151:264-9, W Van Tulder M, Furlan A, Bombardier C, Bouter L. Updated method guidelines for systematic reviews in the cochrane collaboration back review group. Spine (Phila Pa 1976). 2003;28: Wald R, Adhikari NK, Smith OM, et al. Comparison of standard and accelerated initiation of renal replacement therapy in acute kidney injury. Kidney Int. 2015;88: Lim CC, Tan CS, Kaushik M, Tan HK. Initiating acute dialysis at earlier Acute Kidney Injury Network stage in critically ill patients without traditional indications does not improve outcome: a prospective cohort study. Nephrology (Carlton). 2015;20: Lin YC, Lin FY, Shih CM, et al. Early continuous renal replacement therapy in cardiogenic shock patients with severe acute kidney injury undergoing extracorporeal membrane oxygenation. CardioRenal Medicine. 2014;4: Jun M, Bellomo R, Cass A, Gallagher M, Lo S, Lee J. Timing of renal replacement therapy and patient outcomes in the randomized evaluation of normal versus augmented level of replacement therapy study. Crit Care Med. 2014;42: Shum HP, Chan KC, Kwan MC, Yeung AWT, Cheung EWS, Yan WW. Timing for initiation of continuous renal replacement therapy in patients with septic shock and acute kidney injury. Therapeutic Apheresis and Dialysis. 2013;17: Leite TT, Macedo E, Pereira SM, et al. Timing of renal replacement therapy initiation by

18 AKIN classification system. Critical Care. 2013;17: 23. Jamale TE, Hase NK, Kulkarni M, et al. Earlier-start versus usual-start dialysis in patients with community-acquired acute kidney injury: a randomized controlled trial. Am J Kidney Dis. 2013;62: Hu ZJ, Liu LX, Zhao CC. [Influence of time of initiation of continuous renal replacement therapy on prognosis of critically ill patients with acute kidney injury]. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2013;25: Wu SC, Fu CY, Lin HH, et al. Late initiation of continuous veno-venous hemofiltration therapy is associated with a lower survival rate in surgical critically ill patients with postoperative acute kidney injury. American Surgeon. 2012;78: Shiao CC, Ko WJ, Wu VC, et al. U-curve association between timing of renal replacement therapy initiation and in-hospital mortality in postoperative acute kidney injury. PLoS One. 2012;7:e Oh. JH, Shin DH, Lee MJ, et al. Early initiation of continuous renal replacement therapy improves patient survival in severe progressive septic acute kidney injury. J Crit Care. 2012;27:743 e do Nascimento GV, Balbi AL, Ponce D, Abrao JM. Early initiation of dialysis: mortality and renal function recovery in acute kidney injury patients. J Bras Nefrol. 2012;34: Chon GR, Chang JW, Huh JW, et al. A comparison of the time from sepsis to inception of continuous renal replacement therapy versus RIFLE criteria in patients with septic acute kidney injury. Shock. 2012;38: Boussekey N, Capron B, Delannoy PY, et al. Survival in critically ill patients with acute kidney injury treated with early hemodiafiltration. International Journal of Artificial Organs. 2012;35:

19 31. Konopka A, Banaszewski M, Wojtkowska I, Stȩpińska J. Early implementation of continuous venovenous haemodiafiltration improves outcome in patients with heart failure complicated by acute kidney injury. Kardiol Pol. 2011;69: Ji Q, Mei Y, Wang X, et al. Timing of continuous veno-venous hemodialysis in the treatment of acute renal failure following cardiac surgery. Heart Vessels. 2011;26: Fernandez NG, Perez-Valdivieso JR, Bes-Rastrollo M, et al. Timing of renal replacement therapy after cardiac surgery: a retrospective multicenter Spanish cohort study. Blood Purif. 2011;32: Chou YH, Huang TM, Wu VC, et al. Impact of timing of renal replacement therapy initiation on outcome of septic acute kidney injury. Crit Care. 2011;15:R Carl DE, Grossman C, Behnke M, Sessler CN, Gehr TW. Effect of timing of dialysis on mortality in critically ill, septic patients with acute renal failure. Hemodial Int. 2010;14: Shiao CC, Wu VC, Li WY, 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:R Sabater J PX, Albertos R, Gutierrez D, Labad X. Acute renal failure in septic shock. Should we consider different continuous renal replacement therapies on each RIFLE score stage? Intensive Care Med. 2009;35:S Iyem H, Tavli M, Akcicek F, Buket S. Importance of early dialysis for acute renal failure after an open-heart surgery. Hemodial Int. 2009;13: Bagshaw SM, Uchino S, Bellomo R, et al. Timing of renal replacement therapy and clinical outcomes in critically ill patients with severe acute kidney injury. J Crit Care. 2009;24: Manché A, Casha A, Rychter J, Farrugia E, Debono M. Early dialysis in acute kidney

20 injury after cardiac surgery. Interact Cardiovasc Thorac Surg. 2008;7: Wu VC, Ko WJ, Chang HW, et al. Early renal replacement therapy in patients with postoperative acute liver failure associated with acute renal failure: effect on postoperative outcomes. J Am Coll Surg. 2007;205: Andrade L, Cleto S, Seguro AC. Door-to-dialysis time and daily hemodialysis in patients with leptospirosis: impact on mortality. Clin J Am Soc Nephrol. 2007;2: Piccinni P, Dan M, Barbacini S, et al. Early isovolaemic haemofiltration in oliguric patients with septic shock. Intensive Care Med. 2006;32: Liu KD, Himmelfarb J, Paganini E, et al. Timing of initiation of dialysis in critically ill patients with acute kidney injury. Clin J Am Soc Nephrol. 2006;1: Koo J R HJ, Paganini E,. Timing of initiation of dialysis in critically ill patients with acute kidney injury. Clin J Am Soc Nephrol. 2006;1: Tsai H WV, Yang M, Lin Y, Ko W, Wu K. Outcome in the acute liver failure patients treated with renal replacement therapy for acute renal failure: Comparison between early or late dialysis. J Am Soc Nephrol. 2005;16:540A (abstr). 47. Sugahara S, Suzuki H. Early start on continuous hemodialysis therapy improves survival rate in patients with acute renal failure following coronary bypass surgery. Hemodial Int. 2004;8: 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: Demirkiliç U, Kuralay E, Yenicesu M, et al. Timing of Replacement Therapy for Acute Renal Failure after Cardiac Surgery. J Card Surg. 2004;19: Durmaz I, Yagdi T, Calkavur T, et al. Prophylactic dialysis in patients with renal dysfunction undergoing on-pump coronary artery bypass surgery. Ann Thorac Surg.

21 2003;75: Bouman CS, Oudemans-Van Straaten HM, Tijssen JG, Zandstra DF, Kesecioglu J. Effects of early high-volume continuous venovenous hemofiltration on survival and recovery of renal function in intensive care patients with acute renal failure: a prospective, randomized trial. Crit Care Med. 2002;30: Splendiani G, Mazzarella V, Cipriani S, Pollicita S, Rodio F, Casciani CU. Dialytic treatment of rhabdomyolysis-induced acute renal failure: our experience. Ren Fail. 2001;23: Kresse S, Schlee H, Deuber HJ, Koall W, Osten B. Influence of renal replacement therapy on outcome of patients with acute renal failure. Kidney Int Suppl. 1999;S 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: Pursnani ML, Hazra DK, Singh B, Pandey DN. Early haemodialysis in acute tubular necrosis. J Assoc Physicians India. 1997;45: 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: Conger JD. A controlled evaluation of prophylactic dialysis in post-traumatic acute renal failure. J Trauma. 1975;15: Kleinknecht D, Jungers P, Chanard J, Barbanel C, Ganeval D. Uremic and non-uremic complications in acute renal failure: Evaluation of early and frequent dialysis on prognosis. Kidney Int. 1972;1: Kornhall S. Acute renal failure in surgical disease with special regard to neglected complications. A retrospective study of 298 cases treated during the period

22 Acta Chir Scand Suppl. 1971;419: Bosteels V, Verberckmoes R, Vandenbroucke J, Michielsen P. [Importance of early and frequent dialysis for the prognosis of postoperative and post-traumatic anuria]. J Urol Nephrol (Paris). 1970;76: Katz E AS, Manzur F,. Extracorporeal dialysis in acute renal failure Rev Med Chil. 1968;96: Fischer RP, Griffen WO, Jr., Reiser M, Clark DS. Early dialysis in the treatment of acute renal failure. Surg Gynecol Obstet. 1966;123: 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: Balslov JT, Jorgensen HE. A survey of 499 patients with acute anuric renal insufficiency. Causes, treatment, complications and mortality. Am J Med. 1963;34: Parsons FM, Hobson SM, Blagg CR, Mc CB. Optimum time for dialysis in acute reversible renal failure. Description and value of an improved dialyser with large surface area. Lancet. 1961;1: Zarbock A, Kellum JA, Schmidt C, et al. Effect of Early vs Delayed Initiation of Renal Replacement Therapy on Mortality in Critically Ill Patients With Acute Kidney Injury: The ELAIN Randomized Clinical Trial. JAMA. 2016;315: Gaudry S, Hajage D, Schortgen F, et al. Initiation Strategies for Renal-Replacement Therapy in the Intensive Care Unit. N Engl J Med. 2016; 68. Ghahramani N, Shadrou S, Hollenbeak C. A systematic review of continuous renal replacement therapy and intermittent haemodialysis in management of patients with acute renal failure. Nephrology (Carlton). 2008;13:570-8.

23 Table 1 Basic characters of the studies included Study Country Design Sample Male/ Mean Diabet Follow Risk size Female age (yr) es (%) up(days) bias Conger 1975 US Quasi- RCT 18 18/0 23 NR 180 L Pursnani 1997 India RCT 35 NR NR NR NR L Bouman 2002 Netherlands RCT 71 42/ NR 28 L Durmaz 2003 Turkey RCT 44 34/ L Sugahara 2004 Japan RCT 28 18/ L Koo 2006 Korea RCT Abs /40 63 NR NR L Jamale 2013 Mumbai RCT / L Wald 2015 Canada RCT / L Gaudry 2016 France RCT 619 NR L Zarbock 2016 Germany RCT / L Liu 2006 US P /93 56 NR 28 M Sabater 2009 NR P, Abs 32 NR NR NR NR H Shiao 2009 Taiwan P 98 57/ L Bagshaw 2009 Canada P / NR NR M Min Jun 2014 Australia P / NR 90 M Lim 2015 Singapore P / M Parsons 1961 UK R 33 20/13 50 NR 44 M Balslov 1963 Denmark R 96 NR 56 NR NR M Kennedy 1963 UK R 57 29/28 40 NR NR M Fischer 1966 US R 162 NR NR NR 84 M Katz 1968 Chile R 89 NR NR NR NR M Bosteels 1970 Belgium R 149 NR 41 NR NR M Kornhall 1971 Sweden R 193 NR NR NR NR M Kleinknecht 1972 France R 320 NR NR NR NR M Lange 1987 US R 36 22/ M Gettings 1999 US R / NR 45 L Kresse 1999 Germany R /94 60 NR NR M Splendiani 2001 Italy R 27 20/7 52 NR NR M Elahi 2004 UK R 64 48/ NR M Demirkilic 2004 Turkey R 61 48/13 NR 39.3 NR M Tsai 2005 Taiwan R, Abs 98 70/28 57 NR NR M Piccinni 2006 Italy R 80 60/20 56 NR 28 L

24 Wu 2007 Taiwan R 80 59/ NR 100 M Andrade 2007 Brazil R 33 32/ NR NR M Manche 2008 Malta R 71 NR NR M Iyem 2009 Turkey R / L Carl 2010 US R / M Konopka 2011 Poland R 37 NR 65 NR NR M Chou 2011 Taiwan R / NR M Fernández 2011 Spain R / NR M Ji 2011 China R 58 36/ NR M Chon 2012 Korea R 55 38/ NR 90 M Jung 2012 Korea R / M Nascimento 2012 Brazil R 86 55/ NR NR M WU 2012 Taiwan R 73 48/ M Boussekey 2012 France R / L Shiao 2012 Taiwan R / L HU 2013 China R 52 16/ NR 180 M Shum 2013 China R / NR 180 M Leite 2013 Brazil R / NR M Lin 2014 Taiwan R 15 11/ M : Abstract :Not reported : L-low risk bias; M-moderate risk bias; H-high risk bias

25 Table 2. Definitions of early and late RRT Study Early RRT Late RRT Hemodialysis initiated when Conger Hemodialysis initiated to maintain BUN BUN-54mmol/L, scr-10mg/dl(-884umol/l), 1975 <25mmol/L or for clinical indication Pursnani Dialysis initiated before clinical 1997 deterioration Dialysis initiated for clinical deterioration Dialysis initiated within 12h after Dialysis initiated after fulfilling the following Bouman 2002 fulfilling the following criteria: urine output<30ml/h and Cr clearance<20 ml/min on 3-h sample conventional criteria: urea> 40 mmol/l (BUN> 112mg/dL),sK>6.5 meq/l, or severe pulmonary edema Preoperative prophylactic hemodialysis Hemodialysis was performed in all patients Durmaz was performed in all patients and if and if postoperative scr increased by 50% or 2003 postoperative scr increased by 10% urine output was <400 ml/24h Sugahara h urine output< 30ml/hr 2h urine output < 20ml/hr Koo 2006 CVVH (for 48h) started immediately after diagnosis of severe sepsis or septic shock ARF plus 1 of the following conventional criteria for dialysis: refractory volume overload and oliguria. Severe azotemia (BUN>29mmol/L, metabolic acidosis (ph<7.2), or hyperkalemia (sk>6.5mmol/l) Jamale 2013 Serum urea nitrogen and/or creatinine Clinically indicated as judged by treating levels increased to 70 and 7 mg/dl, nephrologists respectively Wald 2015 Patients started RRT within 12 h of Patients started RRT after 12 h of fulfilling fulfilling eligibility eligibility Gaudry Within 6 hours after documentation of 2016 stage 3 acute kidney injury More than 72 hours after randomization Within 12 hours of stage 3 AKI (urine output Within 8 hours of diagnosis of stage 2 <0.3 ml/kg/h for 24 h and/or >3 fold AKI using the KDIGO classification increase in serumcreatinine level Zarbock (urine output <0.5 ml/kg/h for 12h or comparedwith baseline or serumcreatinine of fold increase in serum creatinine 4mg/dLwith an acute increase of at least comparedwith baseline). 0.5mg/dL within 48 hours [to convert to μ mol/l, multiply by 88.4]) Liu 2006 BUN 27mmol/L BUN>27mmol/L Sabater 2009 Rifle Criteria (Risk, Injury) Rifle Criteria (Failure) Shiao 2009 Increase plasma creatinine 1.5 or GFR decrease > 25%; Urine output:< 0.5 ml/kg/h 6 h Increase plasma creatinine 2 or GFR decrease > 50%; Urine output < 0.5 ml/kg/h 12 h

26 Bagshaw 2009 Urea<24.2mmol/L Urea>24.2mmol/L Time between RIFLE-I acute kidney Time between RIFLE-I acute kidney injury Min Jun injury and continuous renal replacement and continuous renal replacement therapy 2014 therapy commencement was <17.6 hours commencement was 17.6 hours Traditional indications for ARRT initiation included: serum potassium 6.0 mmol/l, Lim 2015 Absent of traditional indication serum urea 30 mmol/l, arterial ph < 7.25, serum bicarbonate <10 mmol/l, acute pulmonary oedema, acute uraemic encephalopathy or pericarditis. Parsons Clinical deterioration or BUN increase to 71 BUN increase to mmol/l 1961 mmol/l Balslov 1963 BUN-50mmol/L BUN>67mmol/L Kennedy 1963 BUN 67 mmol/l BUN >67 mmol/l Hyperkalemia, gross hematuria or fluid Fischer Clinical deterioration alone or BUN imbalance, increased BUN (rarely before 1966 increase to < 150 mg/dl (<54 mmol/l) 200mg/dL(54mmol/L)) Katz 1968 Mean NPN 187mg/dL (range, ) Mean NPN 338mg/dL (range, ) Bosteels Before development of uremic 1970 symptoms After development of uremic symptoms Kornhall 1971 NPN 199mg/dL NPN>199mg/dL Kleinknecht 1972 Early and frequent hemodialysis to maintain blood BUN 33mmol/L BUN>59mmol/L and/or severe electrolyte disturbances Lange 1987 BUN<29mmol/L BUN>29mmol/L Gettings 1999 BUN<21mmol/L BUN 21mmol/L Kresse 1999 BUN 34mmol/L, scr 380umol/L, and BUN >34mmol/L, scr 477umol/L, and urine urine output 924 ml/24h output 525 ml/24h Splendiani BUN>59 mmol/l and/or severe electrolyte BUN 33mmol/L 2001 disturbances Elahi 2004 Urine output<100 ml/h for 8h after surgery despite furosemide infusion and irrespective of scr or sk BUN 30mmol/L, scr>=250umol/l, or sk>6 mmol/l despite glucose-insulin infusion and irrespective of urine output Urine output<100 ml/h for 8h despite scr>486 umol/l or sk>5.5mmol/l despite Demirkilic furosemide administration and glucose-insulin infusion and irrespective of 2004 irrespective of scr or sk urine output Tsai 2005 BUN<29 mmol/l BUN>29 mmol/l Piccinni CVVH initiate within 12h of admission CVVH initiate for classic indications

27 2006 to ICU with diagnosis of septic shock Wu 2007 BUN<28.6 mmol/l BUN>28.6 mmol/l Andrade 2007 Immediately on admission At 24h after admission When oliguria occurred and did not Manche respond to fluid replacement and single 2008 dose intravenous diuretics and Urine output <0.5mL/kg/h hyperkalaemia in the absence of oliguria Iyem 2009 As soon as urine output 0.5 ml/kg/h post operatively and a 50% increase in baseline urea and creatinine 48 h after onset of urine output 0.5 ml/kg/h and 50% increase in baseline urea and creatinine Carl 2010 BUN<35.7 mmol/l BUN>35.7mmol/L After full treatment for HF and unsuccessful Konopka As soon as AKI was diagnosed pharmacological treatment of complicating 2011 AKI Chou 2011 RIFLE -0 or -Risk RIFLE-Injury or -Failure Fernández days after cardiac surgery >3 days after cardiac surgery Ji 2011 Within 12h of urine output 0.5 ml/kg/h postoperatively and a 50% increase in baseline urea and creatinine 12 h after urine output 0.5 ml/kg/h postoperatively and a 50% increase in baseline urea and creatinine Chon 2012 Time of inception of CRRT from sepsis Time of inception of CRRT from sepsis >24h 24h and RIFLE criteria (RIFLE-Injury) and RIFLE criteria (RIFLE-Failure) Jung 2012 Based on the median interval between Based on the median interval between the the start time of vasopressors infusion start time of vasopressors infusion and CRRT and CRRT initiation, Interval between initiation Interval between the start time of the start time of vasopressor infusion and vasopressor infusion and CRRT initiation 3.9 CRRT initiation is 1.1 days Nascimento 2012 BUN 26.7 mmol/l BUN>26.7 mmol/l In the Risk (RIFLE-R) stage due to In the Injury/Failure stage (RIFLE-IF) due to anuria refractory to diuretics, or clinical WU 2012 oliguria refractory to diuretics (Increase presentation of metabolic acidosis, plasma creatinine * 1.5 or GFR hyperkalemia, fluid overload, or azotemia decrease > 25%) (Increase plasma creatinine * 2 or more or GFR decrease > 50%) Boussekey 2012 RRT delay 16 hours RRT delay > 16 hours group Shiao 2012 The duration of time between the ICU admission and the initiation of the RRT<4 days The duration of time between the ICU admission and the initiation of the RRT 4days HU 2013 (AKIN 1and 2) creatine increased 200% (AKIN 3) creatine increased more >300% or

28 Shum 2013 Leite 2013 Lin 2014 ~300% from baseline; urine output< 0.5ml/kg/h >12h RIFLE Risk, GFR decrease >25% from baseline Patients initiating RRT less than 24 h after reaching AKIN stage 3 were included in the early RRT group Receiving CRRT <24 h after initiating ECMO (Extracorporeal membrane oxygenation) 354umol/L, or increased 44Pumol/L, urine output <0.3ml/kg/h>24h RIFLE Injury or Failure, GFR decrease >50% from baseline Those after 24 h were included in the late RRT group Receiving CRRT 24 h after initiating ECMO

29 Table 3 Results of meta-analysis Outcomes Studies Patients Overall summary I 2 & p Reference Primary outcome All-cause mortality RR (M-H, R, 95% CI) 0.75 [0.69, 0.82] 69%, < All-cause mortality for AKI by definitions of early and late Clinical indications [0.47, 0.76] 53% , 31, 40, 43, 55, 60 BUN [0.54, 0.78] 0%, , 56 RIFLE criteria [0.46, 1.05] 80% < , 29, 34, 36, 37 All-cause mortality for AKI by type of disease Surgery [0.52, 0.72] 68%, < , 26, 32, 33, 36, 38, 40, 41, 46-50, 54, 56, 57, 59, 65 Internal Medicine Mixed [0.76, 1.03] 0.79 [0.71, 0.89] 60%, %, < , 21, 23, 27, 29, 34, 35, 37, 42, 45, 52, 67 17, 18, 20, 22, 24, 28, 30, 31, 39, 43, 44, 51, 53, 55, 58, All-cause mortality for AKI by type of dialysis CRRT [0.69, 0.82] 58% < , 24-27, 29, 31, 32, 36-38, 45, 47-49, 51, 54 IHD [0.55, 0.83] 71% < , 28, 40, 46, 50, Mixed [0.71, 0.90] 74% < , 22, 30, 33-35, 39, 41-44, 52, 53, 66, 67 All-cause mortality for AKI by disease severity Critically ill patients Others [0.66, 0.84] 0.76 [0.67, 0.85] 74%, < %, < , 18, 20, 22, 27, 30, 35-42, 44, 45, 47-49, 51, 54, 59, 66, 67 19, 21, 23-26, 28, 29, 31-34, 43, 46, 50, 52, 53, 55-58, All-cause mortality for AKI by study type RCT [0.62, 1.08] 62%, , 23, 45, 47, 50, 51, 55, 57, 66, 67 Non-RCT [0.68, 0.81] 71%, < , 24-44, 46, 48, 49, 52-54, 56, All-cause mortality for AKI by sample size < >= [0.51, 0.73] 0.82 [0.76, 0.89] 62%, < %, < , 24, 25, 28, 29, 31, 32, 36, 37, 40-43, 46-52, 55-57, 61, , 18, 20-23, 26, 27, 30, 33-35, 38, 39, 44, 45, 53, 54, 58-60, 62, 66, 67