Timing of Renal Replacement Therapy Initiation in Patients with Septic Acute Kidney Injury; A Systematic Review and Meta-Analysis

Background; Acute kidney injury (AKI) is the most frequent complication seen in patients with septic shock and is an independent risk factor for death. Although renal-replacement therapy (RRT) is standard care for patients with severe septic AKI, the optimal timing of RRT initiation remains controversial. Methods; The PubMed, Cochrane, and Embase databases were searched from their inception to June 2021 to identify the ideal timing of RRT initiation in patients with septic AKI by comparing 28- and 90-day mortality rates. Results; Among a total of six studies including 1,058 patients, the 28-day mortality rate was signicantly lower in the early RRT-treated group compared to the late group [RR=0.69; 95% CI (0.51-0.94); P=0.018]. Moreover, among the ve studies including 938 patients, the 90-day mortality rate was also signicantly lower in the early RRT-treated group than the late group [RR=0.61; 95% CI (0.47-0.80); P=0.01]. In a subgroup analysis for continuous RRT (CRRT), we also found signicantly lower 28- and 90-day mortality rates in the early CRRT-treated group compared to the late group. Conclusion; This study showed that early initiation of RRT might reduce 28- and 90-day mortality compared with late initiation in septic AKI patients.


Introduction
Acute kidney injury (AKI) is a frequent complication among patients in the intensive care unit (ICU) with septic shock [1][2][3], because the kidney is one of the rst organs to sustain damage with AKI occurring in 51% of septic shock patients. Meanwhile, the mortality due to septic AKI remain substantial, despite the consistent efforts and advanced developments of medical technology [4]. Currently, septic AKI is associated with more severe hemodynamic instability, greater requirement of mechanical ventilation, longer hospital stay and higher hospital mortality compared with non-septic AKI [4,5].
Therefore, we conducted a meta-analysis study by collecting the results of recently published trials in order to explore whether the early initiation of RRT decreases the mortality of septic AKI patients.

Materials And Methods
This study was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Metaanalyses (PRISMA) guidelines [19] (Supplemental table 1).

Search strategy
We performed a comprehensive search of PubMed, EMBASE, the Cochrane Central Registry of Controlled Trials, and Web of Science from the inception of each database to June 2021 to identify randomized trials and cohort studies that assessed when RRT was initiated in septic patients with AKI. We restricted our search to clinical studies performed in adult populations and those that were published in the English language. Twelve keywords, which were "acute renal failure," "acute kidney injury," "acute kidney failure," "continuous renal replacement therapy," "renal replacement therapy," "dialysis," "renal dialysis," "hemo ltration," "sepsis," "septic shock," "time to treatment," and "time factors" were used in the search. The bibliographies of the obtained publications and the references of relevant reviews were also checked to ensure that no relevant studies were inadvertently omitted.

Study selection
Two reviewers (HO and NK) independently performed an initial eligibility screen of all retrieved titles and abstracts. Studies that reported original data that speci cally mentioned applying RRT in septic patients with AKI were selected for further review. Full-text reviews were independently performed by the same two reviewers with the following eligibility criteria: 1) observational cohort and/or randomized clinical trial (RCT) design; 2) adult septic patients; 3) diagnosis of AKI; 4) description of factors related to the timing of RRT initiation; and 5) description of mortality outcomes. Any discrepancies were resolved in a consensus discussion mediated by a third reviewer (YC). Our study's protocol was registered in PROSPERO. The authors did not seek speci c institutional review board or ethics committee approval for this study because the study was conducted with publicly available data obtained from online databases.

Assessment of risk of bias
The quality of the included RCTs was assessed using the Cochrane Collaboration Risk of Bias Tool [20] (Supplemental table   2). The quality of the observational studies was evaluated using the Newcastle-Ottawa Scale (NOS) [21] (Supplemental table   3). Studies with NOS scores <7 were considered to have high risks of bias [22,23].

Data extraction and statistical analysis
The author, publication year, study type, study location, RRT modality, RRT timing and mortality outcomes were extracted from each of the included studies. In addition, we contacted the authors of these studies for missing or incomplete data. The primary outcome was the 28-day mortality of patients in whom RRT was initiated. The 90-day mortality was the secondary outcome. A meta-analysis was performed using the package 'meta' and 'metafor' of the R package, version 3.6.0 (The R Foundation for Statistical Computing, Vienna, Austria). Pooled estimates of risk ratios (RRs) with 95% con dence intervals (CIs) were calculated using random effects models based on the restricted maximum-likelihood estimator of the between-study variance to take account for heterogeneity among studies. An additional subgroup analysis of CRRT and the impact of its initiation time on 28-and 90-day mortality was performed with the same protocol. Signi cant statistical heterogeneity was determined by the Q-statistic test (P<.10) and I-squared statistics (I2 ≥ 50%). A P value <.05 was set as the threshold of statistical signi cance. Publication bias was assessed using Egger's regression model, and visualized with a contour-enhanced funnel plot. Additionally, the trim and ll method was performed to investigate potential effects due to unpublished studies.

Study selection and quality assessment
We initially collected 5,566 articles from the three medical databases. After removing the duplicate articles, 4,478 remained.
HO and NK then screened the titles and abstracts of the articles and deleted 4,439 irrelevant studies. Then, 39 articles underwent full-text assessment. Thirty-three articles were excluded as 12 were review or case studies and 21 were either conference papers, not written in English, or not full-text papers. Therefore, only 6 articles were chosen for the nal analysis (Table 1 and Figure 1). Of the 6 articles, one was a RCT [14] and ve were retrospective cohort studies [13,[15][16][17][18]. The results of our quality assessment are shown in Supplemental Table 1 and Supplemental Table 2. The RCT did not have a high risk of bias for sequence generation, addressing outcome data, selective reporting, or funding source. All of the retrospective studies had total bias scores of at least 7 based on NOS, which indicates a low risk of bias.  (Figure 3).

The effect of early CRRT on mortality
Among the six studies, only one included patients undergoing intermittent hemodialysis. However, we explored the impact of early CRRT initiation on 28-and 90-day mortality, and compared 28-day mortality between the early and late groups with 5 studies and 90-day mortality with 4 studies (excluding the study of Shum et al.  [14] found no signi cant difference in 90-day mortality between patients who received early and late RRT. Consistently with the ndings of their meta-analysis on septic AKI patients. Apparently, recent RCTs and meta-analysis did not show early RRT initiation as improving survival rates in critically ill patients with AKI including septic AKI.
However, it does not mean that we need to delay RRT as much as possible. Gaudry et al. [25] stated that their study should not be interpreted as suggesting that a "wait and see" approach is safe for all patients. Indeed, careful surveillance is mandatory when deciding to delay RRT in patients with severe AKI so that RRT can be initiated as soon as any complications are detected. Rather, the issue is how to identify a reasonable time point for beginning treatment. Moreover, Barbar et al. [14] demonstrated that the failure stage was not necessarily intended to identify patients who would require RRT, and a delay of only 48 hours may not be su ciently long enough to allow renal function to recover in some patients or to cause a difference between early and late RRT initiation.
We realized that each study used different de nitions to describe early and late RRT [14,[24][25][26]. Unlikely RCTs, most of the retrospective cohort studies showed that early initiation of RRT decreased mortality compared with delayed initiation [13,15,17,18]. But all the patients included in this study were those who underwent RRT. Hence, the early RRT-treated group might have included several patients who would not need RRT if followed with close observation, and that might explain the better prognosis of the early group. Meanwhile, both Gaudry[25] and Barbar et al. [14] recognized that quite a few patients did not undergo RRT in the late group, with their renal functions spontaneously resolving. Barbar et al. [14] pointed out that 70 patients among the total of 242 patients in the late group (28.9%) did not receive RRT, which means "delay of RRT" in these studies might not be a true "delay" in practical condition, suggesting that could be one of the limitations in the Barbar's study.
With the earlier initiation of dialysis in AKI, we expect to improve acid-base, electrolyte, and uid balance, and thereby prevent more aggravating complications of AKI and perhaps also enhance the removal of toxins such as nonspeci c proin ammatory or anti-in ammatory mediators through convection and adsorption [28]. Moreover, uid therapy and usage of other drugs may be easily done through the uid management of RRT [28]. These points suggest that RRT may be initiated without delay, when deemed necessary.
The strength of this meta-analysis was that it was conducted with only septic AKI patients that underwent RRT as well as CRRT, and we searched for all the related papers published until June of 2021. Several limitations still need to be discussed.
First, small-sized studies were included. Most of the studies on RRT timing were for critically ill patients, not septic AKI patients. Only a few studies included septic AKI patients. Second, substantial heterogeneity was shown even in the few studies, which suggests that the ndings of the current study should be interpreted with caution.

Conclusions
In conclusion, this study found that early initiation of RRT and CRRT might reduce 28-and 90-day mortality compared to late initiation in septic AKI patients. However, continuous studies are required with different criteria to identify the optimal timing of RRT and CRRT in septic AKI patients in the future.

Funding
This research was not supported by any grant.

Availability of data and materials
The data that support the ndings of this study are available from the corresponding authors upon reasonable request.

Ethics approval and consent to participate
This study did not seek speci c institutional review board or ethics committee approval because the study was conducted with publicly available data obtained from online database

Consent for publication
We obtained consent to publish from the participants for the reporting of individual patient data.  Meta-analysis of risk ratios for 28-day mortality in the RRT group. Higher risk ratios indicate higher 28-day mortality in the early RRT-treated group. RRT, renal replacement therapy; RR, risk ratio; CI con dence interval Meta-analysis of risk ratios for 90-day mortality in the RRT group. Higher risk ratios indicate higher 90-day mortality in the early RRT-treated group. RRT, renal replacement therapy; RR, risk ratio; CI con dence interval Figure 4 Meta-analysis of risk ratios for 28-day mortality in the CRRT group. Higher risk ratios indicate higher 28-day mortality in the early CRRT-treated group. CRRT, continuous renal replacement therapy; RR, risk ratio; CI con dence interval Figure 5 Meta-analysis of risk ratios for 90-day mortality in the CRRT group. Higher risk ratios indicate higher 90-day mortality in the early CRRT-treated group. CRRT, continuous renal replacement therapy; RR, risk ratio; CI con dence interval