Effect of care bundles for acute kidney injury: A systematic review and meta-analysis

Purpose Acute kidney injury (AKI) is frequent among in-hospital patients with high incidence and mortality. Implementing a series of evidence-based AKI care bundles may improve patient outcomes by reducing changeable standards of care. The aim of this meta-analysis was therefore to appraise the influences of AKI care bundles on patient outcomes. Materials and methods We explored three international databases (PubMed, Embase, and Cochrane Central Register of Controlled Trials) and two Chinese databases (Wanfang Data and China National Knowledge Infrastructure) for studies from databases inception until November 30, 2022, comparing the impact of different AKI care bundles with usual standards of care in patients with or at risk for AKI. The study quality of non-randomized controlled trials and randomized controlled trials was evaluated by the NIH Study Quality Assessment Tool and the Cochrane risk of bias tool. Heterogeneity between studies was appraised by Cochran’s Q test and I2 statistics. The possible origins of heterogeneity between studies were assessed adopting Meta-regression and subgroup analyses. Funnel plot asymmetry and Egger regression and Begg correlation tests were performed to discover potential publication bias. Data analysis was completed by software (RevMan 5.3 and Stata 15.0). The primary outcome was short- or long-term mortality. The secondary outcomes involved the incidence and severity of AKI. Results Sixteen studies containing 25,690 patients and 25,903 AKI episodes were included. In high-risk AKI patients determined by novel biomarkers, electronic alert or risk prediction score, the application of AKI care bundles significantly reduced the AKI incidence (OR, 0.71; 95% CI, 0.53–0.96; p = 0.02; I2 = 84%) and AKI severity (OR, 0.59; 95% CI, 0.39–0.89; p = 0.01; I2 = 65%). No strong evidence is available to prove that care bundles can significantly reduce mortality (OR, 1.16; 95% CI, 0.58–2.30; p = 0.68; I2 = 97%). Conclusions The introduction of AKI care bundles in routine clinical practice can effectively improve the outcomes of patients with or at-risk of AKI. However, the accumulated evidence is limited and not strong enough to make definite conclusions.


Introduction
Acute kidney injury (AKI) is frequent among in-hospital patients with high incidence and mortality [1].Owing to the absence of effective pharmacotherapies for AKI treatment, existing international guidelines focus on early recognition and timely intervention [2].Advances in diagnostic tools, electronic alerts and novel renal biomarkers, such as tissue inhibitor of metalloproteinases-2 and urine insulin-like growth factor-binding protein 7, have enabled the early identification of AKI [3,4].However, early diagnosis does not improve patient outcomes by physician behavior, possibly because of great variations across routine clinical care.Care bundles, described as a set of simple evidence-based practices, have been proposed as tools to improve patient outcomes by reducing variability in standards of care [5].The core elements of AKI bundles include the amelioration of hemodynamic status, avoidance of nephrotoxic agents, and prevention of hyperglycemia [6].In several clinical situations, care bundles have been applied to patients with or at risk for AKI.Some isolated articles or reviews have reported patient-related benefits of care bundles including lower mortality and inhibition of AKI progression such as the original report by Kolhe et al. [7] and narrative review by Sykes et al. [8].However, in reality, the overall care bundle compliance remains poor [7].On the other hand, other studies report contrary findings.Joslin et al. found significant improvements in recognition, fluid assessment and nephrotoxic cessation following introduction of AKI bundle; however, these improvements were not correlated with enhanced patient outcomes [9].
The objectives of this study were to appraise the effects of AKI care bundles on patient outcomes, summarize the bundle components, and identify the key bundle care implementation strategies, which can improve compliance with AKI care bundles.

Methods
The study protocol followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [10,11].

Search strategy
Two separate investigators (Shuzhen Zhang, Yixin Chen) searched the following electronic databases from database inception until November 30, 2022: PubMed, Embase, Cochrane Central, Wanfang Data, and China National Knowledge Infrastructure.Irrespective of the publication status or language, we searched for studies comparing the effects of AKI care bundles with usual standard of care in patients with or at risk for AKI.Our search strategy included the selection of key words and subject headings related to AKI and care bundles, including: "acute kidney injury", "acute kidney failure", "acute renal failure", "patient care bundle", "care checklist", "prevention bundle", etc. S1 Table shows the details of the search strategy.The references in the studies selected were further used to search for other related studies.

Inclusion and exclusion criteria
Inclusion criteria for this meta-analysis were: 1) in-hospital adult patients with or at high risk of AKI (identified using a laboratory or electronic alert system); 2) patients managed using AKI care bundles; 3) at least one AKI outcome (short-and long-term mortality, AKI incidence, or AKI severity); and 4) (quasi) randomized trials or observational studies.An AKI care bundle was a series of simple evidence-based diagnostic and therapeutic interventions collectively planned and performed in line with the Institute for Healthcare Improvement.The basic elements of the AKI bundle were set as advised by the guidelines of Improving Global Outcomes (KDIGO) or other relevant AKI guidelines [12].
According to the recommendations of the KDIGO Guideline, studies were excluded if:1) they were studies from letters/editorials, conference abstracts, comments, or reviews; 2) studies without sufficient information for analysis; and 3) there were no measured outcomes.

Study selection
Two investigators independently filtered all titles and abstracts to select pertinent articles, retrieved the full-text copies of the relevant studies for closer inspection to determine study eligibility and resolved any disagreement by consulting with a third investigator.
Initially, 4912 citations were identified.86 potentially pertinent full-text studies were read for further appraisal after reduplication and filtering all titles and abstracts.Finally, 16 studies were involved in this meta-analysis.See the PRISMA diagram for the detailed procedure (Fig 1).

Data extraction
Study characteristics (i.e., year, design, setting, and country), the number of participants, the intervention, comparison groups, and the outcomes were retrieved from individual studies by two investigators and summarized in evidence tables.The characteristics of the AKI bundles were also summarized.
Implementation strategies included in care bundles were categorized by referring to the taxonomy developed by the Cochrane EPOC [13].
When studies that met the inclusion criteria lacked significant primary data, we attempted to contact the authors.

Quality assessment and data synthesis
The bias risk in cohort studies was appraised by the Newcastle-Ottawa scale [14].This evaluation scale rates the quality of the included studies pursuant to three groups: the selection of the study groups, the comparability of the groups, and the ascertainment of the study outcomes.Scores of 7-9, 4-6, and <4 were classified as having a low, moderate, or high risk of bias, respectively.We also used the ROBINS-I to evaluate the risk of bias in cohort studies, which address key criteria such as selection bias, exposure measurement, blinding, completeness of outcome data and selectivity of reporting [15].Bias risk in randomized controlled trials (RCTs) was appraised by Cochrane Risk of Bias Assessment Tool [16], review authors' judgements about each risk of bias item for each included study.To identify any potential publication bias, both statistically and visually, Egger regression and Begg correlation tests and funnel plot asymmetry were conducted.
This meta-analysis was completed utilizing the Cochrane Collaboration Review Manager (RevMan 5.3) and the Cochran-Mantel-Haenszel analysis (odds ratio [OR] reported with the 95% confidence interval [CI]).Data were combined using a random effects model, given the expected diversity of clinical approaches and methodologies used in the included studies.Cochran's Q test and I 2 test statistics were applied to evaluate the heterogeneity between studies [17] and analyzed using STATA version 15.0 software (STATA Corp, College Station, Texas).I 2 index less than 25% indicates low heterogeneity, between 25% and 75% indicates moderate heterogeneity, and more than 75% indicates high heterogeneity.Due to heterogeneity in the studies, a random effects model was used to perform meta-analysis.Interactions between subgroups were tested by Meta-regression models, and a p value � 0.05 was considered as statistical significance.Six trial features (study design, country, care bundle compliance, setting, care bundle components and implementation strategies) were defined as covariates.Univariate meta-regression was used to assess the association between each covariate and each outcome (short-or long-term mortality, AKI incidence and AKI severity).Subgroup analyses were carried out according to the source of heterogeneity.
Two studies [9,26] used audit and feedback, and five studies [4,6,21,22,28] used multidisciplinary teams or nephrological consultation as strategies for implementation of the care bundle.

Quality assessment
The bias risks of cohort studies involved in this study low -to-moderate risk.The quality of the involved studies was appraised by the Newcastle-Ottawa Scale (Table 2).We also used the ROBINS-I to evaluate the risk of bias in cohort studies, the detailed results are available in Table 3.Four RCTs were involved in the analysis (Fig 2 ) and assessed using the Cochrane Risk of Bias Assessment Tool.

AKI incidence
Ten studies evaluated changes in the AKI incidence following implementation of the care bundles [4, 6, 18-23, 25, 29] in 19,424 patients.Despite the high inconsistency in the study findings [Q (9) = 54.56;p < 0.00001; I 2 = 84%], in the random effects method results, most studies showed a significant decrease in the AKI incidence, the overall OR was 0.71 (0.53-0.96; p = 0.02) (Fig 3).The midpoint of each line and the length of the line respectively show the mean difference and its 95% confidence interval for each study.The size of squares represent the weight which that study had on the overall summary effect.The middle of the diamond sign shows the summary effect and the horizontal diameter of it represent 95% confidence intervals of the summary effect.We conducted a sub-analysis of studies from European countries (subgroup 1) and other countries (subgroup 2) (Fig 4).The studies from European countries continued to demonstrate high heterogeneity with an I 2 of 83.5%; however, the heterogeneity was markedly decreased for studies from other countries, with an I 2 of 0. Studies from other countries (OR = 0.487; 95%CI, 0.345-0.688;p < 0.001) continued to report decreased AKI occurrence.Studies from European countries (OR = 0.847; 95%CI, 0.621-1.155;p = 0.293) showed a non-significant association between AKI incidence and implementation of the AKI care bundles.Meta-regression analysis showed there were no statistically significant associations (p > 0.05) between AKI incidence and any of the following intervention characteristics: care bundle compliance, study design, setting, care bundle components and implementation strategies.

AKI severity
A pooled analysis of 8 studies (including 4,479 patients) [4, 6, 18-21, 23, 29] revealed that based on the Q and I 2 index, there was moderate heterogeneity in the studies [Q (7) = 20.28;p = 0.005; I 2 = 65%], so in the random effects method results, implementation of a care bundle bring on a significant decreases in moderate severe AKI occurrence (p = 0.01).The overall OR was 0.59 with 95%CI (0.39-0.89) (Fig 6).Meta-regression analysis showed there were no statistically significant associations (p > 0.05) between AKI severity and any of the following intervention characteristics: country, care bundle compliance, setting, care bundle components and implementation strategies.The sub-analysis aimed to distinguish uncontrolled before-after studies (subgroup 1) from prospective observational studies and RCTs (subgroup 2) (Fig 7).These results indicated that the uncontrolled before-after studies continued to demonstrate high heterogeneity with an I 2 of 75.4%; however, heterogeneity was markedly decreased for the prospective observational studies and RCTs, with an I 2 of 0. Publication bias could not be appraised due to the number of included studies is small.
Statistical tests Begg regression (p = 0.033) showed publication bias, and the funnel plot showed asymmetry (Fig 10).Although publication bias was not discovered by Egger correlation (p = 0.712) tests.

Discussion
We implemented a systematic search of the literature and recognized 16 studies involving 25,690 patients and 25,903 AKI episodes that assessed whether care bundles were valid in improving the outcomes of patients with or at-risk of AKI.This meta-analysis did not reveal any strong evidence that could prove that AKI care bundles significantly reduce mortality in all patients.Patients exposed to the AKI care bundles had a significantly reduced AKI incidence as compared with the usual standards of care.Moreover, a lessened incidence of moderate-to-severe AKI was noted in high-risk AKI patients.
Table 3.The ROBINS-I were used to assess risk of bias for the cohort studies.

AKI incidence
Compared to the general incidence of AKI, the utilization of an AKI care bundle in patients was associated with a lower incidence of AKI.Studies conducted in Egypt, the USA, and China showed decreases in AKI occurrence with low heterogeneity [6,21,29].However, Meersch et al. failed to demonstrate any benefit from implementing care bundle in RCTs (OR = 0.70; 95%CI, 0.43-1.15;P = 0.16) [4,[18][19][20].This reduction in terms of patient outcome could not be substantiated using randomized data due to inadequate sample sizes and variability in studied outcomes.
Each individual should receive optimal kidney-sparing treatment from a nephrological perspective [2].The early detection of serum creatinine to diagnose AKI in high-risk patients is generally considered insufficient; furthermore, newer diagnostic tools, such as electronic alerts (E-alerts), stress/damage biomarkers, and risk stratification, have not yet been explored [30].
In recent years, E-alerts have garnered significant attention [31], as they can indicate early or impending AKI episodes and prompt rapid prevention measures along with early clinical evaluation and treatment strategies [32].Regarding biomarkers, Schanz et al. proved that urinary tissue inhibitor of metalloproteinases-2(TIMP-2) and insulin-like growth factor-blinding protein 7 (IGFBP7) are beneficial in screening high-risk populations [4].

AKI severity
Previous studies have demonstrated a significant reduction in moderate-to-severe AKI following the implementing of a care bundle.However, uncontrolled before-after studies conducted by Schanz et al. failed to show any benefit from implementing the care bundle [4,6,23], perhaps attributable to insufficient statistical power to detect differences in AKI progression.
Several technology-based interventions have been investigated to improve AKI outcomes, including the development of a working conceptual model for studying the kinetics of AKI reversal and renal recovery [33].However, a recent cluster-randomized trial [34] found no improvement in clinical outcomes when hospitals were randomized to introduce an alerting system.This could be due to the lack of the associated authoritative guidelines, which results in the lack of preventive or treatment measures in hospitals, and the lack of targeted education.Care bundles can be used to improve care and make services more uniform.Substantial evidence supports the use of care bundles in hospital settings, such as emergency departments, wherein they have greatly improved the outcomes of patients with sepsis and community acquired pneumonia [35,36].AKI care bundles enable non-nephrologists to take action rapidly to prevent and treat AKI [35].Therefore, early recognized AKI can be reversed by ensuring that the patients receive adequate fluid and medication to avoid the worsening of kidney function or any chance of toxicity.

Short-and long-term mortality in AKI
The implementation of AKI care bundles did not result in a significant reduction in short-or long-term mortality based on both randomized controlled trials (RCTs) and non-RCTs; however, the number of involved studies is small, thus the interpretation of findings should be cautiously done.Selecting the appropriate patient may be crucial in successfully implementing an AKI care bundle.According to the studies by Hodgson et al. [22] and Koeze et al. [23], both studies involved in this review demonstrated a significant reduction in mortality following the implementation of an AKI care bundle.However, it should be noted that these studies enrolled large numbers of patients, which may have resulted in inaccurate outcomes.
Generally, inpatients are at a higher risk of developing AKI, which is associated with high mortality [1,37,38].Consensus guidelines for AKI have recommended prompt treatment, including the maintenance of perfusion pressure, improvement of fluid status, avoidance of nephrotoxins, and prevention of hyperglycemia [2,12].Since AKI can be induced by hypovolemia and hypervolemia, optimization of fluid status is crucial [39].The target blood pressure should be controlled at a mean arterial pressure >60-65 mmHg using vasopressors [12,40].In high-risk patients for AKI, the doses of nephrotoxic drugs and other drugs should be strictly controlled [12].Recently, an observational and retrospective study [41] analyzed 5 interventions of the nephrology team with the potential to meliorate AKI outcomes, including the fluid adjustment and nephrotoxic withdrawal, etc.It reported that only fluid management lowered risk of starting renal replacement therapy (RRT) and progression to AKI stage 3, and none of the interventions reduced the risk of death in AKI patients.Hence, in the absence of evidence to endorse the application of other interventions to reduce AKI mortality.It suggested that multidisciplinary approach was demanded due to the multifactorial nature of AKI in hospitalized patients.

Strengths and limitations
Limited systematic studies have been conducted to evaluate the effectiveness of AKI care bundles; prior our study, only Schaubroeck et al. [42] presented a comprehensive review and meta-analysis on patient outcomes associated with AKI care bundle implementation.In particular, we extensively searched Chinese databases in order to include similar studies that could potentially serve as professional guidelines for Chinese hospitals.Our critical assessment of existing studies on AKI care bundles can inform further research planning, while our interpretation of the study results can aid in evaluating the local implementation and generalizability of AKI care bundles.
The effectiveness of this study is affected by the limited number of randomized studies, the small sample numbers in some of the individual studies and the wide differences in endpoints of studies.Furthermore, there were disparities in the quality of clinical care for AKI among included studies, with poorer quality associated with worse outcomes.Given that AKI often occurs under the management of healthcare professionals who are not nephrologists, it is crucial to enhance awareness about this condition among all healthcare professionals.Most studies applied a before-after design which has inherent limitations such as susceptibility to longterm trends bias and potential overestimation of the true effects attributed to AKI care bundles [43,44].

Conclusion
Taken together, implementation of AKI care bundles in conventional clinical practice can effectively improve the prognosis of patients with or at risk of AKI.However, the accumulated evidence is limited and not robust enough to make definable conclusions.

Fig 2 .Fig 3 .
Fig 2. The Cochrane Risk of Bias Assessment Tool was used to assess risk of bias for the randomized, controlled trials.Red, high risk of bias; Green, low risk of bias; Yellow, uncertain risk of bias.https://doi.org/10.1371/journal.pone.0302179.g002

Table 2 . The Newcastle-Ottawa Scales were used to assess risk of bias for the cohort studies.
Each domain was rated on a scale of zero or one star, except comparability, which can be awarded up to two stars.0, High or unclear risk of bias; 1 or 2, low risk of bias (scores � 7-9, 4-6, <4 are considered low, intermediate, and high risk, respectively).