Association of Malnutrition with Risk of Acute Kidney Injury: A Systematic Review and Meta-Analysis

Background Acute kidney injury (AKI) is a complex clinical syndrome of hospitalization that may be affected by undernutrition and metabolic changes. The aim of this meta-analysis was to systematically assess the association between malnutrition and the risk of prevalent AKI. Materials and Methods We searched PubMed, Embase, Ovid MEDLINE, Web of Science, and Chinese databases (WANFANG, VIP, and CKI) from database inception until May 1, 2023, for studies evaluating the association of malnutrition with the risk of AKI. Summary odds ratios (ORs) were estimated using a random-effects model. Results We identified 17 observational studies, which included 273,315 individuals. Compared with patients with normal nutritional status, those with malnutrition had a 125% increased risk of prevalent AKI (pooled ORs, 2.25; 95% confidence interval, 1.80–2.82). Malnutrition was also significantly associated with prevalent AKI across all subgroups when subgroup analyses were performed on covariates such as region, study design, age, sample size, malnutrition assessment method, patient characteristics, covariate adjustment degree, and risk of bias. Meta-regression models demonstrated no significant differences in AKI risk between patients with malnutrition and without malnutrition. Conclusions Our results suggest that malnutrition may be a potential target for AKI prevention. However, well-designed studies with ethnically or geographically diverse populations are needed to evaluate strategies and interventions to prevent or slow the development and progression of AKI in malnourished individuals.


Introduction
Acute kidney injury (AKI) is a complex clinical syndrome caused by a range of factors in various clinical settings and is characterized by a sudden but often reversible decline in the estimated glomerular fltration rate (eGFR) and the resultant accumulation of metabolic waste products [1].Annually, up to 13.3 million people worldwide are afected by AKI, and approximately 1.7 million die of AKI [2].Although renal replacement therapy is widely used in critically ill patients with AKI, these patients still face an increased risk of mortality and irreversible kidney function deterioration, with rapid progression to chronic kidney disease (CKD) [3].Terefore, the identifcation of efective predictive factors for AKI prevention remains critical.
Malnutrition is defned as a physical state of imbalanced nutrition that is signifcantly associated with increased length of stay, complication rates, healthcare costs, and mortality in hospitalized patients [4,5].A recent study that collected data from 238 million emergency department visits reported the prevalence of malnutrition diagnosed by the International Classifcation of Diseases (ICD), 9th Edition, and diagnosis codes in elderly people increased from 2.5% in 2006 to 3.6% in 2014 [6].Meanwhile, approximately 15%-80% of hospitalized patients and more than 50% of renal inpatients were reported to be at undernutrition risk [7,8].AKI is a common complication of hospitalization and is afected by undernutrition and metabolic changes.Terefore, malnourished patients may be more prone to progression to AKI [9].
Emerging evidence has suggested that malnutrition is a predictor of AKI.A retrospective study of 46,549 Chinese inpatients demonstrated that a malnutrition level ≥3, as assessed using the nutritional risk screening 2002 (NRS-2002), was strongly associated with prevalent AKI [10].Another retrospective multicenter study of 3,185 patients with acute coronary syndrome found that NRS-2002 ≥3 was associated with an approximately 1.36-fold likelihood of developing AKI [11].A study of 4,386 patients who underwent coronary angiography showed that moderate malnutrition, evaluated using the Controlling Nutritional Status (CONUT) score and the prognostic nutritional index (PNI), was not associated with an increased risk of AKI [12].Given the contradictory relationship between malnutrition and the risk of AKI in previous studies, a meta-analysis is imperative to assess the association between malnutrition and the risk of incident AKI.Terefore, the present systematic review and meta-analysis aimed at providing information on the association between malnutrition and the risk of AKI based on current evidence, according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.

Registration of Review
Protocol.We reported our metaanalysis review in compliance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [13] (PRISMA checklist; see Supplementary Appendix S1), and the protocol for this systematic review was registered in INPLASY (registration number: INPLASY202320062).

Study Selection.
All articles retrieved using our search strategies were assessed by screening titles and abstracts.Study selection was conducted by two authors independently.In cases of disagreement on the inclusion or exclusion of studies, consensus was reached through discussion with a third author.

Inclusion and Exclusion Criteria. Te inclusion criteria
were observational studies, studies investigating the association between malnutrition and AKI risk, studies on populations of any sex or ethnicity, and studies with a clear diagnosis of malnutrition and AKI.Animal studies, non-English-language studies, and study types such as review, conference, letter, case report, and comments were excluded.
2.5.Data Extraction.Data were independently extracted from the included studies by two authors.Any disagreement was resolved by consensus between the two authors.Te following data were extracted from the included studies: year of publication, country, participant characteristics, sample size, malnutrition assessment tools, adjusted variables, and study quality.Disagreements between the two authors were resolved through discussion until consensus was reached.
2.6.Quality Assessment.In the evaluation of potential biases within the selected studies, two independent reviewers utilized the Risk of Bias in Nonrandomized Studies of Interventions (ROBINS-I) tool [14].Te ROBINS-I tool, comprised of seven domains, facilitated the assessment of bias arising from factors such as confounding, selection of participants, exposure assessment, misclassifcation during follow-up, missing data, outcome assessment, and selective reporting.Te risk associated with each domain was systematically assessed by the reviewers, who assigned one of the following ratings: low, moderate, serious, critical, or no information.In the event of any discrepancies in ratings, a senior investigator was brought in to arbitrate and resolve the disagreement.Detailed descriptions and decisionmaking parameters for each of the ROBINS-I tool's domains are comprehensively outlined in Supplementary Appendix S3.Tis ensures a transparent and replicable process for bias assessment, supporting the robustness of our study's methodology.

Meta-Analysis.
For data analysis, we used the "metaprop" function in the "meta" package of R (version 4.1.3).To enable accurate calculation and input of efect sizes and thus ensure the reliability of our results, we used the "metagen" function.In our research, we dealt with dichotomous data, for which we calculated the odds ratios (ORs) and their corresponding 95% confdence intervals (CIs).In addition, to evaluate the heterogeneity among the studies, we utilized the chi-squared test for Cochran's Q statistic and I 2 .Both of these tools helped us quantify the extent of variability in efect estimates due to heterogeneity rather than chance.We capitalized on both fxed-and random-efects models, depending on the level of heterogeneity.In instances of signifcant heterogeneity, the preference was given to the random-efects model.More specifcally, we employed the Der Simonian and Laird method within the context of the random-efects model.Tis method considers the variance of the efect sizes across the studies, providing a more accurate computation of weights for meta-analysis.Tis approach thereby ofers a more reliable estimate of the overall efect size, enhancing the credibility of our fndings.

2
International Journal of Clinical Practice Moreover, to identify potential sources of heterogeneity, we conducted subgroup and meta-regression analyses.We delineated subgroups based on various factors, including region, malnutrition assessment method, participant demographics, sample size, the degree of covariate adjustment, and study quality.Tis comprehensive analysis enabled us to delve deeper into the infuence of these factors on our results.In addition, to evaluate whether the overall estimate depended on a single study, we conducted a sensitivity analysis by removing any of the included studies.Moreover, we used funnel plots to visualize the publication bias of the study data and further used Egger's test to objectively assess potential publication bias.

Study Selection.
A fowchart of the literature selection process is presented in Figure 1.Trough database searches of PubMed, Embase, Ovid MEDLINE, Web of Science, and Chinese databases (WANFANG, VIP, and CKI), 3593 articles were selected for subsequent fltering.Of these, 565 duplicate studies were excluded.After checking the title and abstract of each paper and excluding inconsistent literature types, 52 studies were found to be pertinent to the research topic.Among them, seven articles were excluded due to inaccessible data, twenty-three articles were excluded after a detailed review of the full text, three were review studies, and two were case reports.Finally, 17 original studies were included in the meta-analysis.

Overall Assessment of Evidence
Quality.Utilizing the ROBINS-I instrument for evaluation, it was discerned that four studies presented a moderate overall risk of bias.Eleven studies were identifed as having a serious overall risk of bias.
In addition, two studies were found to show a serious overall risk of bias.Supplementary Appendix S3 presents the detailed assessment of the risk of bias for each domain.

Association of Malnutrition with Prevalent AKI.
Compared with patients with normal nutritional status, those with malnutrition had a 125% increased risk of prevalent AKI (pooled OR, 2.25; 95% CI, 1.80-2.82).Given their combined I 2 � 97.0% and P < 0.01 and that the included studies were derived from multiple study settings and their populations, relevant efect values were pooled using a random-efects model (Figure 2).Owing to the heterogeneity of the aggregated results, it was necessary to explore the sources of heterogeneity.Subgroup analyses were performed on covariates, such as region, study design, age, sample size, malnutrition assessment method, patient characteristics, covariate adjustment degree, and study quality, demonstrating that malnutrition was signifcantly associated with prevalent AKI across all subgroups.Detailed results are presented in Table 2 and Supplementary Figures 1-8.A univariate metaregression analysis was then used to examine the potential source of heterogeneity, which revealed no signifcant diferences in AKI risk between patients with and without malnutrition in terms of region (standard error (SE) � 0.29, P � 0.88), study design (standard error (SE) � 0.44, P � 0019), sample size (SE � 0.40, P � 0.20), mean or median patient age (SE � 0.29, P � 0.60), malnutrition assessment method (SE � 0.29, P � 0.73), patient characteristics (SE � 0.28, P � 0.62), adjusted/unadjusted confounders (SE � 0.30, P � 0.51), or risk of bias (SE � 0.57, P � 0.47) (Table 3).

Publication Bias and Sensitivity Analysis.
To test the publication bias of the included studies, we used funnel plots and Egger 's test, and the results showed that the funnel plots assessing the risk of publication bias showed asymmetry, but Egger' s test result was t � 0.16 and P � 0.87, suggesting that the included studies had no publication bias (Supplementary Figure 9).Sensitivity analyses were also performed to determine the stability and reliability of the results and the extent to which the individual studies infuenced the results.Te efect size after excluding individual studies was in agreement with the total combined efect size, indicating that our results were robust to a certain extent (Supplementary Figure 10).

Discussion
To the best of our knowledge, the present meta-analysis of the association between malnutrition and the risk of prevalent AKI is the most comprehensive assessment of this association to date.In this meta-analysis of 17 observational studies, with aggregate data from 273,315 patients from diferent regions, we found that malnutrition was signifcantly associated with an approximately 2.25-fold increased risk of prevalent AKI.Te results of our meta-analysis indirectly support the concept that interventions against malnutrition may be efective targets for the prevention of AKI.
To date, the possible mechanisms underlying the observed association between malnutrition and an increased International Journal of Clinical Practice risk of AKI are currently unclear, but they can be explained by several factors.Albumin, a popular biomarker of nutritional status in clinically stable conditions, is the most abundant circulating protein and plays an essential role in antioxidant, anti-infammatory, and antiplatelet aggregation activities [29][30][31].Low albumin levels in malnourished patients may contribute to the development of AKI through the deterioration of endothelial function and oxidative infammatory pathways [32,33].Yu et al. reported a higher incidence of AKI in patients with hypoalbuminemia (serum albumin level <3.4 mg/dL) (10.7%) than that in those with normal albumin levels (4.1%) [34], and the results from a meta-analysis of 68,000 participants across a diverse range of settings confrmed that hypoalbuminemia was an independent predictor for the development of AKI [34,35].In addition, hypercholesterolemia and low high-density lipoprotein levels, which are indicators of underlying malnutrition, have long been regarded as important risk factors for the development and progression of cardiovascular diseases [36][37][38], whereas hypocholesterolemia has been reported to be signifcantly associated with a 1.5-fold risk of 30-day mortality and a 1.3-fold risk of 3-year mortality in patients with coronary artery disease undergoing percutaneous coronary intervention (PCI) [39].Previous studies have also demonstrated that a low preoperative high-density lipoprotein cholesterol concentration was associated with an increased risk of AKI after cardiac surgery, and low cholesterol levels were signifcantly associated with worse survival in patients with AKI [40][41][42][43].
According to the subgroup analysis, the source of heterogeneity is partly explained by the diferences in patient characteristics.Eleven studies explored the association of malnutrition with the risk of CA-AKI among patients undergoing CAG or PCI.CA-AKI is a common phenomenon in patients following contemporary PCI and is characterized by an abrupt decline in eGFR after intravascular administration of iodinated contrast media.Studies reported that CA-AKI had an incidence of 6%-18% in patients  International Journal of Clinical Practice   International Journal of Clinical Practice undergoing PCI [44][45][46] and that it was associated with an approximately 2.0-fold increased risk of the 2-year rate of net adverse clinical events [47].Tus, efective early screening and preventive strategies for those with high-risk CA-AKI are critical.In the present meta-analysis, 11 studies with 24,843 patients were included to explore the role of malnutrition in the risk of CA-AKI; it was found that malnutrition was associated with a 2.09-fold increased risk of CA-AKI.In addition, malnourished patients with CA-AKI had a signifcantly higher risk of all-cause mortality than patients with malnutrition.A multicenter prospective cohort of 2,083 patients undergoing PCI demonstrated that malnourished patients with CA-AKI had a higher risk of allcause mortality than did those without [19].Terefore, clinicians should assess and monitor the nutritional status of patients undergoing PCI.Malnutrition was also associated with a high risk of AKI in patients undergoing abdominal surgery.Sim et al. studied 3,543 patients who underwent colorectal cancer surgery and found that a high preoperative PNI was signifcantly associated with a low risk of postoperative AKI (OR, 0.96; 95% CI, 0.93-0.99,P � 0.003) [48].
Another study of 817 patients who underwent hepatectomy for hepatocellular carcinoma demonstrated that a high preoperative PNI was signifcantly associated with a lower prevalence of postoperative AKI (OR, 0.92; 95% CI, 0.85-0.99;P � 0.021) [49].Tus, more attention should be paid to preoperative nutritional management, which can provide useful information about postoperative AKI and prognosis in patients undergoing abdominal surgery.
Another important source of heterogeneity can be partly explained by the diferences in malnutrition assessment methods.To date, there is a lack of unifed and standard measurement tools for malnutrition; thus, the prevalence of malnutrition varies when diferent measurement tools are used.Fortunately, the results from the subgroup analysis showed that malnutrition was associated with an increased risk of AKI across all groups of assessment tools for malnutrition.Te NRS-2002 score, developed by the European Society for Clinical Nutrition and Metabolism, has been widely used in clinical malnutrition screening in various diseases, including CKD.Te parameters of the NRS-2002 score, such as decreased body mass index and intensive care admission, were also signifcantly associated with the occurrence of AKI [10,50].In our study, malnutrition assessed by NRS-2002 was associated with a 2.38-fold risk of AKI, and it showed a robust ability to predict AKI, with areas under the curve of 0.67 for the univariate model and 0.78 for the multivariate model [10].Te CONUT score, which is calculated using the serum albumin level and lymphocytes, is an efcient and simple tool for detecting malnutrition.Pooled analysis demonstrated that malnutrition assessed using the CONUT score was associated with a 2.37-fold risk of AKI.In addition, Wei et al. found that, compared with the mild malnutrition group, the moderate-to-severe malnutrition group showed a higher risk of CA-AKI [28].Te PNI, which is calculated using the serum albumin level and total lymphocyte count, is another tool for assessing nutrition and refects chronic infammation and immunity [51].Malnutrition assessed by the PNI was associated with a 1.79-fold increased risk of AKI.Yu et al. demonstrated that each decrease in PNI score was associated with a 1.8% increased risk of AKI [34].
In our meta-analysis, the source of heterogeneity was investigated through regression analysis.Intriguingly, our fndings indicate that study design could potentially be a notable source of heterogeneity.Specifcally, there was a signifcant diference between cohort studies and crosssectional studies (SE � 0.44, P � 0.019).Tis could be attributable to the fundamental diferences in methodology between these two types of study design.In cohort studies, participants are followed over time, and data about them is collected at regular intervals, allowing for the observation of changes and the development of outcomes over time.In contrast, cross-sectional studies provide a snapshot of a population at a specifc point in time, which could lead to diferent fndings and interpretations [52,53].Tis discrepancy in design could introduce variability, or heterogeneity, into our meta-analysis.Consequently, future metaanalyses should consider the potential impact of study design and address this source of heterogeneity to ensure the robustness and reliability of their fndings.
Our meta-analysis has some limitations that are strictly inherent to the design of the eligible studies.First, our metaanalysis could not prove causality between malnutrition and the risk of AKI because of the observational design of the included studies.Second, heterogeneity was high in the pooled analyses, and the meta-regression analysis failed to defne the source of heterogeneity, which may have afected the reliability of our results.Tird, the included studies were mostly conducted in China and Turkey, and there was a signifcant publication bias in the funnel plot, which may have weakened our ability to evaluate the strength of the 8 International Journal of Clinical Practice association between malnutrition and the risk of prevalent AKI in other countries.Fortunately, the signifcant association between malnutrition and the risk of prevalent AKI has also been confrmed in several studies conducted in America and Korea [15,48,49].Nevertheless, further welldesigned cohort studies with ethnically or geographically diverse populations are required to confrm whether this further amplifes the increased risk of developing AKI in malnourished patients in other regions.Fourth, we excluded studies in languages other than English and Chinese, which may introduce language bias.Despite these limitations, the present meta-analysis has several strengths.To our knowledge, this is the frst study to examine the correlation between malnutrition and the risk of acute kidney injury (AKI) from a global perspective.Although previous studies have deepened our understanding of the relationship between malnutrition and contrast-induced nephropathy [54], it is crucial to note that contrast-induced nephropathy and AKI are two diferent pathologies with distinct etiologies, pathophysiological mechanisms, and clinical presentations.In our research, we focus on the association between malnutrition and AKI.Tis feld has not been extensively discussed in prior research and is often confounded by other variables, leading to signifcant controversy and ambiguity over the correlation between malnutrition and AKI.Our study capitalizes on data from diverse countries and populations, enabling a more accurate and universal refection of malnourished individuals, thus providing higher value for routine clinical practice; furthermore, the included subjects are likely to be an accurate refection of individuals with malnutrition seen in routine clinical practice.
In conclusion, the results of the present meta-analysis provide evidence that malnutrition is signifcantly associated with an increased AKI prevalence.Our fndings suggest that malnutrition may be a potential target for AKI prevention.However, well-designed studies with ethnically or geographically diverse populations are needed to evaluate strategies and interventions to prevent or slow the development and progression of AKI in malnourished individuals.

Figure 2 :
Figure 2: Association between malnutrition and acute kidney injury risk.

Table 1 :
Main characteristics of included studies.

Table 2 :
Subgroup analyses examining the associations between malnutrition and AKI risk.

Table 3 :
Results of the univariate meta-regression analyses examining possible sources of between-study heterogeneity., coronary angiography; PCI, percutaneous coronary angiography; OR, odds ratio; CI, confdence interval; SE, standard error. CAG