Risk Factors for Acute Kidney Injury in Patients with Coronavirus Disease 2019 (COVID-19): A Meta-Analysis

0.47-2.65, p=0.06, random-effect model). Signicant heterogeneity was found, with I 2 =59% (p=0.06). Six studies provided estimates for the relationship of between COPD and the incidence of AKI in patients with COVID-19, showing obvious statistical signicance, with OR of 1.7 (95% CI, 1.40-2.07, p<0.00001) under xed-effect model. Between-study heterogeneity was not apparent (I 2 =9%, p=0.36). Malignancy was not signicantly associated with AKI in patients with COVID-19 (pooled OR=1.18, 95%CI, 0.97-1.45, p=0.85, under random-effect model), based on 5 studies. These studies showed a signicant heterogeneity, with I 2 =66% and p=0.93. COVID-19: Coronavirus disease 2019; NOS: Newcastle -Ottawa Scale; OR: Odds ratio; RAAS: Renin-Angiotensin-Aldosterone System; SARS-CoV-2: Severe acute respiratory syndrome coronavirus 2; SOFA: Sequential Organ Failure Assessment; ICU: Intensive care unit.


Background
The outbreak of coronavirus disease 2019 , caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has rapidly evolved into a global pandemic. Most patients with COVID-19 have mild symptoms, but about 15.7%-26.0% develop severe symptoms, including acute respiratory distress syndrome, septic shock, and multiple organ failure [1,2]. The kidney may serve as a target for organ injury in COVID-19 because angiotensin-converting enzyme 2, the binding site for SARS-CoV-2, is highly expressed in proximal tubule cells and podocytes. Data from the single center studies in China showed an incidence of AKI ranging from 0.5% to 29% in hospitalized patients with COVID-19 [2][3][4]. In contrast, early studies in the United States have reported a much higher incidence of AKI of 28%-46% [5,6].
AKI is more common among patients with more severe disease, particularly in those recovering in the intensive care unit (ICU), and is considered a negative prognostic factor for survival. In a retrospective study of 333 patients with COVID-19 those who presented with kidney dysfunction had higher mortality rates than patients without kidney involvement (11.2 and 1.2%, respectively) [7]. Attention must be paid to AKI in patients with  Indeed, the actual AKI incidence remains uncertain and may have been underestimated due to the retrospective design of the studies. Unfortunately, with the global outbreak of COVID-19, it is di cult to conduct prospective studies to explicit the incidence of AKI in patients with COVID-19. Therefore, identifying risk factors for AKI in COVID-19 patients will be conductive to early intervention.
To improve strategic intervention and prognosis, it is indispensable to make systemic analysis of factors contributing to AKI and clear identi cation of post-COVID-19 patients who are likely to have AKI. Thus, this meta-analysis was conducted to estimated risk factors and provide more reliable evidence for the clinical setting.

Methods
The protocol for this review was registered on the International Prospective Register of Systematic Reviews (PROSPERO)(ID :CRD42020211933) .

Data Sources and Searches
Relevant articles were identi ed from PubMed, ISI Web of Science, EMBASE, Google Scholar, Scopus and Cochrane library up to September 2020. Searches were conducted by combined text words and medical subject headings (MeSH), including all spellings of "novel coronavirus", "COVID-19", "SARS-COV-2", "acute kidney injury", "AKI", and "risk factor". The reference lists of retrieved articles were also checked to identify additional studies. The searches were performed by 2 investigators (Tian L and Shao X) independently, and eligible studies were veri ed by all authors to determine suitability for inclusion.

Study Selection
We included observational studies that reported risk factors for AKI in patients with COVID-19. Only articles published in the English language were included. We excluded studies that did not utilized the Kidney Disease Improving Global Outcomes (KDIGO) de nition for AKI. Conference, abstracts and letters to journal editors were also excluded on account of the limited data.

Data Extraction and Quality Assessment
Data were reviewed and con rmed by all authors independently. We extracted data of study characteristics, including rst author, the country of origin, year of publication, study design, the study sample, and participant demographic characteristics.
The quality of all included studies was assessed independently by two authors (Tian L and Hang Y), using the validated Newcastle -Ottawa Scale (NOS) [8]. The NOS is based on an accumulative score in each of three categories: selection, comparability, and exposure or outcome. The NOS scores range between 0 and 9 stars. Each study was assigned a nal quality rating of good (7-9 stars), fair (5-6 stars) or poor (0-4 stars). We excluded studies as poor quality.

Risk Factor Analysis
All statistical analyses were performed using the Review Manager Software (RevMan 5.3). Our metaanalysis examined risk factors identi ed in at least two studies. Odds ratio (OR) with 95% con dence interval (CI) were assessed for each risk factor. Heterogeneity of studies was measured using I 2 statistic. When I 2 >50% or P value<0.05 was identi ed for heterogeneity among studies, we used the random effect model; Otherwise, a xed effects model was adopted. Publication bias was de ned as visual asymmetry of the funnel plots.

Male
All included studies provided eligible data for demonstrating the relationship between gender and AKI. The pooled results showed that males had a higher risk of developing AKI than females in patients with COVID-19 (OR=1.62, 95% CI, 1.24-2.13, P<0.05) (Fig S1). There was signi cant heterogeneity among these studies (I 2 =68%).

Race
Race was categorized into of the four groups: White, Black, Hispanic and Asian. The data for White and Black were available in four studies. Patients in White with COVID-19 had a signi cant increase in risk for AKI (OR=1.14, 95%CI, 1.03-1.26, p<0.05), whereas the pooled OR revealed a non-signi cant increase in Black (OR=1.29, 95%CI, 0.96-1.74, P=0.09). There studies reported the relationship between Hispanic and the risk of AKI. The results showed decreased risk of AKI in Hispanic patients with COVID-19, but the result was not statistically signi cant (OR=1.09, 95%CI, 0.46-1.046, P=0.08). Two papers reported information on Asian. The overall effect size was OR=1.18(95% CI, 0.75-1.12), p=0.39, with I 2 =0%, P=0.05( Fig S2).

Mechanical Ventilation
We identi ed six studies that reported available data on mechanical ventilation as risk factor for AKI.

ICU admission
There studies with 7026 patients about ICU admission were pooled in the analysis (Fig 3). No heterogeneity across studies was detected (I 2 =0%, p=0.84), so xed-effect model was applied. Patients in ICU had a strong signi cant increase in risk for AKI (OR=10.57, 95%CI, 9.33-11.98, p<0.00001).

Renin-Angiotensin-Aldosterone System (RAAS) inhibitors
Five studies that included a total of patients provided available data on the association between RAAS inhibitors use and AKI. Heterogeneity across studies was detected (I2=70%, p=0.009), so random-effect model was applied. However, no signi cant difference was found between the combined estimates (OR=1.17, 95%CI, 0.77-1.80, p=0.46) (Fig S3).

Glucocorticoids
There studies provided extractable data to analyze the association between the risk of AKI and glucocorticoids use. Heterogeneity across studies was detected (I2=80%, p=0.007), so random-effect model was applied. However, no signi cant difference was found between the combined estimates (OR=1.64, 95%CI, 0.27-9.91, p=0.59) (Fig S5).
Comorbidity Comorbidities related to AKI were analyzed, including obesity, hypertension, diabetes mellitus, cardiac disease, chronic kidney disease (CKD), cerebrovascular disease, chronic obstructive pulmonary disease (COPD) and malignancy, as showed in Table 2

Publication Bias
Publication bias assessments were carried out on studies that examined male sex. The funnel plot suggested possible publication bias in studies reporting male and risk of AKI, as asymmetric graph was observed. (Fig S6)

Discussion
To better identify COVID-19 patients at risk of AKI, we conducted a comprehensive, systematic review of studies from the past one year. This review sought to identify risk factors associated with AKI in patients with COVID-19, and, as far as we are aware, is the rst meta-analysis conducted to do so. Sixteen risk factors related to AKI in patients with COVID-19 were included and analyzed, of which male, white, mechanical ventilation, ICU admission, vasopressor, obesity, diabetes, hypertension, cardiac disease, CKD, and COPD were most likely associated with AKI in patients with COVID-19, with statistical signi cance.
Studies [19,20] on ICU patients with COVID-19 suggested that male sex is the risk factor for death and disease severity associated with COVID-19. The ndings of our study indicate a higher rate of AKI prevalence in male gender groups of COVID-19 patients. It may be explained by the study of Carrero et al [21] that kidney function declines faster in men than women, possibly owing to unhealthier lifestyles in men and the protective effects of oestrogen or the damaging effects of testosterone. Data from the America and England reported on the disparities in the risk and outcomes of COVID-19, with a higher risk in black people than white people [22,23]. Nimkar et al. also demonstrated African-American race showed higher odds of AKI than the White [11]. However, our study indicated patients in White with COVID-19 had a signi cant increase in risk for AKI, whereas the pooled OR revealed a non-signi cant increase in Black. It is di cult to have de nitive views on the cause of ethnic disparities in COVID-19 until the overall infection rate has been established in different racial groups.
Mechanical ventilation is an important treatment for COVID-19 patients with respiratory failure. Our study indicated COVID-19 patients treated with mechanical ventilation were at a higher risk of AKI (OR=9.44, 95% CI, 5.16-17.27, P<0.05), which was consistent with the results of previous studies (6,10,15,18). A recent study, which is by far the largest cohort of COVI-19 patients with a focus on AKI, reported the rate of AKI was 89.7% among patients on ventilators compared with 21.7% among other patients [6]. This might be explained by the largest cohort which implied AKI is a condition that occurs among COVID-19 patients with respiratory failure. In addition, ICU admission was con rmed as a risk factor of AKI in this meta-analysis. A multicenter prospective observational study, which investigated COVID-19 patients received in 19 ICUs of 16 hospitals, demonstrated the average incidence of AKI was 25.2% [24]. It is common knowledge that patients admitted in ICU are critically-ill. AKI is a common complication of patients in ICU, which can be caused by various conditions [24]. The pathophysiology of AKI in critically ill patients with COVID-19 currently relies on unspeci c mechanisms (hypovolemia, nephrotoxic drugs, high PEEP, right heart failure), a direct viral injury, an imbalanced elevation of pro-in ammatory cytokines elicited by the viral infection, and a procoagulant state [25].
Angiotensin-converting enzyme-2(ACE-2), which has been described as the most likely "receptor" for viral entry into human cells, is abundantly expressed in kidney [26,27]. Likewise, patients on angiotensinconverting enzyme inhibitors or angiotensin receptor blockers (ARB) could be at a greater risk due to the mechanism by which SARS-CoV-2 enters the cell. Nevertheless, in our study, renal-angiotensinaldosterone-inhibiting drugs were no longer an AKI predictor, which was consistent with the results of the study of Hirsch et al [6]. In addition, vasopressor treatment was con rmed as a risk factor of AKI in this meta-analysis. Pathologic studies performed in kidney tissues from ten deceased patient revealed AKI in COVID-19 patients was more likely caused by ischemia rather than viral infection [10]. Benjamin et al [28] have reported sepsis shock was the leading cause for AKI, and vasoactive agents were used to treat shock. This may be the main reason that vasopressor treatment was associated with AKI. Great attention should be paid to the hemodynamics and volume balance of patients with COVID-19.
A variety of comorbidities, obesity, diabetes, hypertension, cardiac disease, CKD and COPD, did present evidence to correlate with AKI in COVID-19 patients. The association between morbid obesity and risk for AKI in individuals with ARDS in the ICU has been previously recognized [29] Our observation of higher odds of AKI in COVID-19 patients with CKD is not surprising, as CKD is a well-known risk factor for AKI [30]. Diabetes and hypertension frequent co-exist with CKD, and can modify kidney disease outcomes [31]. Importantly, both of these conditions have been associated with heightened risks of AKI in several clinical setting [32,33]. Cardiac and renal disease interact in complex bidirectional and interdependent manner in both acute and chronic settings, as shown by a pathophysiological perspective that cardiac and renal disease share a number of common pathways [34]. A COPD cohort has reported the incidence and prevalence of AKI is relatively high in COPD patients [35], which was similar to our research trends.
Our study still had some limitations to consider. The standard for estimating risk factors differed with the study design in variable studies, which could have led to discrepant results. Furthermore, the study inclusion was restricted to published studies, there was a risk for overestimating effects of risk domains due to publication bias. In addition, our study only evaluated the relationship between clinical characteristics of AKI in CIVID-19 patients, not the laboratory ndings. The previous studies have reported elevated D-dimer, higher serum interleukin-6, and higher Sequential Organ Failure Assessment (SOFA) scores were associated with AKI [10,36]. Unfortunately, there was no way to extract available biochemical data form included studies. Nonetheless, to the best of our knowledge, the present study is the rst metaanalysis to investigate the risk factors of AKI in COVID-19 patients.

Conclusion
In conclusion, this meta-analysis has revealed statistically signi cant increases in risk of AKI in COVID-19 patients with mechanical ventilation treatment, ICU admission, use of vasopressor, and CKD. For patients with male sex, obesity, diabetes, cardiac disease, hypertension, and COPD were identi ed as the contributing risk factors. Future molecular epidemiology studies are also warranted to establish the underlying mechanisms linking these risk factors with AKI in COVID patients. Our study strengthened clinical awareness of early warning to identify COVID-19 patients with high-risk for AKI. Abbreviations AKI: Acute kidney injury; ACE-2: Angiotensin-converting enzyme-2; ARB: Angiotensin receptor blockers; CI: Con dence interval; CKD: Chronic kidney disease; COPD: Chronic obstructive pulmonary disease; COVID-  Figure 1 Identi cation process for eligible articles.