A modified renal angina index in critically ill patients with COVID-19

Abstract

Background

The renal angina index (RAI) is a tool that has been validated by several studies in the pediatric population to predict the development of severe acute kidney injury (AKI). The aims of this study were to evaluate the efficacy of the RAI in predicting severe AKI in critically ill patients with COVID-19 and to propose a modified RAI (mRAI) for this population.

Methods

This was a prospective cohort analysis of all COVID-19 patients receiving invasive mechanical ventilation (IMV) who were admitted to the intensive care unit (ICU) of a third-level hospital in Mexico City from 03/2020 to 01/2021. AKI was defined according to KDIGO guidelines. The RAI score was calculated for all enrolled patients using the method of Matsuura. Since all patients had the highest score for the condition (due to receiving IMV), the score corresponded to the delta creatinine (ΔSCr) value. The main outcome was severe AKI (stage 2 or 3) at 24 and 72 h after ICU admission. A logistic regression analysis was applied to search for factors associated with the development of severe AKI, and the data were applied to develop a mRAI and compare it vis-à-vis the efficacy of both scores (RAI and mRAI).

Results

Of the 452 patients studied, 30% developed severe AKI. The original RAI score was associated with AUCs of 0.67 and 0.73 at 24 h and 72 h, respectively, with a cutoff of 10 points to predict severe AKI. In the multivariate analysis adjusted for age and sex, a BMI ≥30 kg/m², a SOFA score ≥6, and Charlson score were identified as risk factors for the development of severe AKI. In the new proposed score (mRAI), the conditions were summed and multiplied by the ΔSCr value. With these modifications, the AUC improved to 0.72 and 0.75 at 24 h and 72 h, respectively, with a cutoff of 8 points.

Conclusions

The original RAI is a limited tool for patients with critical COVID-19 receiving IMV. The mRAI, with the parameters proposed in the present study, improves predictive performance and risk stratification in critically ill patients receiving IMV.

Keywords:

• COVID-19
• acute kidney injury
• renal angina index
• modified renal anginal index

Introduction

The coronavirus disease 2019 (COVID-19) pandemic has led to millions of intensive care unit (ICU) admissions worldwide, with acute kidney injury (AKI) being a common complication [1]. In this and other contexts, AKI is a recognized risk factor for increased morbidity, length of hospitalization, costs, and mortality [2]. Furthermore, AKI is associated with long-term complications such as hypertension and cardiovascular disease and an increased risk of progression to chronic kidney disease (CKD) [3]. Thus, strategies must be developed to identify high-risk patients early and to reduce the risk and severity of AKI.

Traditionally, AKI is defined by the Kidney Disease Improvement Global Outcomes (KDIGO) as the elevation of serum creatinine (SCr) and/or a decrease in urine output [4]. Nevertheless, these functional biomarkers are known to be limited mainly for two reasons: first, they reflect deteriorated renal function when AKI is already established, and second, they can be affected by other nonrenal determinants and have been shown to have low sensitivity and specificity [5]. Given these limitations and the importance of a timely AKI diagnosis to prevent complications, there have been multiple efforts in the last decade to develop risk stratification tools to predict AKI and improve timely AKI care. Recently, the addition of kidney injury and/or functional biomarkers to clinical variables has shown some promising results. However, the clinical application of these biomarkers, especially in developing countries, is limited [6].

The renal angina index (RAI) is a tool derived from and validated by multiple studies in the pediatric population [7–10]. This tool employs clinical parameters from the admission day that are grouped into two categories, risk (ICU admission, hematopoietic cell transplant, invasive mechanical ventilation (IMV), and inotropic support) and injury/damage (changes in SCr values or positive fluid balance), to predict the development of persistent or severe AKI [7]. This tool has been validated in several multicenter studies, yielding a high diagnostic performance in the pediatric population [8–10]. However, the diagnostic performance of this index in the adult population has not been as precise as that in the pediatric population when the original score is kept [11]. Therefore, changes to the original index have been made by adding known risk factors for the development of AKI in the adult ICU population, such as diabetes mellitus (DM), advanced age (equal to or >70 years), CKD or hypertension, as proposed by other authors [12,13]. The RAI for adult patients described by Matsuura et al. [13] includes previously described risk factors to predict the development of AKI, and in this score, vasopressor therapy or IMV represents the worst condition and the highest score, which is then multiplied by the points obtained from the delta creatinine (ΔSCr) value.

Since to our knowledge, there are no reports on the use of this particular index to predict the development of severe AKI in adult COVID-19 patients, the present study aimed to evaluate the efficacy of the original RAI in predicting severe AKI in critically ill adult patients with COVID-19 and to propose a modified RAI (mRAI) for this population.

Materials and methods

This was a single-center, prospective, and observational cohort study performed at the Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán in Mexico City. The study protocol was approved by our Institute’s Research Ethics and Bioethics Committees (NMM-3325-20-20-1 and CEI-011-20160627, respectively).

The cohort comprised all consecutive adult patients with a positive polymerase chain reaction (PCR) test for SARS-CoV-2 who were admitted to the ICU and managed with IMV from March 2020 to January 2021. Patients transferred to another hospital, those with advanced CKD (an estimated glomerular filtration rate ≤30 mL/min/1.73 m²), those with prior kidney transplantation, those with AKI at admission, and those who remained hospitalized on 31 January 2021, were excluded. All data were obtained from a local database that prospectively collected data from the admission of the first COVID-19 patient to the present date.

The collected variables included data on demographic characteristics, prevalent comorbidities, home medications, clinical findings and laboratory parameters at admission and at the start of IMV in the ICU, chest CT scan findings, vasoactive drug support (pressors or inotropes), AKI development, medications administered in the ICU, and outcomes.

Variable definitions

AKI was defined and staged according to the KDIGO guidelines [4] by the SCr criteria without registering urine output [4]. Patients were staged according to the highest AKI degree attained during their hospitalization. Baseline SCr was defined as the mean SCr value corresponding to the 6 months before hospitalization or the minimum SCr value obtained during hospitalization if previous values were not available [14]. The Charlson Comorbidity Index [15] was calculated to integrate comorbidity information.

Outcome

The primary outcome was the incidence of severe AKI, defined as KDIGO stage 2 or 3 AKI [4] within days 2–7 of the ICU stay.

Renal angina index and modified RAI

To achieve the first objective of this study, that is, to evaluate the efficacy of the RAI in predicting severe AKI in critically ill patients with COVID-19, the previously reported RAI for adults was used [12,13]. The condition of each patient was scored as follows: receiving IVM and/or vasopressor therapy, 5 points; having ≥1 comorbidity (DM, advanced age (equal to or >70 years)), CKD, or hypertension, 3 points; and being admitted to the ICU, 1 point. The creatinine score was determined by the difference between the SCr value at ICU admission and that within 24 h and 72 h after ICU admission (delta SCr (ΔSCr)), as follows: a ΔSCr value ≥0.4 mg/dL, 8 points; a ΔSCr value ≥0.3 mg/dL, 4 points; a ΔSCr value ≥0.1 mg/dL, 2 points; and a ΔSCr value <0.1 mg/dL, 1 point. The RAI score was defined as the worst condition score multiplied by the ΔSCr score [13]. All patients were evaluated by using the RAI at 24 h and 72 h after ICU admission. Since all the patients in this cohort were receiving IMV, 5 points were assigned for this risk category. Therefore, the score was practically the same as that obtained for the ΔSCr value.

To obtain a presumptively more useful model for adults receiving IMV, as in our critically ill COVID-19 patients, we developed a new RAI model (mRAI). Candidate variables for the mRAI score were obtained using a logistic regression model. To obtain the final mRAI score, the scores corresponding to the different conditions (the new risk factors obtained in the logistic regression model) were multiplied by the ΔSCr score.

As a secondary objective, the performance of the score proposed by Ortiz-Soriano et al. [16] in critically ill patients was analyzed, including only the available variables (diabetes, sepsis, IMV, or the use of vasopressors/inotropes, as well as the ΔSCr percentage).

Statistical analysis

The distribution of continuous variables was assessed by employing the Kolmogorov–Smirnov test. Descriptive statistics are expressed as numbers (percentages) and medians (interquartile ranges), as appropriate. Baseline patient characteristics between patients with or without AKI were analyzed using the Mann–Whitney U-test. Chi-square or Fisher’s exact tests were used for categorical variables.

The factors associated with severe AKI development were evaluated by univariate logistic regression. All variables with a p value <.05 and all factors previously reported to be associated with AKI were selected for the multivariate analysis. The variables that remained significant in the multivariate analysis as well as the body mass index (BMI) and SOFA scores were assigned a score according to the OR. The Charlson index [15] contributed to the score according to the points obtained individually. The performance of the RAI and mRAI was ascertained using a receiver operating characteristic (ROC).

All statistical tests were two-sided, and a p value below .05 was considered statistically significant. All analyses were performed using SPSS (Statistical Package for the Social Sciences) 25.0 (IBM, Armonk, NY).

Results

From 1 March 2020 to 31 January 2021, a total of 3604 patients were admitted to our health facility for severe COVID-19 infection. Of these, 452 patients who were admitted to the ICU and received IMV met our inclusion criteria (Figure 1). The baseline characteristics of all included patients are shown in Table 1. Severe AKI was diagnosed in 138 (30.5%) patients. All patients developed severe AKI within the first 72 h of ICU admission.

Figure 1. Flowchart of the enrolled patients.

Table 1. Clinical characteristics of patients included in this study.

Characteristics	No AKI (n = 314)	AKI 2–3 (n = 138)	p Value
Age (mean ± SD), years	51 ± 13	54 ± 14	.077
Male, n (%)	222 (71)	98 (71)	.946
Body mass index, kg/m^2 (median, IQR)	29.6 (27–33.2)	30.9 (27.6–35.8)	.006
Charlson index	1 (0–2)	1 (0–3)	.017
Comorbidities, n (%)
Hypertension, n (%)	73 (23)	42 (30)	.106
Diabetes, n (%)	80 (26)	42 (30)	.274
Chronic kidney disease, n (%)	3 (1)	2 (1)	.644
Obesity, n (%)	150 (48)	80 (58)	.046
SOFA score at ICU admission (median, IQR)	4 (4–6)	4 (4–7)	.138
Kidney function (median, IQR)
Baseline SCr, mg/dL	0.7 (0.6–0.8)	0.8 (0.7–1.0)	.000
SCr at IMV initiation, mg/dL	0.8 (0.7–0.9)	1.1 (0.9–1.5)	.000
Laboratory at IMV initiation, (median, IQR)
Leukocytes, ×1000/mm^3	10.4 (8.2–13.3)	11.1 (8.2–14.9)	.661
C-reactive protein, mg/dL	18 (11–26)	21 (14–30)	.432
Lactate dehydrogenase, U/L	420 (319–575)	492 (421–640)	.002
Ferritin, ng/mL	753 (439–1249)	787 (474–1300)	.238
PaO2/FiO2 ratio	107 (84–148)	108 (75–145)	.336
Length of hospitalization (days) (median, IQR)	20 (14–29)	20 (11–30)	.570
Mortality, n (%)	92 (29)	79 (57)	.000

SCr: serum creatinine; SOFA: Sequential Organ Failure Assessment score; ICU: intensive care unit; IMV: invasive mechanical ventilation.

Factors associated with AKI in critical COVID-19 patients

The factors associated with severe AKI development (stage 2 or 3) were analyzed by logistic regression (Table 2). Age, BMI, the Charlson Comorbidity Index score, and the SOFA score at the start of IMV were associated with the occurrence of severe AKI as revealed by univariate analysis. In the multivariate analysis, a BMI ≥30 kg/m² (OR 1.67, 95% CI 1.09–2.55), Charlson Comorbidity Index per point (OR 1.31, 95% CI 1.06–1.62), and a SOFA score ≥6 points at ICU admission (OR 1.94, 95% CI 1.26–2.98) were associated with the occurrence of severe AKI. These three risk factors were used for the construction of the mRAI (Figure 2).

Figure 2. Renal angina index used in the adult population (index made by Matsuura et al. [13]) and the new index proposed for the critical ill adult population with COVID-19. Cr: creatinine; RAI: renal angina index; mRAI: modified renal angina index; ICU: intensive care unit; DM: diabetes mellitus; BMI: body mass index; SOFA: Sequential Organ Failure Assessment score.

Table 2. Logistic regression analysis for the development of severe AKI in critically ill COVID-19 patients.

	OR crude			OR adjusted
	OR	95% CI	p Value	OR	95% CI	p Value
Age, per year	1.015	0.999–1.030	.062	0.998	0.975–1.021	.854
Male vs. female	0.946	0.653–1.578	.740	1.108	0.697–1.761	.665
BMI ≥30 kg/m^2, vs. not	1.508	1.007–2.259	.046	1.666	1.087–2.553	.019
Charlson index, per point	1.245	1.082–1.433	.002	1.313	1.062–1.624	.012
SOFA ≥6 points at ICU admission	1.978	1.299–3.012	.001	1.939	1.262–2.980	.003

OR: odds ratio; 95% CI: confidence interval at 95%; BMI: body mass index; SOFA: Sequential Organ Failure Assessment score; ICU: intensive care unit.

The RAI and mRAI in COVID-19 patients

In the validation of the original RAI, since all our patients were receiving IMV, they automatically had 5 points for the risk category of the RAI, and this was then multiplied by the ΔSCr value. Using a cutoff value of 10 points at 24 and 72 h, AUCs of 0.67 (95% CI 0.60–0.72) and 0.73 (95% CI 0.67–0.78) were obtained, respectively (Figure 2). The sensitivity and specificity of the RAI were 78% and 54% at 24 h, respectively, and increased to 80% and 66% at 72 h.

The newly proposed mRAI considered the sum of the score yielded by a BMI ≥30 kg/m², a SOFA score ≥6 and the points obtained from the Charlson index (Table 2). This score was then multiplied by the score obtained for the ΔSCr at ICU admission (as in the original index). With a cutoff value of 8 points, at 24 h and 72 h, AUCs of 0.72 (95% CI 0.66–0.78) and 0.75 (95% CI 0.69–0.80) were obtained, respectively (Table 3). The specificity improved to 74% and 81% at 24 and 72 h, respectively.

Table 3. Performance of the RAI and mRAI score for prediction of severe AKI.

Score	AUC (Cl)	Sensitivity	Specificity	NPV	PPV
RAI (cutoff 10 points)
24 h	0.67 (0.60–0.72)	78%	54%	85%	43%
72 h	0.73 (0.67–0.78)	80%	66%	89%	50%
mRAI (cutoff 8 points)
8 h	0.72 (0.66–0.78)	70%	74%	85%	54%
72 h	0.75 (0.69–0.80)	69%	81%	86%	60%

RAI: renal angina index; mRAI: modified renal angina index; AUC: area under curve; NPV: negative predictive values; PPV: positive predictive values.

Finally, we compared the performance of the mRAI developed in this study with the other indices previously published by studies in adult populations, observing a slightly better AUC to predict severe AKI (supplementary material Table S1).

Discussion

In the present study, we showed that the original RAI score for adults has fair performance in predicting severe AKI in critically ill patients with COVID-19 who are receiving IMV. However, all our patients had the highest score corresponding to the IMV condition. In the mRAI score, using the variables that showed greater relevance in the multivariate analysis, it was shown that the mRAI had a better performance in this particular population, almost as good as that reported in the RAI derivation and validation study in children [10].

The importance of recognizing patients who are at risk of developing AKI early cannot be underestimated, as it may lead to obvious benefits. The RAI is a tool that has proven to have excellent performance in the pediatric population [7]. Basu et al. [10] analyzed a cohort of 144 patients in a pediatric intensive care unit (PICU), of whom 19% subsequently developed severe AKI. They used existing pediatric AKI literature to assign points to risk variables, i.e., PICU admission (1 point), bone marrow transplant (BMT) (3 points), the use of inotropic support or IMV (5 points), or injury defined as follows: decreased creatinine clearance (CrCl) or fluid overload (FO); no change in CrCl or FO less than 5% (1 point); 0–25% decrease in CrCl or FO of 5–10% (2 points); 25–50% decrease in CrCl or FO of 10–15% (4 points) and >50% decrease in CrCl or FO >15% (8 points). The authors concluded that a score >8 points had an AUC of 0.77, whereas a score of <8 had a high negative predictive value (NPV) of 92% for developing severe AKI. They further validated the score in three different cohorts with similar findings (AUCs between 0.74 and 0.81 and NPVs between 95 and 99%) [10].

Matsuura et al. [13] analyzed 263 patients included in three different cohorts, in which only 8.3% (22 patients) developed severe AKI. The RAI score was calculated using the same principles as those described by Basu et al. [10], with some modifications (the exclusion of patients with a BMT history and the inclusion of patients with sepsis and diabetes in the high-risk group) to adjust risk variables into an adult population. In addition, they only used the SCr change and did not integrate FO in their RAI model. The corresponding RAI score was associated with a modest predictive performance with an AUC of only 0.63 for the development of severe AKI.

In a recent study by Ortiz-Soriano et al. [16], the authors analyzed two independent ICU cohorts, including more than 18,000 adult patients. A derivation cohort of 13,965 patients and a validation cohort of 4,789 patients, of whom only 7.7% (1076 patients) and 8.95% (429 patients), respectively, progressed to severe AKI. The study developed an mRAI score. As in the study of Matsuura et al. [13], the variables for the mRAI score were based on the original pediatric RAI score, with modifications to include known AKI risk factors in the adult ICU population. However, in contrast, they performed a logistic regression model to obtain the score of each variable. The global mRAI score was calculated by the sum of the risk score of each individual mRAI variable (DM, the presence of sepsis, IMV, pressor/inotrope use, percentage change in serum creatinine (ΔSCr) in reference to that at admission and FO percentage within the first day of ICU admission). They found that by employing an mRAI cut off ≥10 points, instead of isolated changes in SCr for the prediction of severe AKI, a better performance was obtained. Here, we analyzed the performance of this score according to the available variables and obtained a performance similar to the RAI proposed by Ortiz-Soriano et al. [16].

In contrast with the studies described above [10,13,16], our study included a younger population (mean age of 52 years old), with the same prevalence of DM2 (∼30%) and a lower prevalence of hypertension (<30%). The severity of the disease, as assessed by the SOFA score, was similar (median 4 points). Nevertheless, our study included a much sicker population, as revealed by a threefold mortality rate compared to that of the other cohorts. This finding could be explained by the poor discriminant accuracy of the SOFA score for mortality prediction in patients with COVID-19 [17], which is in line with the mortality reported in other cohorts of severe COVID-19 patients [18–20].

The incidence of severe AKI in our study was higher (30% vs. 8%) than that reported in the studies discussed above [13,16], which is consistent with the incidence reported for other COVID-19 cohorts [21,22]. This observation could reflect the severity of the medical conditions of our patients and could also explain the lower NPV of the RAI found in this study [23].

Given our inclusion criteria, the whole cohort consisted of patients who were receiving IMV, and most of them received vasopressor support. As we calculated the original RAI score, all the individuals already had 5 points, and the difference in the final score was just a reflection of the ΔSCr value. Since the etiology of AKI in COVID-19 patients is multifactorial and given that some nontraditional AKI risk factors, such as direct viral infection of the kidney tubules, systemic inflammation, the activation of coagulation pathways, and obesity, may play an important role [1,24–26], we decided to develop our own mRAI. The performance of this index yielded a slightly better AUC than the original index, albeit it did not reach statistical significance. Nevertheless, we believe that its performance is quite acceptable and that it could be more specific for this particular population. At this time, we do not have clinical or biochemical tools (biomarkers) that have better discriminant power. Therefore, new tools should continue to be investigated to improve earlier diagnosis of AKI episodes and thus improve patient outcomes.

This is the first study aimed at deriving an RAI score for adult patients with severe COVID-19 infection who are admitted to the ICU. This score is easy to obtain, does not require additional studies that may not be available and could help to identify patients who will benefit from an early nephroprotective strategy to prevent progression to severe AKI.

This study has some limitations, including the relatively low number of patients studied, that it was performed in only one center and that it did not consider FO or urine output. Additional studies in similar populations are still needed to reproduce our results and validate the proposed mRAI.

Conclusions

In conclusion, the original RAI is a limited tool for critically ill COVID-19 adult patients receiving IMV. The mRAI performance yielded a slightly better AUC than the original index, although it did not reach statistical significance. Additional studies are required to determine if the proposed modified index is more specific for adult patients with COVID-19 infection.

Ethical approval

The study was performed in accordance with the Declaration of Helsinki and approved by the research ethics and bioethics committees (NMM-3325-20-20-1 and CEI-011-20160627, respectively) of the Instituto Nacional de Ciencias Médicas y Nutrición, Salvador Zubirán, Mexico City.

Consent form

Written informed consent was obtained from the patients or relatives upon admission to the hospital, for using the information and applying those medical procedures specific for the health contingency.

Author contributions

Concept and study design: O.V.V. and N.T.C.; supervision: O.V.V. and N.T.C.; data collection: N.B.P., A.U.P., and A.C.I.; analysis and interpretations: N.T.C. and O.V.V.; writing: N.B.P. and N.T.C.; critical review: O.V.V. All authors read and approved the final version of the manuscript.

Acknowledgements

An abstract based on this work was presented in poster form at the 2022 American Society of Nephrology Kidney Week in Orlando, Florida. The abstract is available at the following link: https://www.asn-online.org/education/kidneyweek/2022/program-abstract.aspx?controlId=3769929

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All data generated or analyzed during this study are included in this article. Further enquiries can be directed to the corresponding author.

Additional information

Funding

The authors declared that this study has received no financial support.