Bringing doctors together

or

A free, private network for verified physicians

Leading voices

Dr. Peter Hotez

Global Health

View

Dr. Jennifer Coren

General Pediatrics · AMH

View

Dr. Chethan Sarabu

General Pediatrics · Cornell Tech

View

Dr. Philip Schauer

Bariatric Surgery · PBRC

View

Dr. Justin Hovey, FAAP, FACP, DABOM

Obesity Medicine · Noland Hospital Dothan II

View

Dr. Kimberly Gudzune

Obesity Medicine · JHU

View

Dr. Richard Dutton

General Anesthesiology · UMSOM

View

Dr. Sarah Ro

Obesity Medicine · UNC Health Care

View

Dr. Rebecca Andrews

Primary Care · UCHC

View

Dr. Jed Wolpaw

General Anesthesiology · Johns Hopkins

View
  • Discover your community

  • Stay current

  • A profile that represents you

  • Find your people

About RoonPrivacyTerms

    Overview

    Modification of the Cardiac Renal Angina Index for Predicting Adverse Kidney Events After Pediatric Cardiac Surgery.

    Aug 12, 2025DOI: 10.1161/jaha.125.042941Journal ArticleAHA

    Written by: Katherine F MelinkNatalja L StanskiEmily GorrellJeffrey A AltenKevin A PettitStuart L GoldsteinHuaiyu ZangNicholas J OllberdingRajit K BasuShina MenonMegan SooHooKatja M Gist

    Find these authors on Roon

    RB
    Dr. Rajit Basu
    Peds Critical Care · Lurie Children's Hospital

    METHODS

    We performed a secondary analysis of a 2‐center prospective observational study of pediatric patients aged 0 to 18 years undergoing cardiac surgery who were admitted to the Cardiac Intensive Care Unit (CICU) at Cincinnati Children's Hospital Medical Center and Children's Hospital Colorado. Children enrolled at Cincinnati Children's Hospital Medical Center consented to enrollment in the Heart Institute Biorepository from January 2019 to December 2021, and those at Children's Hospital Colorado were enrolled prospectively from April 2021 to June 2022. Patients were included in this study if they had urine samples collected 8 to 12 hours after CICU admission and had postoperative SCr measurements obtained per standard of care. Exclusion criteria included the absence of SCr measurement on postoperative day (POD) 0 or absence of at least 1 POD 2 to 4 measurement, lack of uNGAL measurement, preexisting AKI within 72 hours of enrollment, end‐stage kidney disease or chronic kidney disease with an estimated glomerular filtration rate of <30 mL/min per 1.73 m,2 cardiac arrest, or extracorporeal life support before surgery (Figure 1). An estimated glomerular filtration rate <30 was chosen to account for the normal low value in neonates in the absence of CS‐AKI and aligns with the expected increase up to full kidney maturation in early childhood. Institutional review board approval was obtained at both institutions and included a waiver of informed consent at Children's Hospital Colorado. The data supporting this study's findings are available from the corresponding author upon reasonable request. The Pediatric Cardiac Critical Care Consortium database was used to obtain clinical and demographic data for included patients, which were then merged into a single deidentified database. Electronic medical record review was used to supplement clinical data not included in the Pediatric Cardiac Critical Care Consortium by manual extraction. This manuscript adheres to the Strengthening the Reporting of Observational Studies in Epidemiology.15The total number of patients initially considered for enrollment. After excluding those without uNGAL or serum creatinine values to determine the presence or absence of postoperative acute kidney injury, 476 patients were enrolled for the initial cRAI analysis. cRAI indicates cardiac renal angina index; POD, postoperative day; SCr, serum creatinine; and uNGAL, urine neutrophil gelatinase‐associated lipocalin.Given the need for model modification in our cohort (Supplemental Methods), we redefined the important outcome in this population to be more proximate to cardiac surgery and to cRAI calculation. Thus, for cRAI modification, we used a new primary outcome of either (1) any POD 2 to 4 CS‐AKI or (2) mechanical ventilation ≥3 days. This composite is justified by known kidney–lung crosstalk. Day 2 to 4 CS‐AKI assessment potentially mitigates the intra‐ and early postoperative management strategies that impact creatinine and urine output. Pathologic fluid accumulation is difficult to quantify due to missing postoperative daily weights and inaccuracies in fluid balance, which independently and synergistically contribute to poor AKI‐related outcomes.16 Furthermore, elevated levels of interleukin‐6 and interleukin −8 at 2 hours after cardiopulmonary bypass (CPB) predict CS‐AKI and are associated with acute lung injury (quantified by prolonged mechanical ventilation) in children.17 Finally, integration of a mechanical ventilation metric considers the possibility of a dilutional effect of serum creatinine in fluid overload, manifesting as increased extravascular lung water.1, 2Intraoperative management was at the discretion of the surgical and anesthesia team. Fluid removal in the operating room was performed via modified ultrafiltration at the end of CPB at Cincinnati Children's Hospital Medical Center, and continuously throughout CPB at Children's Hospital Colorado. Urine NGAL was assessed at 8 to 12 hours after CICU admission. Urine NGAL was measured via particle‐enhanced turbidimetric immunoassay (the NGAL Test, Bioporto Diagnostics) on a BNII Nephelometer (Siemens), as previously described.18 A positive uNGAL was defined as ≥125 ng/mL, per the recent US Food and Drug Administration–approved clinical cutoff for AKI risk stratification.18 CS‐AKI was defined using the Kidney Disease: Improving Global Outcomes (KDIGO) SCr criteria,19 with a minimum SCr threshold of 0.5 mg/dL to qualify for CS‐AKI diagnosis.20 Urine output criteria were not used because of the use of peritoneal catheters for passive drainage or prophylactic dialysis in neonates and the ubiquitous use of diuretics in this population. Severe CS‐AKI was defined as KDIGO stage ≥2 on PODs 2 to 4.19 While it is possible that prophylactic dialysis may result in clearance of serum creatinine, raising the concern for CS‐AKI misclassification, the peritoneal volume (10 mL/kg) does not usually result in clearance. Furthermore, patients who received prophylactic dialysis in prior studies were not classified as having CS‐AKI, unless serum creatinine thresholds were met or it was continued for ≥72 hours.21 A baseline SCr was available for all patients and was defined as the lowest SCr in the 3 months preceding surgery. Surgical complexity for cRAI was categorized using the Society of Thoracic Surgeons–European Association for Cardio‐Thoracic Surgery Congenital Heart Surgery scores.22 Vasoactive requirements were quantified using the Vasoactive Inotrope Score (VIS).23 The time of VIS determination was chosen to be 8 hours after CICU admission, given the anticipated onset of low‐cardiac‐output syndrome.24 This may not reflect the peak of low‐cardiac‐output syndrome but reflects a time point proximate to surgery, where vasoactive support is likely increasing. Fluid balance at CICU admission was assumed to be 0 because of the challenges in capturing intraoperative fluid balance. Fluid balance at 8 hours after CICU admission was calculated using the following formula: [total fluid intake−total fluid output]/preoperative (dry) weight, as previously described.25The cRAI was remodeled as a clinical nomogram to predict the new composite outcome using 5 a priori selected risk‐weighted variables: age, Society of Thoracic Surgeons–European Association for Cardio‐Thoracic Surgery Congenital Heart Surgery score, CPB duration, VIS at 8 hours after CICU admission, and fluid balance indexed to preoperative weight (kg; assumed dry weight) at 8 hours after CICU admission. For surgeries without a Society of Thoracic Surgeons–European Association for Cardio‐Thoracic Surgery Congenital Heart Surgery score, they were categorized as “unavailable,” and the score was summarized in the nomogram. Similarly, for cases done without cardiopulmonary bypass, a score was assigned on the basis of “Unavailable” (Figure S1). Multivariable logistic regression using a priori identified variables known to be associated with postoperative CS‐AKI in the CICU was used to generate coefficient estimates and predicted probabilities for the composite outcome, as further described in Supplemental Methods. A clinical nomogram was derived from the multivariable logistic regression model to visualize associations of model terms with their contribution to the predicted outcome and to aid in the scoring of patients for risk stratification (Figure S1 and Figure S2). Potential nonlinear relationships were modeled using restricted cubic spline terms for all continuous predictors with 3 knots placed at the 0.05, 0.50, and 0.95 quantiles as recommended by Harrell.26 Predictor importance was measured via the Wald χ2 statistic penalized by the predictor degrees of freedom. Only predictors with P values <0.2 for the Wald statistic were retained in the final model. Youden's index was used to determine the optimal cutoff for cRAI, with patients above this optimal cutoff predicted to have the new composite outcome (ie, cRAI positive). The optimism corrected area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, positive predictive value, and negative predictive value (NPV) were calculated to assess model predictive performance. Finally, the additive value of uNGAL in the clinical nomogram was assessed using AUROC, sensitivities, specificities, positive predictive value, and NPV.Models were internally validated using bootstrap optimism correction based on 400 resamples (Supplemental Methods).27 A calibration curve was generated to assess the predictive accuracy of cRAI by aligning predicted probability and observed occurrence of the new composite outcome. All analyses were performed using the R software environment version 4.4.0 (R Foundation for Statistical Computing, Vienna, Austria). Multivariable logistic regression models were fit, the nomogram generated, and model validation/calibration performed using the rms package (version 6.8.1).

    RESULTS

    After excluding patients with missing SCr and uNGAL data (n=115), there were 476 patients remaining for analysis (Figure 1). The median age of the cohort was 5 (interquartile range [IQR], 1–30) months, and most patients were boys (n=286 [61%]; Table 1). Patients with single ventricle pathology comprised roughly one third (32%) of the total cohort. CPB was used in most surgeries (93%). Delayed sternal closure occurred in 77 (16%) patients. POD 2 to 4 AKI occurred in 44 (9%), of which 24 cases (5%) were classified as severe. Death was infrequent, occurring in 23 (5%) patients.Demographic, Clinical Characteristics, and Outcomes of Patients by the Presence or Absence of the New Composite Outcome of POD 2 to 4 AKI or Mechanical Ventilation ≥3 DaysContinuous variables reported as median (interquartile range). P values were obtained using Pearson's χ2 test or Wilcoxon rank‐sum test, as appropriate. AKI indicates acute kidney injury; CPB, cardiopulmonary bypass; FB, fluid balance; ICU, intensive care unit; POD, postoperative day; STAT, Society of Thoracic Surgeons–European Association for Cardio‐Thoracic Surgery Congenital Heart Surgery; uNGAL, urine neutrophil gelatinase‐associated lipocalin; UOP, urine output; and VIS, Vasoactive Inotrope Score.Death was defined as death occurring within the surgical hospitalization or within 28 days of surgery if discharged.Table 1 summarizes the cohort by the presence or absence of the new composite outcome (POD 2–4 CS‐AKI or mechanical ventilation ≥3 days). The new composite outcome occurred in 129 (27%) patients. POD 2 to 4 CS‐AKI occurred in 44 (34% of composite outcome, 9% of total cohort), and mechanical ventilation ≥3 days occurred in 101 patients (78% of composite outcome, 21% of total cohort). Patients who experienced the composite outcome were younger and had higher surgical complexity, longer CPB duration and cross‐clamp times, and higher 8‐hour VIS. These patients also had higher POD 0 fluid balance (17.6 [IQR, 5–37] mL/kg versus 9.8 [IQR, 1.2–24] mL/kg; P<0.001) and higher uNGAL concentrations (159 [IQR, 49–567] ng/mL versus 16 [IQR, 10–42] ng/mL; P<0.001), but there were no differences in 8‐hour urine output or rates of POD 0 CS‐AKI between the groups. Patients with day 2 to 4 CS‐AKI had significantly higher uNGAL concentrations compared with those without CS‐AKI (P<0.001). Similarly, patients with mechanical ventilation ≥3 days also had significantly higher uNGAL concentrations compared with those with mechanical ventilation duration <3 days (P<0.001; Table S1). Patients who experienced the composite outcome also experienced worse outcomes, including longer ICU length of stay (15 [IQR, 9–23] days versus 3 [IQR, 2–6] days; P<0.001) and higher mortality rate (12% versus 2%, P<0.001).For cRAI modification, the relative importance of each predictor and its individual effects on the composite outcome are demonstrated in Figures S2A and S2B, respectively. Surgical complexity by Society of Thoracic Surgeons–European Association for Cardio‐Thoracic Surgery Congenital Heart Surgery score was the most important predictor of the new composite outcome, followed in descending order by 8‐hour VIS, age, CPB duration, and fluid balance (Figure S2A). The partial effects plots for composite outcome prediction by variable are shown in Figure S2B and were used to build the nomogram model (Figure S1). The final nomogram highlighting the assignment of points for each predictor to be added together to achieve a “Total Points” value that maps to the predicted risk of the composite outcome is provided in Figure 2. The calibration curve highlights the accuracy of the model predictions when compared with the observed outcomes (Figure S3). While predictor values can be used most efficiently when programmed into an electronic application, a clinical example of the application of this nomogram is provided in Supplemental Results.Points are assigned vertically above the corresponding value of each patient characteristic. The scores from each category are added together to yield a total score, which is converted to a predicted risk score. The optimal cutoff using the Youden index was a predicted score of 0.27, translating into a total score of 183.5. The red line in the figure denotes the optimal cut point. CPB indicates cardiopulmonary bypass; cRAI, cardiac renal angina index; FB, fluid balance; POD, postoperative day; STAT, Society of Thoracic Surgeons–European Association for Cardio‐Thoracic Surgery Congenital Heart Surgery; uNGAL, urine neutrophil gelatinase‐associated lipocalin; and VIS, Vasoactive Inotrope Score.Table 2 outlines the demographic, clinical, and outcome differences between patients who were cRAI negative and patients who were cRAI positive. Among 476 patients, 191 (40%) were cRAI positive. Compared with cRAI negative, these patients were younger (0.7 [IQR, 0.2–4.5] versus 8.9 [IQR, 4–47.6] months; P<0.001) and had higher surgical complexity. CPB duration was longer (181 [IQR, 145–231] versus 125 [IQR, 98–168] minutes; P<0.001), delayed sternal closure was more frequent, and 8‐hour VIS was higher (11 [IQR, 8–17] versus 5 [IQR, 5–8]; P<0.001) in patients who were cRAI positive. Incidence of POD 2 to 4 severe CS‐AKI and uNGAL levels were higher in patients who were cRAI positive. Mechanical ventilation duration and ICU length of stay were longer, and the mortality rate was higher in patients who were cRAI positive. One hundred four of the patients who were cRAI positive (54%) experienced the new composite outcome of POD 2 to 4 CS‐AKI (16%) or mechanical ventilation ≥3 days (45%). In comparison, only 25 (9%) of patients who were cRAI negative experienced the composite outcome (n=13 [5%]) of day 2 to 4 CS‐AKI or n=15 (5%) mechanical ventilation ≥3 days.Demographics, Characteristics, and Outcome Data for Patients Who Were cRAI Positive and cRAI NegativeContinuous variables reported as median (interquartile range). P values were obtained using Pearson's χ2 test or Wilcoxon rank sum test, as appropriate. AKI indicates acute kidney injury; CPB, cardiopulmonary bypass; cRAI, cardiac renal angina index; CRRT, continuous renal replacement therapy; ICU, intensive care unit; POD, postoperative day; STAT, Society of Thoracic Surgeons–European Association for Cardio‐Thoracic Surgery Congenital Heart Surgery; uNGAL, urine neutrophil gelatinase‐associated lipocalin; UOP, urine output; and VIS, Vasoactive Inotrope Score.Death was defined as death occurring within the surgical hospitalization or within 28 days of surgery if discharged.Table 3 displays the predictive performance of uNGAL, cRAI alone, and the cRAI/uNGAL. The optimism‐corrected AUROC for uNGAL to predict the composite outcome was 0.72. The optimism‐corrected AUROC for cRAI to predict the composite outcome was 0.82, with a sensitivity of 0.81 (95% CI, 0.73–0.87), specificity of 0.75 (95% CI, 0.70–0.80), positive predictive value of 0.55 (95% CI, 0.47–0.62), and NPV of 0.91 (95% CI 0.87–0.94) using the optimal cutoff for prediction of 0.27 to define cRAI positive (Youden index=0.56) and corresponds to a total points score of 183.5. Finally, when uNGAL was incorporated into the nomogram for predicting the composite outcome, the optimism corrected AUROC of prediction improved from 0.82 to 0.84, the specificity increased from 0.75 to 0.79 (95% CI, 0.74–0.83) with an increase in positive predictive value from 0.55 to 0.60 (95% CI, 0.52–0.67) and NPV from 0.91 to 0.93 (95% CI, 0.90–0.96). The sensitivity of prediction also increased from 0.81 to 0.84 (95% CI, 0.77–0.90).Performance of the Modified cRAI in Predicting the New Composite Outcome of POD 2 to 4 Acute Kidney Injury or Mechanical Ventilation ≥3 Days With and Without the Addition of uNGALAIC indicates Akaike information criterion; AUROC, area under the receiver operating characteristic curve; cRAI, cardiac renal angina index; NPV, negative predictive value; POD, postoperative day; PPV, positive predictive value; and uNGAL, urine neutrophil gelatinase‐associated lipocalin.uNGAL+ was defined as >125 ng/mL using the US Food and Drug Administration–approved cutoff. Model uNGAL includes uNGAL as the sole predictor. Model cRAI denotes the multivariable model used to construct the cRAI nomogram. Model cRAI/uNGAL refers to the expanded multivariable model incorporating uNGAL into the model cRAI. The optimism‐corrected values in the prediction model were obtained from bootstrap resampling (400 samples).

    DISCUSSION

    In this 2‐center prospective observational study, a new prediction model, cRAI, was modified as a nomogram, which successfully predicted the proximate composite outcome of any POD 2 to 4 CS‐AKI or mechanical ventilation ≥3 days. The cRAI performed well in our cohort to predict this new composite outcome, with improvement in all test characteristics, including a high NPV. Integration of the recently US Food and Drug Administration–cleared uNGAL test improved the performance of cRAI with an increase in sensitivity, specificity, and NPV, and illustrates a feasible, stepwise mechanism for the tailored measurement of uNGAL only in the highest‐risk patients at the bedside. Ultimately, the high NPV of cRAI with (93%) and without (91%) uNGAL could allow clinicians to effectively rule out the composite outcome of CS‐AKI and prolonged mechanical ventilation in patients who are cRAI negative, specifically targeting interventions only to those at risk.The advantage of the presented cRAI nomogram over the original cRAI,14 is its lack of dependence on SCr or urine output metrics for risk stratification. This is particularly relevant, as intraoperative management can alter serum creatinine. While recent research has highlighted the limitations of SCr and urine output in predicting AKI, these markers have remained the mainstay risk criteria in many previously developed predictive tools.28, 29, 30, 31 Assessment of kidney function using urine output is hindered by the frequent use of postoperative diuretics and indwelling prophylactic dialysis catheters for passive drainage and prophylactic peritoneal dialysis in this population. In addition to variations in muscle mass and placental transmission of creatinine (among those who undergo cardiac surgery within the first few days of birth), elevations in SCr after cardiac surgery can be especially misleading as intraoperative ultrafiltration concentrates creatinine and may falsely quantify AKI.4, 32, 33, 34, 35 The redefined cRAI nomogram presented in this study instead incorporates fluid balance, which is a surrogate for evolving AKI and fluid accumulation in the lungs and has been shown to prolong ventilation.36, 37, 38, 39 In our study, fluid accumulation during the first 8 hours of CICU admission was associated with developing the new composite outcome, while early SCr‐defined AKI and urine output were not. Indeed, historical reliance on SCr and urine output early after cardiac surgery to select patients for interventional AKI studies may be another factor in why these studies have failed. Strategies for success may be enhanced by consideration of AKI phenotypes, as suggested by the Acute Disease Quality Initiative (ADQI) consensus conferences. Recommendations for delineating AKI phenotypes include its duration,40 integration of biomarkers,6 and fluid metrics.25 Chawla and colleagues40 defined transient AKI as being present on PODs 1 and 2 and resolved by day 3, and persistent AKI as present on PODs 1 or 2, with persistence on day 3 and beyond. In a multicenter investigation of neonates undergoing the Norwood operation from the Neonatal and Pediatric Heart and Renal Outcomes Network collaborative, we demonstrated that transient AKI was not associated with adverse outcomes. Persistent AKI alone, and with the addition of a fluid metric, was associated with a higher mortality rate and fewer ventilator‐free and ICU‐free days.41 Ostermann and colleagues recommended in the 23rd ADQI meeting that biomarkers should be integrated into the AKI definition.6 Pettit and colleagues evaluated the utility of early AKI phenotypes, integrating uNGAL for predicting subsequent day 2 to 4 CS‐AKI.42 In this study, which used the same data set to derive the cRAI nomogram, it was demonstrated that early elevation of SCr on POD 0 occurred in 4.6% (n=22) of the cohort, and only 2 of those patients went on to have day 2 to 4 CS‐AKI. There was no association with day 2 to 4 CS‐AKI. In contrast, early uNGAL elevation, regardless of creatinine, was associated with a nearly 9‐fold higher odds of day 2 to 4 AKI.42 Interleukin‐6 and interleukin‐8 have been reported to be associated with prolonged mechanical ventilation after cardiac surgery, a marker of acute lung injury.17 Our findings demonstrate that uNGAL is higher in those with prolonged mechanical ventilation. While determining causality is not possible, it is plausible that kidney–lung organ crosstalk is at play in these patients.The variable importance model demonstrated age to be the third most important factor for predicting the composite outcome. Infants aged <3 months scored 5‐fold more points than children aged ≥24 months, aligning with clinical observations in this younger population for higher CS‐AKI risk, and mechanical ventilation ≥3 days. While a separate risk nomogram is likely to yield different results when evaluated only in the youngest patients, the nomogram developed is generalizable to all children and facilitates risk stratification accordingly. Integration of risk stratification tools across different disease types into clinical practice is challenging for various reasons, including resource limitations, differences in electronic health records, and challenges in workflow integration and usability.43 Indeed, creating separate risk stratification tools for different‐aged patients may add to the complexity and lack of use. Integrating risk stratification tools, such as the cRAI, into a cloud‐based platform agnostic to electronic health records provides a reliable and potentially feasible method for routine clinical care. RAIDAR Health is one such example that offers clinical decision support for the original RAI for children admitted to the pediatric ICU.13 Several pediatric hospitals in the United States have licensing agreements with RAIDAR Health for RAI integration (personal communication). After thoughtful design and testing, the cRAI could be deployed through this platform.The prognostic enrichment gained from cRAI has important prognostic and therapeutic implications, providing an opportunity to identify and intervene in high‐risk patients on POD 0. Both the PREV‐AKI (Prevention of Cardiac Surgery by Implementing the KDIGO Guidelines in High‐Risk Patients Identified by Biomarkers) and the BigpAK (Biomarker‐Guided Intervention to Prevent Acute Kidney Injury After Major Surgery) studies demonstrated decreased rates of moderate and severe AKI after both cardiac and noncardiac major surgery, respectively, when AKI prevention bundles were applied only to those identified as high risk for AKI via elevated tubular injury urinary biomarkers.44, 45, 46 A similar study was trialed in children without risk stratification and was not successful (manuscript under review). Among critically ill children admitted to the general ICU, the use of the RAI and targeted uNGAL testing to guide interventions such as furosemide stress tests, fluid restriction, and initiation of continuous renal replacement therapy was shown to decrease patient morbidity and ICU length of stay while increasing survival rates to ICU discharge.13 Our results should not imply that cRAI positive mandates the need for aggressive intervention such as continuous renal replacement therapy. In the TAKING FOCUS 2 (Using RAI, Urinary NGAL, and Furosemide Stress Test to Direct Management of Acute Kidney Injury in the Pediatric Intensive Care Unit) clinical decision support study, 40 of 304 patients who were RAI positive and 38 of 142 patients who were RAI positive/uNGAL positive received continuous renal replacement therapy.13 It is hoped that the cRAI will provide a construct for a more cautious approach to fluid management, diuresis, nephrotoxin medication stewardship, and, if indicated, earlier continuous renal replacement therapy before pathologic fluid accumulation occurs. Despite the relatively low burden of the outcome (CS‐AKI [9%] in this study), the cRAI also provides a mechanism to identify patients most likely to benefit from therapies to mitigate AKI and early augmented fluid removal and who may be appropriate for enrollment in future clinical trials. For example, the recently published PROTECTION (Amino Acid Infusion to Protect Kidney Function After Cardiac Surgery) study demonstrated a reduction in the occurrence and severity of SCr‐defined CS‐AKI in adults who received amino acid infusions during and following CPB.47 A study examining the effectiveness of this intervention in children following cardiac surgery would be enriched using a risk stratification tool like the cRAI, an approach highlighted as a research priority in the 26th ADQI Consensus Conference.8, 9This study investigates the value of the novel tubular injury biomarker, uNGAL, to further enrich the predictive performance of cRAI, as has been encouraged by the 23rd ADQI.6 Given prior promising results in critically ill pediatric populations susceptible to AKI, uNGAL was analyzed with cRAI and compared against the performance of cRAI alone.48, 49, 50 The predictive performance of uNGAL was lower than that of the cRAI alone and with uNGAL. The predictive performance of cRAI mildly improved with the addition of uNGAL, though most notably, all predictive performance parameters increased. Our study revealed that patients who have evidence of ongoing tubular injury (uNGAL positive) and who are cRAI positive are at highest risk for POD 2 to 4 CS‐AKI and mechanical ventilation ≥3 days compared with patients who are cRAI negative or uNGAL negative. As demonstrated in the noncardiac pediatric intensive care setting, the authors propose using cRAI as an initial, cost‐effective risk stratification tool to identify further a subset of patients who may benefit from uNGAL testing when available.51 Our results support using a similar strategy that needs to be tested in clinical practice. Furthermore, given the low uNGAL levels of patients who were cRAI negative (15 [IQR, 10–40] ng/mL) in this study, the cRAI can be used independently at institutions without available biomarker testing and can confer a low likelihood of biomarker elevation and tubular injury when cRAI testing is negative.The strength of this study is its 2‐center design. However, the results are limited by the low incidence of POD 2 to 4 AKI among the patients enrolled (9%). Interestingly, a similar rate of CS‐AKI was seen in our prior work on the first iteration of the cRAI.14 Similarly, the classification of persistent CS‐AKI was also much lower (22%) in the report by Hasson and colleagues, which included neonates undergoing the Norwood operation, where only persistent CS‐AKI was associated with adverse outcomes.41 These data suggest that while CS‐AKI rates are lower than prior reports, these may reflect clinically important events that impact outcomes and negate the effect of intraoperative factors such as modified ultrafiltration that hemoconcentrates SCr in the early postoperative period. It is possible that the patients included in the analysis were biased toward a sicker cohort because they underwent standard‐of‐care laboratory measures of SCr and had an indwelling catheter for uNGAL measurement. Patients whose daily SCr measures were not performed cannot be compared with the included cohort because clinical data were not collected. Future studies should include comparisons with excluded patients to validate the risk score. Finally, we did not evaluate an extensive list of risk factors for the cRAI nomogram. Other factors, including genetic syndromes and extracardiac anomalies, can increase the risk for the composite outcome in this study. Future iterations could include some of these other risk factors.

    CONCLUSIONS

    In conclusion, the cRAI was modified as a nomogram, and predicted POD 2 to 4 AKI or mechanical ventilation ≥3 days in critically ill children following cardiac surgery. Importantly, this tool does not rely on urine output or SCr, 2 historically unreliable tools for AKI prediction in this population. The high NPV of cRAI may make it a useful tool to rule out patients at low risk for adverse kidney‐related outcomes. If successful, cRAI may allow us to better focus efforts on patients who are at the highest risk, including guiding tubular injury biomarker measurement and informing future patient enrollment in clinical trials.

    Sources of Funding

    This study was supported by funding from the American Academy of Pediatrics Resident Research Grant and Cincinnati Children's Hospital Medical Center Research Innovation in Support of Excellence Award (K.F.M.). This study was also supported by NIH/NCATS Colorado CTSA Grant Number UM1 TR004399. Contents are the authors’ sole responsibility and do not necessarily represent official NIH views. Funding for uNGAL testing was partially supported by Bioporto Diagnostics (K.M.G.), although they had no influence on the data or their analysis or the development of the manuscript.

    Disclosures

    K.M.G. receives consultant fees from Bioporto Diagnostics and Potrero Medical. K.M.G. did not receive support for this study from Bioporto other than sample analysis, as stated above. Bioporto has not seen the results, nor did they contribute to the design of the study. S.L.G. is a consultant for Baxter, Medtronic, Bioporto, and Potrero Medical. He is the founder and chief scientific officer for Medibeacon. None of the authors have any relevant conflicts.

    Supporting information

    Data S1Table S1Figures S1–S3Reference 52
    RBDr. Rajit Basu

    Dr. Rajit Basu

    Pediatrics · Pediatric Critical Care
    Lurie Children's Hospital
    rajit