Association of Acute Kidney Injury with Subsequent Sepsis in Critically Ill Children

Abstract

Objective:

Acute kidney injury (AKI) is a major cause of morbidity and mortality in critically ill children. A growing body of evidence has shown that AKI affects immune function, yet little is known about the association between AKI and subsequent infection in pediatric patients. Our objective was to examine the association of non-septic AKI with the development of subsequent sepsis in critically ill children.

Design:

A single-center retrospective cohort study.

Setting:

The pediatric and cardiac ICUs at a tertiary pediatric care center.

Patients:

All patients, ages birth to 18 years without a history of chronic kidney disease, who did not have sepsis prior to or within the initial 48 hours of ICU admission.

Interventions:

None.

Measurements and Main Results:

We analyzed data for 5538 children (median age 5.3 years, 58.2% male), and identified 255 (4.6%) with stage 2 or 3 AKI. Suspected sepsis occurred in 46 (18%) of children with stage 2 or 3 AKI compared to 286 (5.4%) of children with stage 1 or no AKI. On adjusted analysis, children with stage 2 or 3 AKI had 2.05 times greater odds of developing sepsis compared to those with stage 1 or no AKI (95% CI 1.39–3.03, P<.001). Looking at AKI severity, children with stage 2 and 3 AKI had a 1.79-fold (95% CI 1.15–2.79, P=.01) and 3.24-fold (95% CI 1.55–6.80, P=.002) increased odds of developing suspected sepsis, respectively.

Conclusions:

AKI is associated with an increased risk for subsequent infection in critically ill children. These results further support the concept of AKI as a clinically relevant immunocompromised state.

Introduction

Acute kidney injury (AKI) is now recognized as a systemic disease with numerous short- and long-term complications that impact patient morbidity and mortality [1–3]. AKI occurs in 17–56% of critically ill children admitted to pediatric intensive care units (ICUs) worldwide and is associated with a 28-day mortality rate of 11% [4, 5]. While it is generally established that infection represents a frequent cause of death in patients with AKI [6, 7] and that infection and sepsis are well-accepted risk factors for AKI, a growing body of evidence suggests that AKI may be under-recognized as a risk factor for infection. Observational studies of hospitalized adults have shown that AKI is associated with an increased risk for sepsis in the absence of preexisting infection, with hospital-acquired sepsis occurring in up to 45% of patients following AKI [8–13]. In addition to observational data, animal models have been used to identify potential pathogenic mechanisms to support these clinical findings. Several experimental models of AKI, including renal ischemia reperfusion injury, nephrotoxic AKI and rhabdomyolysis-induced AKI have shown attenuated neutrophil recruitment and function following kidney injury, leading to greater susceptibility to invasive bacterial infection in animal models [14, 15].

While some studies have reported rates of infection following AKI in adults, minimal data is available on rates of infection following AKI in children. We hypothesized that critically ill children with AKI will be at higher risk of developing hospital-acquired infections compared to critically ill children without AKI. Accordingly, our objective was to compare rates of sepsis manifesting in critically ill children with and without non-septic AKI at ICU admission, while controlling for severity of illness or other known risk factors for sepsis.

Methods

Source Population and Data Collection

We analyzed data on 8733 pediatric patients with first-time admission to the cardiac or pediatric ICUs at UPMC Children’s Hospital of Pittsburgh from January 1, 2010 to December 31, 2014. Data were obtained from the pediatric high-density ICU (Peds HiDenIC) database, which integrates patient demographics, laboratory results, diagnoses codes, billing codes, and various text elements including clinical notes and discharge summaries. The final data set was de-identified for use in the analysis. This study was approved by the Institutional Review Board at the University of Pittsburgh.

We included children who had sufficient information to categorize AKI status by Kidney Disease Improving Global Outcomes criteria (KDIGO) [16] (n=8143). To have sufficient information to categorize AKI status, children had to have a minimum of one serum creatinine measurement, not including the admission serum creatinine measurement, and/or a minimum of 6 hours of recorded urine output during the exposure window. We excluded children who had sepsis prior to or within 48 hours of ICU admission (n=2125), a history of chronic kidney disease (n=75), or patients older than 18 years of age (n=405). The final study cohort was comprised of 5538 patients (Figure 1).

Figure 1. Flow Diagram of Study Cohort with Exclusions.

Abbreviations: AKI, acute kidney injury; CKD, chronic kidney disease; ICU, intensive care unit

Exposure and Outcome Measures

The primary exposure was stage 2 or 3 AKI in the initial 48 hours of ICU admission compared to stage 1 or no AKI. AKI stage was defined using the KDIGO criteria, utilizing the maximum daily serum creatinine and/or decrease in urine output over a 6-hour rolling window within the first 48 hours of ICU admission [16, 17]. Baseline serum creatinine was defined by the median of all serum creatinine values available in the 6 months prior to ICU admission, excluding the admission creatinine value for the hospitalization. If no serum creatinine value was available prior to admission and the patient did not carry a diagnosis of chronic kidney disease, we assumed an estimated glomerular filtration rate (eGFR) of 100 mL/min per 1.73m² and back-calculate a reference serum creatinine based on height using the Schwartz bedside estimating equations [18–21]. For this estimate, an eGFR of 100 ml/min per 1.73m² was chosen, rather than the standard of 120 ml/min per 1.73m², to avoid over-detection of AKI in our cohort. If baseline serum creatinine and height were not available, we used the admission serum creatinine as the reference creatinine. The secondary exposure was all stages of AKI within 48 hours of ICU admission compared to no AKI.

The primary outcome measure was suspected sepsis within 7 days following the 48-hour exposure window (i.e. days 3–9 after ICU admission). Suspected sepsis was defined as ordering of blood cultures and antibiotics within 24 hours of each other [22, 23]. The secondary outcome measure was an International Classification of Diseases, Ninth Edition (ICD-9) code confirmed sepsis, defined as ordering of blood cultures and antibiotics within 24 hours of each other, plus an ICD-9 diagnosis code of sepsis for the hospitalization (listed in eTable 1 in the Supplement).

Covariates

We performed a literature review to identify risk factors for sepsis, and considered the following susceptibilities and exposures: age, sex, race, comorbid conditions (i.e. cardiac disease, liver disease, solid organ transplant, bone marrow transplant and malignancy), mechanical ventilation, vasopressor use, surgery, and severity of illness at ICU admission using the Pediatric Index of Mortality (PIM) 2 score [24, 25].

Statistical Analysis

Continuous variables are presented as means (standard deviation) when normally distributed or as medians (interquartile range (IQR)) for variables with skewed distribution. Categorical data are presented as numbers (percentages). Characteristics for children with stage 2 or 3 AKI versus children with stage 1 or no AKI were compared using t-test or Wilcoxon rank-sum test for continuous variables and chi-squared test or Fischer-exact test for categorical variables, where appropriate.

We conducted a regression analysis to assess the adjusted association between exposure to stage 2 or 3 AKI and the development of suspected sepsis. Potential confounders for the model were identified as all covariates associated with both the exposure and outcome variables on univariate analysis. To identify a more parsimonious model, some variables were removed based on their independent association with the outcome. After removing each variable, we confirmed that we did not introduce significant bias on our assessment of the association between AKI and the development of suspected sepsis. We also confirmed that model fit was not compromised, based on the likelihood ratio test. The same analysis was repeated for our secondary exposure; all stages of AKI within 48 hours of ICU admission compared to no AKI. After identifying our final models, we evaluated model diagnostics, which included looking for potential outliers, assessing overall model fit, and evaluating discrimination of the model. Potential outliers were evaluated by standardized residuals and logistic model assumptions were tested using partial residual plots and variance inflation factors. Goodness of fit was assessed using Hosmer-Lemeshow methodology. Discrimination ability of the fitted logistic regression model was assessed using the area under the curve (AUC) of the receiver operating characteristic (ROC) plot. Contingency table chi-squared analysis was used to assess the unadjusted association between the primary exposure and the secondary outcome of interest. Statistical significance was set at a p-value of less than or equal to 0.05. Statistical analyses were performed using Stata software, version SE 14.2 (StataCorp, TX, USA).

Results

Cohort Characteristics

A total of 5538 children were included in this analysis, of which 255 (4.6%) had stage 2 or 3 AKI (Figure 1). Of the 255 children with stage 2 or 3 AKI, 201 (78.8%) had stage 2 AKI and 54 (21.2%) had stage 3 AKI. Median age was 5.3 years (IQR 1.5–12.3), 58.2% were male, 78.2% were Caucasian and 16.3% were African American (Table 1). Children who had stage 2 or 3 AKI within the first 48 hours of ICU admission had greater severity of illness based on PIM2 score, higher frequency of mechanical ventilation, vasopressor use, and surgical admission to the ICU (Table 1).

Table 1.

Patient Characteristics for Eligible Study Patients

Characteristics	No AKI or Stage 1 AKI n=5283 (95.4%)	Stage 2 or Stage 3 AKI n=255 (4.6%)	All n=5538	P Value
Age, Median (IQR), years	5.3 (1.5–2.2)	6.3 (1.1–4.8)	5.3 (1.5–12.3)	.2
Males No. (%)	3,071 (58.1)	153 (60.0)	3,224 (58.2)	.6
Race No. (%)
Caucasian	4119 (78.0)	211 (82.8)	4,330 (78.2)	.2
African-American	879 (16.6)	32 (12.6)	911 (16.3)
Other	285 (5.4)	12 (4.7)	297 (5.4)
Mechanical Ventilation No. (%)a	1124 (21.3)	126 (49.4)	1,250 (22.6)	<.001
Vasopressor Use No. (%)a	911 (17.2)	72 (28.2)	983 (17.8)	<.001
History of Solid Organ Transplant No. (%)	341 (6.5)	13 (5.1)	354 (6.4)	.4
History of Heart Failure No. (%)	135 (2.6)	16 (6.3)	151 (2.7)	<.001
History of Liver Failure No. (%)	52 (1.0)	13 (5.1)	65 (1.2)	<.001
History of Malignancy No. (%)	185 (3.5)	8 (3.1)	193 (3.5)	.8
PIM2 Risk of Mortality, Median (IQR)a	0.57 (0.2–1.1)	1.2 (0.5–3.9)	0.64 (0.2–1.1)	<.001
Surgical Admission No. (%)	2,134 (40.4)	129 (50.6)	2,263 (40.9)	.001
ICU Length of Stay, Median (IQR), days	2 (1.0–3.0)	3 (2.0–6.0)	2 (1.0–3.0)	<.001
Hospital Length of Stay, Median (IQR), days	4 (3.0–7.0)	6 (4.0–15.0)	4 (3.0–7.0)	<.001
Hospital Mortality No. (%)	34 (0.6)	7 (2.8)	41 (0.7)	<.001

Abbreviations: AKI, Acute Kidney Injury; ICU, intensive care unit; PIM2, pediatric index of mortality score; IQR, interquartile range

^aMeasured within 24 hours after ICU admission

Regarding determination of reference serum creatinine values, a baseline serum creatinine measurement was available in 3739 children (67.5%). Of the remaining 1799 children, 245 (4.4%) did not have an available height; therefore, the admission serum creatinine measurement was used as the reference creatinine. For the 1554 children with a recorded height (28.1%), the reference creatinine was back calculated from an eGFR of 100 cc/min per 1.73m² using the Schwartz bedside estimating equations. Maximum AKI stage was determined by both serum creatinine and urine output in 7.8% of children, by urine output in 30.8% of children, by serum creatinine in 59.3% of children, and by need for renal replacement therapy in 2.1% of children.

Development of Suspected Hospital-Acquired Sepsis

Suspected sepsis occurred in 46 (18.0%) critically ill children with stage 2 or 3 AKI and 286 (5.4%) children with stage 1 or no AKI on days 3–9 after ICU admission. Of those with stage 2 AKI (n=201) and stage 3 AKI (n=54), 33 (16.4%) and 13 (24.1%) developed subsequent sepsis, respectively. On unadjusted analysis, odds of developing sepsis were 3.85 times higher for children with stage 2 or 3 AKI compared to those with stage 1 or no AKI (95% confidence interval (CI) 2.74–5.41, P<.001). The p-test for trend was significant at P<.001, consistent with our hypothesis of a dose-response relationship between AKI severity and odds of developing suspected sepsis. Other variables that were significantly associated with the development of sepsis included mechanical ventilation and vasopressor use within the first 24 hours of ICU admission, PIM2 score, surgery prior to ICU admission, history of heart failure and history of liver failure. History of malignancy and history of solid organ transplant were not associated with the development of suspected sepsis on univariable analysis. There were too few children with history of bone marrow transplant (n=8) to assess the association between bone marrow transplant and the development of sepsis. Older children had a lower risk of developing suspected sepsis compared to younger children (Table 2).

Table 2.

Variables associated with the development of suspected hospital-acquired sepsis in critically ill children on unadjusted analysis

Outcome: Sepsis	Odds Ratio (95% CI)	P Value
Stage 2 or 3 AKI	3.85 (2.74–5.41)	<.001
Stage 2 AKI versus Stage 1 or no AKI	3.43 (2.32–5.08)	<.001
Stage 3 AKI versus Stage 1 or no AKI	5.54 (2.94–10.46)	<.001
Age	0.96 (0.94–0.98)	<.001
History of Heart Failure	4.56 (3.03–6.85)	<.001
History of Liver Failure	9.15 (5.43–15.4)	<.001
Mechanical Ventilation	5.76 (4.58–7.24)	<.001
Vasopressor Use	2.85 (2.25–3.61)	<.001
Surgery Prior to ICU Admission	4.60 (3.57–5.92)	<.001
PIM2 score	1.64 (1.53–1.75)	<.001

Abbreviations: AKI, acute kidney injury; CI, confidence interval; ICU, intensive care unit; PIM2, pediatric index of mortality score; OR, odds ratio

After covariates adjustment, stage 2 or 3 AKI remained a significant predictor for the development of suspected hospital-acquired sepsis during days 3–9 after ICU admission. Covariates included in the final multivariable regression model were age, PIM2 score, mechanical ventilation within the first 24 hours of ICU admission, surgery prior to ICU admission, a history of heart failure and a history of liver failure. On adjusted analysis, children with stage 2 or 3 AKI had 2.05 times greater odds of developing suspected sepsis compared to those with stage 1 or no AKI (95% CI 1.39–3.03, P<.001) (eTable 2 in the Supplement). The multivariable logistic regression model showed good fit and reasonable discrimination (AUC, 0.83). Looking at AKI severity, both stage 2 and 3 AKI were significant predictors for the development of sepsis when compared to children with stage 1 or no AKI, with adjusted odds of 1.79 (95% CI 1.15–2.79, P=.01) and 3.24 (95% CI 1.55–6.80, P=.002), respectively (Table 3). The overall test for equality across AKI severity levels on the odds of developing sepsis was significant at P<.001, consistent with our hypothesis that the odds of developing sepsis are different as AKI severity increases.

Table 3.

Multivariable regression model for suspected hospital-acquired sepsis in critically ill children with stage 2 or 3 non-septic acute kidney injury

Outcome: Sepsis	OR (95% CI)	P Value
AKIa
Stage 2 AKI versus Stage 1 or no AKI	1.79 (1.15–2.79)	.01
Stage 3 AKI versus Stage 1 or no AKI	3.24 (1.55–6.80)	.002
Age	0.96 (0.94–0.98)	<.001
History of Heart Failure	1.67 (1.04–2.70)	.04
History of Liver Failure	2.06 (1.11–3.81)	.02
Mechanical Ventilationb	2.31 (1.70–3.13)	<.001
Surgery Prior to ICU Admission	4.10 (3.15–5.35)	<.001
PIM2 scoreb	0.94 (0.92–0.97)	<.001

Abbreviations: AKI, acute kidney injury; CI, confidence interval; ICU, intensive care unit; PIM2, pediatric index of mortality score; OR, odds ratio

^aAcute Kidney Injury defined by severity

^bMeasured within 24 hours after ICU admission

Of the 46 children with stage 2 or 3 AKI who developed suspected sepsis, 30 (65%) were surgical admissions, 20 (43%) required vasopressor use and 36 (78%) required mechanical ventilation. The 30 surgical procedures performed prior to ICU admission included: 1 tonsillectomy and adenoidectomy, 1 tympanostomy tube placement, 1 cranial vault remodeling, 2 craniotomies, 2 ventriculoperitoneal shunt placement, 1 lung biopsy, 1 chest tube insertion, 1 exploratory laparotomy, 1 small bowel resection, 1 appendectomy, 1 splenectomy, 1 splenorenal shunt placement, 1 lower extremity fasciotomy, 8 cardiac surgeries, 5 liver transplants and 2 small bowel transplants. Chronic conditions present in this subgroup included: heart failure (11%), congenital heart disease (22%), liver failure (15%), history of malignancy (2%), solid organ transplant (11%), bone marrow transplant (2%) and history of seizures or epilepsy (13%).

Looking at our secondary exposure, no AKI versus all stages of AKI, 720 of the 5538 children (13.0%) developed AKI. Of those with AKI, 465 (8.4%) had stage 1 AKI. Suspected sepsis occurred in 11% (n=80) of children with AKI and in 5.2% (n=252) of children without AKI; equating to 7.3% (n=34) of children with stage 1 AKI, 16.4% (n=33) of children with stage 2 AKI and 24.1% (n=13) of children with stage 3 AKI. On unadjusted analysis, AKI was associated with 2.26-fold increased odds of developing sepsis, compared to children without AKI (95% CI 1.74–2.95, P<.001; eTable 3 in the Supplement). On adjusted analysis, AKI remained a significant predictor for the development of suspected sepsis (OR 1.39 [95% CI 1.04–1.86], P=.03; eTable 4 in the Supplement). However, when evaluating AKI by severity, stage 1 AKI was not significantly associated with the subsequent development of suspected sepsis, with an OR of 1.42 (95% CI 0.98–2.06, P=.06) on unadjusted analysis and 1.00 (95% CI 0.67–1.48, P=.9) following covariate adjustment (eTable 3 and eTable 5 in the Supplement).

Development of Confirmed Sepsis

Evaluating the secondary outcome, we had 38 children meet the ICD-9 code confirmed definition of sepsis: 32 (0.6%) children with stage 1 or no AKI and 6 (2.4%) with stage 2 or 3 AKI. The association between AKI and subsequent sepsis was significant on unadjusted analysis (OR 3.95 [95% CI 1.34–9.70], P=.001). Due to the limited number of children meeting our outcome criteria, adjusted analysis could not be performed.

Discussion

AKI is relatively common among critically ill children and is often associated with longer lengths of ICU and hospital stay, greater duration of mechanical ventilation and higher risk for hospital mortality [5, 6, 26]. While infection is a common cause of death among hospitalized patients with AKI [6, 7], little is known about the impact of AKI on systemic immunity and susceptibility to subsequent infection. To our knowledge, this is only the second study to show a direct association between early non-septic AKI and an increased risk for developing hospital-acquired sepsis in critically ill children. In our cohort, suspected sepsis developed in 18.0% of children with stage 2 or 3 non-septic AKI compared to 5.4% of children with stage 1 or no AKI, equating to a 2.0 times greater odds of developing suspected sepsis following moderate or severe AKI on adjusted analysis. These findings support recent data by SooHoo et al. [27], which demonstrated that children in the cardiac ICU with AKI had a 2.54 times higher odds of developing post-operative infection compared to those without AKI after adjusting for delayed sternal closure, postoperative steroids, presence of a peritoneal drain, cardiopulmonary bypass duration and ethnicity.

Importantly, our study demonstrates that worsening stages of AKI confer greater risk for subsequent infection, with a 79% increase in the odds of developing suspected sepsis in children with stage 2 AKI and a 224% increase in the odds in children with stage 3 AKI. A trend was also noted among children with stage 1 AKI, in which 7.3% developed sepsis compared to 5.2% for those without AKI. Given the high prevalence of AKI among critically ill children and sepsis-related mortality of pediatric ICU patients as high as 40% globally [28], this small but notable difference in rates of sepsis between children with stage 1 AKI and no AKI may have a meaningful impact on ICU-related morbidity and mortality when assessed in a large cohort of critically ill children.

Prior studies assessing the frequency of sepsis following non-septic AKI, found that sepsis occurred in 34–45% of adult patients after the diagnosis of AKI and prior to hospital discharge [8, 10, 29]. When looking at AKI and subsequent infection in neonates following Norwood procedures, SooHoo et al. reported that post-operative infection occurred in 64% of neonates with cardiac surgery-related AKI, compared to 28% of those without AKI [27]. The observed frequency of suspected sepsis occurring in our patient cohort was lower, ranging from 16.4–24.1% in children with stage 2 or 3 AKI and 11% in children with any stage of AKI. While previous studies identified very specific patient cohorts with high rates of AKI and infection, our population included a broader range of children with varying risks for AKI and sepsis. Furthermore, the risk for infection following AKI may differ between pediatric and adult patients, and even between pediatric patients of various ages. Adaptive and innate immune function matures and evolves throughout infancy, childhood and young adulthood, and eventually undergoes remodeling and decline in late adulthood [30]. These changes in immune function over time may influence the association between AKI and subsequent infection, in additional to the evolution of comorbid conditions and AKI etiology with advancing age.

The pathophysiologic mechanisms for immune dysfunction following AKI remain under investigation. Several experimental models of AKI have demonstrated attenuated neutrophil recruitment in remote organs [31]. Further studies have revealed interference of PSGL-1/E-selectin-dependent signally pathways in the setting of AKI, leading to compromised neutrophil rolling and migration [15]. Modification of actin polymerization essential for neutrophil cytoskeleton regulation and cell migration has also been demonstrated [31]. Mouse models of AKI have shown that pre-existing AKI significantly increases the severity of bacterial pneumonia demonstrated by reduced oxygen levels and higher bacterial load compared to mice without AKI, independent of AKI etiology [14]. Certain cell signaling proteins may play a key role in neutrophil dysfunction during AKI. Resistin, a cysteine-rich plasma protein integral for carbohydrate-lipid metabolism, has been shown to impede neutrophil chemotaxis and production of reactive oxygen species via inhibition of the PI3K pathway [32], and is elevated in a dose dependent manner in patients with chronic renal dysfunction [33, 34]. Neutrophil recruitment and function are central to the innate immune response to bacterial and fungal infections, and therefore alteration in neutrophil activity may pose a plausible mechanism for AKI as a risk factor for infection.

Our study has several important strengths. Our cohort included a large diverse population of general and cardiac ICU children, which allowed us to investigate the effect of non-septic AKI on the development of sepsis in a heterogeneous cohort of critically ill children. Importantly, children with sepsis prior to or within 48 hours of ICU admission were excluded from this analysis, which decreased risk for confounding by prior infection. Furthermore, the granular data provided within the Peds HiDenIC database allowed us to establish temporal priority of the exposure variable, thereby strengthening the causal relationship between AKI and subsequent infection. However, there are also several important limitations to our study. First, microbiologic data was not available for our patient cohort; therefore, we used a definition of suspected sepsis rather than culture positive sepsis for our primary outcome. As a result, we may have captured some children without true infection. To address this, we assessed the association between exposure to stage 2 and 3 AKI and the development of “confirmed” sepsis; ordering of blood cultures and antibiotics within 24 hours of each other, plus a ICD-9 code for sepsis for the hospitalization. Although an adjusted analysis could not be performed due to the limited number of children meeting our definition for confirmed sepsis, the unadjusted association remained strong with an OR of 3.95 (95% CI 1.34–9.70, P=.001). Second, based on previously reported data by Mehta et al. and Soohoo et al [8, 27], we looked for the development of sepsis in the 7 days following AKI exposure. However, current data showed that patients may be at increased risk for infection up to 1 year following AKI [13]. As a result, we may have underestimated the number of children who developed sepsis after exposure to stage 2 or 3 AKI, thereby underestimating the association between AKI and subsequent infection. Third, as a retrospective study, we were unable to adjust for all conditions and therapies placing children at increased risk for infection. Specifically, a review of operative reports was utilized to identify type of surgery for the subgroup of patients with AKI who developed subsequent sepsis but was not readily available for the remaining 2233 surgical admissions to the ICU. As a result, adjustment for specific type of surgery, such as splenectomy, could not be performed. However, to account for two major risk factors for nosocomial infection, we excluded patients with chronic kidney disease and adjusted for all-cause surgical admissions to the ICU. Lastly, although children in the final cohort included a heterogeneous mixture of pediatric ICU patients, all came from an academic medical center providing tertiary care. Accordingly, associated comorbidities and severity of illness of the patient population may not be representative of critically ill children nationwide.

Conclusions

Non-septic AKI is associated with an increased risk for subsequent infection in critically ill children. Results of this study and other recently published research further support the concept of AKI as a clinically relevant immunocompromised state. As such, health care providers should have heightened awareness for possible infection in critically ill children following an episode of AKI. Future multicenter pediatric studies are needed to gain a greater understanding of the impact of AKI on patient outcomes as they relate to infection, and to assess if routine screening for infection or evidence of immune dysfunction is indicated in critically ill children with AKI.