Fluid restriction contributes to poor nutritional adequacy in patients with congenital heart disease receiving renal replacement therapy

Abstract

Objective:

Critically ill patients receiving renal replacement therapy (RRT) in the pediatric cardiac intensive care unit (CICU) are at high risk for inadequate nutrition delivery. The objective of this study was to evaluate barriers to adequate energy and protein delivery in critically ill patients with congenital heart disease receiving RRT.

Methods:

This is a single-center retrospective cohort study of patients receiving RRT in the CICU from 2011–2019. Energy and protein adequacy were recorded over the first seven days of RRT. Adequacy was defined as delivery of greater than 80% of the energy and protein targets during this time period. Patients who achieved adequacy were compared to those who did not. Multivariable logistic regression models were constructed to determine factors independently associated with energy and protein adequacy while receiving RRT.

Results:

Sixty patients were included for analysis. Fifty-five patients (92%) achieved energy adequacy and 37 patients (62%) achieved protein adequacy. A higher weight-for-age z-score (WAZ) on admission to the CICU was the only independent predictor of inadequate energy intake (OR 0.07, 95% CI 0.01–0.58; p=0.014); median WAZ was −1.17 vs. +1.24 for those with inadequate vs. adequate energy intake, respectively. Fluid restriction to less than 80% of maintenance fluid at the time of RRT initiation was more likely in patients with higher WAZ. Fluid restriction was the only independent predictor of inadequate protein intake (OR 0.13, 95% CI 0.02–0.7; p=0.018); 5% vs. 30% were fluid restricted in those with inadequate vs. adequate protein intake, respectively. Azotemia was not associated with inadequate protein intake. Initiation of RRT did not allow for liberalization of fluid intake over the time period evaluated.

Conclusion:

Protein delivery was inadequate in 38% of children undergoing RRT in the CICU. Fluid restriction was associated with inadequate protein intake and higher WAZ was associated with inadequate energy intake.

Introduction

Acute kidney injury (AKI) is common in critically ill children, with an incidence ranging 26.9% to 40.3%.^1–3 Post-operative cardiac surgical patients requiring cardiopulmonary bypass are particularly vulnerable to AKI.^4–6 Similarly, patients with medical cardiac disease are at risk for AKI secondary to nephrotoxic medications, renal venous congestion, and low cardiac output leading to inadequate renal perfusion.^7–8 AKI carries significant morbidity and mortality risk in critically ill children with cardiac disease.^4–6 Renal replacement therapy (RRT) is an important intervention in managing AKI, but it is also associated with worse outcomes than those experienced by patients who do not receive RRT.^1,9–10

Inadequate nutrition is one potential contributor to poor outcomes in patients with AKI receiving RRT in the pediatric cardiac intensive care unit (CICU). These patients may have poor baseline nutritional status.^11–13 In addition, oliguria and fluid overload may result in fluid restriction and deferral of adequate nutrition. Concerns about worsening azotemia may prompt suboptimal protein delivery as well.^13–14 While RRT may allow for higher fluid intake, it may also result in altered nutrient metabolism and increased nutrient losses.^15–16

While several studies have been conducted in critically ill children with AKI, there are limited data evaluating the nutritional status and nutrient intake in patients with congenital heart disease (CHD) receiving RRT. Furthermore, identification of the barriers to the delivery of adequate nutrition in this population could provide clinically relevant information in the assessment and management of these patients. The purpose of this study was to describe the nutritional status of a cohort of critically ill patients receiving RRT in the pediatric CICU and to identify factors associated with inadequate nutrition delivery.

Methods

This is a retrospective cohort study conducted at a single tertiary-level pediatric CICU. The CICU involves a multidisciplinary team including a dietitian who performs daily assessments on all patients and in conjunction with the clinical team prescribes and recommends the targets and mode of nutrition delivery. The final decision regarding nutritional intake is made by the supervising physician managing the patient. Standard fluid management practice includes limiting total fluid intake (TFI) to half maintenance in the first 24 hours after surgery involving cardiopulmonary bypass (CPB) and liberalizing to full maintenance volume thereafter. Standard nutritional practices include prioritizing enteral delivery primarily; however, when unable to provide enteral nutrition (i.e., significant vasoactive support, feeding intolerance), parenteral nutrition (PN) is provided if it is anticipated that a patient will not meet at least 80% of goal energy needs within 7 days. For patients with AKI, there is no standard practice of limiting protein intake.

Current Procedural Terminology (CPT) codes were used to identify all patients receiving any mode of renal replacement therapy (RRT) during an inpatient hospital encounter between January 1, 2011 and December 31, 2019. Patients receiving RRT for the indications of acute kidney injury (AKI) or fluid accumulation were included for analysis. Exclusion criteria included prematurity (gestational age < 37 weeks), RRT initiation outside of our CICU, administration of RRT for an indication other than AKI or fluid accumulation (e.g., hyperammonemia), chronic kidney disease (defined as an estimated glomerular filtration rate [eGFR] <60 mL/min/1.73 m²), and/or incomplete documentation. Given the increasing prevalence of adult patients with CHD, the study group elected to include all ages.¹⁷ This study was approved by the Institutional Review Board with waiver of informed consent.

Data were manually extracted from the medical record on standardized data collection forms. Data collected included sex, age at RRT initiation, and anthropometric variables including weight-for-age z-score (WAZ), height-for-age z-score (HAZ), and BMI z-score at admission. Z-scores were calculated using the World Health Organization (WHO) growth standards for patients less than 2 years old and the Center for Disease Control (CDC) growth standards for patients greater than or equal to 2 years old. Diagnostic and procedural characteristics were recorded and categorized for all patients, and operative variables including CPB, cross-clamp, and circulatory arrest times were collected for surgical patients. Severity of illness was captured through calculation of admission Pediatric Risk Index of Mortality 3 (PRISM3) score¹⁸, level of respiratory support and vasoactive-inotropic score (VIS)¹⁹ at the time of RRT initiation, ventilator-free days (VFDs) in the 28 days after RRT initiation, and the initiation of extracorporeal membrane oxygenation (ECMO) at any point during the hospitalization.

Markers of renal function evaluated include the lowest eGFR (calculated using the modified Schwartz formula²⁰) between CICU admission and RRT initiation, blood urea nitrogen (BUN) level at the time of RRT initiation, and urine output in the 24 hours prior to RRT initiation. Fluid balance was recorded from CICU admission to RRT initiation (limited to the seven days prior to RRT initiation) and calculated as a percentage using the equation²¹: [Fluid in (L)-Fluid out (L)]/[CICU admission weight (kg)] * 100%. Patients with a negative fluid balance were assigned a value of 0.0%.

Maintenance fluid requirement was calculated using the Holliday-Segar formula²², and patients were defined as fluid restricted if their TFI on the day of RRT initiation was less than or equal to 80% of maintenance. This threshold was chosen to acknowledge the enhanced degree of fluid restriction placed on patients with AKI within our CICU. Dietitians provide recommendations for energy and protein targets for patients. Energy targets were calculated using the Schofield equation without the addition of a stress factor.²³ The Schofield equation is a method for estimating resting energy expenditure and is recommended in the absence of indirect calorimetry.²⁴ Protein targets were defined as 3 g/kg/day for children less than one year and 1.5 g/kg/day for those older than one year of age based on the Society of Critical Care Medicine and American Society for Parenteral and Enteral Nutrition (SCCM-ASPEN) guidelines.²⁴ Actual energy and protein intake as well as fluid intake and fluid balance were recorded for the first 7 days after RRT initiation, but only if RRT was still being administered. Route of nutrition was also recorded. Adequacy was calculated as: actual intake / recommended intake *100. The primary outcome was energy adequacy by day 7 after starting RRT. For this outcome, we hypothesized that fluid restriction would be a barrier to achieving adequate energy intake. The secondary outcome was protein adequacy by day 7 after starting RRT. For this outcome, we hypothesized that adequate protein intake is not associated with azotemia (defined as a BUN greater than three times the upper limit of normal). Adequacy was defined as achieving > 80% of goal energy or protein intake – a metric identified in contemporaneous nutrition studies and our institutional trigger for the prescription of supplemental PN.^25–26

Descriptive statistics were used to explore the cohort. Continuous data were presented as medians with interquartile ranges (IQRs). Patients who received goal energy intake were compared with those who did not using a Fisher exact test for categorical predictors and Student’s t-test or Wilcoxon rank sum test for continuous predictors. Continuous predictors were also transformed into categorical measures based on data tertiles to identify any nonlinear relationships. Univariate logistic regression was used to estimate association between predictors and energy adequacy. Multivariable logistic regression models were constructed using variables with a univariate p-value < 0.20 as candidate predictors. The stepwise regression algorithm required a p-value of 0.15 to enter the model and p-value < 0.05 to be retained in the model. Results are presented as odds ratios (OR) with 95% confidence intervals. These analyses were also performed to compare those who achieved protein adequacy with those who did not. A p-value of less than 0.05 was considered statistically significant. Analyses were performed using SAS version 9.4 (SAS Institute, Inc., Cary, NC).

Results

Seventy patients received RRT in the CICU during the study time period, and sixty met criteria for analysis. Of the ten excluded patients, two were premature infants, two had CKD, four had inadequate documentation, and two were started on RRT at an outside hospital. The cohort was 67% male with a median (IQR) age of 5.2 (0.4, 16.1) years (Table 1). Twelve patients were over 18 with the majority under the age of 22. Aside from obstructive sleep apnea in one patient, there were no significant comorbidities in this population that were not related to the underlying CHD diagnosis. The median (IQR) CICU admission WAZ was −0.95 (−1.86, −0.02), HAZ was −1.23 (−2.27, −0.29), and BMI z-score was −0.19 (−1.41, 0.73), suggesting baseline malnutrition. 14 patients (23.3%) had a WAZ < −2, 20 patients (33.3%) had a HAZ < −2, and 7 patients (11.7%) had a BMI z-score < −2. For the 27 patients less than 3 years of age who had head circumference documented, the median (IQR) z-score was −1.16 (−2.76, −0.45). Twenty-five patients (42%) had single ventricle anatomy and 47 (78%) were surgical patients. Among the most common surgical procedures were complex biventricular repair (n=9; 15%), heart or lung transplant (n=9; 15%), and Norwood procedure (n=8; 13%). Heart failure was the most common non-surgical diagnosis (n=9 of 13; 69%). Median CPB time in surgical patients was 238 minutes (IQR 179, 318) and cross-clamp time was 118 minutes (IQR 63, 202). The acuity of the cohort was relatively high, with a median admission PRISM3 score of 13 (IQR 7, 20), VIS of 10 (3.2, 17.5) at the time of RRT initiation, VFDs of 0 (IQR 0, 3), and 42 patients (70%) requiring ECMO at some point during hospitalization. All patients had stage 3 AKI using the Kidney Disease: Improving Global Outcome Criteria (KDIGO).²⁷ Median CICU length of stay was 42.5 days (IQR 33.2, 84.2) and 39 patients (65%) died prior to hospital discharge.

Table 1.

Baseline Demographics of Patients Receiving Renal Replacement Therapy

Variable	n (%) or Median (IQR)
Age, y	5.2 (0.4, 16.1)
Male sex	40 (66.7)
Weight*, kg	15.85 (4.45, 51.35)
Weight-for-age z-score*	−0.95 (−1.86, −0.02)
Height*, cm	107.0 (54.0, 156.0)
Height-for-age z-score*	−1.23 (−2.27, −0.29)
BMI z-score*	−0.19 (−1.41, 0.73)
Surgical patients	47 (78.3)
Cardiopulmonary bypass time, min	238 (179, 318)
Cross clamp time, min	118 (63, 202)
Circulatory arrest time, min	0 (0, 2)
PRISM-3 score*	13 (7, 20)
Percent fluid accumulation†, %	10.3 (4.8, 17.1)
Lowest eGFR†, mL/min/1.73 m2	15.75 (11.58, 20.69)
Admission BUN, mg/dL	17 (11, 26)
Pre-RRT BUN, mg/dL	130 (86, 155)
Urine output‡, mL/kg/hr	0.4 (0.1, 1.3)
Time from CICU admission to RRT, d	14.4 (7.4, 36.2)
Indication for RRT
Fluid removal	19 (31.7)
Clearance	26 (43.3)
Fluid removal and clearance	15 (25.0)
Mode of RRT
CRRT	20 (33.3)
IHD	22 (36.7)
PD	18 (30.0)
Vasoactive inotropic score§	10 (3.2, 17.5)
Mechanical ventilation§	59 (98.3)
CMV	52 (88.1)
HFOV	7 (11.9)
Ventilator-free days‖	0 (0, 3)
ECMO during hospitalization	42 (70)

Values are presented as either count (with percentage of whole in parentheses) or median value (with interquartile range in parentheses).

BMI = body mass index; BUN = blood urea nitrogen; CICU = cardiac intensive care unit; CRRT = continuous renal replacement therapy; CMV = conventional mechanical ventilation; ECMO = extracorporeal membrane oxygenation; PRISM = Pediatric Risk of Mortality; eGFR = estimated glomerular filtration rate; HFOV = high frequency oscillatory ventilation; IHD = intermittent hemodialysis; IQR = interquartile range; PD = peritoneal dialysis; RRT = renal replacement therapy

^*Measured at CICU admission

^†Measured between CICU admission and RRT initiation

^‡Measured in the 24 hours prior to RRT initiation

^§At time of RRT initiation

^‖In 28 days following RRT initiation

Figure 1 displays the distribution of energy adequacy, protein adequacy, and fluid balance amongst the cohort over the duration of RRT. The majority of patients (n=55, 92%) achieved energy adequacy and 39 patients (65%) received goal energy requirements for over half of their RRT duration. Most patients (63%) had achieved energy adequacy by the time RRT was initiated. The five patients who did not achieve goal energy were older (median 15.8 vs. 3.2 years) (Table 2). Two of the patients who did not achieve energy adequacy had heart failure, one patient had Ebstein’s anomaly of the tricuspid valve and underwent cone-type repair, one patient had Fontan circulation and presented with hemoptysis, and one patient had undergone biventricular conversion from a Fontan circulation. Patients who achieved energy adequacy had a lower median WAZ (−1.17 vs. 1.24, p<0.001) and a lower median BMI z-score (−0.32 vs. 1.65, p=0.006) compared to those who did not achieve energy adequacy (Table 2). Nine patients (15%) were fluid restricted to less than 80% of maintenance fluid requirement on the day of RRT initiation, and a higher proportion of patients who were fluid restricted did not achieve goal energy requirements (60% vs. 11% in those without fluid restriction, p=0.021).

Figure 1.

The distribution of (a) energy adequacy, (b) protein adequacy, and (c) fluid balance amongst patients over the first seven days of renal replacement therapy. Most patients had adequate energy intake throughout RRT with fewer having adequate protein intake. Energy adequacy and protein adequacy were calculated as actual intake/recommended intake*100. Fluid balance was calculated as [Fluid in (L)-Fluid out (L)]/[CICU admission weight (kg)] * 100%. The middle horizontal line represents the median of each day, and the lower and upper borders of the boxes represent the 25^th and 75^th percentiles, respectively. The lower and upper whiskers represent the 5^th and 95^th percentiles, respectively. Outliers are represented by the black circles.

Table 2.

Characteristics of patients achieving versus not achieving goal energy and goal protein

Variable	Goal Energy			Goal Protein
	Achieved Goal Energy (n = 55, 91.7%)	Did Not Achieve Goal Energy (n = 5, 8.3%)	p-value	Achieved Goal Protein (n = 37, 61.7%)	Did Not Achieve Goal Protein (n = 23, 38.3%)	p-value
Age, y	3.2 (0.3, 14.8)	15.8 (13.6, 20.3)	0.062	1.9 (0.3, 9.6)	15.0 (0.9, 19.7)	0.048
WAZ*	−1.17 (−1.95, −0.29)	1.24 (0.98, 1.75)	<0.001	−1.09 (−1.81, −0.36)	−0.69 (−2.06, 0.92)	0.187
HAZ*	−1.29 (−2.45, −0.32)	−0.63 (−0.74, 0.02)	0.197	−1.20 (−2.17, −0.25)	−1.29 (−2.78, 0.40)	0.399
BMI z-score*	−0.32 (−1.68, 0.58)	1.65 (0.8, 2.7)	0.006	−0.45 (−1.68, 0.54)	0.09 (−1.17, 2.25)	0.086
Fluid restricted	6 (11)	3 (60)	0.021	2 (5)	7 (30)	0.021
Single ventricle	23 (42%)	2 (40%)	1.000	17 (46)	8 (35)	0.432
Percent fluid accumulation†, %	10.2 (4.4, 17.5)	14.5 (5.2, 15.7)	0.872	12.3 (8.0, 20.4)	8.9 (0.7, 15.1)	0.322
Lowest eGFR†			0.223			0.021
<13.28	16 (29)	3 (60)		7 (19)	12 (52)
13.28–19.41	19 (35)	2 (40)		14 (38)	7 (30)
≥19.41	20 (36)	0 (0)		16 (43)	4 (17)
Elevated BUN‡	50 (91)	5 (100)	1.000	34 (92)	21 (91)	1.000
Average intake PN (cc/kg/d)	35.8 (17.9, 51.4)	9.8 (7.7, 10.1)	0.004	43.3 (26.6, 53.2)	16.6 (10.1, 33.2)	<0.001
Year of RRT			1.000			0.026
2011–2016	35 (63.6)	3 (60)		19 (51.4)	19 (82.6)
2017–2019	20 (36.4)	2 (40)		18 (48.7)	4 (17.4)
Ventilator-free days	0 (0, 3)	0 (0, 3)	0.564	0 (0, 5)	0 (0, 1)	0.606
CICU LOS, d	43 (25, 72)	35.4 (33, 53.3)	0.831	49.9 (30, 72)	36.4 (22.6, 48.4)	0.354
Hospital LOS, d	55.8 (32, 88.2)	47.1 (44.2, 53.3)	0.904	60 (32, 82.8)	47.4 (33.8, 88.2)	0.39
Death	36 (65.5)	3 (60)	1.00	25 (67.6)	14 (60.9)	0.781

All values median (IQR) or n (%). BMI = body mass index; BUN = blood urea nitrogen; eGFR = estimated glomerular filtration rate measured as mL/min/1.73 m²; HAZ = height-for-age z-score; IQR = interquartile range; LOS = length of stay; WAZ = weight-for-age z-score

^*Measured at CICU admission

^†Measured between CICU admission and RRT initiation

^‡BUN more than three times the upper limit of normal for age

In multivariable analysis, only WAZ was associated with achieving energy adequacy, with a decreasing odds for achieving goal energy with higher WAZ (OR 0.07, 0.01–0.58; p=0.014) (Table 3). Fluid restriction was highly correlated with WAZ. Patients without a fluid restriction had a lower WAZ (median: −1.2; IQR: −1.9, −0.2) compared to those with a fluid restriction (median: 0.8; IQR: −0.6, 1.4) (p=0.005). When WAZ was excluded from multivariable modeling, fluid restriction became the only factor associated with inadequate energy intake (OR 0.08, 0.01–0.59; p=0.013).

Table 3.

Multivariable Models for Achieving Goal Energy and Goal Protein

Model	OR (95% CI)	Adjusted OR (95% CI)	p-value
Goal Energy (N=60, 55 events, c-statistic 0.96)
Age	0.76 (0.56 – 1.04)
Weight-for-age z-score	0.07 (0.01 – 0.58)	0.07 (0.01 – 0.58)	0.014
Height-for-age z-score	0.65 (0.34 – 1.25)
BMI z-score	0.39 (0.19 – 0.83)
Fluid restricted	0.08 (0.01 – 0.59)
Single ventricle diagnosis	1.08 (0.17 – 6.98)
Percent fluid accumulation	1.00 (0.92 – 1.08)
Lowest eGFR	1.49 (0.83 – 2.66)
Goal Protein (N=60, 37 events, c-statistic 0.63)
Age	0.79 (0.62 – 1.01)
Weight-for-age z-score	0.79 (0.55 – 1.12)
Height-for-age z-score	1.13 (0.85 – 1.49)
BMI z-score	0.70 (0.50 – 0.99)
Fluid restricted	0.13 (0.02 – 0.70)	0.13 (0.02 – 0.70)	0.018
Single ventricle diagnosis	1.59 (0.54 – 4.67)
Percent fluid accumulation	1.03 (0.98 – 1.08)
Lowest eGFR	1.21 (0.96 – 1.52)
Elevated BUN	1.08 (0.17 – 7.01)

Presented as OR and adjusted OR with 95% CIs. BMI = body mass index; BUN = blood urea nitrogen; CI = confidence interval; eGFR = estimated glomerular filtration rate measured as mL/min/1.73 m²; OR = odds ratio

Thirty-seven patients (62%) achieved protein adequacy with only 28 patients (47%) achieving protein adequacy for over 50% of the days they received RRT. A higher percentage of the patients who did not achieve protein adequacy were fluid restricted compared with those who did (30% vs. 5%, p=0.021) (Table 2). Patients who achieved protein adequacy were also younger (median 1.9 years vs. 15 years, p=0.048). Most patients displayed significant azotemia, with a median BUN of 130 (IQR 86, 155) at the time of RRT initiation. Fifty-five patients (95%) had a BUN greater than three times the upper limit of normal for age, and there was no difference in the proportion of patients meeting these criteria who achieved protein adequacy vs. those that did not (92% vs. 91%, p=1.00). However, there was some indication of inferior renal function in those not achieving protein adequacy, with a larger proportion of these patients being in the lowest tertile of eGFR compared to those who achieved protein adequacy (52% vs. 19%, p=0.021). The majority of patients (n=19, 82.6%) with inadequate protein delivery received RRT prior to 2017. In multivariable analysis, fluid restriction was the only independent correlate. Patients who were fluid restricted were less likely to achieve goal protein requirements compared with those who were not restricted (OR 0.13, 0.02–0.70; p=0.018).

Given the influence fluid restriction had on meeting energy and protein needs, we further explored the route of nutrition delivery as well as the nutritional makeup of the TFI. On average, 93% of calories were provided parenterally. On the day of RRT initiation, an average of 5% of calories were provided enterally and this increased to 12% on the final day of RRT evaluated (Figure 2). Eight patients (13%) were receiving enteral nutrition at the time of RRT initiation, with an additional 23 patients initiating enteral nutrition over the RRT period evaluated. Only two patients received the majority (>50%) of their calories enterally. Both of these patients achieved energy and protein adequacy, and one was on full enteral feeds at RRT initiation. Forty patients (67%) received some amount of enteral nutrition prior to RRT initiation, with 31 of those patients receiving more than trophic feeds (for patients ≤ 5 years, ≥ 2 mL/hr; for patients > 5 years, ≥ 5 mL/hr). The majority of TFI was non-nutritive fluids. Patients achieving protein and energy adequacy had a higher total intake from PN (approximately threefold) compared to those with who did not (Table 2). Eighteen of the 31 patients receiving feeds (58%) had enteral nutrition stopped during RRT with most being secondary to feeding intolerance or pre-procedure. Four patients had PN stopped (3 related to metabolic abnormalities and 1 pre-procedure).

Figure 2.

Average composition of total fluid intake (TFI) on each day of renal replacement therapy (RRT). The average daily TFI is on the Y-axis. Enteral nutrition made up the smallest amount of TFI throughout RRT. The proportion of TFI from enteral nutrition increased throughout the duration of RRT. Other fluids included medications, infusions, and blood products.

Potential complications related to feeding strategy, including infection and adverse gastrointestinal events, were also evaluated. There were 4 bloodstream infections identified (3 fungemia and 1 bacteremia) in patients receiving RRT. An additional two patients had bacteremia prior to RRT initiation. The three fungemia infections were in patients receiving solely PN, with the remainder of patients receiving some amount of enteral nutrition. Five of these patients achieved energy adequacy and five achieved protein adequacy. There were eight gastrointestinal adverse events including 2 bowel perforations, 5 episodes of GI bleeding, and 1 episode of NEC. Seven of these patients achieved energy adequacy and 4 achieved protein adequacy. Aside from three episodes of GI bleeding, all other events occurred prior to RRT initiation.

Discussion

In this retrospective cohort study, we have described the nutritional status and highlighted barriers to achieving energy and protein adequacy in critically ill children with CHD receiving RRT. The cohort had relatively high disease acuity with a majority requiring ECMO support, a high mortality rate and long length of stay. Nearly all patients were able to achieve energy adequacy, mostly via the parenteral route, over the first week of RRT. A high percentage of patients received enteral nutrition, albeit low in volume. Patients that achieved energy adequacy had a lower admission WAZ, whereas absence of fluid restriction was associated with protein adequacy.

The overall nutritional status of the cohort was poor with low WAZ, HAZ, and BMI z-score. This profile is similar to previous studies. In a prospective cohort study of 174 critically ill children receiving CRRT, 35% of patients had a weight below the third percentile.¹¹ The incidence of weight below the third percentile was higher in patients with cardiac disease. In another prospective cohort study evaluating data on critically ill children from two multicenter studies, children with congenital heart disease had the highest prevalence of underweight, defined as a BMI z-score < −2.0.²⁸ In both of these studies, malnutrition was associated with poor outcomes. Almost all of the patients in our cohort suffered from chronic illness, which likely contributed to poor baseline nutritional status. Reduced nutrient stores along with increased nutrient demand seen in this acute group of patients likely played a role in the poor outcomes.

Provision of adequate protein during critical illness is necessary to avoid catabolism. Catabolism is a result of increased protein utilization and impaired protein synthesis which impacts wound healing and functional outcomes due to loss of lean body mass.^16,29–30 However, children are at increased risk for catabolism during critical illness due to reduced protein stores and higher baseline requirements. Additionally, RRT can enhance protein clearance, resulting in increased protein requirements.^16,30

Inadequate protein intake in critically ill children with AKI has been described previously.^13–14 Kyle and colleagues¹³ evaluated critically ill children with AKI and found that only 55% and 19% of patients received adequate energy and protein intake, respectively, in the first five days of ICU admission. Furthermore, Zappitelli and colleagues¹⁴ described inadequate protein delivery in a multicenter evaluation of critically ill children receiving continuous renal replacement therapy (CRRT). In our cohort, a higher percentage of patients achieved energy and protein adequacy, but still over two-thirds of the cohort did not achieve protein goals. The importance of protein adequacy in critically ill patients with AKI is increasingly being acknowledged and we noted an era effect with a higher proportion of patients achieving protein adequacy after 2017, potentially corresponding with the release of the SCCM-ASPEN guidelines.²⁴

Barriers to adequate protein delivery have not been described extensively in our patient population. Nutrition may be deferred early in critical illness resulting in a nutritional deficit – however our cohort was analyzed relatively late in the ICU course (RRT started at median 14.4 days, IQR 7.4, 36.2). Additionally, the prescription of protein in the setting of declining renal function is at times restricted to prevent worsening azotemia. However, this was not independently associated with the adequacy of protein intake in our cohort.

While timing and azotemia were not associated with nutritional adequacy, fluid restriction was associated with both inadequate energy and protein intake. Fluid restriction (<80% of maintenance) at the time of RRT initiation was independently associated with inadequate protein delivery. Fluid restriction was also associated with inadequate energy intake on univariate analysis. The balance between fluid restriction and accumulation has been extensively studied.³¹ However the duration of fluid restriction in our cohort may have led to inadequate nutrition delivery. Fluid restriction was likely instituted given the degree of fluid accumulation in the cohort prior to RRT (median 10.3%). Despite the potential benefit of RRT to liberalize fluid intake for nutrition, the proportion of patients who were fluid restricted at the end of RRT was the same (15%), and 18 patients (30%) were restricted for more than 50% of RRT days evaluated. Additionally, a negative fluid balance was only achieved in 30 patients (50%) while on RRT. It is possible that RRT did not provide sufficient fluid removal for providers to liberalize fluid intake, and this may be reflective of the overall hemodynamic state and severity of illness in the cohort. Consequently, in this cohort, RRT did not allow for liberalization of fluid intake and enhanced delivery of nutrition, which may have been a contributing factor to the overall poor outcomes.

The majority of patients were able to achieve goal energy while receiving RRT. This is likely reflective of the relatively late initiation of RRT with respect to CICU admission. In our cohort, a lower WAZ was independently associated with adequate energy intake. A higher CICU admission WAZ may lead some providers to more willingly defer adequate nutrition. The correlation seen between WAZ and fluid restriction supports this, with patients with higher WAZs more likely to be fluid restricted. Adequate energy intake in the setting of critical illness has been associated with improved outcomes^25,32, but barriers to adequate energy intake have been less well described. Zappitelli and colleagues¹⁴ identified younger age as a predictor of higher energy intake in children receiving CRRT. However, in that study the relationship of weight was not explored and there is no evaluation of energy adequacy. In another multicenter prospective cohort study of critically ill children, Mehta and colleagues³² identified younger age, longer duration of PICU stay, and shorter duration of mechanical ventilation to be associated with increased percentage of energy intake. Those who received a higher percentage of goal energy had lower mortality. The improved outcomes in this cohort were associated with increased energy intake via the enteral route, however. In our cohort, the majority of energy was delivered parenterally, which may have been another contributor to poor clinical outcomes.

There are several limitations to this study. This is a retrospective analysis and reliant on accurate documentation in the medical record. Fluid balance measurements do not consider insensible losses. Additionally, this was a single center study and the results may not be generalizable to patients at other institutions. The small sample size and high percentage (92%) of patients who achieved their goal energy limited the power to detect correlates for achieving versus not achieving energy adequacy. There are limitations to the Schofield equation in accurately predicting energy needs, resulting in underestimation or overestimation of the energy targets. None of the patients underwent indirect calorimetry during the study period. Furthermore, protein dosing recommendations in the SCCM and ASPEN guidelines were based on low level evidence and none of the patients underwent assessment of nitrogen balance. We also did not account for energy delivered via peritoneal dialysis, but this was unlikely to significantly contribute given the majority of patients achieving energy adequacy. Finally, energy and protein intake were only evaluated over the time RRT was received. Protein and energy delivery may have been different before and after RRT, and likely played a role in clinically important outcomes.

Practical Application

Many patients admitted to the CICU have chronic illnesses that make them susceptible to malnutrition. Critical illness and AKI can promote nutritional deterioration in this vulnerable population. Fluid restriction was associated with inadequate protein delivery in patients with AKI receiving RRT and a higher WAZ was associated with inadequate energy intake. The use of RRT in this population did not facilitate liberalization of fluid intake for all nutritional goals to be met.

Abbreviations:

AKI

acute kidney injury

BMI

body mass index

BUN

blood urea nitrogen

CDC

Centers for Disease Control

CHD

congenital heart disease

CICU

cardiac intensive care unit

CPB

cardiopulmonary bypass

CPT

current procedural terminology

ECMO

extracorporeal membrane oxygenation

eGFR

estimated glomerular filtration rate

HAZ

height-for-age z-score

IQR

interquartile range

KDIGO

Kidney Disease: Improving Global Outcome

OR

odds ratio

PN

parenteral nutrition

PRISM3

Pediatric Risk Index of Mortality-3 score

RRT

renal replacement therapy

SCCM-ASPEN

Society of Critical Care Medicine-American Society for Parenteral and Enteral Nutrition

TFI

total fluid intake

VFD

ventilator-free days

VIS

vasoactive-inotropic score

WAZ

weight-for-age z-score

WHO

World Health Organization