Introduction
Adequate nutrition in children is essential for growth and development, and thereby in case of critical illness and trauma, for survival and recovery. Critically ill children are at high risk of malnutrition during their stay on the Pediatric Intensive Care Unit (PICU), due to their metabolic response to stress, injury or inflammation and their metabolic alterations during the course of illness. In addition the children are mostly in a poor nutritional status while admitting to the PICU , have little energy reserves and less muscle mass than adults. , Furthermore children tend to have a greater basal metabolic requirements per unit of weight comparatively to adults, because of their need to sustain growth , wherefore malnourishment will have significant consequences.
The prevalence of malnutrition has remained unchanged over the last two decades in the PICU’s1, and happen to be around 75% of the patients days in the Emma Children’s Hospital (AMC, Amsterdam, The Netherlands) . Malnutrition in critically ill children is associated with loss of critical lean body mass, hospitality mortality, morbidity, longer stay in the PICU, an increase in the incidence of complications, more ventilator days and higher nosociomial infection rates. , , , Since the fact that adequate nutrition is associated with better outcomes an physiologic stability , adequate nutrition supplement is requested in the PICU. After the publication of the first pediatric critical care nutrition guidelines by the American Society for Parenteral and Enteral Nutrition (ASPEN) in 2009, the volume of research on this topic has increased substantially, recognizing the importance of nutrition in the outcomes of children with critical illness.
It is a challenging task for the PICU-staff to provide the optimal amount of nutrients to critically ill children, due to the fact that there is an enormous heterogeneity in relation to age, diagnoses, interventions and comorbidities. Furthermore interventions typically for a PICU-setting markedly change the energy needs, such as mechanical ventilation and vasoactive or sedative agents. Therefore, it is overly simplistic to establish a single guideline applicable to all patients. Determination of individual nutritional needs is essential to build customized nutrition therapy, with to goal of augmenting the short-term benefits and minimizing the long-term harmful consequences. The individual nutritional needs can be either measured at the bedside with indirect calorimetry (IC) or predicted by using various equations.
Total energy expenditure (TEE) is the daily energy requirement of subjects, and consists of 3 components: basal energy expenditure (BEE), diet-induced thermogenesis (DIT) and physical activity (PA). For the determination of the nutrition status the resting energy expenditure (REE) is the most frequently used in the PICU. This because of the fact that REE is comparable to BEE but measured in less standardized ambient conditions, with a small influence of DIT and PA, corresponding to the situation in the PICU. REE is 70-80% of the calories used by the body during a period of 24 hours, which can be defined as the energy released to maintain normal basal physiological functioning and determines the metabolic rate. 10 Metabolic states include hypometabolism (when the measured resting energy expenditure [MREE] <90% of predicted), normal metabolism (MREE between 90%–110% of predicted), and hypermetabolism (MREE >110% of predicted). 10, , , , ,
The gold standard for measuring energy expenditure in critically ill children is the use of indirect calorimetry (IC).1 IC measures O2 consumption and CO2 production. Based on these values it is possible to calculate REE with the modified Weir equation: REE(kcal/day) = [3.941(VO2) + 1.106(VCO2)] × 1440. In addition, IC also allows for the estimation of preferential oxygenation of macronutrients (i.e. proteins, fats and carbohydrates) by simultaneous measurement of the respiratory quotient (RQ). Each individual substrate has an inherent RQ which reflects the percentage use of each substrate, like fat (RQ = 0,70), protein (RQ=0,80) and carbohydrate (RQ=1,0). There are three different ways to perform the measurements of the in- and exhaled gases, namely the ventilated hood, by a mouth piece and in a whole-body chamber. The most common and less labour-intensive way is the use of the ventilated hood. Limitations of IC are the high costs and the relatively complex execution which requires trained personnel for its correct use. In addition, technical factors may prevent correct measurements, such as air leaks, incorrect calibration and equipment malfunctioning and should be watched. 13
Determination under- and overfeeding
The caloric intake is adequate when energy intake meets total energy expenditure, including growth and physical activity. Underfeeding can be described as energy intake-energy expenditure ratio < 0.9, while overfeeding is defined as an energy intake-energy expenditure ratio >1.1 with in between (0.9-1.1) is considered as adequate. The parameters nitrogen-balance (N-balance) and the respiratory quotient (RQ) are hypothesized to be useful for easy determination of under- and overfeeding in daily use.
Nitrogen- Balance
The N-balance gives information about the protein status and therefore on under- and over feeding. This due to the fact that there is no storage pool of non-functional proteins, so the N-balance reflects on the actual intake. A N-balance < 0 means that the increase of whole-body proteolysis is greater than the increase in protein synthesis. However, an important limitation of measuring the N-balance is that no correction is made for nitrogen losses through stools, skin, wound nasogastric suction or blood sampling, due to the difficulty of this determinations.30 Therefore using the N-balance as a parameter for adequate feeding is not sufficient.
Respiratory Quotient
IC measurements include additional parameters to the amount of kcal, such as RQ, which is VCO2 divided by VO2 and can be used to differentiate between under-and over-feeding, and to determine the adequacy of dietary macronutrient composition. Underfeeding forces the body to utilize fat stores to meet caloric requirements, which will lower the RQ to <0.85. Overfeeding and carbohydrate intake in excess of oxidation capacity will increase the lipogenesis and the RQ to >1.0. Used in this way, the RQ provides an evaluation tool for nutritional adequacy. However, from studies in critically children the following data were observed: 22,
RQ (<0.85) and underfeeding (n = 40) RQ (>1.0) and overfeeding (n=135)
– Positive predictive value = 60% – Positive predictive value = 93%
– Negative predictive value = 90%. – Negative predictive value = 35%
– Sensitivity = 63% – Sensitivity = 21%
– Specificity = 89% – Specificity = 97%
Table 1: sensitivity, specificity, positive predictive and negative predictive values in generally ill critical patients
It can be concluded that due to poor sensitivity the RQ is unreliable as a parameter for establishing under- and overfeeding in individual children (Table 1). The same results occur while stratifying for burn patients. , The high negative predictive value of underfeeding and the high positive predictive value suggest that with reasonable certainty, a RQ higher than 0.85 excludes underfeeding and a RQ higher than 1.0 correctly classified children as overfed.
In conclusion, it is clear that the RQ is of little value when predicting the adequacy of the under- and overfeeding in general critically ill patient and burn patients, and it is only useful detecting severe under- and overfeeding.
Identification of risk-groups
Children at high risk for alterations in metabolic state can be identified by a change in clinical status, or matching the following criteria set by ASPEN for targeting patients for IC when resources are limited (Table 2). 1 These patients are at high risk for malnourishment and thus in need for IC measurements to determine adequate feeding.
• Underweight (BMI < 5th percentile for age), at risk of overweight (BMI > 85th percentile for age) or overweight (BMI > 95th percentile for age)
• Children with > 10% weight gain or loss during ICU stay
• Failure to consistently meet prescribed caloric goals
• Failure to wean, or need to escalate respiratory support
• Need for muscle relaxants for > 7 days
• Neurologic trauma (traumatic, hypoxic and/or ischemic) with evidence of dysautonomia
• Oncologic diagnoses (including children with stem cell or bone marrow transplant)
• Children with thermal injury or amputations
• Children requiring mechanical ventilator support for > 3 days
• Children suspected to be severely hypermetabolic (status epilepticus, hyperthermia, systemic inflammatory response syndrome, dysautonomic storms, etc) or hypometabolic (hypothermia, hypothyroidism, pentobarbital or midazolam coma, etc.)
• Any patient with ICU LOS > 4 weeks may benefit from IC to assess adequacy of nutrient intake.
Table 2. Criteria for high risk for metabolic alterations per ASPEN guidelines.
In a prospective chart review of 150 consecutive PICU patients with variable diagnoses, Kyle et al. (2012) showed that 3 out of 4 patients on the PICU were candidates for IC based on 1 or more the ASPEN criteria. The most frequent indications for IC, the most favourable diagnosis and further prioritization according to Kyle and colleagues are shown in table 3 .
Most frequent criteria for IC Diagnosis with ≥ 2 criteria Further prioritization
– Suspected to be severely hypermetabolic
– Not meeting nutrition goals
– Under/overweight / obese – Sepsis/septic shock
– Neurological disorders
– Respiratory diseases – Age > 2 years
– PICU LOS* > 5 days
TABLE 3. PRIORITIZATION OF RISK-GROUPS
*Length of Stay
Predicted energy expenditure versus measured energy expenditure
In a prospective cohort study in ventilated critically ill children, performed by Vasquez-Martinez et al (2004) , multiple predictive equations methods (Harris-Benedict, Caldwell-Kennedy, Schofiel, WHO, Maffies, Fleish, Kleiber, Dreyer and Hunter equations) have been compared with measured energy expenditure. Most of the equations reported an overestimation by the predictive equations and a significant difference between de predicted and measured expenditure. The Fleish and Caldwell-Kennedy equations were identified as the best predictors of energy expenditure. In addition, a study of Kyle et al. (2012)26 demonstrated that underfeeding as based on the Schofield equation appears to be a significant problem in the early days of the PICU stay. This is due to the fact that the variability of stress response varies extremely during the first period of critical illness, and standard predictive equations fail to account for effects of the underlying medical condition and the impact of the interventions of the same illness over time in individual patients , , In conclusion, prediction of energy expenditure is not desirable in critically ill children and individualized determination of nutrition requirements are needed.
Use of IC
Since the fact that critically ill children are at risk for malnutrition, IC measurements are recommended. However no consensus exists about the frequency and the required time period for measurements by IC in the general PICU. In different cohort studies MREE did not significantly vary over time ( 1 week) within the same stable patient, independently of diagnosis. 13, , , While other studies report differences within days in a PICU population, especially in multi-trauma patients, due to changes during the illness course. , Additionally de Klerk et al. (2002) concluded that a great variability in Total Daily Energy Expenditure (TDEE) between all the patients admitted to the PICU occurred, contrary to a small variability within the individual patients (n=18). In other words further research about the frequency will be needed.
The essential duration of measurement is investigated by Smallwood et al. (2012) , who examined in a prospective study the agreement between REE obtained in stable PICU patients in a short period of time; 3-, 4- and 5-minute measurements and 24-h measurements. Concluding that in a steady state an abbreviated period was optimally correlated with 24-h REE measurements, but only when achieving a steady state. 27 The steady state is determined as: minute-to-minute VO2 and VCO2 varied by not more than 10%.25 Along with a shortened period of measurements comes the benefit that presumably a higher percentage of the children will meet the steady state and so on, measuring for a period of 3 minutes is the most effective way to measure. However in metabolic unstable-patients (meeting the ASPEN criteria) longer periods from 30-min35, to 2h measurements may be required.
Adequate feeding and clinical outcomes
Studies describing the clinical outcomes according to estimated adequate calorie delivery (>80% of REE) reported different data. Wong et al. (2017) reported in children with Acute Respiratory Distress Syndrome (ARDS) that adequate calories supplement was not associated with better clinical outcomes, only overfeeding did. On the other hand Metha et al. (2012) conducted a multi-centre study involving 500 intubated PICU patients. Patients receiving >66% of prescribed calories via Enteral Nutrition (EN) had a lower PICU mortality than the patients receiving <33% of the prescribed calories, which demonstrates a positive correlation with meeting the adequate feeding calorie-goals and better outcomes. While Betue (2015) and her colleagues concluded that, with a multivariate analysis correcting for malnutrition, severity of disease, age, surgery and route of nutrition, no significant assumptions can be made between being fed below, within or above the target energy range and the patients outcome variables. Taking in mind the estimated EE along with the various outcomes, a strong hypothesis considering the outcomes of adequate calorie delivery according to measured EE cannot be formulated. Nonetheless the ameliorating part of adequate calorie delivery can be expected due to the numerous reported unfavourable consequences of under- and over feeding.
Furthermore Wong et al. described adequate proteins based on estimated EE decreased PICU mortality and increased Ventilated Free Days (VFDS). The outcomes in the study of Wong et al. about the adequate protein intake are in agreement with other observational studies. The study of Weijs et al. (2014) consisted of 1295 mechanically ventilated children, of which a part received <20% intake of the prescribed protein take and the other part >60% of the prescribed goal, where the 60-days mortality was lower in the second group. In addition the study of Metha et al. (2005) demonstrated that in the early stage of disease in non-septic patients high protein was associated with lower mortality, and inadequate protein delivery is significantly associated with mortality, even after adjusting for severity of illness. From this previous reported articles based on estimated values, we can hypothesize the favorable effect of adequate protein delivery.
Due to the time consuming procedure (notably the preparations) and the need for trained staff the impossibility occurs to measure every child admitted in the PICU. Therefore a prioritization in patient population in the PICU has to be established. As mentioned before, ASPEN formulated recommendations for selecting patients at high risk of changes in their metabolic state during their stay on the PICU (Table 1). However, the effectiveness of feeding according to measurements by IC in the patients meeting the criteria, leading to better patient-outcomes, has not been investigated. In addition no prioritization in the leading diagnostic subgroups in the PICU is contrived (cardiac, sepsis, ARDS, burned and AKI) in those criteria. In conclusion having an insight in the patient group in the PICU who will experience the most benefits of IC measurements followed by guided feeding, creates a prioritization for the use of IC.
Therefore, in this study, we have formulated the following research question: which (sub)groups of critically ill children in a PICU benefit most from IC-guided nutritional support, as expressed by primary, secondary and tertiary outcome measures? In this study, short term, primary outcome measures are: Length of Stay (LOS) in the PICU, Ventilated Free days (VFD), mortality and multi-organ dysfunction. Long-term, secondary measures are: linear growth, motoric and psychologic function scales and reported quality of life after discharge from the PICU. And finally, permanent dependency on health care facilities with associated increased costs (were considered tertiary outcomes).
Methods
Search methodology
A literature search was performed to find articles reporting on critically ill children admitted to the paediatric intensive care unit (PICU), who had received feeding as measured by indirect calorimetry (IC). The Pubmed database was searched for original articles published between 1998-2018, meeting the inclusion and exclusion criteria. The reason for limiting the time period to 20 years is that at that time optimal nutrition was brought to attention. Additionally, the subspecialty of pediatric intensive care has only been recognized since 30 years , so research in this field appeared a few years later.
Other databases were not searched due to the fact that Pubmed is marked as the most common database and includes most of the articles of other databases. The search terms were determined by the author, checked by the clinical Librarian and finally approved by a researcher who is an expert on IC and malnutrition at the PICU. To ensure identification of all relevant studies, no filters were used for study population or study design. The search combined synonyms for ‘undernutrition/energy intake’, ‘energy expenditure/calorimetry’, ‘ICU’ and ‘children’. Table 3 shows an overview of the used search terms.
Undernutrition / Energy intake Malnutrition [Mesh] OR Energy Intake [Mesh] OR Nutritional Requirement [Mesh] OR Nutritional Status [Mesh] OR nutrition Assessment [Mesh] OR Nutrition Requirement* [tiab] OR Dietary Requirement* [tiab] OR Malnourishment*[tiab] OR malnutrition[tiab] OR undernutrition*[tiab] OR Caloric Intake [tiab] OR Caloric requirement* [tiab]
AND Energy expenditure / Calorimetry Energy expenditure [Mesh] OR Basal Metabolism [Mesh] OR Calorimetry, Indirect [Mesh] OR Calorimetry, Indirect/methods* [tiab]OR Resting Metabolic Rate [tiab] OR energy expenditure*[tiab] OR metabolic rate*[tiab] OR Energy Metabolism*[tiab] OR Resting Energy expenditure* [tiab]
AND ICU Critical Care [Mesh] OR Intensive Care Units, Pediatric [Mesh] OR Critical Illness [Mesh] OR Critically Ill [tiab] OR pediatric Intensive Care[tiab]OR critical ill*[tiab] OR Intensive Care[tiab] OR PICU*[tiab]
AND Children Child[Mesh] OR Child, Preschool [Mesh] OR Adolescent [Mesh] OR Infant [Mesh] OR Pediatrics [Mesh] OR child*[tiab] OR Infant*[tiab] OR Adolescent*[tiab] OR teen*[tiab] OR pediatric*[tiab] OR paediatric*[tiab]
Table 3. Search strategy
Inclusion criteria
Patients: Children aged < 18 years (excluding neonates) admitted to the PICU. In this study neonates were excluded due to the fact that there are not admitted to the PICU but to a specialized neonatal ICU.
Intervention: IC-guided nutritional support
Comparison:
Outcomes: short term, primary outcome measures are: Length of Stay (LOS) in the PICU, Ventilated Free days (VFD), mortality and multi-organ dysfunction. Long-term, secondary measures are: linear growth, motoric and psychologic function scales and reported quality of life after discharge from the PICU. Tertiary outcomes are: permanent dependency on health care facilities with associated increased costs.
Study designs: Controlled trial, prospective cohort study, case control study
Year of publication: 1998 – 2018
Language: English
Exclusion criteria
Patients: neonates, adults (≥18), age of patients not defined, non-PICU patients
Intervention: no nutritional intervention/feeding
Study design: Retrospective cohort study, case reports; on account of the impossibility of feeding by measurements with a concurrent control group could
Data extraction
All publications were imported in EndNote©, where the articles were sorted by title and abstract by the reviewer. Articles not meeting the in- and exclusion criteria were excluded. The sorting was checked by the previously mentioned researcher. Following title and abstract screening, full text screening was conducted, while excluding articles not meeting the in- and exclusion criteria.
Results
Following the search strategy initially included 151 studies, without any duplications. After excluding languages other than English, 116 articles remained. After screening the title and abstract 112 articles were excluded. After full text screening, no articles remained. Reasons for exclusion were (studies could be excluded for more than one reason): no feeding according to measured energy expenditure by IC n=151, no English n=17, no children n=41, retrospective or case report design n=4, non-PICU patients n=6, no abstract available n=1, non-human study n=1, dated before 1998 n=35 (Figure 1). The main reason for excluding articles was that in none of the studies patients received their feeding based on results of IC measurements.
Discussion
With this review, we wanted to establish for which subgroup of critically ill children in the PICU adequate feeding according to IC measurements was the most beneficial. We searched through Pubmed using the search strategy described in the method section. No studies reported any data pertaining to the research question. Therefore, no statements can be made about which paediatric intensive care unit patients benefit most, in the short- and long-term, from receiving adequate feeding, as measured by IC. Nevertheless, almost all the articles reported about the essence of adequate feeding in the PICU and the consequences of under- and overfeeding on the outcomes of the critically ill children. The statement that using IC is important has been reported multiple times, however no study has compared clinical outcomes of critically ill children, who have been fed using predictive equations or IC.
However a study using IC measurements has been executed in critically ill adults instead of children. Singer et al. (2011) performed a study with the aim to determine whether nutritional support guided by repeated measurements of REE improves the outcome of critically ill patients (n = 56, adults, mean age = 59) as compared to a single weight-based measurement (n =56, adults, mean age = 62). No significant differences between the groups concerning the patient characteristics were found. Looking at the primary outcome, the study demonstrated that hospital mortality was significant lower in the study group ( p =0,023), while in contrast ICU mortality was not significantly different. The researchers applied hospitality mortality as primary outcome, due to the fact that nutritional interventions may be expected to have more impact on the long-term ICU variables due to the fact that it requires more time to become apparent. In addition to the primary outcomes, a longer ICU stay and a longer duration of mechanical ventilation has been found as secondary outcomes of the study group. Another study conducted by Strack van Schijndel and colleagues (2009) reported as main finding that reaching an energy goal guided by indirect calorimetry and the provision of protein intake of at least 1,2g/kg reduces the ICU, the 28-day and the hospital mortality in females, while in men there was no significant correlation. Noteworthy is that in the study by Singer et al, the mortality on the long term (hospital mortality) was lower than the ICU-mortality, while in the study of Strack van Schijndel the particular mortality outcomes were the other way around, nevertheless both significant. Because the metabolism of adult differs from that of children, we cannot transfer these conclusions to critically ill children. Nevertheless it supports the hypothesis that adequate feeding according IC measurements indeed contribute to at least a lower mortality rate in the PICU.
For various reasons researchers and clinicians continue to predict energy expenditure, while it has been clear for a longer period that predicted values will not be sufficient. This literature review was originally conducted to detect the sub-population which will benefit the most in patients outcomes while adequately fed. The reason for this was that while having an insight in the subgroup most benefiting from IC guided feeding, IC measurements can at least be conducted in the target subpopulation when access and time is limited. Since there is no study reporting data about feeding according to measurements, we will give some recommendations based on the literature, obtained from our search, about how to take care of the different subgroups and the feeding in general in the PICU. In the PICU the five most common sub-populations are cardiac patients, septic patients, patients with ARDS, burned patients and patients with acute kidney injury (AKI).
General recommendations; General recommendations, for the nutrition support in the PICU, are given by ASPEN in their latest clinical guidelines.1 We will point out some of their recommendations from which, in our opinion, it is essential to pay daily attention to at the PICU. Since a lot of studies reported that the agreement between predicted energy expenditure and measured energy expenditure was poor, we suggest that it is preferable to always use IC within 24h of admission, to guide prescription of the daily energy goal. This recommendation is based on the findings by Hulst and colleagues (2004) who showed that inadequate feeding in the first few day contributed to 50% of the total caloric and protein deficits. In addition, it is necessary to develop validated screening methods to identify malnutrition in patients, contributing to a quick use of IC. The criteria composed by ASPEN (table 1) for high risk metabolic alterations need to be used as criteria to qualify patients for IC measurements and guided feeding in the PICU, while keeping the further prioritization in mind when there is limited availability: age <2 year, under- or overweight, PICU LOS > 5days.18 The moment, the duration and the frequency of measurements by IC should be carried out as followed. The best moment to conduct the first IC measurement is, as stated before, within the first 24h, and the ideal time of the day was found to be at noon. In addition the duration of measurement only needs to be 30 minutes in which an abbreviated steady state of at least 3- or 4-minutes need to be achieved. REE during a shortened period is perfectly correlated with 24-hous measurements but only when steady state is met. The mean of al steady states values for REE during the 30-minute test improves the accuracy and represents the actual REE better than equations. 27 In children not meeting the criteria by ASPEN (table 1) one measurement at admitting to the PICU will be sufficient for the nutrition supplement for the next week. 26 However in children meeting the criteria, IC measurement is needed every day due to rapid changes in REE.
Cardiac patients; Pre- and postoperative nutrition state is extremely important for good clinical outcomes in cardiac patients. Pre-operatively is it essential to achieve a normal weight-for-age z-score, contributing to better anthropometry and decreasing morbidity and mortality. The metabolic profile in cardiac patients postoperatively is highly variable, especially in the immediate postoperative period due to the fact that the cardiac diagnoses has been shown to affect the resting energy expenditure enormously . Energy requirements normalize to an anabolic state and normo-metabolic response in comparison to healthy controls, 1 week after the surgery. This leads to the recommendation that especially in the first week after cardiac surgery, daily IC measurements are favourable. Additionally the peri-operative nutrition state, where the weight-for-age score acts as an important marker, need to be supported by measuring the adequate caloric target by IC.
Sepsis; In the study of Turi et al (2001)25 an observational case-control study cohort, consisting of 13 infants and 8 children <7 year diagnosed with SIRS or Sepsis and equivalent controls, data reported that no differences in REE between cases and controls occurred. In addition the controls had a similar REE to that reported by other authors studying REE in healthy control infants or children of similar age. Resulting in the assumption that children with SIRS or sepsis do not suffer from hypermetabolism and therefore do not seem to require higher caloric intake than stable non-septic patients. Additionally Oosterveld et al. (2006) reported that MEE was comparable to predicted EE values in children with sepsis. Taking this together we don’t initially recommend IC measurements, as a result of the satisfying value of the predicted EE.
ARDS; The use of the IC is problematic and inaccurate in patients with ARDS due to technical factors that decrease the accuracy of IC. In a study conducted by Wong and colleagues (2017) 38 it became clear that especially inadequate protein was associated with PICU mortality, which bring us to the recommendation that clinicians should optimize protein delivery instead of merely total calories. The N-balance can be used for detecting the adequacy of the protein delivery, while taking into account the losses through stools and fluids. Ideally a compensating mechanism should be developed for optimal correction for the losses.
Burn; A lot of studies reported data and recommendations about burn in adult patients, wherefore no strong recommendations can be formulated about burned patients in the PICU. In a study within burned adult patients it’s found that nutrition therapy should be initiated within 12h with enteral feeding, we recommend applying this in the burned children as well . Thereby REE differs between burned patients due to the size of Total Body Surface Area (TBSA) which is correlated with increasing metabolic rate. Hypermetabolism, due to the production of stress-hormones , is found in the burned critically adults and children, so to avoid overfeeding, IC measurements are needed. For optimizing the feeding protocol it is suggested to use IC at least twice a week, as a result of fast changing REE in this subpopulation (confirmed in adults) due to additional problems which may occur; such as sepsis, illness and degree of wound contamination. 58 In contributing for optimizing the care of critically ill children, Hart et al (2002) did find an interesting significant relation between declining energy expenditure and mortality in the first 2-3 weeks after the injury and later on in week 5. This insight leads to the need of extra attention to a declining REE, to prevent mortality.
AKI; underfeeding is accentuated in children with AKI which is linked to increased morbidity and mortality. AKI is very common in the first 5 days of admitting in the PICU, where especially protein malnutrition is present. To reduce the consequences of AKI, which is not avoidable, it is important that in the first 5 days special notice is given to if AKI developing and that in that case IC measurements are implemented.
We believe that our review has clearly demonstrated that no studies have been published of which the outcome answers our research question. First of all in our opinion the search strategy can be marked as high level. This is the result of the several adjustments made on the search strategy, based on the relevance of the outcome articles at first sight and the adding of mesh-terms and ‘tiabs’ of probably relevant articles found in references of relevant articles. Secondly, we did not exclude articles for which we were not sure whether they did or did not answer our research question, based on the title and abstract. This contributes to a not too premature exclusion and thereby missing of relevant articles. Additionally the selection of the studies by two reviewers independently avoided selection bias.
The limitations of this review are partly due to the fact that the search has been performed in solely one database, as explained in the method section. However we cannot guarantee that no relevant articles exist in other databases, leading to a small selection bias. Thereby by excluding non-English articles, relevant information has potentially been missed. In addition, the recommendations stated in this discussion part are based on the studies obtained by the described research strategy (method section), however this research strategy was not formulated to vouch for creating recommendations, which may lead to recommendations based on a limited part of the existing literature and may potentially not be complete.
A multi-centre randomized clinical trial should be performed where the effect of nutrition support on the patient outcomes should be evaluated. The following patient outcomes should be evaluated: In the short term, primary outcome measures: Length of Stay (LOS) in the PICU, Ventilated Free days (VFD), mortality and multi-organ dysfunction. In the long-term, secondary measures: linear growth, motoric and psychologic function scales and reported quality of life after discharge from the PICU. And finally, permanent dependency on health care facilities with associated increased costs (were considered tertiary outcomes).
The recommended study design would be a case-control cohort study in a time span of 1 year of measuring and feeding according to IC and 10 years of follow up screening on the secondary and tertiary outcomes. The study should have twelve arms; 2 of each before mentioned subgroup, cardiac, sepsis, ARDS, burns and AKI, plus a subgroup ‘others’ consisting of: 1) caloric supply based on daily measurements by IC at noon and 2)caloric supply based on estimations with corrections for physical activity and severity of illness (control group). Patients included in the research should be children (≤18 year) matching the inclusion and exclusion criteria stated in the method section. Besides that patients need to fulfil the following criteria recommended by ASPEN for succeeding the IC measurements; expected stay of at least 48 h, mechanical ventilation, FiO2< 0,60, tube leakage < 10%, hemodynamically stable condition. The RQ could be used for determining the accuracy of the IC measurements and for indicating if the nutrition supplement is correctly in retrospective, using a RQ higher than 0,85 for excluding underfeeding and a RQ higher than 1.0 for correctly classifying overfeeding.
With this study we aim to investigate if feeding according to IC measurements actually leads to better patients outcomes on the short- and long-term instead of feeding according to the current feeding protocol and if so, which patient group benefit the most, so priorities can be set. Adequate protein delivery is not taken into account within this proposal but need to be investigated in a different study. Afterwards a study need to be conducted in which both adequate protein and caloric delivery need to be applied.