Pediatric enteral nutrition therapy for burn victims: when should it be initiated?

Objective To review the scientific evidence regarding the initiation of enteral nutrition in the pediatric burn population. Methods This study was a systematic review and meta-analysis of randomized clinical trials comparing early enteral nutrition and late enteral nutrition in individuals aged 1 month to 18 years with burns. The MEDLINE/PubMed, Embase and Cochrane Library databases were searched using the terms "burns", "fires", "child nutrition disorders", "nutritional support" and related terms. Results Three articles that included a total of 781 patients were identified. There was no significant difference in the mortality rate between the early and late groups (OR = 0.72, 95%CI = 0.46 - 1.15, p = 0.17). Patients who received early enteral nutrition had a 3.69-day reduction in the length of hospital stay (mean difference = -3.69, 95%CI = -4.11 - -3.27, p < 0.00001). There was a higher incidence of diarrhea and vomiting and decreased intestinal permeability in the early group. This group also presented higher a serum insulin concentration and insulin/glucagon ratio as well as lower caloric deficit and weight loss when compared to the control group. Conclusion Analysis of the different intragroup variables suggests the importance of starting nutritional support early. Considering the number of pediatric burn patients, there is a need for robust studies with greater scientific impact.

Burns trigger a series of changes in the body, with manifestations resulting from skin injury and responses in the affected systems, especially the hemodynamic, respiratory and metabolic systems. (6) The metabolic response to burn injury is complex and characterized by hypercatabolism, leading to a negative nitrogen balance and significant loss of musculoskeletal mass. (7) The nutritional demand in pediatric patients with burns is increased by factors of great relevance, including body growth and development, high levels of oxidative stress, an intense inflammatory response and prolonged hypercatabolism, which are characteristic of burn injuries. (8) Thus, enteral nutrition is the preferred food option to achieve adequate nutritional support in these patients who require greater nutritional intake because it maintains the integrity of the intestinal mucosa, reduces the incidence of bacterial translocation and decreases the risk of infectious complications. (9,10) Therefore, nutritional therapy plays a key role in the treatment of these patients, and its initiation is recommended early, after hemodynamic stability. (6) Although the importance of nutritional support in pediatric patients with burns is well established, there are few studies evaluating the best time to start enteral nutrition safely and effectively.
The objective of this systematic review and metaanalysis was to evaluate the scientific evidence regarding enteral nutrition initiation in this population.

Protocol and registration
The protocol for this systematic review was registered in PROSPERO [CRD42017077665] and is available at http://www.crd.york.ac.uk/PROSPERO.

Eligibility criteria
The studies were selected based on the following criteria: randomized clinical trial (RCT) comparing early versus late enteral nutrition therapy initiation in pediatric patients with burns. Enteral nutrition therapy was considered early when the enteral diet was started within the first 24 hours after trauma and considered late when started 48 hours after trauma.

Information sources
The following electronic databases were searched for studies that were published prior to August 2017: MEDLINE (accessed vis PubMed), Embase and Cochrane Central Register of Controlled Trials. No data restrictions were applied when performing this search. Only articles in English, Spanish and Portuguese were considered.

Search strategy
No systematic review addressing this research question in the pediatric population with burns was available in the MEDLINE (via PubMed), Embase or Cochrane databases at the beginning of this study. In August 2017, database searches were performed using the following keywords and related terms to obtain the broadest possible results: "burns", "fires", "child nutrition disorders" and "nutritional support". The search strategies were adapted according to the database used and are shown in detail in Appendix A.

Study selection
Duplicate articles were identified and manually excluded. Retrieved articles were analyzed using the following procedures. Titles and abstracts found in the literature search were analyzed independently by two reviewers based on the inclusion/exclusion criteria. Next, all articles included in the study were subjected to full text analysis to identify eligible articles. Disagreements were resolved by consensus or the opinion of a third judge. Agreement between reviewers was assessed using Cohen's kappa coefficient.

Data extraction and methodological quality assessment
Two independent reviewers extracted data from each study using a table template designed for this purpose. The following data were extracted: study identification, study design, population description, intervention details and results. Primary outcomes included mortality and length of hospital stay. Secondary outcomes included clinical complications, hormonal changes, caloric deficit and/or weight loss. A third reviewer evaluated all studies for the integrity of data extraction.
To assess the methodological quality and risk of bias of the studies, the Cochrane Collaboration's tool (11) was used for each included RCT. The following areas were evaluated: sequence generation, allocation concealment, blinding of participants, blinding of outcome assessment, incomplete outcome data, selective reporting of outcomes and other biases. Each domain was classified as low, medium or high risk of bias.

Data synthesis and analysis
To begin the study inclusion/exclusion process, the StArt program was used. The meta-analysis was performed using the statistical software Review Manager 5.3 (RevMan 5.3).

Literature search
Initially, 4,826 titles related to the search strategy used were identified. After excluding ineligible titles and duplicate abstracts, only three original articles were included. A flowchart showing the study selection process is detailed in figure 1. The two independent researchers who evaluated the studies presented excellent interrater reliability, with a Cohen's kappa coefficient of 0.9.

General characteristics of the studies and populations
Three single-center nonblinded RCTs with a total of 781 subjects were included: 413 patients in the intervention group, who received early enteral nutrition (EEN) within the first 24 hours after a burn, and 368 patients in the control group, who received late enteral nutrition (LEN) 48 hours after a burn.
The main characteristics of the included studies are described in table 1. The studies were conducted in different locations: one in North America, (12) one in Africa (13) and one in Asia. (14) The age of the included patients ranged from 30 days to 18 years, and all were admitted to burn intensive care units (ICUs). All studies (12)(13)(14) reported the progression of diet volume throughout the patients' hospital stay, but Gottschlich et al. (12) did not report the initial feeding volume administered. The volume of fluids used in the replacement and maintenance stages was calculated separately and was not included in the nutritional requirements of these patients. To define the nutritional needs of the included patients, Gottschlich et al. (12) used their institution's protocol, (15) and Venter et al. (13) used the formula published by Solomon (16) and Galveston's derived equation published by Hildreth et al. (17) Khorasani and Mansouri (14) used the formula proposed by Seashore (18) to estimate the energy needs of the study population.
Regarding the characteristics of the study population, only one study (12) reported the sex of the participants, who were predominantly male. In all studies, the EEN and LEN groups were similar in terms of mean age and burned body surface area (BSA). The main characteristics of the studied population are described in table 2.
The main burn types in the studied population were scalding and flame burns. Enteral nutrition was administered via a nasoduodenal tube in one study (12) and via a nasojejunal tube in the other two studies. (13,14) The latter two studies used the oral route in a complementary manner, once the enteral nutrition reached the established goal. The study by Khorasani and Mansouri (14) did not specify the time of nutritional therapy initiation, either for the EEN or LEN group, or the average time to reach the protein-energy goal after nutritional therapy initiation. The main characteristics of the burn types and the nutritional strategy adopted are described in table 3.

Mortality
All included articles reported mortality during the study period. According to the results of the meta-analysis ( Figure 2), there was no significant difference in mortality rate between the EEN and LEN groups (odds ratio [OR] = 0.72; 95% confidence interval [95% CI] 0.46 -1.15; p = 0.17). Only two studies (12,13) described the main causes of death: multiple organ dysfunction and cerebral ischemia.

Length of hospital stay
All three studies reported length of hospital stay; however, Venter et al. (13) did not provide standard deviation data for this outcome, leaving only two studies for the meta-analysis. According to Figure 3, there was a significant difference in the mean length of hospital stay between the EEN and LEN groups (mean difference = -3.69; 95% CI -4.11 --3.27; p < 0.00001), and patients who received early nutritional therapy had a shorter length of hospital stay, by 3.69 days. This result is chiefly due to the study by Khorasani and Mansouri (14) given the inclusion of a larger number of patients.

Clinical complications
Patients in the study by Gottschlich et al. (12) belonging to the EEN group had a higher incidence of diarrhea (63% versus 58%; p = 0.62). Corroborating this finding, Venter et al. (13) observed a higher incidence of diarrhea (2.2% versus 1.0%) and vomiting (1.3% versus 0%) in the EEN group. Venter et al. (13) reported that intestinal permeability decreased significantly on the third day in both groups and remained decreased after 48 hours only in the EEN group (p = 0.02). Khorasani and Mansouri (14) evaluated the presence of paralytic ileus, intestinal obstruction or decreased intestinal perfusion (diarrhea, vomiting, pain and abdominal distension) and did not find these clinical complications in either group during the study period.

Hormonal effect
Different studies have evaluated the effect of EEN on the hormonal parameters of burned pediatric patients. Gottschlich et al. (12) analyzed various hormones, such as cortisol, insulin, glucagon, epinephrine, norepinephrine, dopamine, gastrin, triiodothyronine (T 3 ) and tetraiodothyronine (T 4 ). Only serum insulin (55.1% versus 20.4%; p = 0.0004) and T 3 (40.7% versus 30.2%; p = 0.0162) were significantly higher during the first week in the EEN group. Venter et al. (13) analyzed the effect of glucagon, insulin, growth hormone (GH), cortisol and insulin-1 growth factor. The analyses of these data showed that in the EEN group, there was a significantly higher insulin concentration and insulin/glucagon ratio when compared to those in the LEN group (p = 0.008 and p = 0.04, respectively). The GH concentration was significantly higher in the LEN group compared to that in the EEN group until the 12th day (p = 0.03). There was no significant difference in the other hormones analyzed.

Caloric deficit
Gottschlich et al. (12) found significantly lower caloric deficits in the EEN group in the first (-363 versus -1,960 kcal; p < 0.0001) and second (-898 versus -2,734 kcal; p = 0.0022) evaluated weeks. Venter et al. (13) observed that in the intervention group, there was a lower caloric deficit in the first 26 days (excluding the first two days), but there was no significant difference relative to the LEN group (p = 0.7).

Risk of bias of the included studies
All studies used a randomized controlled design, but Venter et al. (13) did not report how the randomization process occurred. Regarding blinding, the studies were classified as nonblinded but with no effect on the results. All studies described allocation concealment, reported specific outcomes and explained why patients were lost, when this ocurred (Table 4).

DISCUSSION
The importance of enteral nutrition for pediatric patients with burn injuries is known; however, there is still a gap in knowledge on the optimal time to start nutritional support. The guidelines recommend starting enteral feeding within 24 hours (19,20) or even within 12 hours (21)(22)(23) after a burn injury. However, these studies were performed in animal models or have selection bias due to the inclusion of adults. (24,25) This systematic review summarizes the evidence available in the literature over the last 15 years. Three RCTs were found in the languages included, showing that there are few studies on the pediatric population when compared with the number of studies on adults.
One of the difficulties found in this population was estimating the actual caloric expenditure at different recovery times. In the population studied, the caloric expenditure was estimated using different techniques and formulas among the studies. (15)(16)(17)(18)(26)(27)(28) Regarding energy intake, the studies showed lower caloric deficit in the EEN group than in the LEN group, explaining the less pronounced weight loss in the EEN group, as evidenced in the study by Venter et al. (13) Corroborating these findings, Khorasani and Mansouri (14) found lower weight loss in the EEN group, even though Other sources of bias Low There seems to be no other sources of bias

Study identification
Khorasani and Mansouri (14) Study type RCT they did not evaluate caloric intake. In burned patients, weight may undergo several variations mainly due to the fluids used in the recovery and maintenance phases, demonstrating the importance of recognizing longterm effects and of monitoring during the rehabilitation phase. (29) Weight, age, BSA percentage and burn cause are important data to estimate and monitor the metabolic expenditure of burned pediatric patients when there is no possibility of using a calorimeter. In the studies included in this review, the main burn cause was scalding, which is consistent with results from a multicenter study in 2001, in which 44% of burns in individuals younger than 15 years old occurred due to boiling liquids. (30) Reinforcing the lack of studies with robust and reliable results, we highlight the study by Khorasani and Mansouri, (14) which reported no information on the time of nutritional therapy initiation, negatively affecting the comparison with other studies; however, mortality and length of hospital stay decreased significantly in the EEN group when compared to the LEN group, which was only observed in that study. This result is mainly due to the larger sample size and sample homogeneity compared to the other studies. In agreement with these findings, a multicenter study with 153 burned adults that evaluated the effect of EEN found a significant reduction in the length of ICU stay (p = 0.03). (31) This finding, therefore, has a considerable impact on hospitalization cost reduction and higher hospital bed turnover.

Risk of bias assessed by the authors Justification of the assessment
In addition to mortality, this systematic review evaluated other outcomes. Among the gastrointestinal complications observed in the studies, vomiting and diarrhea had an equal prevalence in the EEN and LEN groups. These events are, in part, characteristic of burn patients due to a marked decrease in immune function, a sudden change in metabolism and the beginning of antibiotic use.
Another evaluated complication, which may contribute to the impairment of nutritional status, is increased intestinal permeability, which is one of the initial characteristics of damage to the intestinal barrier, contributing to an increase in the occurrence of gastrointestinal complications and sepsis. (32) Venter et al. (13) found a significant decrease in the lactulose:rhamnose ratio, which remained decreased after 48 hours only in the EEN group.
The considerable release of inflammatory mediators in response to metabolic changes caused by stress also compromises the nutritional status of burned pediatric patients. (31) However, the monitoring frequency of these markers, such as insulin, T 3 , cortisol, glucagon, and GH, differs between studies. The assessment of insulin values in these patients is important because insulin is an anabolic hormone with a modulating effect on the immune response, in addition to a trophic effect on mucous membranes and the skin, improving the barrier against microorganism invasion and translocation. (33)(34)(35) The higher insulin concentration in the EEN group in both studies suggests that this strategy facilitates wound healing and decreases the risk of inflammation.
The thyroid hormones T 3 and T 4 , whose reduction can be caused by stress from trauma and vary according to burn severity, (36) were evaluated only by Gottschlich et al., (12) who observed a significant increase in T 3 in the EEN group, suggesting a better prognosis when compared to the LEN group.
Numerous changes in the metabolism of burned patients occur due to cytokines, which stimulate the hypothalamus to increase thermoregulation. These changes generate an increase in the production of stress hormones, causing lipolysis and proteolysis. (37) Classic studies in animals showed a decrease in the production of these stress hormones when the animals were fed within 2 hours after a burn. (24,25) Serum levels of cortisol and glucagon are considered sensitive indicators of stress, are proportional to the injury extent and may persist in high concentrations for up to 3 years after a burn. (38) This result was confirmed in a pediatric population, in which the increase in cortisol levels was positively correlated with BSA. (39) Gottschlich et al. (12) and Venter et al. (13) measured cortisol in their patients and found no significant difference in the levels of this hormone between the groups. In contrast, for glucagon, Venter et al. (13) observed a lower concentration of this hormone in patients from the EEN group, demonstrating a reduced hypermetabolic response when compared to the LEN group.
The catabolic properties of GH during periods of stress and trauma increase gluconeogenesis and lipolysis. (40) In the study by Venter et al., (12) a significantly higher concentration of GH was observed in the LEN group compared to the EEN group until the 12th day, suggesting that patients fed early do not depend on glycogenolysis or gluconeogenesis to obtain their energy requirements, contributing to the maintenance of body composition and, consequently, avoiding the loss of lean mass.

CONCLUSION
After performing this systematic review and given the number of cases of burned pediatric patients, the need for robust studies with greater scientific impact in this population is evident. The studies reviewed herein, despite their limitations, do not rule out the hypothesis that early nutritional support brings great benefits to this population. Analysis of the intragroup variables suggests the importance of starting nutritional support early, as reducing this time period has a significant impact on injury recovery and length of hospital stay and assists in the normal development and growth of these children and adolescents.