Nutritional support for critically ill children
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
The objective of this review is to assess the impact of enteral and total parenteral nutrition on clinically important outcomes in critically ill children. There are two primary hypotheses:
1. the mortality rate of critically ill children fed enterally or parenterally is different compared to children who are given no nutrition;
2. the mortality rate of critically ill children fed enterally is different compared to children fed parenterally.
If data are available, we will conduct subgroup analyses to examine whether the treatment effect is altered by:
a. age (infants less than one year versus children greater than or equal to one year);
b. type of patient (medical (purpose of admission to ICU is for medical illness without surgical intervention immediately prior to admission) versus surgical (purpose of admission to ICU is for postoperative care)).
If there are other clinical trials that examine nutrition more distinctly, the following secondary hypotheses will also be examined:
3. the mortality rate is different in children who are given enteral nutrition alone versus enteral and parenteral combined;
4. the mortality rate is different in children who are given both enteral feeds and parenteral nutrition versus no nutrition.
Background
Theoretical considerations
Feeding in the critically ill child has not been extensively investigated and remains a controversial topic within paediatric intensive care. There are no clear guidelines as to the best form or timing of nutrition in critically ill infants and children. The balance between relative starvation of children with its harmful effects on muscle strength, wound healing, and immune function must be weighed against the harmful effects associated with intravenous nutrition and overfeeding which occurred in older studies (Cerra 1980). Children have increased energy requirements per kilogram of body weight and decreased energy reserves compared to adults, and critical illness disrupts normal metabolic demands by increasing the risk of cell death and impairing growth (Dimand 1999). Current nutritional goals include minimizing the effects of an increased metabolic rate and subsequent energy depletion but not to the extent where overfeeding may worsen the clinical situation (Zaloga 1994). The Recommended Nutrient Intake goals are not appropriate for critically ill children since those goals were designed for healthy children and do not take into account differences in activity, growth, and the changes in metabolism related to critical illness (Dimand 1999). Critical illness often results in abnormal metabolism at several different levels within the human cell (Fink 2001; Hotchkiss 2003; Mizock 1984). Each of these mechanisms makes it difficult to predict food requirements in critically ill children (Dimand 1999). Patients with critical illness and a very high metabolic rate can be markedly catabolic, meaning they can digest their own muscle protein to create more energy, resulting in weight loss and a poor outcome (Cerra 1980; Joffe 2001). Although it seems intuitive that providing nutrition will always be of benefit and malnutrition is harmful, it does not necessarily follow that feeding can improve a patient's outcome (Heyland 1998; Joffe 2001; Klein 1997; Koretz 1995).
Paediatric experience
Children with critical illness are not capable of normal growth and therefore it is unrealistic, and potentially harmful, to try to promote growth in these patients (Chwals 1994). Predicting the food requirements in any patient is difficult and recent studies show that the energy used at rest in critically ill children is significantly less than the predicted energy requirement (Briassoulis 2000; White 2000). Attempts to provide enough nutrition can place the patient at risk of overfeeding and increased carbon dioxide production (which can contribute to ventilatory compromise), abnormal fat production and subsequent liver dysfunction (Chwals 1994; Zaloga 1994). Ideally nutrition should be targeted to support basal metabolic requirements, prevent autodigestion of protein, and not to promote growth. Traditionally, enteral or gut feeding via a naso‐gastric or more recently, naso‐jejunal tube has been promoted within 48 to 72 hours of admission to the paediatric intensive care unit (PICU), depending on the nature and severity of the presenting illness. In those patients in whom enteral nutrition is often delayed because of unstable heart rate or blood pressure, parenteral or intravenous nutrition (TPN) has been instituted days to weeks after admission. There are very few studies examining this practice and controversy still exists as to whether either mode of nutrition is superior in children, and at what point nutrition is necessary to minimize further morbidity. Critically ill children are often fluid restricted and are therefore limited in the volume of nutrition which can be provided; this often delays feeding, especially in young infants. Whether this delay harms has yet to be determined.
Adult experience
Feeding in adult critical care patients is also controversial, but somewhat better studied in terms of general guidelines. Gut barrier dysfunction is thought to be a major contributor to the development of multiple organ dysfunction syndrome (Doig 1998), the most common cause of death in the intensive care unit (Marshall 1996). Enteral nutrition may be superior to TPN based on the surrogate outcomes of gut barrier function and infectious complications (Hadfield 1995; Moore 1992). Furthermore, parenteral nutrition has been associated with increased complications compared with enteral nutrition in adult patients with abdominal trauma (Lipman 1998). Total parenteral nutrition enhances the body's response to certain toxins (Fong 1989), and when compared to intravenous fluid infusion, TPN increased mortality in the severely ill adult patient in a recent meta‐analysis (Heyland 1998). Although information from the above studies and several adult meta‐analyses regarding feeding is helpful, we cannot extrapolate that information directly to children since they adapt differently to critical illness.
Objectives
The objective of this review is to assess the impact of enteral and total parenteral nutrition on clinically important outcomes in critically ill children. There are two primary hypotheses:
1. the mortality rate of critically ill children fed enterally or parenterally is different compared to children who are given no nutrition;
2. the mortality rate of critically ill children fed enterally is different compared to children fed parenterally.
If data are available, we will conduct subgroup analyses to examine whether the treatment effect is altered by:
a. age (infants less than one year versus children greater than or equal to one year);
b. type of patient (medical (purpose of admission to ICU is for medical illness without surgical intervention immediately prior to admission) versus surgical (purpose of admission to ICU is for postoperative care)).
If there are other clinical trials that examine nutrition more distinctly, the following secondary hypotheses will also be examined:
3. the mortality rate is different in children who are given enteral nutrition alone versus enteral and parenteral combined;
4. the mortality rate is different in children who are given both enteral feeds and parenteral nutrition versus no nutrition.
Methods
Criteria for considering studies for this review
Types of studies
We will include randomized controlled trials (RCTs), (completed or ongoing).
Types of participants
We will include trials of paediatric patients aged 1 day to 18 years of age that are cared for in a paediatric intensive care setting and who receive nutrition within the first seven days of admission. We will include studies involving both paediatric and adult participants if data are available for paediatric cases. We will analyse those patients aged less than one year separately from those older than one year if such data are available, given that infants are believed to have higher nutritional requirements compared to older children. Furthermore, medical patients are often studied separately from surgical critical patients. If there are no studies that differentiate medical from surgical patients, we will not consider these groups separately in the analysis.
Types of interventions
Patients must be randomized to receive either:
-
enteral feeding versus no feeding;
-
total parenteral nutrition versus no feeding;
-
enteral versus total parenteral nutrition;
-
enteral versus enteral with supplemental parenteral nutrition.
Other areas of paediatric nutrition, such as immediate versus delayed feeding and immunonutrition versus normal nutrition, will not be addressed in this review.
Types of outcome measures
Primary outcomes
30‐day mortality; if this is not available PICU mortality will be used.
Secondary outcomes
-
length of stay in the PICU;
-
length of stay in hospital;
-
number of days on ventilator;
-
morbid complications including nosocomial infections.
We will not consider nutritional outcomes, quality of life assessments, or economic implications.
Search methods for identification of studies
Electronic searches
We will search the following bibliographic databases: the Cochrane Central Register of Controlled Trials (CENTRAL), (current issue of The Cochrane Library), OVID MEDLINE (1966 to present), OVID EMBASE (1988 to present), OVID All Evidence Based Medicine Reviews (includes Cochrane DSR, ACP Journal Club, DARE, and CCTR), ISI Web of Science ‐ Science Citation Index Expanded (1965 to present), WebSPIRS Biological Abstracts (1969 to present) and WebSPIRS CAB Abstracts (1972 to present)) and trial registries (ClinicalTrials.gov, CenterWatch Clinical Trials Listing Service, Current Controlled Trials, GlaxoSmithKline Clinical Trial Register, National Clinical Trials Registry and the National Research Register; all found at: http://www.ualberta.ca/ARCHE/litsearch.html#trials).
The MEDLINE search strategy (1966 to June 2003, Week One) is available in Appendix 1. We will modify this strategy as necessary for the other databases.
Searching other resources
We will review reference lists of all selected articles, handsearch relevant conference proceedings (the British Association for Parenteral and Enteral Nutrition, the European Society of Parenteral and Enteral Nutrition, and the American Society of Parenteral and Enteral Nutrition), and contact primary authors, experts in the area, and manufacturers of enteral nutrition products.
We will not limit the search by language or publication status.
Data collection and analysis
Study selection
The selection of studies will involve two steps. First, two authors will independently screen the results from searching to identify citations with potential relevance (AJ, NA). Second, we will obtain the full text of selected articles. Two authors will independently decide on trial inclusion using a standard form with predetermined eligibility criteria (AJ, NA). We will resolve any disagreements through discussion with a third author (LL).
Assessment of quality
Two authors will independently assess the methodological quality of all included studies using the Jadad 5‐point scale (Jadad 1996) (NA, LH). The Jadad scale is the only quality assessment tool, to the best of our knowledge, that has been validated and it incorporates components that are directly related to the control of bias including randomization (0 to 2 points), double blinding (0 to 2 points), and reporting of withdrawals and dropouts (0 to 1 point). We will provide overall quality scores according to the Jadad scale. In addition, we will describe and display the quality information by individual components (ie generation of random sequence, blinding, loss to follow‐up, and allocation concealment (Schulz 1995)) which will be classified as adequate, inadequate, unclear or not used. We will examine the effect of methodological quality through sensitivity analyses as described below under "Data Analysis". In addition, we will record whether or not the studies used an intention‐to‐treat analysis, as well as the funding source. We will resolve differences through discussion with a third party (BV).
Data extraction
Two authors will extract data from each study (NA, LL, or BV) and discrepancies will be resolved through discussion and by referring to the original paper. We will request unpublished data from authors when necessary. We will use a standard form that will describe the following: characteristics of the study (design, method of randomization, withdrawals or dropouts); participants (age, gender); intervention (type, dose, route of administration, timing and duration of therapy, co‐interventions); control (agent and dose); outcomes (types of outcome measures, timing of outcomes, adverse effects); and results. We will request additional data from study authors as required.
Data analysis
We will conduct separate analyses for the four comparisons: enteral feeding versus standard care; total parenteral nutrition versus standard care; enteral versus total parenteral nutrition; and enteral versus enteral with supplemental parenteral nutrition. We will express dichotomous data (for example, mortality) as a relative risk and we will calculate an overall relative risk with 95% confidence intervals. We will likely express complications as risk differences due to low event risks. We will derive the number needed to treat for dichotomous data to help clarify the degree of benefit for a range of baseline risks. We will convert continuous data to the mean difference and we will calculate an overall weighted mean difference (with 95% confidence intervals). We will summarize time to event data (for example, length of stay in hospital, number of days on a ventilator) by the log hazards ratio (Parmar 1998) and we will calculate an overall log hazards ratio. If this is not possible due to insufficient reporting of individual cases in the RCTs, then we will treat survival data as continuous data.
We will calculate results using random effects models. We will quantify heterogeneity using the I‐squared statistic (Higgins 2002). The I‐squared statistic estimates the percent variability due to between study differences. If a sufficient number of trials are included in the study, we will assess possible sources of heterogeneity with either subgroup or sensitivity analyses or both. We will examine the following clinical subgroups: age (infants less than one year, children greater than or equal to one year); and surgical patients (purpose of admission to PICU is for postoperative care) versus medical patients (purpose of admission to PICU is for medical illness without surgical intervention prior to admission). The age subgroup is based on the fact that infants are at higher risk of catabolism and are generally fed more aggressively than older children. The subgroup based on surgical versus medical patients is based on inherent differences between these populations and the precedence in the literature to examine these populations separately (Heyland 1998; Heyland 2001; Marik 2001). If a study does not provide the data or results by age, or has a different age categorization, we will contact authors for additional data for the subgroups of interest. If data are unavailable from the authors and the report, the study will not be included in the subgroup analysis. We will conduct the following sensitivity analyses: methodological quality of included trials; intention‐to‐treat status; and funding source (medical or pharmaceutical companies versus other). We will also calculate fixed‐effect estimates as a sensitivity analysis. Subgroup and sensitivity analyses will be performed on the primary outcome.
We will test for asymmetry using the funnel plot visually and quantitatively (ie the rank correlation test (Begg 1994), the trim and fill method (Duval 2000), and weighted regression (Egger 1997)) depending on the number of trials included in the review. One source of asymmetry is publication bias.
