Clinical Nutrition 33 (2014) 191e197
Contents lists available at ScienceDirect
Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu
Review
Nutrition biomarkers and clinical outcomes in critically ill children: A critical appraisal of the literature Chengsi Ong a, *, Wee Meng Han a, Judith Ju-Ming Wong b, Jan Hau Lee c, d a
Department of Nutrition and Dietetics, KK Women’s and Children’s Hospital, Singapore Department of Pediatric Medicine, KK Women’s and Children’s Hospital, Singapore c Children’s Intensive Care Unit, Department of Pediatric Subspecialties, KK Women’s and Children’s Hospital, Singapore d Office of Clinical Sciences, Duke-NUS Graduate School of Medicine, Singapore b
a r t i c l e i n f o
s u m m a r y
Article history: Received 29 October 2013 Accepted 23 December 2013
Background & aims: Malnutrition can significantly affect clinical outcomes in critically ill children. In view of the limitations of anthropometry, nutrition-related serum biomarkers have been used to assess the degree of malnutrition in the pediatric intensive care unit. The aim of this review is to critically appraise the use of nutrition-related serum biomarkers in predicting clinical outcomes in critically ill children. Methods: We searched major databases (MEDLINE, EMBASE, CINAHL, Cochrane Library) using MeSH terms and key words related to “biomarkers”, “nutrition” and “critically ill children”. All studies that explored the relationship between any nutrition-related serum biomarker and clinical outcomes in critically ill children (1 daye18 years) were included. The clinical outcomes of interest were duration of intensive care unit or hospital stay, duration of mechanical ventilation and mortality. Results: We found one randomized controlled trial and 15 observational studies involving 2068 children. In these 16 studies, 16 different nutritional biomarkers and two nutrition indices were examined. Albumin (n ¼ 7), magnesium (n ¼ 4), transferrin, prealbumin and calcium (n ¼ 3 respectively) were the most commonly studied biomarkers. Seven biomarkers (25-hydroxyvitamin D, albumin, calcium, magnesium, total protein, transferrin, triglycerides) and two indices (modified nutritional index and Onodera’s prognostic nutritional index) had positive associations with clinical outcomes. However, no biomarkers or nutrition indices consistently predicted clinical outcomes. Conclusions: Current medical literature does not provide convincing data to demonstrate any association between nutrition-related serum biomarkers and clinical outcomes in critically ill children. Further research is required to identify novel and clinically robust nutrition-related biomarkers. Ó 2014 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Keywords: Nutrition biomarkers Albumin Nutrition indices Critical care Children Pediatric intensive care unit
1. Introduction Malnutrition is associated with poor clinical outcomes in critically ill children. In the pediatric intensive care unit (PICU), children with malnutrition have been demonstrated to have increased duration of mechanical ventilation, length of stay (LOS) and mortality.1,2 Furthermore, malnutrition is common among critically ill
Abbreviations: CRRT, continuous renal replacement therapy; ICU, intensive care unit; PICU, pediatric intensive care unit; LOS, length of stay; OPNI, Onodera’s prognostic nutritional index; MNI, modified nutritional index; RSV, respiratory syncytial virus. * Corresponding author. Department of Nutrition and Dietetics, KK Women’s and Children’s Hospital, 100 Bukit Timah Road, Singapore 229899, Singapore. Tel.: þ65 63941640; fax: þ65 63941642. E-mail address: [email protected] (C. Ong).
children with rates reported to be as high as 53%.3e5 Critically ill children often have lower nutrition stores compared to critically ill adults, and are thus likely to be at higher risk of malnutrition. Timely and accurate nutrition assessment and prescription is often necessary to optimize clinical care in these children.6 To help determine malnutrition risk and nutrition adequacy, various modalities of nutritional assessments are currently being used. These modalities include measurements of weight and height, mid upper arm circumference and triceps skin fold thickness. These measurements provide estimates of body fat and muscle stores, and have been shown to correlate with clinical outcomes.2 Indeed, these measurements have been shown to have low inter and intra-observer variability.7 However, there are potential challenges in ensuring these measurements are performed in the critical illness setting. In some critically ill patients, specialized bed scales are required to measure body mass and they may be
0261-5614/$ e see front matter Ó 2014 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved. http://dx.doi.org/10.1016/j.clnu.2013.12.010
192
C. Ong et al. / Clinical Nutrition 33 (2014) 191e197
inaccurate due to dynamic fluid balance.8,9 Given these limitations, alternative measurements to assess nutritional status have been proposed. Serum biomarkers have been used to evaluate the adequacy of nutrient intake or nutritional status during hospital admissions.10 Serum biomarkers, measured as part of routine blood tests, are objective and relatively convenient to use. In critically ill adults, serum biomarkers such as albumin, pre-albumin, transferrin and retinol binding protein have been used to assess nutritional status.10,11 However, the concentrations of these biomarkers can potentially be altered in critical illness, leading to misinterpretation.12 In an effort to reduce the bias associated with measurements of these biomarkers, investigators have combined certain laboratory parameters to generate nutrition indices that more accurately reflect nutritional status whilst accounting for stress conditions in critical illness. The prognostic inflammatory and nutritional index and the nutritional risk index are two examples of such indices.13,14 Both have been proposed to correlate with mortality and hospital LOS in acutely ill adults. Despite the widespread use of these nutrition-related serum biomarkers in hospitalized patients, their prognostic value in relation to clinical outcomes remains unclear. There appears to be a limited prognostic value in the adult population, but this relationship has not been evaluated in critically ill children.10 The aim of this review is to identify and critically appraise the use of nutrition-related serum biomarkers or indices and their correlation with clinical outcomes in critically ill children. 2. Methods 2.1. Search strategy We searched the following electronic databases: MEDLINE, EMBASE, CINAHL and the Cochrane Library (including the Cochrane database of systematic reviews). All databases were searched from the earliest available date up until June 2013. A combination of MeSH terms and key words addressing “biomarkers”, “nutrition” and “critically ill children” were used. We further examined the reference lists of short-listed articles for other relevant studies. 2.2. Inclusion criteria All full reports of randomized controlled trials (RCT), prospective or retrospective observational studies and reviews involving children from birth to 18 years old, admitted to the PICU were included. We included studies that examined the association of any nutrition-related serum biomarker, either individually or grouped with any other parameters, with pertinent clinical outcomes (e.g., LOS, mortality and duration of mechanical ventilation). We then examined whether the primary aim of the studies was to explore the association between the biomarker(s) and clinical outcomes, and critically appraised them for strengths and limitations. We defined nutrition-related serum biomarkers as any biomarker measured in the blood that has been associated with general nutritional status (protein or caloric malnutrition) and specific nutritional deficiencies (e.g., vitamins and minerals). We did not restrict studies based on underlying disease or indications for admission to the PICU. Studies involving exclusively premature infants (gestational age < 36 weeks), urine biomarkers, and those that did not examine pertinent clinical outcomes were excluded. In addition, we decided to exclude serum glucose from our review as the effect of glucose homeostasis on clinical outcomes in critically ill children has been extensively studied and we refer readers to the available studies and excellent reviews on this topic.15e18
3. Results We found 762 potential studies which we screened for duplicity and relevance. 16 studies involving a total of 2068 children satisfied our inclusion criteria (Table 1). There was one randomized controlled trial (RCT) exploring the effect of early feeding using an immune-modulating formula versus a standard formula on nutrition biomarkers in 50 children on mechanical ventilation.19 The remaining 15 studies were prospective observational20e32 or retrospective in design.33,34 The clinical outcomes studied in all these studies include PICU or hospital LOS, duration of mechanical ventilation, post-operative infection rates and mortality. Four of the 16 studies involved a specific PICU population, while the rest involved multidisciplinary PICU patients. Two studies involved children with congenital heart disease after cardiac surgery.27,34 Castillo et al. conducted a prospective observational study on critically ill children receiving continuous renal replacement therapy (CRRT).24 Mezoff et al. examined nutritional biomarkers in infants with a primary diagnosis of respiratory syncytial virus (RSV) infection.30 Of the 16 studies, 11 were primarily designed to explore the association between biomarkers and clinical outcomes.22,23,25e32,34 The other five studies did not explore this association as their primary study aim.19e21,24,33 3.1. Types of biomarkers A total of 16 different biomarkers and two nutrition indices were identified. The most commonly studied biomarker was serum albumin, which was measured in eight studies,20e22,24e27,30 followed by magnesium, which was measured in four studies.22,26,32,33 Prealbumin (transthyretin),19e21 transferrin19e21 and calcium22,23,32 were measured in three studies, while 25-hydroxyvitamin D was measured in two studies.29,31 The remaining 11 biomarkers were explored in only one study each.19,22,26,28,30,32 Mezoff et al. studied the use of a nutritional screen to identify patients with RSV infection at risk for adverse clinical outcomes based on their nutritional status. This nutritional screen comprised of a scoring system which included comorbidities, clinical and dietary history, serum biomarkers (hemoglobin, hematocrit, total lymphocyte count, albumin) and anthropometric measurements.30 Three other studies, by combining nutritional and inflammatory biomarkers into mathematical equations, investigated the association of nutrition indices (e.g., Onodera’s prognostic nutritional index and the modified nutritional index), with clinical outcomes.20,21,34 In the following sections, we focused on the results of the most commonly studied biomarkers and biomarkers that showed a significant association with clinical outcomes (Table 2). 3.1.1. Albumin Eight studies examined the association between serum albumin and clinical outcomes.20e22,24e27,30 Five were conducted in a multidisciplinary PICU cohort,20e22,25,26 two were conducted in specific medical PICU cohorts,24,30 and one was conducted in children undergoing cardiac surgery.27 Of the five studies that were conducted in a multidisciplinary PICU cohort, only one study found a significant association between albumin and clinical outcomes.25 Durward et al. found that hypoalbuminemia on admission was related to a longer LOS (median 4.9 vs. 3.6 days, p ¼ 0.006). In patients with RSV infection, Mezoff et al.’s nutritional screen, conducted within 48 h of admission, found that hypoalbuminemia, in the presence of a significant comorbidity (e.g., gastrointestinal disorders, renal failure, diabetes), could identify patients with RSV infection at risk for prolonged duration of mechanical ventilation.30
Table 1 List of studies and clinical outcomes. Study
Study population and design
Outcomes
Main results
Comments
Cardenas- 145 children (mean Rivero age 5.34 0.47 y) Prospective observational 198923 study
Yes
Ionized calcium on admission
Mortality
Low ionized calcium (2.3 mg/dL) was associated with higher mortality (38 vs. 8%, p < 0.01) and longer PICU LOS (mean 11.9 vs. 4.53d, p 0.05). Non-survivors had significantly lower ionized calcium (4.2 vs. 4.9 mg/dL, p < 0.05), total protein (4.4 vs. 5.8 g/ dL, p < 0.05) and albumin (2.9 vs. 3.8 g/dL, p < 0.05). Low albumin (
Contents lists available at ScienceDirect
Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu
Review
Nutrition biomarkers and clinical outcomes in critically ill children: A critical appraisal of the literature Chengsi Ong a, *, Wee Meng Han a, Judith Ju-Ming Wong b, Jan Hau Lee c, d a
Department of Nutrition and Dietetics, KK Women’s and Children’s Hospital, Singapore Department of Pediatric Medicine, KK Women’s and Children’s Hospital, Singapore c Children’s Intensive Care Unit, Department of Pediatric Subspecialties, KK Women’s and Children’s Hospital, Singapore d Office of Clinical Sciences, Duke-NUS Graduate School of Medicine, Singapore b
a r t i c l e i n f o
s u m m a r y
Article history: Received 29 October 2013 Accepted 23 December 2013
Background & aims: Malnutrition can significantly affect clinical outcomes in critically ill children. In view of the limitations of anthropometry, nutrition-related serum biomarkers have been used to assess the degree of malnutrition in the pediatric intensive care unit. The aim of this review is to critically appraise the use of nutrition-related serum biomarkers in predicting clinical outcomes in critically ill children. Methods: We searched major databases (MEDLINE, EMBASE, CINAHL, Cochrane Library) using MeSH terms and key words related to “biomarkers”, “nutrition” and “critically ill children”. All studies that explored the relationship between any nutrition-related serum biomarker and clinical outcomes in critically ill children (1 daye18 years) were included. The clinical outcomes of interest were duration of intensive care unit or hospital stay, duration of mechanical ventilation and mortality. Results: We found one randomized controlled trial and 15 observational studies involving 2068 children. In these 16 studies, 16 different nutritional biomarkers and two nutrition indices were examined. Albumin (n ¼ 7), magnesium (n ¼ 4), transferrin, prealbumin and calcium (n ¼ 3 respectively) were the most commonly studied biomarkers. Seven biomarkers (25-hydroxyvitamin D, albumin, calcium, magnesium, total protein, transferrin, triglycerides) and two indices (modified nutritional index and Onodera’s prognostic nutritional index) had positive associations with clinical outcomes. However, no biomarkers or nutrition indices consistently predicted clinical outcomes. Conclusions: Current medical literature does not provide convincing data to demonstrate any association between nutrition-related serum biomarkers and clinical outcomes in critically ill children. Further research is required to identify novel and clinically robust nutrition-related biomarkers. Ó 2014 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Keywords: Nutrition biomarkers Albumin Nutrition indices Critical care Children Pediatric intensive care unit
1. Introduction Malnutrition is associated with poor clinical outcomes in critically ill children. In the pediatric intensive care unit (PICU), children with malnutrition have been demonstrated to have increased duration of mechanical ventilation, length of stay (LOS) and mortality.1,2 Furthermore, malnutrition is common among critically ill
Abbreviations: CRRT, continuous renal replacement therapy; ICU, intensive care unit; PICU, pediatric intensive care unit; LOS, length of stay; OPNI, Onodera’s prognostic nutritional index; MNI, modified nutritional index; RSV, respiratory syncytial virus. * Corresponding author. Department of Nutrition and Dietetics, KK Women’s and Children’s Hospital, 100 Bukit Timah Road, Singapore 229899, Singapore. Tel.: þ65 63941640; fax: þ65 63941642. E-mail address: [email protected] (C. Ong).
children with rates reported to be as high as 53%.3e5 Critically ill children often have lower nutrition stores compared to critically ill adults, and are thus likely to be at higher risk of malnutrition. Timely and accurate nutrition assessment and prescription is often necessary to optimize clinical care in these children.6 To help determine malnutrition risk and nutrition adequacy, various modalities of nutritional assessments are currently being used. These modalities include measurements of weight and height, mid upper arm circumference and triceps skin fold thickness. These measurements provide estimates of body fat and muscle stores, and have been shown to correlate with clinical outcomes.2 Indeed, these measurements have been shown to have low inter and intra-observer variability.7 However, there are potential challenges in ensuring these measurements are performed in the critical illness setting. In some critically ill patients, specialized bed scales are required to measure body mass and they may be
0261-5614/$ e see front matter Ó 2014 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved. http://dx.doi.org/10.1016/j.clnu.2013.12.010
192
C. Ong et al. / Clinical Nutrition 33 (2014) 191e197
inaccurate due to dynamic fluid balance.8,9 Given these limitations, alternative measurements to assess nutritional status have been proposed. Serum biomarkers have been used to evaluate the adequacy of nutrient intake or nutritional status during hospital admissions.10 Serum biomarkers, measured as part of routine blood tests, are objective and relatively convenient to use. In critically ill adults, serum biomarkers such as albumin, pre-albumin, transferrin and retinol binding protein have been used to assess nutritional status.10,11 However, the concentrations of these biomarkers can potentially be altered in critical illness, leading to misinterpretation.12 In an effort to reduce the bias associated with measurements of these biomarkers, investigators have combined certain laboratory parameters to generate nutrition indices that more accurately reflect nutritional status whilst accounting for stress conditions in critical illness. The prognostic inflammatory and nutritional index and the nutritional risk index are two examples of such indices.13,14 Both have been proposed to correlate with mortality and hospital LOS in acutely ill adults. Despite the widespread use of these nutrition-related serum biomarkers in hospitalized patients, their prognostic value in relation to clinical outcomes remains unclear. There appears to be a limited prognostic value in the adult population, but this relationship has not been evaluated in critically ill children.10 The aim of this review is to identify and critically appraise the use of nutrition-related serum biomarkers or indices and their correlation with clinical outcomes in critically ill children. 2. Methods 2.1. Search strategy We searched the following electronic databases: MEDLINE, EMBASE, CINAHL and the Cochrane Library (including the Cochrane database of systematic reviews). All databases were searched from the earliest available date up until June 2013. A combination of MeSH terms and key words addressing “biomarkers”, “nutrition” and “critically ill children” were used. We further examined the reference lists of short-listed articles for other relevant studies. 2.2. Inclusion criteria All full reports of randomized controlled trials (RCT), prospective or retrospective observational studies and reviews involving children from birth to 18 years old, admitted to the PICU were included. We included studies that examined the association of any nutrition-related serum biomarker, either individually or grouped with any other parameters, with pertinent clinical outcomes (e.g., LOS, mortality and duration of mechanical ventilation). We then examined whether the primary aim of the studies was to explore the association between the biomarker(s) and clinical outcomes, and critically appraised them for strengths and limitations. We defined nutrition-related serum biomarkers as any biomarker measured in the blood that has been associated with general nutritional status (protein or caloric malnutrition) and specific nutritional deficiencies (e.g., vitamins and minerals). We did not restrict studies based on underlying disease or indications for admission to the PICU. Studies involving exclusively premature infants (gestational age < 36 weeks), urine biomarkers, and those that did not examine pertinent clinical outcomes were excluded. In addition, we decided to exclude serum glucose from our review as the effect of glucose homeostasis on clinical outcomes in critically ill children has been extensively studied and we refer readers to the available studies and excellent reviews on this topic.15e18
3. Results We found 762 potential studies which we screened for duplicity and relevance. 16 studies involving a total of 2068 children satisfied our inclusion criteria (Table 1). There was one randomized controlled trial (RCT) exploring the effect of early feeding using an immune-modulating formula versus a standard formula on nutrition biomarkers in 50 children on mechanical ventilation.19 The remaining 15 studies were prospective observational20e32 or retrospective in design.33,34 The clinical outcomes studied in all these studies include PICU or hospital LOS, duration of mechanical ventilation, post-operative infection rates and mortality. Four of the 16 studies involved a specific PICU population, while the rest involved multidisciplinary PICU patients. Two studies involved children with congenital heart disease after cardiac surgery.27,34 Castillo et al. conducted a prospective observational study on critically ill children receiving continuous renal replacement therapy (CRRT).24 Mezoff et al. examined nutritional biomarkers in infants with a primary diagnosis of respiratory syncytial virus (RSV) infection.30 Of the 16 studies, 11 were primarily designed to explore the association between biomarkers and clinical outcomes.22,23,25e32,34 The other five studies did not explore this association as their primary study aim.19e21,24,33 3.1. Types of biomarkers A total of 16 different biomarkers and two nutrition indices were identified. The most commonly studied biomarker was serum albumin, which was measured in eight studies,20e22,24e27,30 followed by magnesium, which was measured in four studies.22,26,32,33 Prealbumin (transthyretin),19e21 transferrin19e21 and calcium22,23,32 were measured in three studies, while 25-hydroxyvitamin D was measured in two studies.29,31 The remaining 11 biomarkers were explored in only one study each.19,22,26,28,30,32 Mezoff et al. studied the use of a nutritional screen to identify patients with RSV infection at risk for adverse clinical outcomes based on their nutritional status. This nutritional screen comprised of a scoring system which included comorbidities, clinical and dietary history, serum biomarkers (hemoglobin, hematocrit, total lymphocyte count, albumin) and anthropometric measurements.30 Three other studies, by combining nutritional and inflammatory biomarkers into mathematical equations, investigated the association of nutrition indices (e.g., Onodera’s prognostic nutritional index and the modified nutritional index), with clinical outcomes.20,21,34 In the following sections, we focused on the results of the most commonly studied biomarkers and biomarkers that showed a significant association with clinical outcomes (Table 2). 3.1.1. Albumin Eight studies examined the association between serum albumin and clinical outcomes.20e22,24e27,30 Five were conducted in a multidisciplinary PICU cohort,20e22,25,26 two were conducted in specific medical PICU cohorts,24,30 and one was conducted in children undergoing cardiac surgery.27 Of the five studies that were conducted in a multidisciplinary PICU cohort, only one study found a significant association between albumin and clinical outcomes.25 Durward et al. found that hypoalbuminemia on admission was related to a longer LOS (median 4.9 vs. 3.6 days, p ¼ 0.006). In patients with RSV infection, Mezoff et al.’s nutritional screen, conducted within 48 h of admission, found that hypoalbuminemia, in the presence of a significant comorbidity (e.g., gastrointestinal disorders, renal failure, diabetes), could identify patients with RSV infection at risk for prolonged duration of mechanical ventilation.30
Table 1 List of studies and clinical outcomes. Study
Study population and design
Outcomes
Main results
Comments
Cardenas- 145 children (mean Rivero age 5.34 0.47 y) Prospective observational 198923 study
Yes
Ionized calcium on admission
Mortality
Low ionized calcium (2.3 mg/dL) was associated with higher mortality (38 vs. 8%, p < 0.01) and longer PICU LOS (mean 11.9 vs. 4.53d, p 0.05). Non-survivors had significantly lower ionized calcium (4.2 vs. 4.9 mg/dL, p < 0.05), total protein (4.4 vs. 5.8 g/ dL, p < 0.05) and albumin (2.9 vs. 3.8 g/dL, p < 0.05). Low albumin (
