Karen Stephens1, Meike Orlick2, Shannon Beattie2, Audrey Snell2, Kim Munsterman2, Leah Oladitan2, Susan Abdel-Rahman3,4,5. 1. Nutrition Services, Children's Mercy Hospital, Kansas City, Missouri, USA. 2. Children's Mercy Hospital, Kansas City, Missouri, USA. 3. UMKC School of Medicine, Kansas City, Missouri, USA. 4. Innovation in Health Care Delivery, Children's Research Institute, Kansas City, Missouri, USA. 5. Section of Therapeutic Innovation, Children's Mercy Hospital, Kansas City, Missouri, USA.
Abstract
BACKGROUND: Anthropometric z-scores used commonly for diagnosis and determining degree of malnutrition, specifically body mass index (BMIz), weight-for-length (WLz), and mid-upper arm circumference (MUACz), are not wholly concordant, yet the proposed thresholds for classification are identical. This study was designed to critically examine MUACz thresholds and their ability to correctly classify nutrition status. METHODS: This was a 2-year, prospective single-center study of children ≤18 years seen by registered dietitians within a large pediatric institution. The sensitivity, specificity, and predictive performance of the malnutrition classification thresholds were estimated against clinician-based classification. RESULTS: Sixty-one dietitians enrolled 10,401 patients with distributions of z-scores for weight (-0.5 ± 1.9), length (-0.8 ± 1.6), BMI or WL (-0.1 ± 1.8), and MUAC (-0.4 ± 1.5), suggesting participants were smaller and shorter than the reference U.S. POPULATION: Distributions of MUACz were broad and overlapped between nutrition classification groups, an observation that extended to BMIz and WLz as well. Consequently, existing thresholds do not accurately classify 100% of children. Misclassification rates increase, with increasing severity ranging from 8% in children with no malnutrition to 71% in children with severe malnutrition. Algorithm- and manually-based refinement of thresholds result in mixed improvements and can be explored by the reader with the associated supplement. CONCLUSION: The sensitivity of proposed MUACz thresholds systematically decreases with increasing severity of malnutrition and will require optimization if we aim to limit the number of children at risk of misclassification. Indicators for overnutrition remain to be addressed but are explored herein.
BACKGROUND: Anthropometric z-scores used commonly for diagnosis and determining degree of malnutrition, specifically body mass index (BMIz), weight-for-length (WLz), and mid-upper arm circumference (MUACz), are not wholly concordant, yet the proposed thresholds for classification are identical. This study was designed to critically examine MUACz thresholds and their ability to correctly classify nutrition status. METHODS: This was a 2-year, prospective single-center study of children ≤18 years seen by registered dietitians within a large pediatric institution. The sensitivity, specificity, and predictive performance of the malnutrition classification thresholds were estimated against clinician-based classification. RESULTS: Sixty-one dietitians enrolled 10,401 patients with distributions of z-scores for weight (-0.5 ± 1.9), length (-0.8 ± 1.6), BMI or WL (-0.1 ± 1.8), and MUAC (-0.4 ± 1.5), suggesting participants were smaller and shorter than the reference U.S. POPULATION: Distributions of MUACz were broad and overlapped between nutrition classification groups, an observation that extended to BMIz and WLz as well. Consequently, existing thresholds do not accurately classify 100% of children. Misclassification rates increase, with increasing severity ranging from 8% in children with no malnutrition to 71% in children with severe malnutrition. Algorithm- and manually-based refinement of thresholds result in mixed improvements and can be explored by the reader with the associated supplement. CONCLUSION: The sensitivity of proposed MUACz thresholds systematically decreases with increasing severity of malnutrition and will require optimization if we aim to limit the number of children at risk of misclassification. Indicators for overnutrition remain to be addressed but are explored herein.
Keywords:
body mass index; children; malnutrition; mid-upper arm circumference; overnutrition; pediatrics; sensitivity; undernutrition; weight-for-length