Utility Of Mid-Upper Arm Circumference In Diagnosing Malnutrition In Children With Cystic Fibrosis

Abstract

Background:

Standardized diagnostic indicators for malnutrition using growth percentile z scores [weight-for-length (WFL) or Body mass index (BMI)] and mid-upper arm circumference (MUAC) z scores are being used in clinical practice, however their application to the pediatric cystic fibrosis (CF) population is not well described. In this study we aim to compare growth percentile z scores and MUAC z scores in diagnosing and classifying malnutrition in children with CF and assess the relationship between their degree of malnourishment with corresponding pulmonary function tests (PFT).

Methods:

In this retrospective observational outpatient study of 49 pediatric CF patients, data was collected on baseline characteristics, anthropometrics and PFTs over 12 months. Agreement in malnutrition diagnoses was quantified by Cohen’s Kappa statistics. Pearson test assessed the correlation between MUAC and BMI z scores as well as PFTs and anthropometrics. Serial anthropometrics and PFTs were obtained and compared for a subset of patients (n=28).

Results:

Growth percentile and MUAC z scores were positively correlated in diagnosing malnutrition (Pearson’s correlation r = 0.87) but MUAC z scores identified more patients as malnourished compared to growth percentile z scores (49% vs. 12%, Cohen’s kappa of 0.22 (95% CI: 0.04, 0.40). There was no significant relationship between anthropometrics and PFTs. MUAC z scores increased significantly over time, but BMI z scores did not show this trend.

Conclusions:

Our small-scale data suggests a promising role for MUAC z scores in classifying malnutrition and in measuring changes in nutrition status over time in pediatric CF.

Introduction

Cystic fibrosis (CF) is a genetic disease usually diagnosed in childhood that results in progressive decline in lung function, malnutrition and other complications. For children with CF, early optimization of nutrition is critical to long-term health, as undernutrition and poor growth status has been associated with poor pulmonary function and increased mortality^1,2. Ongoing assessment of growth status, and prevention and identification of malnutrition remain one of the important roles of the CF health care team in the management of these pediatric patients.

Assessing nutrition status in children with CF relies heavily on anthropometric measurements and growth percentiles, with emphasis on early detection of growth failure. Currently the CF Foundation Nutrition Guidelines recommend children age 0–2 years maintain weight-for-length (WFL) above the 50^th percentile, and those ages 2 to 18 years maintain body mass index (BMI) above the 50^th percentile using Centers for Disease Control (CDC) growth curves^3–5. Conversely, undernutrition has been defined as failure to meet the 50^th percentile for WFL or BMI-for-age⁵. Recognizing the limitations of BMI in differentiating lean body mass or fat free mass (FFM) and fat mass (FM), several studies have examined nutrition status with body composition measures such as handgrip strength⁶, dual-energy X-ray absorptiometry⁷ and mid-upper arm circumference^8,9.

Within the nutrition community there has been recent efforts to standardize the approach to diagnose malnutrition in children and adults. In 2014 the American Society of Parenteral and Enteral Nutrition (ASPEN) along with the Academy of Nutrition and Dietetics published standardized guidelines, which have been endorsed by American Academy of Pediatrics, for the diagnosis and classification of pediatric malnutrition, utilizing z scores for single data entry points of growth percentiles¹⁰. Then in 2017 z scores for MUAC measurements in children two months through 18.5 years were published¹¹, enabling diagnosis and classification of malnutrition possible using single datapoint of arm anthropometry as well.

For children with cystic fibrosis who are at increased risk for malnutrition, these additional tools such as MUAC z scores are a welcome advancement in clinical practice. This study compares standardized diagnostic indicators of malnutrition in children with CF and assesses the relationship between degree of malnourishment with corresponding lung function measures. It specifically investigates whether MUAC z score is a better predictor of malnutrition in the pediatric CF population than BMI-for-age/weight-for-length. Project objectives are:

1. To compare MUAC z scores to BMI-for-age/weight-for-length z scores in diagnosing and classifying malnutrition in the pediatric CF population during outpatient clinic visits.

2. To explore the relationship between malnutrition classification (as identified by MUAC z scores, and BMI-for-age/weight-for-length z scores), and pulmonary function status.

Subjects and Methods

Study design and study population

This observational retrospective study was conducted at the outpatient pediatric CF clinic at an accredited care center that serves approximately 70 pediatric patients with CF ranging from infants to 19 years old. Inclusion criteria were children with confirmed diagnosis of CF via sweat chloride or genetic testing or both, and those who had anthropometric measurements obtained during a clinic visit between the study timeframe of November 2017 to December 2018 (Figure 1). Age criteria was limited to ages 2 months through 18.5 years, as these are the ages for which MUAC z scores have been validated¹¹. Exclusion criteria included diagnosis of CFTR Related Metabolic Syndrome or inability to obtain anthropometric measurements during clinic visits. The study was approved by our Institutional Review Board (Project ID 1236644).

Figure 1:

Study design

Data Collection / Growth percentile and MUAC measurements

The data was collected on the age, gender, number of patients with the most common CF mutations (homozygous delta F508), pancreatic insufficiency and liver disease. Heights and weights were obtained for each patient as routine practice for every clinic visit by the CF clinic staff. Weight was measured using the anthropometric platform scale (Detecto Physician scale with height rod) and height (in centimeters) was measured using stadiometer attached to the scale. The height and weight were measured with minimal clothing, without shoes and their heads were positioned with the Frankfurt plane parallel to the floor. Body mass index (BMI) was calculated by dividing weight in kilograms by height in meters squared (kg/m²). The growth percentiles were plotted in the electronic medical record (EMR, Epic), utilizing World Health Organization (WHO) growth curve standards for children 0–2 years old¹², and CDC growth curve standards for children 2–20 years old¹³. Growth percentile z scores are automatically calculated within EMR.

During regularly scheduled nutrition visits as part of the quarterly multidisciplinary team visit, the registered dietitian (RP) obtained MUAC measurements in each patient, utilizing paper measuring tapes applied to the non-dominant arm. Arm circumference was measured at the midpoint between the scapula acromion and the ulna olecranon with the relaxed arm. Measurements were entered into an online pediatric calculator (www.peditools.org) to calculate MUAC z scores, which utilizes data from the CDC National Health and Nutrition Examination Survey (1999–2012)¹⁴.

The dietitian tracked growth percentile z scores, MUAC z scores and corresponding PFTs for each clinic visit in a database, de-identifying the data for analysis.

Malnutrition Classification and Diagnosis

Previously published malnutrition classifications using single data points of MUAC z scores and BMI-for-age/Weight-for-length z scores were utilized: −1 to −1.9 z score is equivalent to classification of mild malnutrition; −2 to −2.9 z score is a classification of moderate malnutrition; and −3 or lesser z score is classification of severe malnutrition^10,11.

Pulmonary Function Test measurements

During regularly scheduled outpatient visits, the respiratory therapist or clinic staff obtained pulmonary function test (PFT) measurements using the NDD Easy One Air PC spirometer (2.0.1.4). PFTs were obtained only for those patients old enough to perform the tests. Data collected included percent predicted values of forced vital capacity (FVC), forced expiratory volume in first second (FEV1), FEV1/FVC ratio and forced expiratory flow (FEF) 25–75, peak expiratory flow (PEF) and forced inspiratory vital capacity (FIVC), the parameters derived from forced expiratory maneuvers, and reliable indicators of CF related lung disease. Of these numbers, we have reported FEF25–75, an early indicator of CF related lung disease, FVC that quantifies lung volume loss and FEV1, the most reliable maker of CF related mortality. The technique used to perform the tests as well as the reference values and PFT interpretation were in accordance with the recommendations of the American Thoracic Society¹⁵.

Statistical Analysis

Quantitative variables were summarized as means ± standard deviation (SD) and qualitative variables as percentages. Distributions of quantitative variables were assessed using Q-Q plots and histograms. Agreement in malnutrition diagnoses (0 = no malnutrition, 1 = mild malnutrition, 2 = moderate malnutrition, 3 = severe malnutrition) based on MUAC and BMI z scores was quantified by Cohen’s Kappa statistic and interpreted according to Landis and Koch (1977)¹⁶. The correlation between MUAC and BMI z scores was calculated using Pearson’s correlation. To assess the relationships between malnutrition and pulmonary function, the correlations between pulmonary function metrics and MUAC and BMI z scores were calculated. In addition, Welch’s two-sample t-test was used to compare means of pulmonary function metrics between malnourished and adequately nourished subjects as determined by MUAC and BMI z scores. A total of 6 PFT values were evaluated in analyses of relationships with each malnutrition metric raising the possibility of committing a Type I error. Thus, for each analysis of PFT values versus each malnutrition metric, we also present p-values adjusted for multiple testing using the¹⁷ procedure to control the Type I error rate at 0.05.

Twenty-eight patients had multiple malnutrition and PFT scores reported at two to four time points from 5 to 50 weeks after baseline. As a secondary exploratory analysis, trends in these metrics over time were evaluated with a linear mixed effect model that related each metric to time reported in weeks with inclusion of a random intercept for each subject to account for correlation of observations within each subject. Because this was an exploratory analysis, no adjustment for multiple testing was conducted.

Statistical analyses were conducted using R Statistical Software Version 3.6.3. All statistical tests were two-sided and assessed relative to a significance level of 0.05.

Results

Patient demographics

The cohort consisted of 49 children, 31 males and 18 females, and 65% of patients were younger than 12 years of age (n=32). Majority of the patients were pancreatic insufficient (89%) and close to 50% were homozygous for delta F508, the most common CF genetic mutation. During the study period of 12 months, 49 children came to outpatient CF clinic and had weight, height and MUAC measurements taken at least once and 28 children had two to four anthropometric and PFT values measured over time (Figure 1).

Classification of Malnutrition

MUAC and BMI/WFL z scores were highly, significantly correlated (r = 0.82 [0.70, 0.89] p < 0.001, Figure 2). However, there was poor agreement in malnutrition categories based on the two scores with a Cohen’s kappa of 0.22 (95% CI: 0.04, 0.40).

Figure 2.

Correlation between MUAC and BMI/WFL Z-scores

MUAC z and BMI/WFL showing significant correlation (r = 0.82 [0.70, 0.89] p < 0.001)

Definition of abbreviations: MUAC: mid upper arm circumference, BMI/WFL: body mass index/weight for length

The number of subjects in each malnutrition category based on MUAC and BMI/WFL are shown in Table 2. MUAC z scores identified 49% of our patients as malnourished, however BMI/WFL z scores only identified 12% as malnourished. Additionally, there was disagreement as to degree of malnourished status. Both methods identified one patient as severely malnourished, however MUAC z scores identified a larger number of patients as moderately and mildly malnourished patients. In contrast, BMI z scores classified 87.8% of patients as adequately nourished and failed to pick up the mild or moderate malnourished classification.

Table 2:

Comparison of malnutrition classifications as identified by MUAC z scores and BMI/WFL z scores*

Malnutrition category	MUAC z score	BMI/WFL z score
No, N (%)	25 (51.0%)	43 (87.8%)
Mild, N (%)	16 (32.7%)	4 (8.2%)
Moderate, N (%)	7 (14.3%)	1 (2.0%)
Severe, N (%)	1 (2.0%)	1 (2.0%)

(*N=49, Cohen’s kappa of 0.22 (95% CI: 0.04, 0.40)

Definition of abbreviations: MUAC: mid upper arm circumference, BMI/WFL: body mass index/weight for length

Malnutrition and Pulmonary Function

Of the 49 subjects, 11 subjects were too young to perform PFTs. An additional four subjects were dropped from the PFT analyses due to poor effort, leaving 34 subjects for analysis of the relationship between malnutrition and PFT metrics (Figure 1).

The %predicted FEV1 and %predicted FEF 25–75 were positively related to MUAC z scores (r = 0.386 [0.055, 0.64], and r = 0.443 [0.123, 0.679], respectively) (Table 3). The % predicted FVC (r = 0.37 [0.043, 0.63]), %predicted FEV1(r = 0.37 [0.04, 0.63]), and %predicted FEF 25–75 (r = 0.34 [0.01, 0.61]) were positively correlated with BMI/WFL z scores. (Table 3). These correlations were statistically significant based on raw p-values but not after adjusting for multiple testing.

Table 3:

Comparison of lung function parameters and malnutrition metrics (N=34)

	MUAC z scores (N=34)			BMI/WFL z score (N=34)
PFT values (N=34)	Corr*(95% CL)	P value	Adjusted P value	Correlation (95% CI)	P value	Adjusted P value
%predFVC	0.312 (−0.03–.588)	0.073	0.292	0.372 (0.039–0.631)	0.03	0.18
%predFEV1	0.386 (0.055–0.64)	0.024	0.120	0.371 (0.038–0.63)	0.031	0.180
%predFEV1/FVC	0.188 (−0.161–0.495)	0.288	0.576	0.111 (−0.236–0.433)	0.533	1.000
%pred FEF25–75	0.443 (0.123–0.679)	0.009	0.054	0.342 (0.005–0.61)	0.048	0.192

^*Pearson’s correlation coefficients (Corr) and 95% confidence limits (LCL, UCL) for PFTs and MUAC z scores and WFL or BMI z scores. Raw p-values (P-value) and p-values adjusted for multiple testing (Adj. P) using the Holm’s procedure are presented. We report results for the 4 primary PFTs of interest but note that p-values were adjusted across 6 PFTs we evaluated.

Definition of abbreviations: PFT: pulmonary function testing, MUAC: mid upper arm circumference, BMI/WFL: body mass index/weight for length, FVC: forced vital capacity, FEV1: forced expiratory volume in one second, FEF 25–75: forced expiratory flow over the middle one half of the FVC

Pulmonary function was then compared between nourished and malnourished subjects as determined by MUAC and BMI/WFL z scores. Based on MUAC measurements, mean %predicted FVC, % predicted FEV1, and %predicted FEF 25–75 were 12 to 25 points higher for adequately nourished subjects versus malnourished subjects. These differences were statistically significant based on raw p-values but not after adjusting for multiple testing. For anthropometric measurements based on BMI, the number of subjects considered to be malnourished was very small (n = 4) such that power to detect differences by malnourishment status was low. The % predicted FVC was the only PFT metric to differ significantly (raw and adjusted p-value < 0.05) between malnourished and adequately nourished based on BMI z-scores

Twenty-eight subjects had two to four anthropometric measurements (both MUAC z scores and BMI/WFL growth curves) and PFT values measured over time. Analyzing these subjects showed that MUAC z scores increased significantly over time but BMI/WFL z scores did not (Table 4). For the PFT metrics, there was not enough evidence to identify a trend over time for any of the metrics (Table 4).

Table 4:

Changes in malnutrition metrics and lung function parameters over 12 months (N=28)

Metric	Slope	SE	P. Value
MUAC z score	0.015	0.005	0.004
BMI/WFL z score	0.003	0.003	0.306
%PredFVC	0.054	0.073	0.459
%PredFEV1	−0.048	0.078	0.533
%PredFEV1FVC	−0.119	0.062	0.057
%PredFEF25–75	−0.145	0.139	0.291

Results of linear mixed effect regression model evaluating change in each metric over time. Slope is an estimate of the change in the metric per week.

Definition of abbreviations: PFT: pulmonary function testing, MUAC: mid upper arm circumference, BMI/WFL: body mass index/weight for length, FVC: forced vital capacity, FEV1: forced expiratory volume in one second, FEF: forced expiratory flow over the middle one half of the FVC

Discussion

Our study explores the use of MUAC in diagnosing and classifying malnutrition in children with CF. In our cohort MUAC and BMI/WFL z scores were highly correlated in diagnosing malnutrition, however they had poor agreement in categorizing malnutrition. MUAC z scores identified more children in the undernourished category than BMI/WFL z scores, with sensitivity to the mild and moderate malnourished category. Additionally, when tracked over time, MUAC z scores changed significantly, though no change was observed in corresponding BMI/WFL z scores. There was a positive relationship between higher MUAC and higher BMI/WFL z scores with PFT metrics as well as some difference in PFT metrics between the nourished and malnourished children but not statistically significant once adjusted for multiple variables.

MUAC has been widely used across nations, especially in resource limited settings, as an early indicator of malnutrition¹⁸ and as an effective marker to predict mortality due to malnutrition in children and adolescents¹⁹. MUAC is easy to measure, requires simple inexpensive equipment and is less prone for error with less interindividual variability compared to other anthropometric measurements²⁰. In 2014, the Academy of Nutrition and Dietetics together with the ASPEN published standardized guidelines for identifying and categorizing malnutrition in children using single anthropometric data points including BMI, WFL and MUAC z scores¹⁰. In 2017, Abdel – Rahman et al published the age specific reference values for MUAC z scores in children two months through 18.5 years¹¹. Using these new reference values, our study is the first to assess malnutrition in a small cohort of children with CF.

A few prior studies have reported the utility of MUAC in early diagnosis of malnutrition in children with CF^8,21. A study by Stapleton et al showed that BMI percentile identified only 3/31 as malnourished in their cohort whereas z adipose or z muscle mass identified 15/31 subjects as malnourished⁸. Similarly, a study by Chaves et al comparing the BMI with MUAC percentiles in children 6–18yrs with cystic fibrosis has shown that MUAC identified a larger number of children with malnutrition compared to BMI (25 vs. 14 patients, out of 48 children)²¹. In our study 47% of children were characterized as mild or moderately malnourished based on their MUAC z scores, however BMI/WFL identified only 10% of children in that category, demonstrating the potential ability of MUAC z scores to serve as an earlier indicator for suboptimal nutrition than weight and height-based anthropometrics. If MUAC z scores are more sensitive in identifying the decline from nourished to mildly malnourished status, this knowledge is critical for the CF care team as well as the family. Early dietary intervention at this stage can more easily prevent the progression of malnutrition and the associated complications. Conversely, failure to identify a child’s deterioration in nutrition status leads to missed opportunities for timely dietary intervention.

Serial measurements of MUAC may pick up smaller improvements in lean body mass than BMI or weight change and may be a better method for assessing nutritional status overtime²². In our study we completed multiple MUAC measurements for 28 patients. Patients who were classified as malnourished per their MUAC z score or growth percentiles received aggressive dietary intervention, and some patients improved over time. Our limited data showed that MUAC z scores improved faster over time compared to corresponding BMI/WFL growth curves. In a study assessing the outcomes of 12-month nutrition interventions for children with CF ages 5–17 years, Groleau et al also found a statistically significant improvement in upper arm z scores, but no changes in BMI²³. There is value in tracking changes in both body composition and growth curves. Improvements in MUAC z scores rather than weight have been utilized in resource-limited countries to evaluate the effectiveness of nutrition interventions in acute malnutrition^24,25. Unlike weight, MUAC is not affected by fluid shifts or hydration status²⁶. This may be of utility for the CF liver disease population where the presence of ascites makes weight and BMI unreliable markers of nutrition status.

The relationship between malnutrition and pulmonary function impairment is bidirectional. The increased demand for breathing and recurrent pulmonary infections can lead to nutritional failure. On the other hand, poor nutrition due to nutrient loss from gastrointestinal malabsorption and poor oral intake can result in increased susceptibility to infections, poor respiratory muscle strength, and impaired pulmonary function^27,28. Chaves et al have also shown significant correlation between BMI and MUAC percentiles and lung function impairment, particularly lower MUAC percentiles showing a better association with moderate to severe lung function impairment. Our study also showed positive association between the malnutrition indicators and pulmonary function at baseline, with lower FEV1 and FEF25–75 in those patients with lower MUAC and BMI/WFL z scores. Also, based on MUAC measurements, FVC, FEV1, FEF 25–75 were 12 to 25 points higher for adequately nourished subjects versus malnourished subjects. However, these associations did not remain statistically significant once adjusted for multiple testing likely due to small sample size. It could also be related to overall better nourishment status and lung function in our cohort. Lastly, serial measurements did not show any significant changes in lung function metrics over time. Due to the small sample size and limited number of follow up points, only relatively large changes in lung function would have been detectable in this study.

As the nutrition community moves toward standardized malnutrition diagnosing in children, the CF care team is empowered with new tools, such as MUAC z scores, to assess this high-risk population. Our study shows that improvement in MUAC is more sensitive compared to BMI or weight changes with a nutritional intervention, which may help alleviate the stress of the family and reassure the clinical team. Additionally, children with ascites or significant stunting have posed challenges for nutrition assessment using BMI, however we have found serial MUAC measurements to be helpful in tracking nutrition status over time. Since this study, our clinic has aimed to standardize quarterly MUAC measurements for all CF patients to be completed by the dietitian with each quarterly nutrition assessment, offering patients yet another tool to track their progress. However, clinicians would benefit from evidence-based recommendations on frequency of MUAC measurements as well as expected improvement in body composition over time, as currently there are no standardized practices.

The study has several limitations. The sample size is small, the cohort is young (70% were less than age 12 years) making the PFT data even more limited. It will be interesting to compare the MUAC data with PFT data in a larger and more diverse sample. In addition, most children in the study are mildly malnourished and this could be the reason why significant correlation was not seen between malnutrition measurements and PFT data. Also, we have limited our anthropometrics to only MUAC and not measured triceps skinfold thickness or lower limb circumference which could provide more comprehensive assessment of body composition. Even though we suggest the utility of MUAC in children with liver failure or ascites, our study is limited by lack of data on body edema or ascites status at the time of MUAC measurement. Also, in our analysis of serial MUAC and PFT measurements, length of time between data points varied greatly, between six weeks and one year between follow ups. This needs to be standardized in future studies.

In conclusion, we found significant correlation between malnutrition metrics measured with MUAC z scores being slightly more sensitive is measuring mild and moderate malnutrition. MUAC z scores offer promising utility, both as a method to classify malnutrition and to measure changes in nutrition status over time. Our study did not show any significant relationship between lung function metrics and malnutrition classifications, likely secondary to small sample size.

Table 1:

General characteristics of our cohort (N=49)

Age, years, Mean ± SD	9.4 ± 5.2
Male gender, N (%)	31 (63%)
Homozygous delta F508 CF mutations, N (%)	24 (49.0%)
Pancreatic insufficient, N (%)	44 (89.8%)
History of liver disease, N (%)	6 (12.2%)
BMI z score, Mean ± SD	0.26 ± 1.3
MUAC z score, Mean ± SD	−0.86 ± 1.3
FVC, % predicted, Mean ± SD (N=34)	103.6 ± 17.1
FEV1, % predicted, Mean ± SD (N=34)	89.9 ± 17.6

Definition of abbreviations: MUAC: mid upper arm circumference, BMI/WFL: body mass index/weight for length, FVC: forced vital capacity, FEV1: forced expiratory volume in one second

Abbreviations

BMI

body mass index

CF

cystic fibrosis

MUAC

mid-upper arm circumference

PFT

pulmonary function test

WFL

weight-for-length

FFM

fat free mass

FM

fat mass

CDC

Centers for Disease Control

FVC

forced vital capacity

FEV1

forced expiratory volume in first second

FEF

forced expiratory flow FEF

PEF

peak expiratory flow

FIVC

forced inspiratory vital capacity