Assessing effectiveness of a novel mid-upper arm circumference Z-score tape in a community setting in Guatemala

Accurate and timely identification of children suffering from, or at risk of, severe acute malnutrition (SAM) is essential to direct them to appropriate care. The deleterious and potentially irreversible effects of undernutrition on growth and cognitive development have been well documented in children under the age of five, and an estimated 45% of mortality in this age range is attributable to undernutrition [1‐4]. Although most intervention work focuses on the critical window from conception to 24 months, an expanding body of evidence suggests that growth catch-up can occur during middle and late childhood and adolescence [5]. This period of life is marked by significant physical, neurological, and social development. Executive neurological function and frontal lobe development is thought to occur in late childhood and adolescence, and adequate nutrient intake is highly associated with cognitive performance and maintenance, as well as educational achievement [6‐8]. Physiologically, underweight school-aged children are more likely to suffer from delayed onset of puberty, reduced bone density, deficient muscular development, and poorer overall health which may impact their ability to work later in life [9]. The effects of malnutrition persisting into adolescence and the reproductive years can extend to subsequent generations, adversely affecting fetal development and the epigenome of their children [10]. Thus, undernutrition in school-aged children and adolescents exerts considerable strain on the health, education, and economic systems in low and middle income countries [11].

Undernutrition is most prevalent in low-resource settings where unskilled staff, inadequate tools, and insufficient time hinder nutritional screening program success. Detection methods should, therefore, be simple for delivery at the community level [12]. Various anthropometric measurements have been utilized to identify high risk cases, of which Mid-Upper Arm Circumference (MUAC), Weight-for-Height (WH) Z-scores in children under 2, and Body Mass Index (BMI) Z-score for ages 2–19 are the most widely used and debated [13, 14]. MUAC, as measured in mm, is an easier method to deploy in a community setting. WH/BMI Z-scores are utilized more often in health care settings where the tools required for obtaining weight and height are accessible. The cutoffs using WH/BMI Z-scores are well-established: SAM is identified by a Z-score < − 3, and moderate acute malnutrition (MAM) is identified as a Z-score between − 2 and − 3 [15]. Cutoffs for SAM and MAM using traditional MUAC tapes are much less consistent. A systematic review in 2013 concluded that MUAC could be adequately used as a stand-alone criterion for hospital admission and discharge due to SAM, but the MUAC cutoffs ranged from 110 to 130 mm for children aged 6 to 59 months [16]. Furthermore, independent studies offer contradictory evidence as to which measurement is more predictive of mortality due to malnutrition. Chiabi et al. concluded that MUAC was more predictive of mortality in children with SAM than WH Z-score [17], whereas Grellety and Golden recently determined that current MUAC methods based on mm cutoffs for children under the age of 5 inadequately identified individuals with the highest risk of mortality based on WH [18]. Briend et al. found that MUAC was preferable to WH but did not add predictive value [19]. The disparities may be attributable in part to the various cutoffs used by each group. Other studies have attempted to determine more appropriate cutoffs in various populations using an endpoint other than mortality, but these were based solely on the MUAC expressed in mm, and/or made for use in children ages 0–5 [20‐29]. Accordingly, trade-off between ease of implementation in resource-limited settings, unacceptable rates of detrimental false negatives, and the burden of false positives needs to be addressed so that better protocol can be established and scaled.

To address limitations of traditional MUAC, several groups independently developed reference growth curves for age-specific MUAC Z-scores using the Health Examination Survey (HES) and the National Health and Nutrition Examination Survey (NHANES) [30‐32]. Application to an East African population offered evidence that MUAC Z-score growth curves were at least as effective as BMI Z-score in 5–19 year-olds with or without comorbid HIV infection. Data from U.S. cohorts extended the MUAC Z-score to include children aged 2 months to 20 years, and provided the first indication that MUAC Z-score and BMI Z-score thresholds may require refinement [30, 33]. To facilitate screening, the Kansas City group also constructed MUAC Z-score tapes that use easily interpretable colors to indicate nutritional risk status. The MUAC Z-score tapes have been evaluated in practice by registered dietitians on a group of over 10,000 patients seen at their institution between October 2015 and October 2017 [33].

The MUAC Z-score tape has yet to be evaluated in the hands of health volunteers in a resource-limited community setting. Increasing accessibility of the MUAC tapes for community health workers and volunteers is an important step to increase program coverage, both in detecting SAM and treating uncomplicated SAM in a community setting [34]. Notably, there is some evidence to suggest that mothers are able to more accurately take and read traditional MUAC for their children than community health workers, which indicates household community monitoring may be as or more effective than community health worker screening [35]. The present study examined the effectiveness of the MUAC Z-score tape described above [32, 33] in the hands of non-medical volunteers to identify children over the age of 5 who would benefit from nutritional rehabilitation (Z–score < –2).

The quantitative and qualitative findings suggest that the paper-based MUAC Z-score tape is a viable, low-cost alternative for assessing nutritional status among the community served by Children International in Guatemala. The present method of identifying cases of moderate and severe acute malnutrition (BMI Z-score) places the onus on time-constrained medical personnel and other field staff with little or no medical experience. It also requires special equipment that must either be taken into rural communities or kept at a community center that is often many miles from the homes of beneficiaries. The single-step MUAC Z-score tape is much faster and easier to understand than collecting height and weight, calculating BMI, converting BMI to an age- and gender-relevant Z-score, and assigning nutritional risk based on that value.

These results further suggest that MUAC Z-score measurements completed by a corps of health volunteers is an effective way to screen for malnutrition. Volunteers with little to no medical experience were able to achieve 91.00% agreement with field medical staff, 88.26% agreement with BMI Z-scores, and 87.10% agreement in identifying underweight candidates for the nutrition rehabilitation program according to BMI Z-scores. However, only about half of the individuals with SAM were identified by both methods (Fig. 3, top left corner). While there are few observations in the extreme nutritional risk categories on either end, the model results provide evidence that probability of disagreement is more likely to occur the more extreme the measurement. This finding is consistent with a previous observation in U.S. children which concluded that Z-score thresholds may require refinement to improve concordance [33].

Importantly, this prior investigation also reported rates of concordance between both BMI Z-score and MUAC Z-score with clinical assessments of malnutrition raising questions of which measure is more sensitive for malnutrition [33]. There was no significant association between gender, age, or stunting (height-for-age Z-score < -2) with probability of disagreement. Only BMI Z-score (quadratic term) was found to be associated with probability of disagreement.

Our findings contribute to the current body of knowledge concerning the effectiveness of MUAC Z-score tape in identifying malnutrition in school-aged children and adolescents in the following ways. Firstly, this study represents one of the first instances of external validation of age-specific MUAC Z-scores using the novel tape in children aged 5 to 19 years in a community setting. Second, these results suggest that measurements completed by volunteers with little to no medical experience are in high agreement with those completed by the field medical staff and with the nutritional risk status as predicted BMI Z-scores. Third, a qualitative survey showed that the MUAC Z-score tape is easy to use and increases understanding of nutritional risk, although there is room for improvement in training and in the characteristics of the tape itself.

Field staff in Guatemala noted that the MUAC Z-score measurements which were discordant with BMI Z-scores occurred at the borders of two colors. This misclassification of some of the MAM and SAM cases indicates that a protocol to classify borderline cases may be required to enhance concordance between the measures. Other possible solutions include repeating measurements, or measuring the opposite arm as well and recording the more extreme of the two. Also noted, the nature of discordance did depend on nutritional status to some extent. MUAC Z-score was more likely to categorize an undernourished child into a more severe classification and an overnourished child into a less severe classification, a finding that coincides with previous studies [33]. Recommendations from the Guatemala field staff to improve the training and ultimately the effectiveness of the MUAC Z-score tape in a community setting include offering a reference pamphlet to the volunteers, which shows the exact version of the MUAC Z-score tape are using, with both sides and corresponding colors. They also noted that the practice sessions were very helpful and one of the most important aspects of the training. Having a doctor, nutritionist, or someone with medical experience leading the process was important during the training. Another benefit of the trainings noted by the field staff was that empowering volunteers and community leaders who have no previous formal medical training foments trust that may improve the quality and efficiency of community projects.

Of note in this study was the high prevalence of “at risk” of overweight and overweight. In contrast to other countries in Latin America where overweight and obesity are rapidly increasing, Guatemala has one of the lowest prevalences of overweight, and one of the lowest prevalences of inactive people over the age of 15 [39]. However, other studies have pointed to the nutrition transition that is occurring in Guatemala toward processed (e.g., canned foods, cheeses, refined sugar) and highly processed foods (e.g., chips, soft drinks, sweetened breakfast cereals) is associated with increases in BMI [40, 41]. Because Guatemala has such a high rate of childhood stunting, the risk of the double burden of malnutrition (overweight or obesity coexisting with undernutrition) is alarming [42]. The MUAC Z-score tape used in this study not only identifies underweight, but the entire spectrum of nutritional status. The use of this specific tape has indicated that there may be a growing concern of overweight and obesity in the population of children and youth served in Guatemala, which may have implications for the field medical staff and nutrition program.

There are several limitations of the current study. One limitation is that the community volunteers were asked to measure children and youth in their communities, and were not randomly assigned participants. This may have introduced selection bias, especially if children were related or lived in close proximity, or if those children or youth who were not included were fundamentally different in ways that were related to the outcome of agreement. Additionally, there were 71 children or youth who did not complete the follow-up measurements, which could have skewed the results. Another limitation is the window of time between MUAC Z-score tape measurement and height and weight collection. Weight is affected by diet, activity level, and illness. The extent to which these environmental factors may have changed a child’s weight and thus BMI over the course of 15 days were not recorded in this study; however, their impact is likely low in this narrow timeframe. When considering the limitations of the generalized linear mixed model, we had far fewer observations in the extreme tails of the distribution than near the center, making those estimates less certain. Interpretation near the ends of or beyond the observed range of BMI Z-scores should be done with caution. We also did not have a determination of nutritional status as determined by a medical professional; we compared only anthropometric measurements. Therefore, knowing the accuracy or performance of each method is not known, but as screening methods they yield similar results. Lastly, the MUAC Z-score tape collates MUAC Z-score data delineated by age in months into ½ year or full year increments. Practical implications limit the real estate available on the MUAC Z-score tape, thus the age categories are not as granular; however, the tapes can be customized to expand or contract age ranges as was done for Children International.

To the best of our knowledge, this is the first study to assess the effectiveness of the MUAC Z-score tape in a community setting using non-medical volunteers. Volunteer-implemented nutritional risk screening with the MUAC Z-score tape was effective at detecting SAM and MAM in this study.

Future directions for this work include investigating scalability of the MUAC Z-score tape to detect malnutrition in locations such as Guatemala that require cost-effective strategies to address the heavy burden. Use of MUAC Z-score tape to monitor rehabilitation progress and determine success also remains to be investigated.