Newborn Mid-Upper Arm Circumference Identifies Low-Birth Weight and Vulnerable Infants: A Secondary Analysis

Background: Low birth weight (LBW) infants are at increased risk of morbidity and mortality. Identification of LBW may not occur in settings where access to reliable scales is limited. Mid-upper arm circumference (MUAC) may be an accessible, low-cost measure to identify LBW and vulnerable infants. Objectives: We explored the validity of newborn MUAC in identifying LBW and vulnerable newborns in rural Sierra Leone. Methods: This study was a secondary analysis of infant data from a randomized controlled clinical trial of supplementary food and anti-infective therapies compared with standard care for undernourished pregnant women. Data for singleton liveborn infants with birth measurement and 6-mo survival data were included in this analysis. The primary outcome was validity of MUAC in identifying low-birth weight (LBW) neonates. Secondary outcomes included validity of MUAC and head circumference (HC) in identifying weight-for-length z-score (WLZ) <-2, length-for-Age z-score (LAZ) <-2, neonatal mortality, and mortality within the first 6 mo of life. Results: The study population included 1167 infants, 229 (19.6%) with LBW. Birth MUAC (r = 0.817) and HC (r = 0.752) were highly correlated with birth weight. MUAC (AUC: 0.905; 95% CI: 0.884, 0.925) performed superiorly to HC (AUC: 0.88; 95% CI: 0.856, 0.904) in identifying LBW. The MUAC for identifying LBW was 9.6 cm (sensitivity: 0.86; specificity: 0.78). Neither MUAC nor HC reliably identified newborns with WLZ <-2 or LAZ <-2. MUAC ≤9.0 cm was the ideal cutoff for neonatal mortality (sensitivity: 53.3%; specificity: 89.7%; HR: 9.57; 95% CI: 1.86, 49.30). Birth anthropometrics did not reliably identify infants at risk of death in the first 6 mo of life. Conclusions: MUAC was used successfully to identify LBW infants and infants at risk of neonatal mortality in Sierra Leone. Further evidence is needed to support increased use of newborn MUAC measurement to identify LBW infants and infants at risk of neonatal mortality in community settings where scales are not available. Primary trial was registered at clinicaltrials.gov as NCT03079388. Lay Summary: Mid-upper arm circumference (MUAC) can be used to identify infants with low birth weight and infants at risk for neonatal mortality, with an MUAC ≤9.0 cm indicating the highest risk. This study was a secondary analysis of data from a prospective, randomized, controlled clinical effectiveness trial in which we compared the impact of a package of nutritional and anti-inflammatory interventions with the standard of care in undernourished pregnant women in Sierra Leone. Full details of the study design and interventions administered have been described previously (26, 27). The primary outcome for this analysis was validity of MUAC in identifying LBW neonates. Secondary outcomes included validity of MUAC and head circumference (HC) in identifying WLZ <−2, LAZ <−2, neonatal mortality, and mortality within the first 6 mo of life. A total of 1489 women were enrolled in the primary clinical trial, and pregnancy outcomes were obtained from 1417 women. Women were enrolled at an average of 24 wk gestation, as assessed by fundal height, and randomly assigned to the intervention package or standard of care. Standard of care included corn-soy flour and intermittent preventative treatment for malaria. The intervention included replacing the flour with ready-to-use supplementary food and added azithromycin as well as testing and treatment for vaginal dysbiosis. Clinic staff and participants notified the study coordinator at the time of delivery. A birth measurement team was dispatched to measure infants within 48 h of delivery. Birth measurements were taken as soon as was feasible, with a goal of within 72 h after birth. Infant weight, length, HC, MUAC, and morbidity were assessed at birth, 6 wk, 3 mo, and 6 mo of life. Nude weight was obtained using a Seca 334 infant digital scale accurate to 5 g. Recumbent length was measured in triplicate using a Seca 417 rigid height board; if the measurements differed by >2 mm, a fourth measurement was taken, and the 3 closest measurements were recorded and averaged. HC and MUAC of the left arm were obtained using a standard insertion tape accurate to the nearest 0.1 cm. Subsequently, if an infant was identified as deceased, the age at time of death was recorded. Pregnant women with undernutrition defined by an MUAC ≤23 cm and a fundal height <35 cm as a proxy for length of gestation were enrolled from 43 government antenatal clinics in Pujehun and the Western Rural Area Districts of Sierra Leone (26, 27). Exclusion criteria were known gestational diabetes, hypertension, or severe anemia. All singleton live births born to mothers enrolled in the described clinical trial with complete birth measurements and follow-up survival data to 6 mo of life were included in the current analysis. Data were collected directly on clinic cards and then double entered into a spreadsheet database (Microsoft Access) and cross-checked for discrepancies. All discrepancies were resolved by examination of the original data card. No imputations or estimations were performed for missing data. Data were then anonymized. Once the content of the database was determined, it was locked for analyses. Descriptive statistics were used to characterize the study population. Anthropometric parameters were converted to z-scores using the 2006 WHO growth standards (28). WLZ cannot be calculated for infants with lengths <45 cm; therefore, incalculable WLZ was evaluated as an additional risk category. All available values were included in the analysis. Pearson correlation coefficients and linear regression analyses were performed to evaluate the relation of birth weight to MUAC and HC. Nonparametric receiver operator characteristic (ROC) curves were used to calculate 95% CIs of AUCs, as this allowed for the assessment of the performance of MUAC and HC to identify LBW, wasted, stunted, and underweight infants over a range of possible values (7). The Youden J statistic (J = sensitivity + specificity − 1), which assumes false positives and false negatives to be equally undesirable, was calculated to evaluate effectiveness and identify the optimal anthropometric cutoff (29). Time-to-event values for mortality stratified by the identified anthropometric cutoffs were analyzed using both the Kaplan–Meier method with log-rank test and Cox proportional hazards regression, with the latter adjusted for maternal intervention received in the primary clinical trial and infant sex. Statistical analysis was performed using IBM SPSS Statistics version 25 and GraphPad Prism version 8.3.0. This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving research study participants were approved by the Sierra Leone Ethics and Scientific Review Committee (SLESRC) and from the Human Research Protection Office at Washington University in St. Louis(ID# 201611119). Informed consent for participation in the primary clinical trial and secondary use of data was obtained for eligible and interested women by a signature or thumbprint if the participant was unable to write. Participants received nutritional supplementation for the duration of pregnancy and incentives for postpartum followup visits. Participants directly benefited from the clinical trial in receiving nutritional supplementation and quality antenatal care. Results of the primary clinical trial were disseminated to local communities. The primary clinical trial was registered at clinicaltrials.gov ({"type":"clinical-trial","attrs":{"text":"NCT03079388","term_id":"NCT03079388"}}NCT03079388).