Precision of Measurements Performed by a Cadre of Anthropometrists Trained for a Large Household Nutrition Survey in Ethiopia

BACKGROUND: Well-trained anthropometrists are essential for the delivery of high-quality anthropometric data used to evaluate public health nutrition interventions. Scant data are currently available on the precision of data collected by large teams of anthropometrists employed for nutrition surveys in low-income country settings. OBJECTIVES: The purpose of this study was to assess the precision of child midupper arm circumference (MUAC) and length/height measurements taken by fieldworkers training for nutrition survey deployment. METHODS: Following 3 d of training, an anthropometry standardization exercise was conducted in small teams of trainees at 7 sites in the Amhara region of Ethiopia. In groups of 2–4, trainee anthropometrists (n = 79) each measured 16 children aged 6–47 mo (n = 336) twice for MUAC and length/height. Both intraobserver and interobserver precision were analyzed using technical error of measurement (TEM), relative TEM, coefficient of reliability (R), and repeatability metrics. Bland–Altman limits of agreement were calculated for intraobserver measurements. RESULTS: Intraobserver TEM was between 0.00 and 0.57 cm for MUAC (Bland–Altman 95% limits of agreement: −0.50 to 0.54 cm) and between 0.04 and 2.58 cm for length/height measurements (Bland–Altman 95% limits of agreement: –1.43 to 1.41 cm). Interobserver TEM was between 0.09 and 0.43 cm for MUAC and between 0.06 and 2.98 cm for length/height measurements. A high proportion of trainees achieved intraobserver R >0.95 (MUAC: 95%; length/height: 97%). Most teams also achieved interobserver R >0.95 (MUAC: 90%; length/height: 95%). CONCLUSIONS: Large numbers of anthropometrists (>75) in low-income settings can attain satisfactory precision in anthropometry following training and standardization. These protocols permit researchers to assess trainees, identify individuals who have not achieved the desired level of precision, and retrain or adjust roles prior to survey deployment.