RESUMEN
BACKGROUND: Weight-for-age is a commonly used indicator of the health of children and communities. We determined the accuracy of health volunteers' weight measurements in a nutrition project in Vietnam. OBJECTIVE: To report the accuracy of the volunteers' weight measurements and to assess the likely effect of any inaccuracies. METHODS: Save the Children /USA trained health volunteers to weigh children (6-36 months old) every other month from December 1999 to August 2000. Trained researchers randomly rechecked 257 weights (range, 24-114 per session). We computed nondirectional and directional differences between the weights measured by volunteers and those measured by researchers. RESULTS: The weights recorded by volunteers were lower than those recorded by researchers by an average of 30 g (p < .05). Almost all of the error occurred during the first weighing session, at which the average weight recorded by volunteers was 280 g below that recorded by researchers (p = .01). The error at subsequent weighings was minimal (< 20 g below reference at each session). CONCLUSIONS: One-time directional error suggests bias. Perhaps some communities (or families) influenced the volunteers to report weights lower than those actually observed to justify the programmatic food supplements or to give the impression at subsequent weighings that the level of malnutrition had been successfully reduced from that at the first session. Careful supervision of measurements of weight at baseline is essential.