Accuracy and reliability of weight change determinations in a model infant weighing less than 750 grams

We designed a study to address the accuracy and reliability of the weighing process for tiny infants ( < 750 gm). A model was made with intravenous bags and stockinette and the following equipment was added: umbilical artery and venous catheters, endotracheal tube, pulse oximeter probe, electrodes, and orgastric tube. The models weight was changed in intervals of 2 to 43 gm and it was then weighed on two different types of scales. The measured weight on each scale was compared with the predicted weight (calculated as initial + added mass), and the mean absolute difference (MAD) was determined. All measurements were obtained by the same two nurses. For the bedside scale the MAD was 5 gm (range - 23 to + 10; n = 30). Similar results were obtained with a sling scale; the MAD was 3 gm (range - 13 to + 6 n = 30). The range of differences noted suggests that discrepancies between measurements may lead to a difference between weights that is in the range of 4% to 7% body weight. Interobserver variability was noted to affect the reliability of the weight determinations. When five experienced neonatal intensive care units nurses weighed the model the MAD was 9 gm (-33 to + 22; n = 15) and 6 gm (range - 11 to + 28; n = 15) for the bedside and sling scales, respectively. For the predicted weight, these discrepancies represent potential errors of 7.4% and 5.2%, respectively. These values are likely to underestimate the potential errors seen clinically inasmuch as we could not duplicate infant movement or anxiety of the person weighing the infant. These results should be considered when protocols are developed for weighing extremely small preterm infants.