Objectives: To report on the ultrasound quality assurance program for the National Institute of Child Health and Human Development Fetal Growth Studies and describe both its advantages and generalizability.
Methods: After training on an ultrasound system and software, research sonographers were expected to capture blank (unmeasured) images in triplicate for crown-rump length, biparietal diameter, head circumference, abdominal circumference, and femur length. A primary expert sonographer was designated and validated. A 5% sample (n = 740 of 14,785 scans) was randomly selected in 3 distinct rounds from within strata of maternal body mass index (round 1 only), gestational age, and research site. Unmeasured images were extracted from selected scans and measured with the ultrasound software by an expert sonographer. Correlations and coefficients of variation (CVs) were calculated, and the within-measurement standard deviation (ie, technical error of the measurement), was calculated.
Results: The reliability between the site sonographers and the expert was high, with correlations exceeding 0.99 for all dimensions in all rounds. The CV % values showed low variability, with the percentage differences being less than 2%, except for abdominal circumference in rounds 2 and 3, in which it averaged about 3%. Correlations remained high (>0.90) with increasing fetal size; there was a monotonic increase in technical errors of the measurement but without a corresponding increase in the CV %.
Conclusions: Using rigorous procedures for training sonographers, coupled with quality assurance oversight, we determined that the measurements acquired longitudinally for singletons are both accurate and reliable for establishment of an ultrasound standard for fetal growth.
Keywords: obstetric ultrasound; obstetrics; quality assurance; quality control; reliability; ultrasound.