Reliability and interobserver variability of ultrasonographic measurement of gastric emptying rate

Abstract

Scintigraphy and real-time ultrasound are valid techniques to measure parameters of gastric emptying. However, scintigraphy involves exposure to ionizing radiation, while the precision of real-time ultrasound has not been previously evaluated. The objectives of the present study were to determine the inter observer and day-to-day variability of liquid gastric emptying rate measured by real-time ultrasonography in healthy volunteers and to compare the gastric emptying rate of males and females. Twenty healthy volunteers had ultrasonographic measurement of gastric emptying rate after ingestion of 300 ml beef broth. Nine subjects had a concurrent assessment by a second observer. Nine subjects had two studies performed on different days by the same sonographer. The T1/2 for 20 subjects was 24.77 +/- 6.84 min with no difference between the half-emptying time for males (25.89 +/- 6.99 min) and females (24.02 +/- 6.94 min). The Pearson and intraclass correlation coefficients for observations made by two observers were 0.83 and 0.625, with a difference due to observer of 2.37 min +/- 5.26 (NS). The test-retest reliability across successive days was 0.136, representing considerable day-to-day variability within subjects. The variability between subjects was also large, explaining up to 79% of the total variance. We conclude that ultrasound is a useful method to evaluate gastric emptying with good interobserver agreement. Due to substantial day-to-day variability, sample sizes larger than previously suggested are required to demonstrate clinically important changes in gastric emptying rate in clinical trials.