Reliability in ultrasound measurements of plantar aponeurosis thickness.

• Ultrasonography is a reliable technique for plantar aponeurosis thickness measurements in healthy adults. • Intra-rater and inter-rater agreement varying substantially depending on the number of images used. • The error seems to be related to the sonogram acquisition process. • To minimize error, average values of a minimum of two images per subject should be used. • The moderate agreement between raters ratifies the need of all measurements being made by the same rater or group of raters. Given the increasing research interest in ultrasound plantar aponeurosis (PA) thickness measurements, this study aimed to analyze the inter and intra-rater reliability of independent sonograms and to identify the error related to the image analysis procedure. Twenty-one healthy men participated in this study. Imaging of PA consisted of two independent sonograms per subject. Two raters (R1 and R2) evaluated each sonogram twice using standardized steps. Precision of the image analysis procedure was analyzed using the Bland and Altman plot and Intraclass Correlation Coefficient (ICC). Agreement estimates and ICC were used to assess absolute and relative inter and intra-rater reliability. Reliability of PA thickness was found to depend strongly on the number of images acquired per subject. Intra-rater agreement for single measurements were 0.696 (R1) and 0.495 (R2), whereas average measurements yielded values of 0.821 (R1) and 0.662 (R2), respectively. Precision within a sonogram varied from ICC values of 0.873 to 0.960 (intra-rater) and 0.670 to 0.822 (inter-rater). Most part of the error in PA thickness measurements seems to be related to the sonogram acquisition process and not to the visual inspection of the image. To minimize error, average values of a minimum of two images per subject should be used. The moderate agreement between raters found in this study ratifies the need of all measurements being made by the same rater or group of raters. If a single rater evaluates all subjects, performing multiple measurements over one image does not seem to affect ICC as much as acquiring multiple images. [ABSTRACT FROM AUTHOR]