Validation of ultrasonography of the thyroid gland for epidemiological purposes.

Ultrasonography of the thyroid is often used in epidemiological surveys, thus thorough characterization of the interobserver variation of the different parameters obtained is important. Various methods have been used for measuring thyroid volume, and different formulas have been used for calculation of thyroid volume from the measured dimensions. In this article, two principles of thyroid volume measurement are described in detail: the wellknown method based on the three axes of each lobe and a new principle based on planimetry in two planes. The interobserver variation of the examination and the measuring procedure in itself were tested on 25 participants in a population study. A comparison of postmortem ultrasonography of the thyroid and results of an autopsy was performed. Good correlation and agreement between observers was found for thyroid volume (r = 0.98) and prevalence of thyroid nodules (kappa = 0.72), whereas echogenecity and echopattern showed little agreement. The correlation of thyroid volume by ultrasonography to autopsy results was satisfactory (r = 0.93), but the volume tended to be slightly underestimated even when using the formula pi/6(= 0.52)*length*width*depth. No major differences were found between the performance of the two principles of volume calculation. We conclude that when the measuring procedure is well defined, results of ultrasonography are comparable between observers for thyroid volume and prevalence of thyroid nodules, but not for echogenecity or echopattern. The formula of length*depth*width*pi/6 is suitable for thyroid volume measurement.