Interobserver variation for ultrasound determination of thyroid nodule volumes

Thyroid ultrasound is used in the routine clinical assessment and the follow-up of thyroid disorders. The follow- up of patients with thyroid nodules is mostly based on thyroid nodule volume determinations performed by different observers. However, for the judgment of treatment effects there is uncertainty about the interobserver variation of thyroid nodule volume measurements by ultrasound because there are no prospective blinded studies available comparing the interobserver variation in thyroid nodule volume measurement. The aim of our study was therefore to determine the variation of thyroid nodule volume determinations for different observers. We conducted a prospective blinded trial. Our study population consisted of 42 probands (8 men, 34 women) with an uniform distribution of thyroid nodule sizes (25 uninodular and 17 multinodular thyroid glands). We compared the results of 3 ultrasonographers with certified experience in thyroid ultrasound. The interobserver variation for the determination of thyroid nodule volume (n = 38) was 48.96% for the ellipsoid method and 48.64% for the planimetric method. The interobserver variation for determining thyroid volume (n = 40) was 23.69% for the ellipsoid method and 17.82% for the planimetric method. A regression analysis revealed that the probability for the identification of the same nodule in nodular thyroids by all sonographers is 90%, if the nodule is at least 15mm in greatest diameter. Future investigations should not describe changes in nodule volume less than 50% as therapy effects because only volume changes of at least 49% or more can be interpreted as nodule shrinkage or growth. Reporting of nodule volume modification 50% or more and lack of information for ultrasound procedures introduce a bias in studies evaluating the effects of nodule treatments. The clinical interpretation of a shrinking/growing thyroid nodule based on volume determinations by ultrasound is not well established because it is difficult to reproduce a two-dimensional image plane for follow-up studies.