MON-572 Assessing Error in the Referencing of Thyroid Nodule Size by Longest Dimension (LD) and by Volume (Vol)

Nodule size is commonly referenced by the volume (Vol) of the nodule or by its longest dimension (LD), determined ultrasonically. Vol directly reflects the nodular mass in that a doubling (100% increase) of the nodular mass, for example, manifests as a 100% Change in the nodular Vol but only about a 30% Change in the LD. Referring to nodule size by its LD, though, is probably more intuitive, easier to remember and more convenient since ultrasound reports and clinical parlance typically refer to nodule size by its LD, and perhaps other dimensions, but do not routinely provide a calculated Vol. Though extremely useful in the evaluation of thyroid nodules, the sizing of nodules ultrasonographically is subject to error at multiple steps, e.g., selecting and freezing the images at which the nodular dimensions are at their maximum, drawing the longest axis for each of a nodule’s 3 diameters [Anterior-Posterior, AP; Left-Right, LR; and Superior-Inferior, SI], and the placement of markers at the measuring points of the often vague nodular margins. To determine the magnitude and frequency of sizing errors that affect the calculation of nodular Vol (Vol = 4/3 x pi x AP/2 x LR/2 x SI/2) and LD, three pairs of observers each evaluated the ultrasonographic images of 34 nodules, determining the Vol and LD of each nodule. The “Change” in the Vol and LD for each nodule was determined by randomly regarding one member of the pair as having sized a nodule at a nominal Time 1, and the other member regarded as having sized a nodule at a nominal Time 2. The calculated “Change” in LD was defined as the LD at Time 2 minus the LD at Time 1, and likewise for “Change’ in Vol. Since the set of images evaluated by each observer was the same for each observer at each nominal Time point, there was no actual change in the LD or Vol of any nodule. Any observed change could only be attributable to observer variability in measurements. The %-Change in LD was calculated as 100 x [(LD at Time 2) minus (LD at Time 1)]/(LD at Time 1). The %-Change in Vol was similarly calculated. The greatest observed %-Changes in LD were a 36% increase and a 43% decrease. The greatest observed %-Changes in Vol were a 105% increase and an 81% decrease. The top 5%, 10% and 20% of calculated rank-ordered LD increases were, respectively, errors of 26% or more, 22% or more and 10% or more. The top 5%, 10% and 20% of calculated rank-ordered Vol increases were, respectively, errors of 64% or more, 50% or more and 26% or more. The larger %-errors for Vol changes compared to LD changes expectedly follow from the need to measure (and accumulate error from) 3 dimensions, not 1, to measure Vol. This study demonstrates that it is important to be aware that the %-error reported for changes in nodule LD or Vol may be considerable and not uncommon even for nodules that have not actually changed.