Assessment of the interobserver and the intraobserver reproducibility for the detection of renal cortical defects in adults and children using [99mTc]Tc-MAG3.

Abstract

Background: One can assess cortical defects on the early images of [99mTc]Tc-MAG3 renography. We aimed to assess interobserver and intraobserver reproducibility for detecting renal cortical defects using [99mTc]Tc-MAG3 for adults and children; identify causes for poor inter- and intraobserver reproducibility and to assess the effect of the kidney to background ratio (KTBR) on reproducibility.

Methods: One hundred adult and 200 pediatric renograms were included. The observers reviewed the summed 1-minute posterior images for the first four minutes to detect cortical defects. Interobserver reproducibility between three observers and intra-observer reproducibility for two observers were determined. Agreement was tested using percentage agreement, Krippendorff's reliability coefficient alpha and Cohen's kappa statistic. The association between KTBR and agreement was evaluated.

Results: Interobserver agreement on the 1-2 minutes images was 78 (95% CI: 74.8-82.7%) and 79.7 (95% CI: 75.9-83.5%) for left and right kidneys respectively. Intraobserver percentage was 89.7% (95% CI: 86.2-93.1%) for the senior and 80.7% (95% CI: 76.2-85.2%) for the junior observer. In 13.5% (27) of the adult and 4.5% (19) of the pediatric kidneys the difference in image interpretation between the observers would have had a clinical impact. If the KTBR is ≤2, the percentage agreement was between 61.5% and 64.8%. In cases with a KTBR >2, the percentage agreement was between 83.6% and 87.1%.

Conclusions: The percentage interobserver agreement was moderate. Disagreement between normal and abnormal cases were infrequent. The interobserver reproducibility was decreased when the KTBR was ≤2.