Interobserver reproducibility in reporting on renal cortical scintigraphy in children: a large collaborative study

To evaluate the interobserver reproducibility in reporting on technetium-99m (99mTc) dimercaptosuccinic acid (DMSA) scan in children.

Sixty 99mTc-DMSA scans, issued from three centres, were distributed by e-mail to nuclear medicine physicians from the five continents interested in paediatric nuclear medicine. Observers had to choose, for each kidney, among four answers: normal, abnormal, equivocal or poor quality. An additional question was the location of the lesion if any: upper part, mid part, lower part. The responses had to be returned by e-mail.

Sixty-one observers, with an experience of approximately five or more 99mTc-DMSA/month, contributed to the study. Median agreement was 93%. The agreement was less than 80% in 29 kidneys (24%) but only in 13% (16 kidneys) was there disagreement between normality and abnormality, the remaining cases being related to 'equivocal' responses. Disagreement was mainly related to the following patterns: (i) normal variants: pear-shaped kidney, hypoactive poles contrasting with important parenchymal mass, triangular kidney, unusual shape of the columns of Bertin; (ii) congenital abnormalities: hydronephrosis, normal duplex kidney; (iii) small defects.

Interobserver reproducibility can be considered as good among a wide number of observers from the five continents. Disagreement among observers could be reduced by taking the normal variants into account.