This study was designed to assess intra- and interobserver variability
and diagnostic accuracy of nuclear medicine physicians in their evalu
ation of baseline and captopril renograms. Methods: The diagnostic per
formance of three experienced nuclear medicine physicians according to
their interpretation of baseline, captopril and paired renograms was
assessed. To this end, the readers evaluated the renograms of 28 hyper
tensive patients in whom a diagnosis of renovascular hypertension was
suspected on the basis of clinical clues. Ail patients also underwent
angiography. The readers were unaware of the angiographic diagnosis. R
esults: Thirteen of 28 patients proved to have renal artery stenosis (
8 unilateral, 5 bilateral) on renal angiography. The concordance in th
e renographic diagnoses between the three readers was reasonably good,
with an intraobserver agreement and kappa (observed agreement correct
ed for chance) ranging from 64% to 89% and from 0.52 to 0.75, respecti
vely, and an interobserver agreement and kappa ranging from 68% to 86%
and from 0.61 to 0.82. The sensitivity of their interpretation of pai
red baseline plus captopril renograms in relation to the angiographic
diagnosis is poor and below 50%, The post-test probability of RAS in c
ase of a negative renographic study was found to be rather similar to
the pre-test probability (prevalence) of 46%. Blinding readers to whic
h renogram was obtained after captopril imaging increased their accura
cy. Conclusion: The intra- and interobserver agreement between experie
nced nuclear medicine physicians who evaluate renograms was found to b
e reasonably good. Blinding readers as to which renogram is tile pre-
and post-captopril image seems to enhance their diagnostic accuracy in
instances of positive scans.