C. De Sadeleer et al., A multicenter trial on interobserver reproducibility in reporting on Tc-99m-DMSA planar scintigraphy: A Belgian survey, J NUCL MED, 41(1), 2000, pp. 23-26
Citations number
15
Categorie Soggetti
Radiology ,Nuclear Medicine & Imaging","Medical Research Diagnosis & Treatment
Conflicting opinions have been expressed regarding reproducibility in Tc-99
m-dimercaptosuccinic acid (DMSA) planar renal image interpretation. The pur
pose of this investigation was to determine the level of interobserver vari
ability among a large group of Belgian nuclear medicine physicians who eval
uated a randomly selected series of DMSA planar scintigraphic examinations
performed on children and adults. Methods: All Belgian nuclear medicine cen
ters (n = 82) were invited to participate in a reproducibility study on Tc-
99m-DMSA scintigraphy. Tc-99m-DMSA scans obtained on 10 adults and 40 child
ren were randomly selected from the databases of 2 hospitals. Those partici
pating in this investigation (65 centers = 79%) received a series of comput
er disks containing 50 Tc-99m-DMSA studies. To avoid potential problems rel
ated to unfamiliar display, the disks were formatted to be interpretable us
ing the participants' own computer systems. Each participant was then free
to use his or her usual display (hard copies, contrast enhancement, color s
cale, gray scale, and so forth). For each kidney, the observers had to choo
se between the following answers: normal, abnormal, equivocal, and lack of
quality. Results: Forty-two responses were obtained from a wide variety of
institutions and from observers with different levels of experience in inte
rpreting Tc-99m-DMSA scintigraphy. Altogether, the following data were obta
ined: 60.8% normal, 25.2% abnormal, 7.0% equivocal, and 3.2% lack of qualit
y. The median percentage of agreement (overall reproducibility) for the 42
observers was 92%. When the results of all 42 observers were compared, the
median agreements on normality and abnormality were 93.5% and 90.5%, respec
tively. In a small number (n = 4) of kidneys, reproducibility was poor and
ranged from 51% to 70%. Except for 2 outliers, all observers had almost the
same level of performance. Conclusion: A large number of Belgian nuclear m
edicine physicians participated in evaluating a large randomly selected sam
ple of Tc-99m-DMSA studies, and excellent interobserver agreement was found
.