The present study aimed to assess the interobserver variation in histo
pathological grading of 88 slides by four experienced pathologists who
, before the start of the study, reached consensus about the method of
grading through a joint session behind a discussion microscope. The r
esults were compared with two previous studies, the first assessing th
e observer variation under normal conditions, the second doing the sam
e after theoretical agreement about which morphological characteristic
s were relevant for grading. In the present study the pathologists fir
st made a provisional diagnosis on low power view, followed by a final
diagnosis on high power view, scoring various morphological character
istics as in the second study. The unweighted and weighted group kappa
values for the final diagnosis were 0.41 and 0.71 respectively in the
present study (after the consensus meeting), compared to 0.33 and 0.7
0 respectively in the second study (after only theoretical consensus),
and 0.28 and 0.55 in the first study (without consensus). We conclude
that a consensus meeting appears to be useful to improve the interobs
erver agreement.