Dear Colleagues
I would be grateful for any advice/references that may help with the task
described below:
I have 10 raters who assessed each 7 biopsys for several characteristics.
The raters also gave a VAS for the overall impression of abnormality. The
same set of raters assessed exactly the same biopsys in both 2005 and 2006
(1 year gap) so we have no learning effect to consider and no disease
progression. Although this is not the main purpose of the study, we would
like to quantify the agreement of the VAS between the 2 years.
My initial thought were to use the methods in section 5 of Bland & Altman
(1999), by considering the 10 judges as replicates on the same subject and
the years the “methods” being compared. However, this of course does not
take account of any within rater variability. I wondered if I should
modify section by adding a rater-year interaction and within rater term-
leaving me with no replicates.
Thank you
Susan Charman
|