Improving accuracy in predicting outcome in the idiopathic glomerulopa
thies requires standard clinical and laboratory data observed over tim
e and the application of certain basic biostatistical tests. In our in
itial model we examined categories of glomerular disease that progress
i.e. membranous, membranoproliferative, diffuse proliferative, focal
sclerosing and IgA nephropathy, We determined that a combination of se
verity and persistence of proteinuria above certain levels over fixed
time frames plus the knowledge of any change in glomerular filtration
rate during these periods resulted in slopes of creatinine clearances
that were the same across this histologic spectrum. This modeling of d
isease was then applied in a more rigorous fashion to our patients wit
h idiopathic membranous nephropathy (IMGN), Multivariate logistic regr
ession analyses was used to test all other potential clinical and labo
ratory predictors of chronic renal failure. These additional factors d
id not improve our ability to predict outcome. This algorithm was subs
equently validated on two independent IMGN data bases. Overall accurac
y of prediction estimated within 6 months of presentation was maintain
ed at >85%. The role of this type of evaluation in both clinical pract
ice and research is discussed.