Skip to main content
. Author manuscript; available in PMC: 2011 Dec 1.
Published in final edited form as: Arch Neurol. 2010 Aug 9;67(12):1485–1490. doi: 10.1001/archneurol.2010.189

Table 4.

GEE Models of Specific Cognitive Decline in Healthy Elders as a function of Aβ

Predictor Memory Language Visuospatial
Baseline Aβ42 Beta p Beta P Beta p
Aβ42 (Q2) −.03 .02 −.02 .07 −.02 .09
Aβ42 (Q3) −.03 .02 −.02 .14 −.02 .03
Aβ42 (Q4) −.01 .37 −.01 .14 −.02 .09
Aβ42 Change Beta p Beta p Beta p
Aβ42 (Dec/Stable) −.02 .06 −.01 .17 −.01 .18
Baseline Aβ40 Beta p Beta p Beta p
Aβ40(Q2) −.03 .02 < −.01 .62 −.01 .24
Aβ40(Q3) −.02 .09 < −.01 .91 −.01 .21
Aβ40(Q4) −.02 .10 −.02 .12 −.02 .08
Aβ40 Change Beta p Beta p Beta p
Aβ40(Dec/Stable) −.01 .26 .01 .33 −.01 .53

Note. Q = Quartile. Dec/Stable= Decreasing/Stable. The reference group for models examining baseline values is the lowest quartile. The reference group for models examining change as a predictor is increasing values. Beta values pertain to the interaction term in the GEE model (predictor * time) and reflect rate of cognitive decline in comparison to that in the reference group with negative betas indicating faster decline and positive betas indicating slower decline.