Table 3.

Slope and Breakpoint Parameters for Linear and Segmented Linear Regression Models Through Pattern Deviation Data as a Function of GCIPL Thicknesses, Extracted Over Locations With Drasdo Displacement Applied and Without Displacement

	Overall	Cluster 1	Cluster 2	Cluster 3
Linear regression
Slope
Drasdo displacement	0.048	0.047	0.035	0.025
No displacement	0.045	0.034	0.039	0.025
P value	0.21	0.04*	0.36	0.90
R2
Drasdo displacement	0.051	0.051	0.047	0.065
No displacement	0.040	0.010	0.046	0.069
Segmented linear regression
Slope 1
Drasdo displacement	0.00007	0.012	0.0038	0.0036
No displacement	−0.00056	−0.11	−0.0020	0.0068
Breakpoint
Drasdo displacement	−9.83	−12.00	−10.00	−11.51
P value†	<0.0001*	<0.0001*	<0.0001*	0.06
No displacement	−10.76	−29.00	−10.93	−8.00
P value†	<0.0001*	0.31	<0.0001*	<0.0001*
Slope 2
Drasdo displacement	0.122	0.087	−0.086	0.051
No displacement	0.107	0.045	−0.096	0.067
P value	0.07	N/A	0.16	N/A
R2
Drasdo displacement	0.069	0.057	0.060	0.079
No displacement	0.053	0.012	0.061	0.089

R², coefficient of determination.

Clusters refer to VF clusters per Choi et al.,²² with cluster 1 being the most central and cluster 3 being most peripheral. The P values for breakpoints of segmented linear regression models indicate results of Davies’ tests, whereas all other P values indicate results of extra sums-of-squares F tests comparing parameters between Drasdo and no displacement regression models. Not applicable (N/A) P values indicate those where at least one segmented linear regression model was not significant per Davies’ test results, and therefore comparison between Drasdo and no displacement segmented regression models was not performed. Across all analyses, asterisks flag significant results (P < 0.05).