Freezing of Gait in Parkinson’s Disease: Risk Factors, Their Interactions, and Associated Nonmotor Symptoms

Table 4

Results of multivariate nominal logistic regression on freezing of gait grouping.

Predictor	β estimate	Standard error	LR χ^2	value
Model 1 (AUC = 0.83, R^2 = 0.28)
MDS-UPDRS Bal-gait score^1 ≥4	1.09	0.23	25.92	<0.0001
LED^2	0.19	0.05	20.59	<0.0001
Cognitive impairment^3	0.23	0.23	1.03	0.31
MDS-UPDRS Bal-Gait score ≥4 × cognitive impairment	−0.45	0.23	3.95	0.047
Model 2 (AUC = 0.83, R^2 = 0.29)
MDS-UPDRS Bal-Gait score^1 ≥4	1.30	0.31	28.13	<0.0001
MDS-UPDRS motor complications score	0.41	0.12	21.75	<0.0001
Cognitive impairment^3	0.22	0.25	0.82	0.36
MDS-UPDRS Bal-Gait score × cognitive impairment	−0.73	0.31	7.44	0.006
MDS-UPDRS motor complications score × cognitive impairment	−0.25	0.12	6.32	0.012

Variables with significant univariate predictive acuracy were entered in the multivariate logistic regression model. (Negative β estimations are the result of the class level design of the 1/−1 for 0/1 response variables in the SAS system). Effect likelihood ratio chi-square test. ¹MDS-UPDRS PIGD score minus freezing score. ²Levodopa equivalent dose (100 mg units). ³MCI or dementia.