Table 5.

Multivariate logistic regression models for fall status, frequency, and severity

	Model information	Parameter information
	Correct prediction	Coefficient	SE	Wald	p value	Exp(b)	Low	High
A Model I ‚fall status‘	0.78
Retrospective fall status		1.34	0.26	26.1	< 0.001	3.45	2.28	6.34
CV of base of support		0.06	0.02	10.8	0.001	1.06	1.03	1.10
CV of stride time		0.53	0.16	5.99	0.001	1.71	1.25	2.32
Phase synchronization index		− 0,22	0.09	6.00	0.014	0.802	0.67	0.96
B Model II ‚frequent falls‘	0.92
Retrospective fall status		1.36	0.33	17.2	< 0.001	3.91	2.27	8.49
MOCA		0.17	0.08	5.0	0.025	1.19	1.02	1.38
ABC-d		− 0.02	0.01	5.3	0.021	0.98	0.96	1.00
CV of stride time		0.51	0.18	8.3	0.004	1.66	1.18	2.34
Phase synchronization index		− 0.25	0.10	5.8	0.016	0.78	0.64	0.96
Ambulatory bout #		− 0.01	0.00	4.0	0.046	0.99	0.99	1.00
Daily intensity		0.42	0.21	3.9	0.047	1.52	1.01	2.30
Medication: non-opioid pain reliever		0.54	0.22	4.3	0.038	1.55	1.39	2.22
Medication: anticoagulant		0.43	0.21	3.8	0.47	1.26	1.07	1.98
C Model III ‚severe falls‘	0.91
Gait velocity		− 0.07	0.03	8.4	0.004	0.93	0.89	0.98
Ambulatory bout alpha		− 33.6	8.4	6.4	0.010	0.10	0.01	0.11

Outcomes of the three multivariate logistic regression models for the categories ‘fall status’ (no falls vs. falls), ‘fall frequency’ (occasional vs. frequent falling), and ‘fall severity' (Hopkins grades I&II vs. III&IV). Regression analyses was performed on patient data only. Only parameters that significantly contributed to the model output are displayed

FGA functional gait assessment, MOCA Montreal cognitive assessment, SF-12 short form 12, CV coefficient of variation, SE standard error