Table 3.

The components of predictive models using multivariate logistic regression (backward stepwise method). Model 1 only contains gait characteristics with p < .05, while model 2 only contains clinical and demographic measures with p < .05. The odds ratio indicates the factor by which the fall probability increases with a decrease of 1SD in the variable across all subjects.

Model	Independent variables	p value	SE	Odds Ratio
1*	R thigh angle at TD	0.016	2.218	1.9
	R AP GRF after TD	0.007	0.993	4.56
	R VT GRF after TD	0.005	0.266	0.19
	MMT from TD to LO	0.004	0.033	0.37
2	Weight	0.028	0.014	1.57
	Fall history	0.047	0.402	1.48

^*The equation for this model is: p(fall) = 1/(1 + exp(6.44 + 5.36 × R-thigh-angle + 2.66 × R-AP-GRF − 0.75 × R-VT-GRF − 0.09 × MMT)).

TD: Right touchdown; LO: Left lift off; AP: anteroposterior; VT: vertical; and SE: standard error.