Table 5.

Extended multiple regression analysis: prediction of future falls, n = 135

Predictorsa	Wald	P value	OR (95 % CI)
History of near fallsb	6.33	0.012	3.03 (1.28–7.17)
Retropulsion (NRT)c	7.43	0.006	3.53 (1.43–8.72)
Comfortable gait speed <1.1 m/s (10MWT)	4.64	0.031	2.55 (1.09–5.98)

Multiple logistic regression analysis backward method (Wald); Nagelkerke pseudo R square: 0.299; Hosmer and Lemeshow test: P = 0.903; tolerance: ≥0.85

Sensitivity/specificity, 0.58/0.87

CI, confidence interval; FOGQsa, Freezing of Gait Questionnaire, self-administered version; NRT, Nutt Retropulsion test; OR, odds ratio; 10MWT, 10-Meter Walk test; m/s, meter per second

^aIndependent variables in the analysis were: history of falls past 12 months, history of near falls, history of FOG (FOGQsa item 3), comfortable gait speed <1.1 m/s (10MWT), retropulsion (NRT), retropulsion (UPDRS item 30), abnormal tandem gait (TG)

^bDichotomous question (Yes/No): Are you ever close to falling, but you manage to grab on to something/someone at the last minute so that your body does not hit the ground?

^cScores ≥1 on the NRT (unexpected shoulder pull) were categorized as having retropulsion