Objective: To characterize preparation for compensatory stepping in people with Parkinson's disease (PD) compared with healthy control subjects, and to determine whether levodopa medication improves preparation or the execution phases of the step.

Design: Observational study.

Setting: Outpatient neuroscience laboratory.

Participants: Nineteen participants with idiopathic PD tested both in the on and off levodopa states and 17 healthy subjects.

Intervention: Moveable platform with posterior translations of 24cm at 56cm/s.

Main Outcome Measures: Compensatory steps forward, in response to a backward surface translation (24 cm amplitude at 56 cm/s), were categorized according to the presence of an anticipatory postural adjustment (APA) before stepping: no APA, single APA, or multiple APAs. The following step parameters were calculated: step latency, step length, center of mass (CoM) average velocity, and CoM displacement at the step initiation.

Results: Lateral APAs were evident in 57% and 42% of trials for people with PD in the off and on medication states, respectively, compared with only 10% of trials for control subjects. Compared with subjects with PD who did not have APAs, those subjects with PD who did make an APA prior to stepping had significantly later (mean ± SEM, 356 ± 16ms vs 305 ± 8ms) and shorter (mean ± SEM, 251 ± 27mm vs 300 ± 16mm) steps, their CoM was significantly farther forward (185 ± 7mm vs 171 ± 5mm) at foot-off, and they took significantly more steps to regain equilibrium. Levodopa did not affect the preparation or execution phase of compensatory stepping. Poor axial scores and reports of freezing in the United Parkinson's Disease Rating Scale were associated with use of 1 or more APAs before compensatory stepping.

Conclusions: Lateral postural preparation prior to compensatory stepping in subjects with PD was associated with inefficient balance recovery from external perturbations.