Control of lateral balance in walking. Experimental findings in normal subjects and above-knee amputees.

In walking the human body is never in balance. Most of the time the trunk is supported by one leg and the centre of mass (CoM) 'falls' to the contralateral side. In dynamical situations the velocity of the CoM should be acknowledged as well in the 'extrapolated centre of mass' (XcoM). Centre of pressure (CoP) position was recorded by a treadmill with built-in force transducers. Lateral CoM and XcoM position were computed by filtering the CoP data. Subjects were six above-knee amputees and six matched healthy controls. They walked at approximately 0.75, 1, and 1.25m/s for 2min. Amputees showed asymmetric gait with shorter stance (60%) at the prosthetic side versus 68% at the non-prosthetic side and a wider stride (13+/-4cm, mean+/-S.D.) compared to controls (9+/-3cm). At foot placement CoP was just lateral to the XcoM. The margin between average CoP and XcoM at foot contact was only 1.6+/-0.7cm in controls, 2.7+/-0.5cm in amputees at the prosthetic side and 1.9+/-0.6cm at the non-prosthetic side. Next to this 'stepping strategy', CoP position was corrected after initial contact by modulating the lateral foot roll-off ('lateral ankle strategy') in non-prosthetic legs up to about 2cm. A simple mechanical model, the inverted pendulum model, can explain that: (1) a less precise foot placement (greater CoP-XcoM margin) results in a wider stride, (2) this effect can be reduced by walking with a higher cadence, and (3) a greater margin at one side, as with a leg prosthesis, should be compensated by a shorter stance duration at the same side to achieve a straight path. This suggests that not in all cases symmetric gait should be an aim of rehabilitation.