A major concern for patients after spinal cord injury is whether or not they will ever walk again. For completely paralysed patients this prospect is slim, but the outlook for those with partial cord lesions is more hopeful. Any programme, such as that suggested in the paper by Wirz et al (this issue, pp 93–96), which may increase the likelihood of subsequent ambulation is to be welcomed.1

Control of locomotion in primates is predominantly supraspinal. In the thalamic macaque monkey electrical stimulation of the posterior subthalamic region or the midbrain tegmentum just ventral to the inferior colliculi produced stepping movements.2 This was abolished in the suspended animal if the cord was completely transected, even when reflexes had returned. However, if pathways in the ventral white matter on one or both sides of the cord were spared, stepping and walking did occur.

Locomotor training of patients with unilateral lower limb paralysis supported over a moving treadmill was found to improve power in the paralysed limb to make walking possible with minimal splintage and this was attributed to enhancing spinal interneuronal locomotor networks.3 However, as the ventral part of the cord was intact, at least in part, this improvement could equally well arise from the training of higher centres, if not also to continuing recovery in spinal neurons, although unlikely in those patients who had started training many years after injury.

When this locomotor training was applied to patients with complete spinal cord injury, only four out of 10 showed an EMG response.4 The failure of the remainder to respond was attributed to the drugs they had been taking—the adrenoceptor antagonist prazosin or cannabis. Support for this was obtained by giving intrathecal adrenaline (epinephrine) which increased and clonidine which decreased motor performance. It could also be possible that those who responded may have had some tracts functioning that could not otherwise be detected by clinical or electrophysiological techniques, but nevertheless allowed conduction to supraspinal centres.

The present study shows that in the partially paralysed patients there was an improvement in the amplitude of gastrocnemius EMG on completion of the treadmill training which was still maintained a mean of 2 years later, especially as all these patients continued to walk during this time. The completely paralysed patients could not walk so the smaller improvement in gastrocnemius EMG was lost when they no longer received treadmill training. The improvement during training never reached functional levels in these patients and advocating such intensive treatment without showing a functional benefit could be psychologically harmful. There may be some benefit if these patients were to proceed to a functional electrical stimulation walking programme after discharge from the treadmill training.

As has been advocated, it is important to compare novel interventions in treatment with conventional techniques.5 Many treadmill studies lack adequate spinal cord injury control groups to demonstrate the value of such training and although spinal interneuronal responses may be stimulated, the vital contribution of higher centres cannot be ignored. With some current research focused on spinal neuronal regeneration, such as by the transplant of oligodendroglia, olfactory, or Schwann cells, any technique to increase the functional capacity of the receiving circuits must surely be encouraged.