To the Editors:

We appreciated the comments from Hall and colleagues about our recent experiment into the effect of imagined movements in people with chronic upper extremity pain. Hall et al report several interesting phenomena relating to motor imagery in people with CRPS. First, they have observed that patients with CRPS take longer to imagine a movement of their affected extremity than they do for their unaffected extremity. Our group and others have quantified that effect (1, 2), but further interrogation led us to conclude that pain probably does not cause the delay, because experimentally induced hand pain, and the expectation of hand pain, has the opposite effect (3, 4). A delay similar to that observed in CRPS patients can be induced in healthy controls by experimentally disrupting proprioceptive input (5), which seems consistent with the idea reported by McCabe et al (6) that disruption of cortical representation of the affected area might underpin the pain. Clearly, more studies are needed.

No less interesting, and arguably more clinically important, are the observations by Hall et al relating to the exacerbation of pain with imagined movements and with left/right extremity recognition in patients with CRPS. They report that ∼33% of their cohort of patients with CRPS reported increased pain in the affected extremity when they imagined moving the ipsilateral but unaffected extremity. CRPS can affect sensory processing across the entire ipsilateral hemibody even if the symptoms are confined to a single extremity (7). With regard to pain provocation, we agree that this may reflect the enhanced sensitivity of the neural mechanisms underpinning pain. Another possibility is that a change in cardiovascular or respiratory function, both of which can be evoked by imagined exercise (8), causes the effect. Alternatively, McCabe et al have reported that patients with CRPS are vulnerable to situations that involve incongruence between motor commands and sensory feedback (6); perhaps that is the mechanism.

The exacerbation of pain during left/right extremity judgments in 2 of 15 CRPS patients is important because left/right extremity judgments involve implicit motor imagery, and the participants do not necessarily feel the mental maneuvers of the extremity. We have also seen patients who get worse with implicit motor imagery; although in our experience, they are rare. Our treatment approach to these patients has been to make an even more conservative exposure to movement by preceding graded motor imagery (itself a more conservative version of mirror therapy [9, 10]) with the observation of movement in others. We adopted this approach on the basis of extensive work that demonstrates that motor processes can be activated by watching others do functionally-relevant movements (the so-called mirror neuron system [for review, see ref.11]). To start rehabilitation with movement observations remains within the theoretical paradigm of graded exposure. Anecdotally, this strategy has been successful for these patients, but we have no empiric data at this stage.

Finally, CRPS is a diagnosis based on clinical signs and symptoms and in part on the exclusion of other conditions; it is not a mechanism-based diagnosis. As such, it may well encompass several distinct disorders. We have found that some patients respond well to graded motor imagery, others to tactile discrimination training (12), and some do not respond to either. We agree that we need to closely monitor patients, and that we need more data from well-designed experiments and clinical trials if we are to determine who will respond to what. The experiences reported by Hall et al seem critical to planning this ongoing pursuit.

G. Lorimer Moseley PhD*, Frank Birklein MD, PhD†, Jacobus J. van Hilten MD, PhD‡, Johan Marinus PhD‡, * Oxford University, Oxford, UK, † University of Mainz, Mainz, Germany, ‡ Leiden University Medical Center, Leiden, The Netherlands.