The concept of mechanism-based treatment of chronic pain is based on the belief that specific signs and symptoms reflect the mechanisms underlying pain generation and pain persistence (Woolf et al., 1998). Neuropathic pain syndromes are generally quite distinct in their clinical presentation and the results of diagnostic tests. Diagnostic criteria are widely accepted, making diagnosis-based clinical trials a practical approach for both clinical researchers and government regulatory agencies responsible for approving new therapies. However, the rationale for mechanism-based treatment algorithms becomes stronger as we learn more about the molecular basis for pain and accumulate evidence that different clinical disorders share the same underlying mechanisms.

Most published norms for quantitative sensory testing (QST) have been oriented toward detection of neuropathy in the distal limb rather than toward exploring the function of the pain system (Dyck et al., 1993). The pain-oriented human somatosensory testing studies published to date have used center-specific protocols, often without validated reference values from a healthy painfree population. Haanpää and her colleagues provide one exception; they studied over 100 healthy volunteers to gather a reference group for their patients with herpes zoster (Haanpää et al., 1999). Testing of pain-related sensation, such as quantification of dynamic allodynia, have used a wide variety of tools, ranging from a simple foam paintbrush to an electrical toothbrush. Few multicenter clinical trials of investigational drugs for chronic pain have used QST as an outcome measure, and have not created normal reference values (Wallace et al., 2002a, b). Developing standardized protocols for clinical somatosensory testing that include all types of primary afferents and multiple body regions is thus extremely valuable. The first steps toward a broader application of quantitative sensory testing in multicenter clinical trials are training investigators at multiple sites and establishing normal values as reference points.