control status of the subjects indicated that about two-thirds of all veterans (cases and controls) had at least one abnormal neurological finding (most frequently reduced strength of the lower extremities); however, there were no significant differences in the frequency of neurological findings in the cases and controls. The examining neurologists and study investigators reviewed the findings on each subject individually and concluded that the clinical and laboratory findings were nonspecific and not sufficient to diagnose any known neurologic syndrome in any subgroup of subjects (Haley et al, 1997a). Nevertheless, the authors expressed the view that the neuropsychological abnormalities seen in Gulf War veterans are likely the result of neurotoxic exposures associated with service in the Gulf War.

This study also has come under criticism. For example, despite the assertion of neurological damage in the 23 most symptomatic subjects, there were no statistically significant differences on neurological examination between cases and controls, and a panel of neurologists concluded that the confusion–ataxia syndrome was not specifically defined by their standard clinical neurological exams. Furthermore, the changes noted on evoked potentials and saccadic velocity were nonspecific compared to larger control populations. Moreover, evoked potentials have limited utility in evaluating patients with possible neurological disorders; their major purpose is to look for evidence of demyelination, which would not be useful in persons suspected of having OPIDN—classically, a distal axonal neuropathy, not a permanently demyelinating disorder of the brain stem and spinal cord. Although a statistically significant difference in interpeak latencies related to PB was reported on the somatosensory evoked potentials in the 23 cases compared with the 20 healthy controls, there were no significant differences between the results in veterans and the authors’ own laboratory’s established normative values from healthy people. Some “statistically significant” differences are expected by chance alone when a large number of tests are performed on a small number of patients, and in this instance, no adjustment was made for “multiple comparisons.”

Some findings had uncertain clinical relevance (e.g., asymmetric nystagmus velocities following caloric stimulation in patients complaining of arthralgia, myalgia, and weakness). Electromyography shows abnormalities in individuals with OPIDN, but single-fiber EMG results were normal in the Haley study. Thus, the clinical data presented provide little evidence with which to conclude that these patients had a neuropathic process, and it is difficult to interpret the findings of a battery of neurological tests under these conditions. For example, five veterans with confusion–ataxia and three with arthromyoneuropathy had peripheral neurophysiological tests (nerve conduction studies, electromyography, single-fiber EMG, and quantitative sensory tests) that were normal, whereas most of the veterans with syndrome 3 (symptoms consistent with peripheral neuropathy) had no findings consistent with this peripheral neuropathy, which calls into question the validity of the symptoms as a measure of neurological damage. The principal limitation of this nested case-control study, in