Epidemiological case definitions of peripheral neuropathy: experience from two neurotoxicity studies

The purpose of this investigation was to develop and apply case definitions of peripheral neuropathy (PN) derived from a set of individual measures of peripheral nerve function obtained in two epidemiological studies.

In the first study, retired workers of an industrial plant who were either previously exposed to mercury or not underwent a set of peripheral neurological tests. In the second study, persons living in an arsenic contaminated community in rural Georgia, USA, and unexposed comparison subjects were tested. In both studies all participants received a neurological physical examination, functional quantitative tests of vibrotactile thresholds and standing stability, and nerve conduction measurements. Two types of analyses were performed. First, the effect of exposure group status was examined for each individual test. For tests with continuous outcomes, the effect of exposure group was estimated while controlling for covariates. For tests with categorical outcomes (all of the neurological examination outcomes), exposure group effects were estimated by comparison of the proportion of abnormal test results among the exposed and unexposed groups. Second, case definitions of PN were constructed using combinations of results of the tests performed. Each of the continuous measures was categorized as normal, equivocal, or abnormal on the basis of adjusted standardized scores. Separate "abnormality scores" were constructed for 1) the nerve conduction tests and 2) the functional quantitative tests and physical examination findings. Five case definitions for PN were constructed based on combinations of these two abnormality scores.

Generally, small differences were observed between exposed (n=79 arsenic exposed and 85 mercury exposed) and unexposed (n=84 arsenic unexposed and 118 mercury unexposed) groups when results of each clinical, functional or electrophysiological test were examined separately. Clearer group differences were observed in the proportions of each group who met case definitions for PN, however. For example, for both studies, the largest difference in nerve conduction measures between exposed and unexposed groups was approximately 0.33 SD, while the composite case definition showed an odds ratio of more than 3.1 for the mercury study and 5.1 for the arsenic study.

These results suggest improved efficiency (and avoidance of the multiple-comparisons problem) for detecting peripheral nerve effects when case definitions of PN are constructed rather when results of individual tests of PN function are compared.