Analyzing multivariate neurobehavioral outcomes in occupational studies: a comparison of approaches

Neurobehavioral studies often employ test batteries and confront issues of multiple testing and comparability between batteries. We have organized our battery of 12 tests into areas of neurobehavioral function to reduce the number of reported results, provide greater statistical power, and improve interpretability of the results. We explored several different organizational and statistical methods of creating summary scores including a priori groupings based upon clinical experience and factor analysis. We compared the sensitivity of these summary scores to performance changes associated with exposure to styrene in the manufacture of reinforced plastics. Our results demonstrated dramatic increases in power to detect exposure related changes compared to using individual test scores. Furthermore, the various methods provided generally compatible and comparable results. We encourage other neurobehavioral investigators to pursue and refine this approach.