Effects of graded dosages of alcohol on nine computerized repeated-measures tests

We report a controlled laboratory validation experiment to provide an empirical sequel to the general background of the development of the Automated Performance Test System (APTS) that was presented in Turnage et al. (1992). The purpose of this study was to index performance deficit against various Blood Alcohol Concentrations (BACs) of small (0.05% BAC), medium (0.10% BAC), and large (0.15% BAC) dosages as well as a placebo (0.00 BAC) condition. Blood alcohol concentrations, measured in four different ways, were highly reliable and the method with the highest intercorrelation was whole blood. When taken singly, eight of the nine tests produced significant (p < 0.001) relationships with the disparate blood alcohol levels, which were essentially monotonic. A multiple regression analysis suggested that most of the tests were behaving similarly and that two or three tests accounted for 52 to 54% of the variance attributable to alcohol that was explained by the longer battery. While all tests appear valid, some of them appeared more sensitive than others (viz., code substitution, manikin, and choice reaction time). From the standpoint of these tests, greater changes occurred in cognitive function between the placebo and 0.05 level than between the 0.05 and 0.10 levels. However, the greatest reduction in performance occurred between 0.10 and 0.15, and the relatively abrupt nature of this change implies that sharp decrements in cognitive performance occur at that point. There were individual differences in resistance to alcohol, and there is strong inference that these differences would be reliable if they were tested again. We believe that further development and study of such techniques is warranted for use in fitness-for-duty testing and development of a dose equivalency index.