Measurement of Drug Liking in Abuse Potential Studies: A Comparison of Unipolar and Bipolar Visual Analog Scales

Abuse liability studies usually measure drug liking using 100-mm visual analog scales (VAS), presented as unipolar (liking measured on entire scale) or bipolar (liking and disliking measured with a neutral midpoint). These 2 types of VAS were compared using drug liking ratings from a randomized double-blind crossover study of immediate-release and controlled-release oxycodone in 2 cohorts of nondependent recreational opioid users. Cohort 1 (n = 19) received intact oxycodone 40 mg, intact OxyContin (R) 40 and 80 mg, crushed OxyContin (R) 40 mg, and placebo, while cohort 2 (n = 16) received intact oxycodone 20, 40, and 80 mg and placebo. In general, bipolar and unipolar ratings were positively correlated (r = 0.72) for all values (n = 2,477). E-max for both scales generally had higher correlation than individual responses for active drug or placebo. The correlation for individual scores after placebo treatment for the 2 scales was poor in both cohorts (r = -0.11, cohort 1 and r = 0.17, cohort 2). Both scales performed similarly within the context of the study, but bipolar scales can also assess disliking, which may be a consideration depending on anticipated drug effects. Appropriate participant training on the use of these scales is also necessary to reduce variability.