Assessing Impairment Risk Produced by Use of a CNS Drug When Normal Ranges Are Uncertain

• Activity Number: 631
• Type: Topic Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Eugene Laska*+
• Companies: Nathan Kline Institute, NYU School of Medicine
• Keywords: normal range ; increased risk ; maximally selected ; test of symmetry ; McNemar

Abstract

Chemistry and laboratory tests in RCTs are usually contextualized in terms of coverage intervals whose limits distinguish normal values from those likely to signal potential for disease. Pre-post observations are often summarized in a shift table, which estimates the transition probabilities from normal to abnormal. In many situations, a treatment induced risk has no clear reference intervals; e.g., the safety risk of impaired driving following use of a centrally acting agent such as a hypnotic. Absent an adverse treatment effect, the probability of increased and decreased risk should be the same; the pre-post difference variables are exchangeable and their distribution is symmetric. Since small differences do not signal safety risk they should have minimal impact on a hypothesis test of symmetry. A reasonable strategy is to systematically exclude observations whose absolute values fall below a threshold, for every possible threshold and calculate a McNemar statistic based on the observations that remain. We give the distribution of the maximum (maximally selected) McNemar, which gives rise to a p-value that can be used to judge whether to reject the null hypothesis of symmetry.