An Efficient Algorithm for Critical Value Determination in Binomial Maximized Sequential Probability Ratio Testing, with an Application for the Vaccine Safety Datalink

• Activity Number: 492
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Brock Stewart*+ and Rongxia Li and Eric Weintraub and Michael M. McNeil and Frank Destefano
• Companies: Centers for Disease Control and Prevention and Chenega Government Consulting and Centers for Disease Control and Prevention and Centers for Disease Control and Prevention and Centers for Disease Control and Prevention
• Keywords: Sequential Analysis ; SPRT ; maxSPRT ; Vaccine Safety ; critical values

Abstract

The Sequential Probability Ratio Test (SPRT) technique was developed by Abraham Wald in the 1940's to control Type-I error in repeated testing environments. Since that time SPRT has been applied in a wide variety of settings, from clinical trials to reliability testing. More recently, Kulldorf [Sequential Analysis 30:58-78] introduced the maxSPRT method that expands the original SPRT to allow composite alternative hypotheses. Computing critical values for maxSPRT is computationally intensive and can be time-consuming for many real-world scenarios. Kulldorf has already presented a table of pre-computed critical values under a wide range of parameters. However, researchers may require other critical values. We present a time-efficient algorithm to compute maxSPRT critical values that utilizes combinatorial results to reduce the search domain and quickly produce the desired result. We evaluate the algorithm under a range of parameter settings relevant to studies conducted by the Vaccine Safety Datalink.