Masters candidate in Biostatistics, Xiao Ma, will present:

“SAS Macros for Estimating Reference Intervals From Meta-analyses Using Aggregate and Individual Participant Data”

Plan B Adviser: Lianne Siegel

Abstract: Reference intervals are the key to estimating whether a clinical measurement value is “normal” or not. Four methods have been recently proposed for calculating the normal reference interval of a subject from a meta-analysis of normally or lognormally distributed outcomes: a frequentist random effects model, a Bayesian random effects model, an empirical method, and a mixture distribution method. However, all the methods are implemented solely through R and built-in SAS procedures for meta-analysis are limited. This thesis briefly introduces the reader to the concepts of these methods for estimating reference intervals in meta-analysis and presents a set of macros to realize the frequentist and empirical methods for aggregate and individual participant data and the mixture distribution method for aggregate data in SAS. It then illustrates the required inputs of the five SAS macros and provides an example about estimating the reference interval for a liver stiffness measurement in healthy children to show the application and the invocation of these macros. These user-friendly macros enable users with limited SAS programming skills to estimate a reference interval from a meta-analysis.

Keywords: meta-analysis; normative data; random effects model; reference interval; SAS macros