Open Access Journal of Biostatistics and Biometrics

Analysis of Sensitivity Index for Assessing Generalizability in Clinical Research

*Yuanyuan Kong
Department Of Biostatistics, Capital Medical University, Beijing 100050, China

*Corresponding Author:
Yuanyuan Kong
Department Of Biostatistics, Capital Medical University, Beijing 100050, China
Email:kongyuanyuan2000@163.com

Published on: 2018-12-14

Abstract

In clinical research, it is often of interest to determine whether the observed clinical results from the original target patient population (e.g., adults) can be generalized to another similar but different patient population (e.g., pediatrics or elderly). Under certain assumptions, it can be shown that the effect sizes adjusted for standard deviation between two patient populations differs by a factor, which is usually referred to as a sensitivity index. If the confidence interval of the sensitivity index is totally within a pre-specified equivalence limits, then we conclude that the clinical results from the original target patient population can be generalized to the similar but different patient population. In this article, the generalizability of clinical results is assessed by evaluating the sensitivity index under different models that (1) shift in location parameter is random, (2) shift in scale parameter is random, and (3) shifts in both location and scale parameters are random. An example concerning an asthma clinical trial is presented for illustration of the proposed method for assessing generalizability of the observed clinical results from adults to elderly population

Keywords

Shift in location parameter; Shift in scale parameter; Sensitivity index; Generalizability

Introduction

For marketing approval of a new drug product, the United States (US) Food and Drug Administration (FDA) requires that at least two adequate and well-controlled clinical trials be conducted to provide substantial evidence regarding the effectiveness and safety of the drug product under investigation. The purpose of requiring at least two clinical studies is not only to assure the reproducibility, but also to provide valuable information regarding generalizability. Generalizability is referred to as one of the following situations. First, it is referred to as whether the clinical results of the original target patient population under study (e.g., adults) can be generalized to other similar but different patient populations (e.g., pediatrics or elderly).Second, it evaluates whether a newly developed or approved drug product in a region (e.g., United States or European Union) can be approved at another region (e.g., countries in Asian Pacific Region) as it is a concern that differences in ethnic factors could alter the efficacy and safety of the drug product in the new region. Third, for case-control studies, it is often of interest to determine whether the developed or established predictive model based on a database at a medical center can be applied to a different medical center that has similar but different database for patients with similar diseases under study. In practice, since it is of interest to determine whether the observed clinical results from the original target patient population can be generalized to a similar but different patient population, we will focus on the first scenario. Statistical methods for assessment of generalizability of clinical results, however, can also be applied to other scenarios.