ABSTRACT
Good statistical practice (GSP) in pharmaceutical/clinical research and development is deŒned as a set of statistical principles and/or standard operating procedures for the best biopharmaceutical practices in design, conduct, analysis, evaluation, reporting, and interpretation of studies at various stages of pharmaceutical research and development (see, e.g., Spriet and Dupin-Spriet, 1992; Wiles et al., 1994; Chow, 1997). The purpose of GSP is not only to minimize bias but also to minimize variability that may occur before, during, and after the conduct of the studies. More importantly, GSP provides a valid and fair assessment of the drug product under study. The concept of GSP in pharmaceutical/clinical research and development can be seen in many regulatory requirements, standards/speciŒcations, and guidelines/guidances set by most health authorities, such as the U.S. Food and Drug Administration (FDA) and the Committee for Proprietary Medicinal Products (CPMP) in the European Community (CPMP, 1990). For example, the U.S. regulatory requirements for pharmaceutical/clinical research and development are codiŒed in the U.S. Code of Federal Regulations (CFR), while the U.S. Pharmacopeia and National Formulary (USP/NF) and National Committee for Clinical Laboratory Standards (NCCLS) include standard procedures, test and sampling plans, and acceptance criteria and speciŒcations of many pharmaceutical compounds (USP/NF, 2000; NCCLS, 2001). In addition, the FDA also develops a number of guidelines and guidances to assist the sponsors in drug research and development. These guidelines and guidances are considered gold standards for achieving good laboratory practice (GLP), good clinical practice (GCP), current good manufacturing practice (cGMP), and good regulatory (review) practice (GRP). The concept of GSP is well outlined in the guideline on Statistical Principles for Clinical Trials issued by the International Conference on Harmonization (ICH, 1997). As a result, GSP not only provides accuracy and reliability of the results derived from the studies but also ensures the validity and integrity of the studies.
