Statistical Thinking for Non-Statisticians in Drug Regulation
- PDF / 32,692 Bytes
- 2 Pages / 612 x 792 pts (letter) Page_size
- 19 Downloads / 176 Views
525
Book Review Mark A. DeWyngaert, PhD Managing Director, Huron Consulting Group New York, New York
Correspondence Address Mark A. DeWyngaert, PhD, Huron Consulting Group, 1120 Avenue of the Americas, New York, NY 10036 (email: mdewyngaert@ huronconsultinggroup.com).
KAY, RICHARD: STATISTICAL THINKING FOR NON-STATISTICIANS IN DRUG REGULATION. Derbyshire, UK, 2007, 276 pages. The growing complexity and reliance on understanding of statistical concepts to assist regulators underlies modern clinical trials. The intricacy of clinical trial design leads to elaborate analysis as called for in the statistical aspects of the regulatory guidelines issued by the European Medicines Evaluation Agency, International Committee on Harmonization, and the Food and Drug Administration. Too often, these methodologies are like Sanskrit to the uninitiated and can confound even well-educated industry executives. Many critical decisions regarding new drug development can only be attempted after a variety of statistical tools have been deployed and the results put into context to remove known, as well as potential, study-induced bias. The regulations (and regulators) themselves also may dictate statistical pathways of analysis and overall study design. Thus there is a need to understand the value of the multiple approaches that statisticians may take in their quest. Important questions are often misunderstood by not only nonstatisticians but also other statisticians. There remains a need for all of us involved in the regulation and management of drug studies to better understand the mean-
ing of these statistical tools, their limitations, and their appropriate use. Modern clinical trials often involve large populations of individuals meeting specific study criteria, who are often not in one location. This leads to a number of concerns related to the proper comparison and grouping of data from multiple study sites and an equally large number of clinical investigators. We also are often challenged with multiple endpoints in multiple-arm trials, which can, if not corrected, provide misleading results. A final point is the impact of publications and the role of reviewers to overlook possible mistakes in analysis or statistical plan design. There is a strong need for clear advice that will enhance communications between statisticians and nonstatisticians. Too often the nonstatistician is answered in an arcane language filled with jargon that requires a more precise understanding to avoid mistakes in judgment. Since many regulatory requests require a team effort in responding to their concerns, we need to develop a common understanding of the terms and analytical tools to guide both the design and analysis of clinical data. The book addresses a significant number of statistical issues in a clear manner from basic principles of randomization to meta-analysis and publication concerns. The chapters are filled with references to current regulatory is-
Drug Information Journal, Vol. 42, pp. 525–526, 2008 • 0092-8615/2008 Printed in the USA. Al
Data Loading...
