lab rat



Working Together







            William R. Fairweather, PhD, PSTAT(R)

I have been a practicing statistician since 1968. I hold a several academic degrees, including a doctorate in statistics. I am accredited as a Professional Statistician (PSTAT®) by the American Statistical Association. I have operated Flower Valley Consulting, Inc., a statistical consulting firm, since 1999, specializing in clinical and preclinical study designs and analyses. Before entering private practice, I served as Associate Director of the Office of Biometrics and Epidemiology at the Center for Drug Evaluation and Research (FDA). I have been identified by the FDA as an Expert Regulatory Statistician. I now routinely prepare materials for submission to US and European regulatory agencies on behalf of drug, biologics and device sponsors. I have served on several Data Safety Monitoring Boards for clinical trials, and as an expert witness in court. I am the author of a number of scientific publications in refereed journals, and I have lectured frequently on biopharmaceutical statistics in the US, Great Britain, Israel, Belgium, Germany, and Italy.

My professional activities encompass the periods before, during, and after a typical clinical trial. Before the trial, I help prepare the protocol. I develop a clear understanding of the medical or biological issues underlying the trial and clarify any applicable regulatory issues. I learn what is known about previous trials or studies with the test product. With this information, I am able to address the current trial with a solid background. Among other things, I can make hypothesis definitions that avoid excessive statistical penalties. This usually involves much discussion with sponsors, who may not be aware of such issues and their effect on the trial. Relevant response variables, baseline variables and on-study covariates must be identified. Definition of sample size is rarely straightforward in early phase trials. Some idea of the analyses ultimately to be employed is needed along with the likely logistic issues (or financial constraints) in conducting the trial. Some trials can be modified as they are being conducted, and this feature must be accounted for in advance. I develop the randomization scheme to correspond to the trial design.

I usually write the entire Statistical Analysis Plan (SAP), which is required for most Phase 2 and 3 clinical trials. The SAP must be formally finalized prior to unblinding the data. It covers all aspects of data handling and analysis of the trial. It usually contains table templates to be populated with trial data.

The protocol and the SAP are roadmaps to conduct of a perfect trial. Most trials are not perfect. After the trial, I participate in ensuring the quality of the data. I review the procedures of the sponsor or CRO that collected the data. I look to see whether data values have been screened for logical and allowable values (eg, ranges). If these procedures have not been performed, I may have to perform them. I tabulate and graph the raw data to identify outliers and impossible data, such as patients becoming younger over the course of the study. I check whether the predefined randomization was indeed followed and whether the number and type of patient is that defined in the protocol. Where possible, I get clarification or correction of faulty data from the sponsor. If the protocol was not strictly followed, I may have to address the impact of these deviations on the final analysis.

Data analysis computer programs require specific data formats for input. I prepare data for analysis from data files prepared by the sponsor or a CRO. They are not usually appropriate for analysis software. There are usually multiple files for each analysis. Merging of files is usually required and it is essential that each patient’s record is appropriately associated with him or her. This can be an arduous effort, and I try to advise sponsors in advance how to prepare the data files to expedite handling at this stage.

I design, modify, and run computer programs to summarize and analyze the data in accordance with good scientific and regulatory practice. I conduct the predefined analyses of the SAP and analyses suggested by the data that yield unexpected findings. I estimate the relevant study parameters. I verify the underlying statistical assumptions, including the definition of the mathematical model.

I prepare the statistical report for the sponsor and for the regulatory agency, including tables and graphs. These may be different reports, depending on sponsor interests. The report for FDA puts emphasis on issues of concern to FDA staff. I prepare the data files in SAS Transport format, as required by FDA, for submission with the report.

The software package that I use for data analysis has been validated in accordance with FDA requirements. If I need to manipulate data, I always do it under program control, not by typing in the changes. When I write computer programs for specific analyses I “signpost” the code to identify the objectives and methods used. I keep a log of my technical activities on each project. These practices help me reproduce any data manipulations or analyses, even after long periods of inactivity.



home | services | working together | resume | bibliography | email