Planning, implementing, and managing multi-centered clinical trials cannot be successfully achieved without a systematic and statistical foundation. From the early inquiries regarding what is being studied, who is being studied, and how many are needed in the study, to the determination of Bayesian or frequentist analysis or response variables with fixed and/or random effects, statistical methodology can provide the solutions which reduce resource consumption, simplify submissions, and insure compliance.

Past, present, and future trends indicate greater reliance on a systems approach to solving issues surrounding the approval of investigational medical devices. FDA has come to recognize that statistics are a part of the early plan for a study at the IDE stage and not an element to be left to the end at the time of the PMA. Consequently, FDA's in-house statisticians are reviewing projects earlier in their submission history bringing with that review a strong influence on the approval process.

From study design and quality assurance methods to statistical model parameterization and statistical model parameterization and statistical interpretation, data assumptions and distributions need to be addressed.

Some of the uncertainties statistical methodology can allay revolve around:

• Patient selection and/or treatment bias
• Number of patients needed to establish a difference or equivalence
• Independent or dependent correlated data
• Clustered or stratified data
• Balanced or unbalanced designs
• Tests of hypotheses and p-values or confidence intervals
• Phase I or Phase II or Phase III type of analysis
• Exactor asymptotic tests
• Random and/or fixed effects
• Binomial or Poisson distribution

The decisions which emanate from the above dichotomies require the expertise and knowledge which can only be acquired through years of experience as found with Buckman Company, Inc.