64 - Statistics

Abstract

An understanding of basic statistics is important for all clinicians to be able to apply scientific evidence to medical practice, and some important statistical concepts are reviewed in this chapter. First, data types and their relation to the normal distribution are discussed, with an aim toward determining the type of statistical test that can be appropriately applied for hypothesis testing. Sensitivity, specificity, and receiver-operator characteristic curves are reviewed, along with type I and II errors. The importance of using confidence intervals and effect size to describe differences, rather than p-value alone, is critically discussed. Measures of correlation and regression are briefly reviewed. Broad categories of experimental designs are described, such that the reader many understand the strengths and weaknesses of the breadth of published studies. Finally, issues related to negative results are discussed, including statistical power and equivalence versus noninferiority testing.