Search

Further statistics in dentistry: If one were to go by the explosion of interest in evidence-based clinical practice in the past decade of the second millennium, one could be forgiven for thinking that the idea was new. In fact, a quick search of Medline revealed 9,306 references to 'evidence-based medicine' (EBM) and 291 when the search was restricted to dentistry. It is claimed (Sackett et al., 1996)1 that the origins of EBM date back to mid nineteenth century Paris or earlier although the name EBM was coined in 1992. The inventor of the randomised controlled clinical trial, Sir Austin Bradford Hill, in the 1950s set out the statistical foundations of EBM.

In order to introduce the concepts underlying multiple linear regression, it is necessary to be familiar with and understand the basic theory of simple linear regression on which it is based.

Further statistics in dentistry: The clinical trial is a planned experiment, strictly on human subjects, which is conducted with a view to investigating the efficacy of one or more treatments for a given condition. It is possible to use statistical techniques to make inferences about the population of patients who will present to the practitioner in the future using information obtained from the sample of patients in the trial. Consequently, the results of the trial may be expected to influence the way in which patients with the condition are treated in the future.

To facilitate statistical discussion, it is essential to have an understanding of factors, effects, interactions, confounding, bias, estimation and hypothesis testing, to name but a few terms commonly used in statistical investigations in dentistry. These terms were explained in the previous paper in this series. This second paper is concerned with describing and differentiating between the various research designs that may be adopted.

Consider the situation in which there is a single quantitative variable of interest that is measured on each individual on several different occasions. Typically, these occasions are defined time points (usually including pre-treatment as well as various post-treatment times), so that each individual contributes a series of readings. The main objective of the study may be to compare the responses on this variable when each individual has been assigned to one of two or more treatments groups.

This new series of articles is designed to supplement, rather than replace, the material contained in the earlier series on statistics in dentistry published in the British Dental Journal and subsequently made available in book form.(1) With the increasing availability of calculators, personal computers and computer statistical software packages, the need for details of the 'nuts and bolts' of statistical theory has diminished, but the need for an understanding of how and when such theory should be applied, and how studies should be designed to make best use of it, has probably increased. It is with these thoughts in mind that this new series is being offered.