Statistics in medical research
Linsell L., Hardy P.
© Cambridge University Press 2017. Introduction Statistics is fundamental to the research process, from the initial stages of planning and design, to data collection and analysis, and then the final stage of reporting and interpretation. In this chapter we give an introduction to the main concepts of statistical analyses in medical research and describe the methods of analysis that are used in common situations. The methods described are relevant to observational studies and randomised controlled trials (RCTs). We do not give details of how to carry out the statistical tests. These can be found in many statistics textbooks, and a list of suitable reference books is provided at the end of the chapter. Why We Need Statistics in Medical Research In a research study we collect data from a sample of participants which is drawn from the wider population of interest. Observed data from the sample is used to make estimates about the true population (this is known as statistical inference) - for example, the proportion of women who develop gestational diabetes during pregnancy. There is an intrinsic variability in the information collected both between and within individuals. For example, participants in a research study will respond differently to the same intervention or exposure to the same risk factor, or measurements made on the same participant (such as heart rate or blood pressure) will vary with each reading. It is this variability that is at the core of statistics; separating out random variability or noise from real variability is a key function of statistical analysis. Statistics can thus be used to: • express the level of uncertainty about the estimated value • estimate the magnitude and direction of changes in the measurement of interest • assess the validity and applicability of the results of a research study. What Statistical Analysis Is Used For The most common aims of statistical analyses in medical research are to: • Explore and describe the distribution and characteristics of the data using summary statistics and graphs • Perform comparative analyses assessing differences between two or more groups • Estimate the prevalence or incidence of a medical condition, outcome or some other value either in the general population or in a particular group of individuals who share similar characteristics.