Learning outcomes

To understand the goal of a study (e.g. equivalence, difference), the importance of proper study design, of actions to be taken to avoid bias, and of appropriate sample size.

To recognise the type of design (e.g. parallel group, crossover, factorial, blocked) and the type of output variable (quantitative, qualitative, independent data, paired data). Detect the presence of potential covariables.

To deduce the kind of statistical analysis to be performed and do the analysis using R (statistical package) if the level of complexity of the experimental design and/or statistical model allows it.

To talk with a statistician and anticipate the information he needs as part of the design and analysis of a study.

Goals

This training presents the statistical methods most commonly used in clinical and preclinical trials. Following the completion of the training, students should be able to perform classical statistical analyses and think critically about the way experiments are designed and statistical analyses are performed.

Content

The training consists in 4 modules:

M1 – Introduction. Basic principles and vocabulary used in applied statistics. Role of statisticians in clinical and preclinical trials. Main steps to the design of experiments and analysis of results. Normal distribution and theory (central limit theorem), location and dispersion parameters, statistical intervals, graphical representations for univariate data.

M2 – Comparison of means. T-test and analysis of variance in the case of independent data (complete randomised design), paired data (random block design). Multiple comparisons of means (pairwise, vs. control group). Difference versus equivalence testing approaches (understanding paradigms, study objectives, decision rules).

M3 – Regression Analysis. Introduction to the simple linear regression and various validity criteria. Statistical intervals (confidence, prediction) applied to bivariate data analysis. Polynomial models (several independent variables, linear and quadratic effects) and experimental designs (e.g. central composite design).

M4 – Comparison of proportions. Introduction to 2 x 2 contingency table and related statistical tests. Confidence limits of a proportion (binomial distribution) and comparison of mean proportions (normal approximation). Various statistical metrics used in clinical and preclinical trials (e.g. sensitivity, specificity).

Exercices

There are many exercises during the training sessions. Students answer to the questions with the support of the teacher and perform the statistical calculations using R (statistical package). Students also do some homework and can find the solutions on WebCampus.

Assessment method

There is a 2-hour written exam, which usually consists in multiple-choice questions (MCQ, 4 points) and 3 exercises (16 points). Students have access to their lecture notes, except for the MCQ.

The exact modalities of the evaluation are likely to be modified during the preparation of the examination schedules, depending on the practical constraints with which the faculty administration may be confronted, or in the event of illness / force majeure / encroachment with an internship, or because of the health situation related to the coronavirus.

Sources, references and any support material

Training package (slides) and R codes associated with exercises are available on WebCampus.

R Development Core Team (2005). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL: http://www.R-project.org

Text Books:

Dagnelie P. [2013]. Statistique théorique et appliquée. Tome 1. Statistique descriptive et bases de l'inférence statistique. Bruxelles, De Boeck, 517 p.    ISBN  978-2-8041-7560-3

Dagnelie P. [2011]. Statistique théorique et appliquée. Tome 2. Inférence statistique à une et à deux dimensions. Bruxelles, De Boeck, 736 p.    ISBN  978-2-8041-6336-5

Language of instruction

Français