Design and debugging databases for statistical analysis

Any form of data analysis requires the prior creation of a database to house the study information collected in one format or other (questionnaire, clinical history, etc.). The design of such databases should be optimised to allow adequate statistical analysis without the drawing of wrong conclusions. In addition, prior to analysis, debugging or filtering of the variables is required in order to avoid doubling the effort made in extracting the results. The present study offers a series of suggestions for database design and debugging, to ensure that the later statistical analyses are based on the revised data.