Evidence-based nutrition — the impact of information and selection bias on the interpretation of individual studies

Abstract

This paper in the series on evidence-based nutrition guides the nutrition scientist and practitioner regarding how to judge whether information or selection bias has occurred in reported studies. Information bias occurs when there are either random or systematic differences in the measurement of exposures, outcomes and possible confounders (effect modifiers). These variables are categorised for nutrition studies and the types of questions to ask in order to assess potential bias in obtaining this information are explained using appropriate examples. Selection bias occurs if the relationship between exposure and outcome for those who participate in the study is different to that for those from the population who are eligible to participate but do not. The paper shows how factors affecting inclusion of subjects at the beginning of the study might introduce bias and gives guidelines for appropriate sampling, indicating how knowledge of the potential sources of bias in nutrition studies should be used in the design and analysis of the study. The paper furthermore gives practical guidance on how to determine sample size, using variance, size of the hypothesised effect, level of statistical significance and power in these calculations. The paper explains the real problem of bias in nutrition studies and how to evaluate and deal with it, but also gives examples where bias may not be a problem.