Interpretation of observational studies: the good, the bad and the sensational

Abstract

A Nutrition Society member-led meeting was held online on 18th January 2021 to discuss the role of observational studies in developing public health policy and dietary guidelines. In addition, participants debated media reporting of observational studies and the implications for public perception and trust in science. Speakers outlined the benefits of observational studies and how they fit within the suite of research tools available for estimating dietary intakes and determining their impact on health and disease risk. However, there are clear limitations, such as conscious and unconscious bias, measurement error, confounding and representativeness of populations. Researchers can overcome some of these issues with careful design, awareness of inter-individual variation, open and transparent reporting of findings, and hypothesis-driven statistical analysis to avoid multiple testing errors. Although there is evidence that data provided by nutritional epidemiology can be misleading, strong and thoughtful methodology including pre-registration, risk of bias assessment, awareness of confounders, and evidence grading can minimise potential bias, particularly when conducting systematic reviews. Translation of relative risk into population health impact is important and feeds into the need for responsible lay communication of results via mass media, especially regarding assumptions about cause and effect. Although use of mass media can bring benefits to academia, responsible dissemination is essential and starts with the press release. In conclusion, nutritional epidemiology is an important tool for exploring the risk/benefits of dietary patterns and contributing to health improvement via dietary guidelines, evidence-based policy and responsible lay communication provided its limitations are fully understood.