Commentary: Big data bring big controversies: HDL cholesterol and mortality

No matter how large a study, how richly it is phenotyped or how intriguing the findings that emerge, it behoves us to be sceptical in interpreting evidence arising from observational epidemiological studies. Various intractable sources of error, including reverse causality and residual confounding, hinder interpretations on potential causal relationships between exposures and outcomes in observational study settings.1
Source: International Journal of Epidemiology - Category: Epidemiology Source Type: research