Abstract

In this talk, I review the evolution of occupational epidemiology over the last 40 years. Methodologically, the field is almost unrecognizable compared to what was ‘standard practice’ 40 years ago. Methodological changes include the use of new study design and statistical methods, such as counterfactual theory, directed acyclic graphs (DAGs), IPW, g-estimation, g-computation, multiple imputation for missing values, sensitivity analysis, and bootstrapping. Biomarkers and various molecular and omics measures are increasingly used for exposure assessment, and exploration of mechanisms. The exposures and outcomes under study have also evolved, e.g. with increased consideration of psychosocial factors, work organisation, musculoskeletal problems, mental health and neurological disease. Despite all of these changes, many of the fundamentals of occupational epidemiology remain the same. The discovery of new causes of occupational disease continues to be lead by astute observers (including astute clinicians and astute workers), rather than by ‘bigdata’ or ‘omics’ methods. The strategy for investigation of possible occupational causes of disease continues to require a variety of study designs and approaches, including ‘descriptive’ studies, and triangulation across study designs and populations (albeit while utilising new molecular biology and statistical techniques). The causal assessment of occupational exposures and their health effects continues to require a wide variety of types of evidence in humans and animals, as well as mechanistic evidence. Forty years later, the Bradford-Hill considerations have been augmented but not been replaced, and the IARC ‘rules’ for combining various types of evidence remain the state-of-the-art. Reports of the death of ‘traditional epidemiology’ (and its replacement by ‘modern epidemiology’ and ‘causal inference’ methods) have been exaggerated.