Why the quality of exposure assessment matters in human observational studies and consequent hazard and risk assessment

Abstract

Occupational exposures can vary dramatically and therefore their assessment and assignment in epidemiological studies often come with considerable error. In prior eras of occupational epidemiology, exposure was treated as static; often in a binary (yes/no) or ordinal (no/low/medium/high) manner. In modern times the need for better quality exposure assessment and assignment is clear, particularly in situations where occupational exposures and outcomes are less convincingly associated. High quality exposure assessment is imperative for informative human observational studies that permit better hazard classifications and quantification of exposure response relationships. A wide range of exposure assessment methods exist and have been applied; varying from self-reports general and specific job-exposure matrices, and case-by-case expert-based assessments to elaborate quantitative algorithms using the collective information of thousands of workplace measurements. The selection of method largely depends on the type of epide-miological study performed (retrospective versus prospective; case-control versus cohort; general population versus worker-based), the agent under consideration, and the extent of access to experts and measurement data. In addition, each method has its own specific limitations and possibilities, and will not be feasible for use in every study. Formal methods to evaluate exposure assessment quality in epidemiological studies are largely non-existent, but recent work within the IARC Monographs Programme and elsewhere have provided a direction for a more systematic approach. Several systematic literature reviews of human observational studies have shown that incorporating the quality of exposure assessment can produce a more convincing and informative evaluation when reviewing a body of evidence. As part of a mini-symposium on the importance of exposure assessment in epidemiology for hazard identification and risk characterization, this presentation will provide a brief introduction to exposure assessment and assignment for human observational studies, as well as directions for a more formal approach to their evaluation.