Abstract
Background: Non-participation, losses to follow up and other types of study-specific selection mechanisms can be serious concerns in epidemiological studies. There are also selection processes that result in non-random groupings and changes in the composition of populations. These are continuously on-going irrespectively of whether they are subject to sampling in empirical studies. Such population selections are often overlooked, but may lead to lack of comparability of exposed and unexposed populations or decrease study validity in other ways. The overall aim of this study was to develop a simple but general framework for classifying various types of selection mechanisms of relevance for epidemiological research. Methods: We classify selection mechanisms in three dimensions: i) selection at the population level vs. selection that is study-specific, ii) type of mechanism (selection causing exposure vs. selection in population at risk), iii) timing of the selection (pre-exposure, during exposure or post-outcome). Results: Examples from the epidemiological literature of selection mechanisms are discussed and classified according to the three dimensions of the proposed framework. Conclusions: Increased mechanistic understanding of when, how, and why confusion of effects can occur because of selection is an important step towards improved validity of epidemiological research.
A novel framework for classification of selection mechanisms in epidemiological research