Analogous to disease lumping and splitting, exposures can be conceptualized in a number of ways with regard to aggregation and disaggregation, depending on the etiologic hypothesis. The goal is to group together the contributors that share a specific etiologic pathway, leaving none out, but to exclude exposures that are irrelevant to that etiologic pathway. Focusing on only a subset of contributors to a given exposure index would constitute a lack of sensitivity in exposure assignment, and including some irrelevant elements in the construction of that index would represent a lack of specificity. In either case, relative to the optimal exposure category, it is misclassification.
For example, if we were interested in the possible effect of caffeine on risk of miscarriage, then the ideal measure of caffeine would be comprehensive and include all sources of dietary caffeine, such as coffee, tea, and chocolate, as well as caffeine-containing medications. Under the hypothesis that caffeine is the crit ical agent, study of caffeine from coffee alone would constitute underascertain-ment of exposure, and the exposure that was assigned would be lower to the extent that women were exposed to unmeasured sources of caffeine. A closely related but distinctive hypothesis however, concerns the possible effect of coffee on risk of miscarriage, in which constituents of the coffee other than caffeine are considered as potential etiologic agents. To address this hypothesis, aggregation of caffeinated and decaffeinated coffee would be justified. Under that hypothesis, coffee alone is the appropriate entity to study. Once the hypothesis is clearly formulated, then the ideal measure of exposure is defined, and the operational approaches to assessing exposure can be compared with the ideal.
Nutritional epidemiology provides an abundance of opportunities for creating exposure indices and demands clear hypotheses about the effective etiologic agent. Much research has been focused on specific micronutrients, such as beta-carotene or dietary fiber, and with such hypotheses, the goal is to comprehensively measure intake of that nutrient. An alternative approach is to acknowledge the multiplicity of constituents in foods, and develop hypotheses about fruit and vegetable intake, for example, or even more holistically, hypotheses about different patterns of diet. A hypothesis about beta-carotene and lung cancer is distinct from a hypothesis about fruit and vegetable intake and lung cancer, for example, with different demands on exposure assessment. Exposure indices must be defined with sufficient precision to indicate which potential components of exposure should be included and which should be excluded, and how the measure should be defined.
In practice, there are circumstances in which exposure is considered in groups that are not optimal for considering etiology but are optimal for practical reasons or for considering mitigation to reduce exposure. For example, there is increasingly clear evidence that small particles in the air (particulate air pollution) exacerbate chronic lung and heart disease and can cause premature mortality (Kat-souyanni et al., 1997; Dominici et al., 2000). The size and chemical constituents of those particles differ markedly, and their impact on human disease may differ in relation to those characteristics as well. Technology now allows isolation of small particles, < 10 microns or < 2.5 microns, so that it is feasible to regulate and monitor compliance with regulation for the particle sizes thought to be most harmful to human health. It is not feasible however, to monitor the chemical constituents of those particles and thus regulations do not consider particle chemistry. We seek to reduce exposure to particulates, accepting that the effect of the mixture of particles with greater and lesser amounts of harmful constituents is accurately reflected in their average effect. Perhaps stronger associations would be found for subsets of particles defined by their chemical constitution, but the measured effect of particulates in the aggregate is still useful for identifying etiology and suggesting beneficial mitigation of exposure.
Was this article helpful?
Studies show obesity may soon overtake tobacco as the leading cause of death in America. Are you ready to drop those extra pounds you've been carrying around? Awesome. Let's start off with a couple positive don't. You don't need to jump on a diet craze and you don't need to start exercising for hours each day.