A method that does not require any surveillance data whatsoever is a survey of sick (or recently sick) individuals to determine when they "emitted'' behaviors that may show up in surveillance data.
A recent survey of pediatric patients is an example of a survey designed specifically to understand the informational value of biosurveillance data (Johnson et al., 2005). In this study, Johnson interviewed caregivers of children who were being seen in emergency rooms for fevers, respiratory ailments, and gastrointestinal symptoms. She asked the children's care-givers whether they had purchased over-the-counter medications (and which ones and when), visited physicians, called for appointments and several other actions that would leave footprints in routinely collected data. The survey instruments used in that study are available at http://rods.health.pitt.edu/ PedsEDStudy.htm. We discuss this study and one other survey explicitly designed to understand the informational value of biosurveillance data in detail in the next chapter.
There are several surveys that were conducted for other purposes, but remain of interest both for the results they obtained relevant to understanding biosurveillance data as well as for providing methodologies for conducting such studies. We discuss the results of the studies in the next chapter because they were studies of use of over-the-counter medications.
Note that surveys do not directly measure the effect of disease outbreaks on surveillance data. Their interpretation relies on assumptions that the behavior reported actually occurred and that it was recorded in an information system. Surveys, however, can provide insights that other methods cannot. For example, the results of surveys can be used to estimate the fraction of sick individuals in a population that will appear in surveillance data, thus enabling evaluators to understand the relationship between the size of a spike in surveillance data and the size of an outbreak (Johnson et al., 2005). A single survey can produce these estimates for many types of surveillance data; thus, it is a highly efficient research design. Surveys are complementary to other studies, confirming the measurements of timing and strength of a signal that can be expected in different outbreak situations.
Was this article helpful?