asthma have used medical records and maternal reports, but analyses of each source tend to
be based on different samples of children. Although these different studies permit
comparison of overall prevalence estimates and some insight into distribution (the effects
of demographic or other predictors) on asthma, they cannot be used to directly compare
estimates from the different sources because the designs of the studies vary. For example,
a study of medical records will typically include different children than a health survey
against which it has been compared.
What cannot be determined by a comparison of
prevalence estimates based on two or more separate study samples is the extent to which
the sets of children identified as cases by the two sources overlap. Furthermore, with
different samples, it is not possible to say with confidence how the two sets of cases
differ in terms of other characteristics. With data from both reporting sources for a
consistent sample, we are able to address both questions.