Biomedical Engineering Reference
In-Depth Information
become a fundamental statistical tool in epidemiologic studies because of its
eciency relative to the cohort study. An outcome-dependent sampling
(ODS) scheme is a retrospective sampling scheme like the case-control study
where one observes the exposure/covariates with a probability, maybe un-
known, that depends on the outcome variable. The principal idea of an
ODS design is to concentrate resources where there is the greatest amount
of information. By allowing the selection probability of each individual in
the ODS sample to depend on the outcome, the investigators attempt to
enhance the eciency and reduce the cost of the study.
Although the case-control design for binary outcomes is a mainstay
of epidemiology research, an increasing number of studies are conducted
using the ODS design with a \continuous" outcomes as the eld of epi-
demiology expanding and evolving. In particular, analytical epidemiology
investigations are often designed to characterize the study population, such
as with respect to disease prevalence, and investigate the potential eect
of an exposure on various health outcomes. Without a more appropriate
method for handling the ODS with a continuous outcome, many inves-
tigators have chosen to analyze data using a dichotomized outcome that
is dened based on whether the measurement is above or below a certain
cuto point (e.g. hypertension and neuro development abnormality). Draw-
backs of dichotomizing or categorizing a continuous outcome include a loss
of eciency, an increased risk of misclassication bias, and a decrease in
the external validity of the analysis since the results may be sensitive to
the choice of cut point.
An example of using ODS design is a study from the Collaborative
Perinatal Project (CPP) to access the relationship between maternal poly-
chlorinated biphenyls (PCB) level and children's health development. The
CPP is a propespectively designed study to provide precise data for studies
of a wide variety of neurological outcome and birth detects
2
. Subjects were
enrolled through 12 university aliated medical clinics, with the centers
contributing unequal numbers of subjects. In all, 55,908 pregnancies were
registered, representing the experience of about 44,000 women. The study
subjects are children who were born into the CPP. Eligible children met
the following criteria: a) live born singleton, b) availability of a 3 ml third
trimester maternal serum specimen, and c) non missing data for at least one
of 8 specied study outcomes. Of the CPP children, 44,075 met all of the
eligibility criteria. Since it was too expensive to assay the PCB exposure for
the entire study population of 44,075 subjects, the investigators decided to
obtain exposure measurements for an ODS from the population. In partic-
Search WWH ::
Custom Search