Vital statistics data is a readily available
source of information on prenatal
care usage in the United States. How to
examine vital statistics and other types of
data for the effect of prenatal care on pregnancy
outcome has generated a number of
different methodologies.1-4 One of the difficulties
encountered in using such data
has been the problem of overcoming the
preterm bias effect. The preterm bias effect
is associated with observing better
outcomes in women who receive prenatal
care than in women who receive no care.
This artifact occurs because women receiving
no care may deliver prematurely
and may not have an opportunity to enroll
in prenatal care.5 By conditioning the effect
of the age of entry into prenatal care
on the gestational age of delivery, however,
the preterm bias effect can be
avoided. Nonetheless, other problems encountered
in vital statistics data may invalidate
the results of the conditional approach.
In this study, we demonstrate the effect
that the possible misclassification of
gestational age in vital statistics data has
on the interpretation of the conditional approach
and discuss the problematic nature
of the evaluation of prenatal care.