The magnitude of the
estimates remained unchanged, suggesting that the effects
of selective fertility, if present, are likely minimal.
Other biases and limitations of our study are those
typical of population-based studies that rely on vital
statistics data. The birth certificate data are prone to
some degree of under-reporting of certain variables (eg,
smoking during pregnancy, medical and obstetric riskfactors), which could introduce systemic or random
bias.33,34 The vital statistics data are often collected after
the termination of the pregnancy, thereby introducing a
misclassification of certain risk factors (such as smoking)
in pregnancy. This misclassification is likely to be differential
in nature, and if present, will bias the effect
measures away from the null.35 The possibility of our
results being affected by residual confounding due to
unmeasured factors (such as cocaine use) may have also
affected the associations noted here. Conversely, the
strengths of this study include the large populationbased
study cohort and our controlling for a variety of
potential confounding variables.
A cesarean birth is an important risk factor for
placenta previa and placental abruption in a subsequent
pregnancy. The presence of a dose-response
pattern in the risk of placenta previa with increasing
number of prior cesarean deliveries, coupled with a
biologically plausible association, provides compelling
evidence in support of the association. Irrespective
of the method of delivery in the first pregnancy,
a short interpregnancy interval appears to be associated
with increased risks of both placenta previa and
placental abruption. The effects of these associations
on adverse pregnancy outcomes remain unknown.