To improve the quality of data
collected, patients’ medical records were
abstracted by reviewers blinded to the
Papanicolaou smear result and group
assignment. In addition, 20% of the
charts were randomly selected and
abstracted by a separate reviewer to
ensure accuracy of data extraction.
Furthermore, to determine HPV status,
we relied on the available Papanicolaou
smear data. Although HPV typing was
only available for women with ASCUS
Papanicolaou smears, we also included
in our exposed group women with LSIL
or HSIL Papanicolaou smear results and
assumed the infection with HR-HPV, as
these results are highly associated with
HR-HPV and current clinical guidelines
do not require HPV typing.15 It is
possible that few of these patients did not
truly have HR-HPV. In addition, it is
possible that women with a normal
Papanicolaou smear may have had either
a false negative cervical cytology test or
acquired HPV later on during the pregnancy.
We attempted to eliminate this
bias by requiring at least 2 normal
Papanicolaou smears, which increase
sensitivity and specificity of the test. HRHPV
is also associated with high risk
sexual behavior, but this history could
not be reliably determined by review of
the obstetric record. This potential
confounder should be included in potential
future studies.