Interpretation
In this study, a total of 555 pregnant women had a short cervix. About
6000 deliveries are recorded yearly in our hospital with a short cervix
incidence of about 9.3%. A previous study conducted in our hospital
between May 2010 and May 2015 found that the incidence of short cervix
was 0.45% for a gestation period of 20 and 24
weeks15. It is possible that a variation in the number
of subjects studied is responsible for the large difference in the
recorded incidence. In the current study, we included patients with a
short cervix found at 11-35 weeks of gestation. To minimize severe
complications of preterm birth, programs that promote early detection of
this condition are often implemented in our hospital. Among such
programs, transabdominal ultrasonography is the use to measure the
cervical length for pregnant women who agree to undergo antenatal
ultrasound examination. Subjects with a cervical length of less than
25mm as well as those that showing signs of miscarriage including
bleeding and contraction were subjected to transvaginal ultrasonography
for a more accurate assessment. We observed that, among women with a
short cervix, BMI before pregnancy was related to PTB. In addition,
women who were obese or overweight before pregnancy were at higher risk
of PTB compared to those with normal BMI, which is consistent with
previous studies10,16,17. Of note, we found that
autoimmune disease (SLE and APS) increased the risk of PTB <
37 weeks13. However, PCOS and hypothyroidism
previously associated with PTB were not significantly correlated with
occurrence of PTB18,19. In the current study,
prepregnancy hypertension was associated with PTB < 34 weeks
but not PTB < 37 weeks. Analysis of other complications
including polyhydramnios, gestational diabetes mellitus and blood
pressure state, showed that they were not significant risk factors of
PTB among patients with a short cervix (Table 1) . These
findings were differed from those reported in a previous two-year
retrospective study. This may be due to the fact that the population
used in the latter included general pregnant women16.
This further indicates that patient’s baseline information can be used
to establish a prediction model for PTB. Among the factors defined in
the risk history variable, only PTB was a significant predictor and this
is in agreement with a previous report17. It is should
be noted that induced abortion has also been reported as a risk factor
for PTB20. We reasoned that the majority of patients
(73.0%) in this study were primipara, and this may explain the
discrepancy. Moreover, all patients in this study had a short cervix,
unlike in other studies. We further observed that LEEP did not increase
the risk of PTB in line with what has previously been
reported21. However, a previous meta-analysis found
that LEEP is a risk factor for PTB (<32/34, <28
weeks) 22. Numerous studies have demonstrated a
relationship between HPV infection and PTB23. In the
current study, however, this could not be clarified owing to the small
data collected. Analysis of obstetric data showed that in the 65% of
patients that underwent IVF-ET, no collinearity was observed between
twin pregnancy and IVF-ET. This suggested that twin pregnancy and IVF-ET
are both independent predictors of PTB < 37 weeks among
patients with a short cervix. However, IVF-ET was not a predictor of PTB
< 34 weeks. In this study, parity was not a risk factor of
PTB, and its relationship with PTB is still unclear, and hence requires
a more systematic assessment in the future17,24.
Cervical length (CL) detected by ultrasound is one of the most commonly
used parameters for early detection of PTB4. In the
current study, we analyzed the relationship between CL, gestational age,
amniotic fluid sludge and overall PTB. We found that the risk of PTB
(< 37 weeks or < 34 weeks) increased with decline in
CL and with the gestational age at 1st diagnosis of
short cervix, however sludge was not a predictor of PTB. Previous works
have focused on CL measurements in the second trimester of pregnancy
with a short cervix during this period found to be significantly
associated with PTB6. However, prediction potential of
CL in the first or third trimester of pregnancy remains
unclear25. Because a high rate of short cervix
(< 25mm) occurring at 36 weeks of gestation and the low rate
of short cervix at 16 weeks of gestation26, we
included pregnancies with a short cervix at first and third trimesters.
Considering severe complications of preterm birth, we estimated the risk
of PTB at first and third trimesters. Our results showed that CL during
this period was a significant predictor of PTB. Particularly, we
observed a low and high risk of preterm at third and first trimesters,
respectively. This could be due to; (i) the population with a short
cervix screened under this study was at high risk of PTB, in which PTB
< 37 weeks was 33.7% and PTB < 34 weeks was 16.9%,
(ii) a combination of factors including CL and other factors such as
gestation age played an important role in the prediction of
PTB16. This is in line with a previous study that
implicated a decrease in CL and the gestational age as a risk factor of
PTB < 35 weeks27.In our study, the
predictors of PTB < 37 weeks and PTB < 34 weeks were
not consistent. Notably, PTB < 37 weeks was influenced by more
factors than PTB < 34 weeks. It was also evident that CL and
gestational age of 1st short cervix were both
significant predictors of PTB < 37 weeks and PTB <
34 weeks.
In the current study, patients were diagnosed with a short cervix (CL ≤
25mm) using transvaginal ultrasonography (TVCL). An obstetrician was
able to assess a patient’s risk of PTB < 37 weeks or PTB
< 34 weeks using nomograms based on her BMI before pregnancy,
autoimmune disease, prepregnancy hypertension, history of PTB, twin
pregnancy or singleton pregnancy, mode of fertilization, frequency of
pregnancy, gestational age, and CL. TVCL screening at 17-23 weeks allows
efficient interventions including potential cost-effectiveness of
screening with limited harm28. A combination of TVCL
screening and our nomogram predictions during the second trimester may
be more helpful to an obstetrician’s decision regarding intervention. A
routine measure of cervical length through transvaginal sonographs is
controversial and may not be suitable. However, our hospital’s protocol
for cervical length measurement (as described in the methods of this
report) is recommended during the first and third trimesters. This
protocol, together with our nomograms could help obstetricians discover
high-risk populations of preterm births during this period thereby
allowing drug therapy or monitoring of changes in the cervical length
for prevention of preterm births.
In the current study, we did not analyze CL changes over time although
has been hypothesized to be related to PTB risk. A previous study found
that changes in transvaginal sonographic CL over time was not a
clinically useful test to predict PTB in women29.
Therefore, a continuous assessment of CL after our nomogram prediction
is recommended in order to improve the sensitivity of prediction. Some
factors including gestational diabetes mellitus, polyhydramnios, and
gestational hypertension have been implicated in PTB risk although the
current study did not find them to be significant predictors of overall
PTB16. Future studies may benefit from a prospective
cohort design that allows researchers to collect more detailed
information on characteristics of study subjects and to analyze the
subtypes of PTB. Moreover, multiple-center studies are further suggested
in the future to improve the accuracy of nomogram prediction.