Keywords
preterm delivery - cervical cerclage - cervical length - cerclage height - proximal
residual length - history indicated cerclage - physical exam indicated cerclage -
ultrasound indicated cerclage
Preterm birth, defined as a delivery between 20 0/7 and 36 6/7 weeks, is the leading cause of neonatal morbidity and mortality in the United States.[1] The rate of extreme prematurity in the U.S. has remained unchanged despite a decrease
in late preterm deliveries.[2]
There are multiple etiologies of spontaneous preterm delivery, including preterm labor,
preterm premature rupture of membranes, and cervical insufficiency. The latter, whose
pathophysiology is still poorly understood, is the inability of the uterine cervix
to retain a pregnancy in the absence of clinical signs and symptoms of labor; and
its diagnosis is made based on a history of painless cervical dilation after the first
trimester with subsequent expulsion of the pregnancy in the second trimester.[3]
Several nonsurgical and surgical modalities have been proposed for treatment of cervical
insufficiency. Cerclage placement, being one of them, may be indicated based on a
patient's history (history of two or more second trimester losses or history of cerclage
placement), physical examination findings (open cervix on physical exam), or a history
of preterm birth and findings of short cervix on ultrasound (less than 2.5 cm).[3]
The original descriptions for cerclage placement emphasized placing the suture “as
high as possible to approximate the level of the internal cervical os.”[4] Multiple studies have suggested that differences in cerclage position may contribute
to the variability observed in cerclage efficacy.[5] One example is the work of Guzman et al who performed a precerclage measurement
and a postcerclage measurement and then found and association between an upper cervical
length (CL) < 10 mm and delivery before 36 weeks of gestation; however, unlike our
study, only patients with physical exam indicated cerclage were included.[6] Other authors have suggested that in the case of ultrasound indicated cerclage,
differences in cerclage position may influence the time from cerclage placement until
delivery.[7]
No prior research has looked into the relationship of cerclage height (cerH: length
from the cerclage to the external cervical os), proximal residual length (PRL: length
of cervix that remains closed proximal to the cerclage), and CL (distance from external
to internal cervical os) after cerclage placement ([Fig. 1]), to gestational age at delivery and rate of preterm delivery (< 34 weeks) for all
types of cerclages segregated based on indication.[8] With this in mind, we studied the relationship of CL, cerH, and PRL to the gestational
age at delivery and rate of delivery at less than 34 weeks for cerclages placed by
any of American College of Obstetricians and Gynecologists' (ACOG) recommended indications
(history, physical exam, and ultrasound).
Fig. 1 The diagram of CL, cerH and PRL. CL, cervical length; cerH, cerclage height; PRL,
proximal residual length.
By studying the effect of cerclage position in all indicated cerclages separated by
indication, we expected to differentiate if cerclage position would affect efficacy
in some but not all cerclages.
Methods
We performed a historical cohort study that included all patients who had a cerclage
placed at Maimonides Medical Center (MMC), and who were subsequently followed at the
hospital's perinatal diagnostic center, from 2006 to 2016. Exclusion criteria were
age younger than 18 years, multiple gestation, known uterine anomaly at the time of
cerclage placement, history of cervical excision procedure, iatrogenic preterm delivery
(< 37 weeks), cerclage placement for indications other than those recommended by ACOG,
as well as a lack of delivery information and/or pre- and postcerclage cervical measurements.
All data were downloaded from the hospital's ultrasound and coding program (ASOBGYN
[AS software for Obstetrics and Gynecology]). Every chart coded under “cerclage” or
“short cervix” was reviewed (n = 1,167). Of these, 379 had a cerclage placed during the index pregnancy. Among those,
a total of 267 patients were excluded either because they lacked sonographic records
before or after placement (n = 211), delivery information (n = 14), such as gestational age at delivery or indications for delivery (e.g., iatrogenic
or spontaneous), had the cerclages placed for an indication other than the ones approved
by ACOG guidelines (n = 38), or had incomplete demographic information (n = 4). The remaining 112 patients were used for analysis ([Fig. 1]).
As per hospital protocol, routine CL screening is performed in every patient at the
time of anatomy ultrasound to diagnose patients with a short cervix (CL of less than
or equal to 2.5 cm). For patients at increased risk of preterm delivery (history of
preterm delivery, history of second trimester loss, or history of cerclage placement
in prior pregnancy) screening starts at 14 to 16 weeks of gestation. We considered
the CL measurements prior to (and in the case of multiple reads, the most proximate
to) the procedure the precerclage CL, and we used those scans to confirm the indication
for placement. We performed postcerclage measurements between 1 and 4 weeks after
the cerclage placement.
We used the standardized method of obtaining CL as described by Berghella[8] and as recommended by the cervical length education and review (CLEAR) program.
The images collected between 2006 and 2016 at the perinatal unit and saved in ASOBGYN
software were used for the measurement of cerH, CL, and PRL ([Fig. 2]). Given differences in the number of sutures (n = one or two) placed, the stitch closest to the internal os was used for measurement
of the previously defined lengths ([Fig. 3]).
Fig. 2 Total charts reviewed and total number of charts used for analysis. ACOG, American
College of Obstetrics and Gynecologists; Hx, history; PE, physical exam; US, ultrasound.
Fig. 3 PRL in blue, cerH in yellow, and red arrows are pointing toward the stitches for
a 2 knot cerclage. cerH, cerclage height; PRL, proximal residual length;
The indications for cerclage were history of cervical insufficiency (history of one
or more second trimester losses or history of cerclage placement; n = 63), physical examination findings (open cervix on physical exam; n = 20) or a history of preterm birth (at less than 34 weeks of gestation) in conjunction
with findings of short cervix on ultrasound (less than 2.5 cm; n = 29).
In the initial analysis cerH, CL, and PRL were looked at as predictors of delivery
time. However, it was found that cerH and PRL were too tightly linked to be analyzed
separately. A general linear model (GLM) was used to predict power-transformed age
at delivery (AD) from cerH, CL, and indication for cerclage (IC); interactions among
these terms were investigated, as were the possible utility of polynomial terms in
cerH and CL. Subanalyses were then conducted and stratified by the indication for
cerclage. Model residuals were inspected for skew and for outliers. Adjusted R
2 values (i.e., corrected for the over-fitting problem) are reported as measures of
strength of association. Significance level was set at 0.05. In a secondary analysis,
logistic regression was used to predict delivery ≤ 34 weeks, in the same manner as
above; the Nagelkerke's pseudo R
2 statistic is reported. The Hosmer–Lemeshow lack of fit test was applied.
Results
One hundred and twelve patients were included in the final analysis. These were broken
down by indication for cerclage: history indicated (n = 63), physical exam indicated (n = 20), and ultrasound indicated (n = 29; [Fig. 2]). In the GLM analysis including all cerclage indication types, a significant CL
by indication for cerclage (p = 0.034) was detected; there was also a significant indication for cerclage main
effect (p = 0.003).
The cerH by indication for cerclage did not reach statistical significance (p = 0.090). However, when stratified by the different indications, the effect of cerH
on age at delivery was apparent for history indicated (adjusted R
2 = 0.18, p < 0.001) and physical exam indicated (adjusted R
2 = 0.43, p = 0.004) cerclages but not for ultrasound indicated cerclages (adjusted R
2 =0.08, p = 0.206). Though there was statistical significance for the history indicated group,
the adjusted R
2 value suggests that the strength of the association is not high. For cases, where
the indication for cerclage is a physical exam, the evidence (based on a small subsample)
that age at delivery is positively associated with cerH was more robust but still
modest. For ultrasound indicated cerclages, the association is much weaker. CL was
not a significant predictor in any model. [Fig. 4] shows for each indication, the curvilinear regression function (adjusted for mean
CL for cases with that indication) for each indication, overlaid over raw data points.
Fig. 4 The curvilinear regression function (adjusted for mean cervical length for cases
with that indication), overlaid over raw data points for Physical exam indicated cerclages.
Logistic regression predicting delivery ≤ 34 weeks (n = 29 early deliveries) produced similar results; the CerH effect was significant
for history indicated (pseudo-R
2 = 0.31, p = 0.002) and physical exam indicated (pseudo-R
2 = 0.71, p = 0.024), but not ultrasound indicated cerclages (pseudo-R
2 = 0.34, p = 0.090; [Figs. 5] and [6]).
Fig. 5 The curvilinear regression function (adjusted for mean cervical length for cases
with that indication), overlaid over raw data points for History indicated cerclages.
Fig. 6 The curvilinear regression function (adjusted for mean cervical length for cases
with that indication), overlaid over raw data points for Ultrasoud indicated cerclages.
There is modest evidence (based on a small subsample) that age at delivery is positively
associated with cerH for cases where the indication for cerclage is physical exam.
Though there is also statistical significance for the history indicated group, it
is not clinically significant. For sonogram indicated the association is (if nonzero),
is possibly much weaker.
Discussion
We found that when a cerclage is placed for a dilated cervix or a poor obstetrical
history, the location of the stitch may influence the age at which the fetus is delivered.
Specifically, the higher the cerclage stitch is placed, the more advanced the gestational
age at delivery. This is statistically significant for both groups but likely only
clinically significant in the group with the dilated cervix. We did not see similar
associations for cerclages placed for ultrasound indications.
Cervical cerclage is a technique used to prolong pregnancy in patients with a clinical
diagnosis of cervical insufficiency. Prior studies have compared the different techniques
used for cerclage placement and found an increase in cerH of approximately 2.7 mm
with Shirodkar versus McDonald.[5]
[6] However, it was felt at the time that this slight gain did not justify exposing
the patient to increased risks given no difference in prognosis or delivery time between
them.[9] Other studies have assessed the relationship between cerclage position as seen on
transvaginal sonogram and preterm delivery and have found no relationship[10]; and some more recent investigations have studied the relationship between cerH
and preterm delivery in ultrasound-indicated cerclages only, and found no relationship.[11]
Thus, there is still uncertainty about whether height influences outcomes. To explore
this issue further, we performed a retrospective cohort study of cerclages placed
at one institution (MMC). To our knowledge, (using search Google Scholar and PubMed
as search engines, with search terms, “age at delivery” and “cerclage height”) this
is the first study analyzing cerclages placed for all the ACOG recognized indications,
and addressing the relationship of cerclage location with gestational age at delivery.
Prior research has questioned the utility of history indicated cerclage[12] by suggesting that up to 50% of patients that receive a history indicated cerclage
will deliver after 37 weeks without any intervention. Thus, these patients may represent
a mix of women with and without a true indication for cerclage. Our results show that
for some women with history indicated cerclage, position may be directly and positively
related to gestational age at delivery, suggesting that at least for some, the procedure,
if done properly can have a salutary effect.
Our data also suggest that in patients with physical exam indicated cerclages, trying
to achieve a greater cerH (by, for example, retrograde filling the bladder, preoperative
amnioreduction, placement of an intrauterine Foley's catheter) might be associated
with a prolongation of pregnancy. This could be clinically significant, and may indicate
the need for more aggressive techniques for higher stitch placement.
We, in agreement with previous reports,[11] did not find an association in patients with ultrasound indicated cerclage. One
possible explanation for this finding is that the small sample size resulted in a
bias toward the null. It is possible that a sufficiently large subset of these patients
is not helped by cerclage, masking our ability to show a benefit among the others.
Limitation
We do need to acknowledge some limitations in our study. There was a variation in
time between stitch placement and subsequent ultrasound (1–4 weeks). Though this variation
may have made the PRL measurement unreliable as a predictor since that measurement
can change over time as a cervix shortens, it is unlikely that CerH and CL were influenced,
since these measurements are distal to a static stitch. Interestingly, our data shows
CL is not a predictor of age at delivery. This may support the practice of not checking
routine CL measurements after cerclage placement.
Conclusion
In conclusion, in women with a cerclage performed for a dilated cervix, our results
suggest that obstetricians should try to place the stitch as high as is safely possible.
That may also be the case in women with a history indicated cerclage. The advantage
of a higher stich is less clear in other settings.