Effect of transabdominal versus transvaginal cerclage on preterm birth and neonatal outcomes among patients with a history of cervical insufficiency.

After adjusting for potential sources of bias, we found that patients who received a TAC in our study population had an estimated mean 13-percentage point reduction in absolute risk of early preterm birth compared with patients who received a TVC (RD= −13.1%, 95% CI: −26.6%, −0.5%). The estimated number needed to treat is 8, meaning that for every 8 eligible patients who receive a TAC, 1 additional early preterm delivery would be prevented. This effect was stable across multiple sensitivity analyses and among patients with and without history of failed TVC. This study differs from the MAVRIC trial in several ways. First, MAVRIC was a prospective randomized trial conducted among 111 patients with history of ≥1 sPTB or miscarriage between 14 and 28 weeks with a TVC in situ . By contrast, this observational cohort study of 188 patients included those who had history of ≥1 deliveries with a TVC in situ up to 34 weeks as well as those with history of ≥1 deliveries before 28 weeks with or without a TVC in situ . Compared to the MAVRIC trial population, patients in this study had fewer risk factors, on average, for recurrent sPTB or late miscarriage. For instance, 24% of patients in our cohort had ≥2 prior second trimester miscarriages (13% in the TVC group, 37% in the TAC group) compared to 95% overall in MAVRIC, and patients had a median of 1 prior sPTB or fetal loss before 28 weeks compared with a mean of 3 in MAVRIC. Additionally, all patients in MAVRIC had history of failed TVC. In our study, 26 patients who received TACs (30%) did not have history of failed TVC. Second, in MAVRIC, all TACs were placed via open laparotomy, while in this study all but one TAC was placed via laparoscopy. A recent meta-analysis of 83 studies including both laparoscopic and open TACs (comprising 3,398 patients) found comparable fetal outcomes across approaches but fewer wound infections and shorter hospital stays for TACs placed laparoscopically pre-conception, as well as reduced blood loss and procedure-related fetal loss for those placed laparoscopically during pregnancy, compared to open TACs. 26 While we are unable to directly compare outcomes from the two approaches, our data reinforce existing literature on the safety and efficacy of laparoscopic TACs for this patient population. 14 , 16 , 27 Third, the primary outcome in the MAVRIC trial was sPTB before 32 weeks. The primary outcome in this study was spontaneous delivery (preterm birth or miscarriage) before 34 weeks. In MAVRIC, TAC resulted in a 77% reduction in the risk of sPTB before 32 weeks compared to TVC. In this study, the effect estimate was more modest on both the relative scale (RR 0.30 versus 0.23 in MAVRIC) and absolute scale (RD −13% versus −25% in MAVRIC). A sensitivity analysis using a parallel outcome definition to MAVRIC yielded similar findings. The smaller magnitude of effect in our study is reflective of the lower overall incidence of the outcome in our study population. Overall, these results confirm the trial’s findings of TAC efficacy in an obstetric patient population with a lower average risk profile than the trial participants. Fourth, a primary purpose of cerclage is to improve neonatal outcomes through a reduction in prematurity-related complications. 28 - 31 Expanding beyond MAVRIC, we report neonatal outcomes through 1 month of life. Among livebirths in our cohort, Apgar scores and birthweight were comparable across cerclage types. However, we observed a higher unadjusted risk of NICU admission in the TAC group. Interestingly, a larger proportion of infants admitted to the NICU in the TVC group were born early preterm (47% versus 18% in the TAC group, respectively) and the median length of stay was longer (7 days versus 5 days, respectively). These data, coupled with the fact that prematurity was the most common admission diagnosis in the TVC group, suggest that the burden of NICU admission in the TVC group was likely disproportionately related to early prematurity. Meanwhile, in the TAC group, NICU admission seemed to be predominantly related to mandatory Cesarean delivery. This is supported by the fact that respiratory distress was the most common cause of admission in the TAC group; respiratory disorders of the newborn are known complications of scheduled C-sections. 32 It is important to recognize that these descriptive statistics do not support or refute a potential causal effect of cerclage type on neonatal outcomes, which would require adjustment for incidence of live birth. While we initially sought to perform an adjusted analysis to estimate the effect of TAC versus TVC on NICU admission, we were unable to do so due to dataset limitations. Additionally, MAVRIC reported a protective effect of TAC on fetal loss. We did not observe a meaningful difference between groups in the unadjusted analysis and estimating the adjusted effect on isolated fetal loss was not possible due to sample size. Finally, in this study, surgical placement complications were rare (1.6%), adding to existing literature demonstrating the safety of the laparoscopic approach. 14 - 16 However, these included one uterine perforation in the TAC group. By comparison, in MAVRIC, all predefined serious adverse events (N=6) occurred in the TVC group (cervical tears, cardiomyopathy, ICU admission for sepsis). However, the definition of surgical complications and serious adverse events differed between studies. Additional considerations relevant for clinicians and patients are the short- and long-term risks associated with mandatory Cesarean delivery. 33 - 35 We observed a similar incidence of delivery complications across groups, but these included two uterine dehiscences and one uterine rupture in the TAC group. This study suggests a more modest superiority of primarily laparoscopic TAC relative to TVC compared to MAVRIC. As discussed above, our study population was comprised of patients who, on average, had fewer risk factors for recurrent sPTB compared to the MAVRIC trial participants. These data may help guide clinical decision-making for patients who are considering laparoscopic TACs but who have fewer risk factors for recurrent sPTB or miscarriage compared to MAVRIC participants. Key to this decision-making is the risk of complications. Although serious complications related to cerclage placement and delivery were rare in the TAC group, they were significant, including uterine perforation, dehiscence, and rupture (although the degree to which the delivery events were related to cerclage is uncertain). Of note, the TACs in this study were placed by experienced surgeons at a highly specialized clinical center. The observed incidence of complications may not generalize to settings where TAC is performed less frequently. More data on the efficacy and short- and long-term safety of TAC relative to TVC from a broader sample of clinical centers is needed before guidance can be given on expanding use in women with lower average risk for recurrent sPTB than the MAVRIC trial population. Currently, ACOG and SMFM guidelines recommend TAC for patients who have either failed a TVC previously or who have cervical contraindications to TVC placement. 8 , 9 In our institution, patients are sometimes offered TACs without history of failed TVC if they have other strong indications for placement. All women are counseled extensively on the risks and benefits of the procedure via a shared decision-making process. Beyond history of sPTB or fetal loss before 28 weeks due to cervical insufficiency, additional considerations for placement in this cohort included multiple prior preterm deliveries or losses, history of cervical surgery or trauma, concurrent planned uterine surgery at the time of placement, and use of ART. While the results of this study cannot conclusively determine the efficacy of this practice due to small sample size, we found a similar effect size of TAC relative to TVC among patients with and without history of failed TVC in this study. Although we acknowledge the decision to offer TAC in these patients may be controversial, we feel it is important to share data on efficacy of the procedure relative to TVC in a population who would not have been eligible for MAVRIC. These results must be weighed against the risks of the procedure via an individualized, patient-centered decision-making process for determining whether a patient should be offered a TAC. Until more data is available in a broader variety of clinical settings, clinicians should continue to follow established guidelines for TAC placement. Further research in larger datasets pooled across multiple institutions, or ideally prospective studies, are needed to conclusively determine whether the potential benefits of TAC outweigh the risks of abdominal surgery and Cesarean delivery among patients with fewer risk factors for recurrent sPTB or miscarriage compared to patients in MAVRIC. In particular, more data is needed to determine the safety and comparative effectiveness of TAC relative to TVC among patients without history of failed TVC but with other strong indications. While a randomized trial would be useful, such a study may be difficult to perform, as demonstrated by MAVRIC. These studies should additionally consider neonatal outcomes as primary outcomes. We observed an unadjusted increase in NICU admission risk among infants born via scheduled C-section before 37 weeks in the TAC group; future studies should consider the optimal timing of delivery for TAC patients, taking into account the risk of labor with a TAC in situ . Finally, although TAC is a more resource-intensive procedure compared with TVC, they may remain in place for multiple pregnancies. Future studies could consider the risks, benefits, and cost-effectiveness of TAC placement across the reproductive lifespan. Strengths of this study include a relatively large sample size of TAC patients, application of a robust causal inference framework, and description of neonatal outcomes through 28 days of life. There are several limitations. First, the small sample size resulted in wide confidence intervals for the primary effect and limited interpretation of subgroup effects. Second, we used a composite outcome to avoid bias due to competing events. Although we expected no effect of cerclage type on causes of iatrogenic delivery (e.g., preeclampsia), there were 2 iatrogenic early preterm deliveries in our sample and both occurred in the TVC arm. As these accounted for just 8% of early preterm deliveries, their impact was likely minimal. Third, due to sample size, we could not separately estimate the effect of TAC versus high and low TVC. In MAVRIC, effect estimates were similar for TAC compared to each of these interventions. Finally, our quaternary care center is a highly specialized center with a predominantly high-risk patient population. However, given that patients who are eligible for cerclage are inherently high-risk patients, and the procedures are usually performed only at specialized care centers, we believe our results generalize to the patient population these interventions are intended to serve. Finally, we make several strong assumptions for causal interpretation of our effect estimates (see Supplemental Methods ); however, the primary result was robust to several sensitivity analyses. Laparoscopic transabdominal cerclage is effective compared to transvaginal cerclage for the prevention of early preterm delivery in an obstetric population with fewer risk factors, on average, for recurrent sPTB compared to the MAVRIC trial patient population. Subgroup analyses among those without history of failed TVC suggest that there may be cases that fall outside of the strict ACOG/SMFM criteria for TAC placement who might stand to benefit from consideration of the procedure. As a more modest absolute risk difference was observed in this study relative to the trial, these results must be considered against the risks of abdominal surgery and Cesarean delivery for determining which patients should be offered the procedure via a shared decision-making process.

After applying eligibility criteria, the cohort consisted of 188 patients: 87 (46%) received a TAC and 101 (54%) received a TVC ( Figure 2 ). The mean age at cerclage placement was 34±5 years. The majority of the cohort (N=185, 98%) had ≥1 prior spontaneous deliveries before 28 weeks; the remaining 3 patients had ≥1 prior spontaneous deliveries before 34 weeks with a TVC in situ . Sixty-five patients (35%) had ≥2 prior spontaneous deliveries before 28 weeks and 84 patients (45%) had history of at least one failed TVC ( Table 1 ). Six patients (3.1%) were lost to follow-up before 34 weeks of gestation: two in the TAC group (2.3%) and four in the TVC group (4.0%). Patients who were lost to follow-up were similar across baseline characteristics compared to those who were not lost to follow-up ( Supplemental Table 4 ). Table 1 shows the distribution of baseline characteristics stratified by cerclage type. Compared to patients who received a TVC, patients who received a TAC were older (35±5 versus 32±6 years, respectively) and had more prior spontaneous deliveries between 14 and 28 weeks (49% versus 22% with ≥2, respectively). Table 2 shows characteristics of cerclage placement. 18 TACs (21%) were placed during pregnancy; the remainder were placed preconceptionally. All but one (99%) were placed laparoscopically. Procedural complications occurred in 1% of TVC placements and 2% of TAC placements; in the TAC group, these included one uterine perforations and one uterine vein laceration/broad ligament arterial bleed. Seventy percent of the TAC group and 23% of the TVC group had history of ≥ 1 failed TVC. Supplemental Table 3 shows characteristics of the 26 patients who received a TAC without history of a failed TVC. Four patients (15%) had ≥2 prior spontaneous deliveries <28 weeks in the setting of cervical insufficiency. Of the remaining 22 patients with 1 prior spontaneous delivery <28 weeks, 100% had at least one additional reason why TAC placement was considered by clinicians, defined as concurrent planned uterine surgery (45%), history of cervical surgery or trauma (73%), or assisted reproductive technology in index pregnancy (45%; 4 patients had history of multi-gestation loss before 28 weeks). Twelve patients (55%) had ≥2 additional reasons. Table 3 shows pregnancy, delivery, and neonatal outcomes among those with follow-up through delivery (N=182). Risk factors for early preterm delivery are shown in Supplemental Table 5 . Early preterm delivery (<34 0/7 weeks, including fetal losses) occurred in 9.4% of the TAC group (95% CI: 3.2%, 16%) and 15% of the TVC group (95% CI: 8.3%, 23%). There were seven fetal losses (3.8%): four in the TVC group (4.1%) and three in the TAC group (3.5%), and six of these occurred in the second trimester. The median gestational age at delivery in the TAC group was 37.1 weeks (IQR: 36.4, 37.7 weeks) and in the TVC group was 38.1 weeks (IQR: 36.4, 39.1 weeks), reflecting scheduled Cesarean delivery of TAC patients near term (median 37.5 weeks). This was also reflected in the GA distributions across groups ( Figure 3 ): in the TVC group, more infants were born very preterm and later term compared to the TAC group, where the majority were born between 35-37 weeks. In the adjusted analysis, the estimated risk of early preterm delivery for the TAC group was 5.5% (95% CI: 2.0%, 9.4%) and for the TVC group was 18.7% (95% CI: 6.8%, 31.4%), RD = −13.1% (95% CI: −26.6%, −0.5%), RR= 0.30 (95% CI: 0.10, 0.94) ( Table 4 , Figure 4 ). The estimated effect of TAC versus TVC for early preterm delivery was similar among patients without history of failed TVC [RD = −12.6% (95% CI: −21.6%, −4.1%)] and among patients with history of ≥1 failed TVC [RD = −13.8% (95% CI: −43.5%, 10.5%)]. The estimated effect was also similar using a comparable outcome definition to MAVRIC, delivery before 32 6/7 weeks (RD = −13.6%, 95% CI: −25.9%, −0.9%). Two patients in the TAC group (2.5%) experienced uterine dehiscence and one patient (1.2%) experienced uterine rupture; 2 of these events occurred at sites of uterine scars from prior cesarean sections. In the TAC group, 17 patients had Cesarean sections before their scheduled delivery date: 5 occurred before 34 0/7 weeks of gestation and 12 occurred between 34 0/7 and 36 6/7 weeks of gestation. All occurred in the setting of preterm contractions and other signs of labor necessitating early delivery. Significant complications within 1 week of delivery occurred in 21% of the TAC group and 27% of the TVC group ( Table 3 ). The most common delivery complication in both groups was blood loss >1 liter (11% overall). Two patients in the TAC group (2.5%) experienced uterine dehiscence and one patient (1.2%) experienced uterine rupture. Two patients in the TVC group (1.1%) experienced cervical lacerations. Additionally, 10 patients requested removal of TAC at delivery; eight out of a desire for sterilization and two because of complications ( Table 2 ). Among livebirths with available neonatal records (N=173), Apgar scores and birthweight were comparable across cerclage groups ( Table 3 ). Higher proportions of neonates born to mothers with TACs relative to TVCs were triaged in the NICU (46% vs. 34%, respectively), admitted to the NICU (35% vs. 23%) and had NICU admissions ≥5 days (21% vs. 14%). However, the median number of days of NICU admission was higher in the TVC group (5 vs. 7 days, respectively). This may reflect differences in the gestational ages at which infants were admitted; 10 out of 21 infants (48%) admitted to the NICU in the TVC group were born before 34 weeks, compared with 5 out of 28 infants (18%) in the TAC group. There was one neonatal death: an infant in the TVC group who was liveborn at 22 4/6 weeks and died within an hour of life due to complications of extreme prematurity. The most common diagnosis for NICU admission in the TVC group was prematurity (76%) and in the TAC group was respiratory distress (71%). In both groups, after these two diagnoses, hypoglycemia was the third most common diagnosis (20% overall). Supplemental Table 5 shows all NICU admission diagnoses. NICU admission in the TAC group was higher among those born via unscheduled C-section (N=15/17, 88%) than those born via scheduled C-section (N=13/64, 20%). Among those delivered via scheduled C-section, NICU admission was higher among infants delivered before 37 weeks (N=5/12, 42%) compared with those delivered on or after 37 weeks (N=8/52, 15%) ( Supplemental Figure 2 ). The effect estimate was robust to several sensitivity analyses; there was minimal impact of excluding preconception TACs, including censoring weights, or including additional variables in the treatment weight model ( Table 4 , Supplemental Figure 3 ).

We designed and analyzed this observational cohort study using the “target trial” framework. 17 - 21 We refer the reader to Supplemental Table 1 for an expanded discussion of methods and how we attempted to align design and analysis of our observational study with a hypothetical, corresponding randomized trial (a “target trial”) to minimize bias. We identified two cohorts of singleton pregnancies at the Brigham and Women’s Hospital (BWH) in Boston, MA between January 1, 2001 and February 28, 2021. Patients were identified within the Research Patient Data Registry using Current Procedural Technology (CPT ® ) and International Classification of Diseases (ICD) codes ( Supplemental Table 2 ). All patients with a code for TAC and a sample of patients with a code for TVC who were members of an existing BWH cohort, LIFECODES, were considered for inclusion. LIFECODES is an ongoing prospective pregnancy cohort and specimen bank, described further elsewhere. 22 , 23 Patients are recruited by consent during their initial consultation with a maternal-fetal medicine specialist at BWH and can therefore be considered a representative sample of the same population of high-risk obstetric patients from which the TAC patients were identified. The use of these data for research was approved by the Mass General Brigham (MGB) Institutional Review Board. Inclusion criteria as well as treatment, outcome, and baseline characteristics were collected for eligible patients by review of medical records using REDCap electronic data capture tools hosted at MGB. 18 , 19 All included patients had a cerclage placed in the index pregnancy and were ≥18 years old at the time of cerclage placement. Patients were included if they had 1) history of ≥1 prior spontaneous deliveries (birth or fetal loss) between 13 6/7 and 28 0/7 weeks in the setting of documented or presumed cervical insufficiency, and/or 2) history of ≥1 failed TVC, defined as spontaneous delivery in the setting of cervical insufficiency between 13 6/7 and 34 0/7 weeks of pregnancy with a TVC in situ . Patients were excluded if they had a contraindication to transvaginal or transabdominal cerclage, including history of trachelectomy or Mullerian anomaly. We included the first pregnancy for which patients met eligibility criteria. TVCs (either MacDonald or Shirodkar type) were placed before 16 0/7 weeks of gestation in the absence of cervical changes detected by manual or ultrasonographic examination. Removal of TVC was planned for ≥35 0/7 weeks of gestation, unless obstetric indications necessitated earlier removal. TACs were placed via laparoscopy or open laparotomy (surgical technique as described elsewhere) 16 either before conception or during pregnancy before 16 0/7 weeks of gestation in the absence of cervical changes detected by manual or ultrasonographic examination. Delivery by Cesarean section was planned for ≥37 0/7 weeks, unless obstetric indications required earlier delivery. The cerclage could either remain in place at delivery (for subsequent pregnancies) or be removed at provider or patient discretion. All patients who were being considered for or seeking TAC met with at least two subspecialist obstetrician-gynecologists (maternal fetal medicine and gynecologic surgery) in separate appointments to discuss risks (including the need for Cesarean delivery with its associated and future risks), benefits, and indications for the procedure. Each patient therefore had a minimum of two hours of one-on-one consultation, and the decision to offer TAC required independent consensus of the providers. At our institution, women are considered for possible TAC if they have history of second trimester spontaneous birth or fetal loss in the setting of cervical insufficiency and/or have a history of failed TVC (defined as above). Some women are considered for TAC without a history of failed TVC; these women typically have other compelling indications for the procedure (see Supplemental Table 3 ). The decision about whether to proceed with TAC versus TVC is made via a shared decision-making process that values biomedical principles of non-maleficence, beneficence, and patient autonomy. In situations where patients independently requested TAC placement, if either the consulting MFM or the performing surgeon did not believe that the request was clinically appropriate, then a supportive recommendation was not provided and the procedure was not performed. The primary outcome was early preterm delivery, a composite outcome defined as spontaneous delivery, iatrogenic delivery, or fetal loss on or after 16 0/7 and before 34 0/7 weeks of gestation. Spontaneous delivery was defined as delivery of a live or deceased infant in the setting of labor, spontaneous rupture of membranes, or cervical incompetence. Iatrogenic delivery was defined as induced labor or Cesarean section due to maternal or fetal causes other than those defined for spontaneous delivery. Fetal loss was defined as delivery of a fetus showing no signs of life, including due to fetal demise, premature rupture of membranes, cervical insufficiency, stillbirth, or miscarriage. Although we were primarily interested in the effect of cerclage type on spontaneous delivery and fetal loss, we chose to use a composite outcome because iatrogenic delivery is a competing event for these outcomes. Other methods exist to correct for bias due to competing events, 26 , 27 however these require strong assumptions or are difficult to interpret clinically in the context of preconception exposures. 24 Under the assumption of no effect of cerclage type on iatrogenic delivery, the effect of treatment on the composite outcome will tend to be a conservative estimate for the effect on spontaneous birth/fetal loss. Secondary pregnancy outcomes included composite preterm delivery after 16 0/7 and before 32 6/7 weeks (for comparison with MAVRIC), gestational age at delivery, fetal loss after 16 0/7 weeks, and fetal loss after 16 0/7 and before 24 0/7 weeks. Obstetric outcomes included complications of cerclage placement, Cesarean section, and delivery complications within 1 week of delivery. Neonatal outcomes included triage and admission to the neonatal intensive care unit (NICU), duration of NICU admission, NICU admission diagnoses, neonatal death within 28 days of life, Apgar scores, and birthweight. We estimated the effect of TAC versus TVC on the primary outcome of early preterm delivery among subgroups of patients with and without history of failed TVC. We additionally characterized reasons for TAC consideration among the patients who received a TAC without history of failed TVC. For the primary outcome, patients were followed from the time of cerclage placement until delivery, fetal loss, 34 0/7 weeks of gestation, or loss to obstetric follow-up, whichever occurred first. Patients who were lost to follow-up were censored at the time of their last visit or ultrasound note with an obstetric provider for which notes were accessible. We assumed that loss to follow-up was non-informative with respect to preterm delivery. As a sensitivity analysis, we adjusted for potential sources of selection bias ( Supplemental Methods ). Patients were followed through 1 week after delivery for secondary delivery outcomes. Infants were followed through 28 days of life for neonatal outcomes. The analysis was restricted to patients who conceived and remained pregnant at 16 0/7 weeks of gestation to account for variation in timing of cerclage placement from preconception to early gestation. This variation in timing of placement could introduce selection bias due to our inability to observe patients who planned for a cerclage but failed to conceive and/or had an early miscarriage, as shown in Figure 1 . A similar restriction was performed in MAVRIC, where patients were randomized both pre- and post-conception and the analysis was restricted to those who remained pregnant at 14 weeks. We adjusted for use of ART in the index pregnancy and insurance status (as a proxy for income) as possible sources of this selection bias. See Supplemental Methods for further discussion of this potential bias and relevant assumptions. Inverse probability weights were used to adjust for baseline confounding by common causes of cerclage type and early preterm delivery and for common causes of cerclage type and pregnancy survival to 16 0/7 weeks. We considered the following baseline confounders: ≥2 prior spontaneous deliveries (birth or miscarriage) between 13 6/7 and 34 0/7 weeks, 5 history of failed TVC, insurance status, age at cerclage placement, prior cervical surgery, Black or African American race, use of assisted reproductive technology (ART) in the index pregnancy, 21 and year of delivery or loss to follow-up to account for possible time trends in preterm birth 22 (see Supplemental Figure 1 for a representative Directed Acyclic Graph, DAG 24 , 25 ). Race was included because of evidence that Black or African American women experience increased risk of preterm birth compared to non-Hispanic white women and because of racial disparities in obstetric care. 3 , 23 - 25 Because history of failed TVC was a strong indication for TAC, weights were stabilized by history of failed TVC. To estimate risk, we fit a weighted, pooled logistic marginal structural model for the outcome with a product term between cerclage type and follow-up time, a term for history of failed TVC, and a product term for history of failed TVC and cerclage type. 29 The predicted hazards from this model were used to estimate the absolute cumulative risk of early preterm delivery before 34 0/7 weeks of gestation with corresponding risk differences (RD), risk ratios (RR), and cumulative risk plots. To estimate the primary effect within the subgroups of patients with and without history of failed TVC, we used a product term in the marginal structural model as described above. Percentile-based 95% confidence intervals for all effect estimates were estimated using nonparametric bootstrapping. Data analysis was performed using R 2021.09.0 (R Foundation for Statistical Computing, Vienna, Austria). See Supplemental Methods for a full discussion of the statistical analysis. We performed the following sensitivity analyses: 1) excluding TACs placed during pregnancy; 2) adjusting for potential selection bias due to loss to follow-up using censoring weights; 3) estimating the effect on preterm delivery before 32 completed weeks of gestation for more direct comparison with MAVRIC; 4) adjusting for the number of prior spontaneous deliveries after 13 6/7 weeks and before 28 0/7 weeks as an additional baseline confounder.

Cervical cerclage is a surgical procedure whereby suture or synthetic tape is used to reinforce the cervix to prevent spontaneous preterm birth (sPTB) or late miscarriage. 1 Cerclages may be placed transvaginally (TVC) or transabdominally (TAC). According to the American College of Obstetricians and Gynecologists (ACOG), history-indicated TVCs may be placed in patients with a history of second-trimester pregnancy loss or cerclage placement in the setting of painless cervical dilation. 2 They may be considered in women with history of unexplained second-trimester delivery in the absence of labor or abruptio placentae. Data on the efficacy of history-indicated TVCs compared to no cerclage for prevention of sPTB are mixed, with two trials showing no significant difference in outcomes among women with history of preterm birth 3 , 4 and one trial showing a 4% decreased risk of delivery <33 weeks. 5 Because of the need for abdominal surgical placement and Cesarean delivery, TAC is considered potentially more morbid than TVC and is therefore reserved for higher risk patients or those with cervical contraindications to TVC. 6 - 8 The Society for Maternal Fetal Medicine (SMFM) recommends offering TAC to patients with history of spontaneous singleton delivery before 28 weeks of gestation with a history- or ultrasound-indicated TVC in situ . 9 Observational studies have suggested a protective effect of TAC compared to TVC for patients with history of failed TVC. 10 - 12 The first and only randomized trial, the Multicentre Abdominal vs. Vaginal Randomised Intervention of Cerclage (MAVRIC) trial, was published in 2020 after taking 6 years to recruit 111 patients. To be eligible for inclusion in MAVRIC, patients had to have a history of ≥1 sPTB or miscarriage between 14 and 28 weeks with a TVC in situ . MAVRIC found that TAC reduced the risk of sPTB before 32 completed weeks relative to both high TVC (from 38% to 8%) and low TVC (from 33% to 8%) (RR for low TVC 0.23, 95% CI: 0.07-0.76). 13 MAVRIC participants represent a very high risk obstetric population: included patients had an average of 3 prior sPTBs or fetal losses before 28 weeks, 95% had ≥2 prior second trimester miscarriages, and all had history of ≥1 failed TVC resulting in sPTB or fetal loss before 28 weeks. 13 In practice, TAC is sometimes offered to patients with lower risk profiles. Additionally, while all TACs in MAVRIC were placed via open laparotomy, the minimally invasive laparoscopic approach has been widely adopted. 14 - 16 It is not clear whether the strong effect sizes observed in MAVRIC are generalizable to a patient population with fewer risk factors, on average, for recurrent sPTB/fetal loss, or to laparoscopically placed TACs. Furthermore, in our practice, patients without history of failed TVC (an inclusion criterium in MAVRIC) or cervical contraindications to TVC are sometimes offered TAC, elaborated on below. There is a need for high-quality data on the magnitude of effect in these patients to inform decision-making in the context of known risks of abdominal surgery and Cesarean delivery. The impact of TAC on neonatal outcomes is additionally not well described. To inform these questions, we estimated the effect of TAC versus TVC on the risk of early preterm delivery among patients with a history of spontaneous preterm birth or fetal loss before 28 weeks and/or history of a failed TVC before 34 weeks. We additionally estimated the effect in a subgroup without history of failed TVC. We describe risks of relevant surgical, obstetric, and neonatal outcomes.