Background
Precocious cervical ripening, as defined by cervical shortening on transvaginal sonography, has prompted a broad evaluation of secondary strategies (such as cerclage, vaginal progesterone, or a cervical pessary) to prevent preterm delivery. However, there is still a lack of direct comparisons between individual treatments or their combinations.
Objective
We sought to compare at-risk patients and screening patients who had been treated with cervical pessary alone with patients who had been treated with pessary plus vaginal progesterone.
Study Design
This is a pre- and postintervention cohort study from a preterm labor clinic where placement of a cervical pessary has been the standard treatment since 2008 for at-risk women defined by (1) a history of spontaneous preterm birth at <37 weeks of gestation, (2) conization, or (3) a cerclage because of a previous short cervical length of <3rd percentile and, additionally, with a cervical length of <10th percentile in the ongoing pregnancy. Patients who did not meet the criteria for the “at risk” group, but who had a cervical length of <3rd percentile comprised the screening group. From July 2011 onward, vaginal progesterone (200 mg, suppositories) was prescribed in addition to the pessary. Both at-risk patients (n = 55) and screening patients (n = 51) were treated at the time of diagnosis. The primary outcome was the rate of preterm deliveries at <34 weeks of gestation. Secondary outcomes included deliveries at <28, <32, and <37 weeks of gestation, the days from start of therapy until delivery, a composite index of neonatal outcome, and the number of days in the neonatal intensive care unit. Primary and secondary outcomes were compared between groups with the use of multivariable models to adjust for possible confounders.
Results
Delivery at <34 weeks of gestation occurred in 17 of 53 patients (32.1%) who were treated with pessary plus progesterone, compared with 13 of 53 patients (24.5%) who were treated with pessary alone ( P = .57). Similarly, there was no difference in the rate of preterm delivery at <28, <32, or <37 weeks of gestation. The composite poor neonatal outcome was 15.1% in the pessary group vs 18.9% in the combined group ( P = .96). The mean duration of stay in the neonatal intensive care unit was 46.5 days (range, 9-130 days) in the combined vs 52.0 days (range, 3-151 days) in the pessary group ( P < .001).
Conclusion
In this cohort study, treatment of precocious cervical ripening with cervical pessary plus vaginal progesterone did not reduce the rates of preterm delivery at <28, <32, <34, or <37 weeks of gestation compared with pessary alone. The neonatal intensive care use was shorter in patients who received additional vaginal progesterone, although there was no difference in composite poor neonatal outcome. These preliminary results may serve as a pilot for future trials and provide a basis for treatment until larger trials are completed.
Preterm birth (PTB) is the major cause of perinatal morbidity and death in high resourced countries, with a prevalence of 5.3% (Latvia) up to 12% (United States). Primary prevention rarely has been shown to be effective apart from a few studies that sought to reduce either physical stress by public health interventions or smoking by smoke-free legislation. After the introduction of transvaginal sonography (TVS) as a tool to identify and follow women who are at risk for PTB, secondary preventive concepts such as progesterone or a cervical pessary were reintroduced as a treatment for women with precocious cervical ripening. In addition, the indication for a cerclage was reevaluated on the basis of TVS results.
In our preterm labor clinic, we have placed cervical pessaries in singleton pregnancies with a short cervix since October 2008. Starting in July 2011, we added the administration of vaginal progesterone based on the publication of Hassan et al hoping that the 2 approaches might provide complementary secondary preventive effects in high-risk pregnancies and pregnancies that, with screening, had a short cervical length (CL). The aims of this study were to develop intermediate information for patients whose treatment cannot await the results of large randomized controlled trials (RCTs) and to serve as a pilot for planning future trials.
Materials and Methods
A pre- and postintervention cohort study was conducted from October 2008 to December 2014 in women with singleton pregnancies at increased risk for PTB by both history and cervical shortening and a screening group without risks but who were found to have isolated cervical shortening. Within this period, 13,179 examinations were performed in 4393 patients who had been referred to our outpatient unit for various reasons that included increased risk factors for PTB. The risk group of this study population was characterized by a history of PTB at <37 weeks of gestation (n = 35), of whom 26 of 35 patients (74.3%) had a previous PTB at <32 weeks of gestation, a history of surgical conization (n = 18), a cerclage because of a previous short CL at <3rd percentile (n = 11) or a combination of cerclage and previous conization (n = 2; Table 1 ). In total, 6 patients in the pessary group and 5 patients in the combined group had both a previous PTB and a conization or cerclage, and they were considered for calculations in both categories. Patients who did not meet criteria for the “at-risk” group, but who had a CL at <3rd percentile, comprised the screening group that was detected during the second-trimester screening scan at our ultrasound unit. The gestational age of referral ranged between 12 and 27 + 1 weeks in the risk group and 17 + 1 and 27 + 6 weeks in the screening group. A perforated Arabin cerclage pessary (Dr Arabin GmbH & Co KG, Witten, Germany) was applied. The sizes were chosen according to previously published recommendations related to height and upper and lower diameters. In our study population with only singleton pregnancies, the height was always 21 mm, and the upper diameter was 32 mm, except for patients with severe funneling for whom an upper diameter of 35 mm is recommended. Dependent on the obstetric history the lower diameter was 65 mm in women without and 70 mm in women with a previous vaginal delivery. The device is CE (Conformité Européenne) approved for the prevention of preterm labor (MEDCERT 0482, certificate 10610 GB 412, 150 324). The pessary was placed when the CL was <10th percentile (eg, 34.9, 29.2, and 25.2 mm at 16, 24, and 28 weeks of gestation, respectively) in the risk group and when the CL was <3rd percentile (eg, 30.5, 23.4, and 18.7 mm at 16, 24, and 28 weeks of gestation, respectively) in screening patients ( Table 1 ). Only patients with a start of treatment at ≥12 and <28 weeks of gestation were included for further analysis. For the interval of 12-16 weeks of gestation, we extrapolated the Salomon percentiles by using the percentiles of Gramellini et al (32.8 mm for the 3rd percentile and 37.5 mm for the 10th percentile at 12 weeks of gestation). The success of pessary treatment depends on both standardized cervical sonography and the skills of the obstetrician in charge because there is a well-defined learning curve. Therefore, patients were included only if they had been diagnosed and treated by 1 experienced specialist who also followed all high-risk referrals of this study. All patients had been examined by TVS according to the Fetal Medicine Foundation in the first trimester and by the method described by Iams et al from 16 weeks of gestation onwards that was redefined by To et al. Funneling was classified as a V-, Y- or U-shaped dilation of the internal os with at least a width of 5 mm.
Variable | Risk group a | Screening group b | Total population c | ||||||
---|---|---|---|---|---|---|---|---|---|
Pessary (n = 27) | Pessary and progesterone (n = 28) | P value | Pessary (n = 26) | Pessary and progesterone (n = 25) | P value | Pessary (n = 53) | Pessary and progesterone (n = 53) | P value | |
Age, y d | 31.4 ± 6.13 | 32.7 ± 3.68 | .347 | 30.3 ± 6.74 | 31.7 ± 5.14 | .408 | 30.9 ± 6.40 | 32.2 ± 4.41 | .203 |
Body mass index, kg/m 2 d | 23.3 ± 3.67 | 22.8 ± 4.76 | .657 | 24.0 ± 3.77 | 23.6 ± 3.29 | .634 | 23.7 ± 3.70 | 23.2 ± 4.11 | .506 |
Smoker, n (%) | 3 (11.1) | 2 (7.1) | .669 | 2 (7.7) | 2 (8.0) | 1 | 5 (9.4) | 4 (7.5) | 1 |
Obstetric history, n (%) e | |||||||||
Nulliparous | 4 (14.8) | 4 (14.3) | 1 | 14 (53.8) | 18 (72.0) | .293 | 18 (34.0) | 22 (41.5) | .548 |
Obstetric history of preterm birth | |||||||||
≤32 Weeks of gestation | 14 (51.9) | 12 (42.9) | .671 | — | — | — | 14 (26.4) | 12 (22.6) | .822 |
32-37 Weeks of gestation | 5 (18.5) | 4 (14.3) | .671 | — | — | — | 5 (9.4) | 4 (7.6) | .822 |
History of surgical conization or treatment for threatening preterm birth with a cervical length of <3rd percentile by a cerclage, n (%) | |||||||||
Cervical cerclage | 5 (18.5) | 6 (21.4) | .665 | — | — | — | 5 (9.4) | 6 (11.3) | .689 |
Surgical conization | 9 (33.3) | 9 (32.1) | .665 | — | — | — | 9 (17.0) | 9 (17) | .689 |
Both, conization and cerclage | 0 | 2 (7.14) | .665 | — | — | — | 0 | 2 (3.8) | .689 |
Course of pregnancy | |||||||||
Median gestational age at start of therapy, wk + d (range) | 19 + 3 (12 + 0–27 + 1) | 19 + 0 (12 + 0–26 + 4) | .707 | 24 + 0 (17 + 1–27 + 6) | 23 + 6 (17 + 1–27 + 6) | .614 | 21 + 5 (12 + 0–27 + 6) | 22 + 2 (12 + 0–27 + 6) | .948 |
Cervical length at start of therapy, mm d | 21.3 ± 6.40 | 17.1 ± 12.0 | .113 | 15.3 ± 7.42 | 13.4 ± 6.45 | .312 | 18.4 ± 7.49 | 15.4 ± 9.91 | .078 |
Cervical length Z-score at start of therapy d | –2.49 ± 0.55 | –2.72 ± 0.96 | .283 | –2.55 ± 0.57 | –2.78 ± 0.55 | .148 | –2.52 ± 0.55 | –2.75 ± 0.79 | .088 |
Funneling at start of therapy, n (%) | 11 (40.7) | 11 (39.3) | 1 | 15 (57.7) | 17 (68.0) | .637 | 26 (49.1) | 28 (52.8) | .846 |
Tocolytics during course of pregnancy, n (%) | 4 (14.8) | 3 (10.7) | .705 | 7 (26.9) | 4 (16.0) | .543 | 11 (20.8) | 7 (13.2) | .438 |
Antenatal corticosteroids, n (%) | 11 (40.7) | 14 (50.0) | .676 | 13 (50.0) | 12 (48.0) | 1 | 24 (45.3) | 26 (49.1) | .846 |
Premature preterm rupture of membranes at <37 weeks of gestation, n (%) | 5 (18.5) | 7 (25.0) | .798 | 7 (26.9) | 4 (16.0) | .543 | 12 (22.6) | 11 (20.8) | 1 |
Gestational age at premature preterm rupture of membranes, wk + d ± SD d | 29 + 6 ± 6 + 0 | 29 + 5 ± 6 + 0 | .965 | 31 + 5 ± 3 + 0 | 34 + 0 ± 2 + 6 | .251 | 31 + 0 ± 4 + 2 | 31 + 2 ± 5 + 2 | .867 |
Mode of delivery, n (%) | |||||||||
Spontaneous | 15 (55.6) | 15 (53.6) | .787 | 12 (46.2) | 11 (44.0) | 1 | 27 (50.9) | 26 (49.1) | .882 |
Vaginal operative | 1 (3.7) | 0 | .787 | 1 (3.85) | 1 (4) | 1 | 2 (3.77) | 1 (1.89) | .882 |
Cesarean | 11 (40.7) | 13 (46.4) | .787 | 13 (50.0) | 13 (52.0) | 1 | 24 (45.3) | 26 (49.1) | .882 |
a Defined by previous preterm birth at >16 and <37 weeks of gestation, a history of surgical conization, or a previous cerclage because of a short cervix, all of which were characterized by a short cervical length at <10th percentile that was detected by transvaginal sonography
b Defined by a cervical length at <3rd percentile
c Separate for the treatment with either pessary alone or pessary combined with vaginal progesterone
e Six patients in the pessary group and 5 patients in the combined group had a previous preterm birth and a conization or cerclage; here they are considered in both categories.
Patients and involved specialists were informed about pessary insertion and possible side-effects (such as vaginal discharge), and the treatment followed a prescribed protocol. During the insertion, the pessary was squeezed, introduced longitudinally, and unfolded only in the upper fornix whereby the smaller inner ring was directed towards the cervix. The proximal part of the pessary’s dome was pushed carefully until the cervix was surrounded; the anterior part of the pessary was then pressed slightly towards the sacrum. The pessary remained until 37 weeks of gestation, premature preterm rupture of membranes (PPROM), or regular contractions that suggested active labor. In 1 patient, the pessary was removed because of mechanical irritation, and a smaller version was inserted. Patients who complained about discharge were reassured that this should not be confused with infection or PPROM. Patients were seen 1 week after pessary insertion, at which time a TVS was performed by placing the transducer on top of the anterior cervix, thus avoiding shadowing by the pessary and cervical manipulation. Thereafter, screening patients were followed by their own gynecologist if there were no further problems. High-risk pregnancies with previous PTB, perinatal losses, or radical conization were followed within our outpatient unit until at least 32 weeks of gestation.
From July 2011 onward, informed consent included information on recent studies of vaginal progesterone that was administered as 200-mg vaginal suppositories in the evening; thereafter, patients received combined therapy. In total, 53 patients received a “pessary only,” and 53 patients received both pessary and vaginal progesterone ( Table 1 ). For the whole study, we calculated a comparability score of 11 of 12 points, 1 point for midyear interval (eg, the time interval between different policies) and 2 points for each other variable (geographic setting, health care setting, health care providers, confounding interventions impact, consensus statements impact).
Exclusion criteria were major fetal abnormalities, PPROM before the start of therapy, ballooning membranes (beyond the external os), vaginal bleeding, and painful or regular contractions before the start of therapy.
The primary outcome was the rate of PTB at <34 weeks of gestation. Preterm cesarean delivery at <37 weeks of gestation was indicated because of severe preeclampsia (n = 1), suspected chorioangioma (n = 1), placenta previa (n = 1), suspicious fetal heart rate monitoring without regular contractions (n = 1), PPROM (n = 12), chorioamnionitis (n = 3), regular contractions with >5 cm dilation combined with abnormal presentation (n = 3), pathologic fetal heart rate pattern (n = 6), or prolonged labor (n = 1). Secondary outcomes included rates of PTB at <28, <32, and <37 completed weeks of gestation, birth weight, admission to the neonatal intensive care unit (NICU), the number of days in the NICU, the composite poor neonatal outcome defined by perinatal or neonatal death, respiratory distress syndrome more than grade II, bronchopulmonary dysplasia, intraventricular hemorrhage grade III or IV, and necrotizing enterocolitis. Days of neonatal hospitalization were analyzed separately. All data were retrieved from our hospital data system and double checked with paper charts. Statistical analyses were performed with R (version 3.1.1 for Windows) with the use of the packages survival, political science computational laboratory, and Microsoft Office Excel 2007 (Microsoft Corporation, Redmond, WA).
Baseline characteristics were analyzed with the use of t -tests and χ 2 -tests. Logistic regression was used to compare the primary and secondary outcomes between the pessary group and the group with pessary and additional vaginal progesterone. For birthweight and prolongation of pregnancy, a linear model was used. For the comparison of the number of days in the NICU, a zero-inflated Poisson regression model was used with the assumption that the treatment had an effect on the number of NICU days (Poisson part of the model) but not on the rate of NICU admission per se (n = 30). For all models, confounders such as the Z-score of CL at the time of treatment initiation, nulliparity, and the history of PTB were included as covariates. In contrast to absolute values, Z-scores of CL consider the normal range of values for each gestational week. The Z-scores were calculated based on the formula by Salomon et al. For the zero-inflated Poisson regression the use of corticosteroids was included as an additional covariate in the Poisson parts of the models. In a separate analysis for the comparison of the number of days in the NICU (zero-inflated Poisson model), gestational age or birthweight was also included as an additional covariate. A variable that included the population group (risk vs screening) was included in all models for the total population. Probability values of <.05 were considered to indicate statistically significant differences between the 2 groups.
Results
The profile of the 106 patients is shown in Table 1 ; 53 women received a pessary, and 53 women received a pessary plus vaginal progesterone. Primigravid women were more common in the screening than the risk group (62.7% vs 14.5%). All 29 patients in both treatment arms who delivered between 24 and 33 + 6 weeks of gestation received corticosteroids that were administered only when the CL was <15 mm or a delivery <34 weeks of gestation was indicated. Patients who delivered at ≥34 weeks of gestation received corticosteroids in 21 of 76 cases (27.6%), in 11 of 40 patients (27.5%) in the pessary group, and 10 of 36 cases (27.8%) in the pessary plus progesterone group. There were no significant differences with respect to baseline characteristics ( Table 1 ). Figure 1 shows the CL at the start of therapy separately for screening and risk populations and both treatment groups within the percentiles that were used for indication.
The rate of delivery at <34 weeks of gestation (primary outcome) was not significantly different: 13 of 53 patients (24.5%) were treated with pessary only, and 17 of 53 patients (32.1%) were treated with pessary and vaginal progesterone (odds ratio [OR], 1.29; 95% confidence interval [CI], 0.53–3.15; P = .57; Table 2 ).
Outcome | Risk group | Screening group | Total population | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Pessary (n = 27) | Pessary and progesterone (n = 28) | Odds ratio | 95% Confidence interval | P value | Pessary (n = 26) | Pessary and progesterone (n = 25) | Odds ratio | 95% Confidence interval | P value | Pessary (n = 53) | Pessary and progesterone (n = 53) | Odds ratio | 95% Confidence interval | P value | |
Primary outcome | |||||||||||||||
Delivery at <34 wk, n (%) | 6 (22.2) | 10 (35.7) | 1.63 | 0.44–6.2 | .46 | 7 (26.9) | 7 (28) | 0.88 | 0.23–3.23 | .84 | 13 (24.5) | 17 (32.1) | 1.29 | 0.53–3.15 | .57 |
Secondary outcomes | |||||||||||||||
Delivery, n (%) | |||||||||||||||
<28 Wk | 1 (3.7) | 4 (14.3) | 2.33 | 0.21–53.76 | .51 | 2 (7.7) | 1 (4.0) | 0.36 | 0.01–4.43 | .44 | 3 (5.7) | 5 (9.4) | 0.98 | 0.18–5.61 | .98 |
<32 Wk | 3 (11.1) | 6 (21.4) | 1.11 | 0.18–6.91 | .9 | 5 (19.2) | 5 (20) | 0.89 | 0.20–3.85 | .88 | 8 (15.1) | 11 (20.8) | 1.1 | 0.37–3.30 | .86 |
<37 Wk | 10 (37.0) | 14 (50) | 1.25 | 0.36–4.26 | .72 | 13 (50) | 10 (40) | 0.56 | 0.16–1.80 | .33 | 23 (43.4) | 24 (45.3) | 0.91 | 0.40–2.05 | .83 |
Composite poor outcome, n (%) | 3 (11.1) | 6 (21.4) | 1.18 | 0.20–7.18 | .85 | 5 (19.2) | 4 (16.0) | 0.76 | 0.16–3.41 | .71 | 8 (15.1) | 10 (18.9) | 1.03 | 0.34–3.09 | .96 |
Admission to neonatal intensive care unit, n (%) | 6 (22.2) | 9 (32.1) | 1.51 | 0.41–5.66 | .53 | 9 (34.6) | 6 (24.0) | 0.54 | 0.14–1.89 | .33 | 15 (28.3) | 15 (28.3) | 0.94 | 0.39–2.25 | .89 |
Neonatal admission time, d a | 51.8 ± 37.7 | 47.6 ± 43.1 | 0.39 b | 0.31–0.50 | < .01 | 52.1 ± 39.1 | 45.0 ± 54.5 | 0.61 b | 0.51–0.73 | < .01 | 52.0 ± 37.1 | 46.5 ± 46.1 | 0.57 b | 0.50–0.65 | < .01 |
Birthweight, g a | 2919 ± 953 | 2668 ± 1010 | –162.9 c | –695.7–369.8 | .54 | 2602 ± 907 | 2805 ± 966 | 277.2 c | –267.3–821.65 | .31 | 2763 ± 935 | 2734 ± 982 | 21.5 c | –355.9–398.9 | .91 |
Prolongation, d a , d | 128 ± 46.8 | 113 ± 52.6 | –9.41 c | –35.74–16.91 | .48 | 83.1 ± 45.3 | 93.7 ± 45.1 | 7.69 c | –18.4–33.78 | .56 | 106 ± 50.9 | 104 ± 49.7 | –1.77 c | –20.71–17.05 | .85 |