Research ajog.

org

1 OBSTETRICS 56
2 57
3 Cervical pessary placement for prevention of 58
4 59
5 preterm birth in unselected twin pregnancies: 60
6 61
7 a randomized controlled trial 62
8 Q11 Q1 Kypros H. Nicolaides, MD; Argyro Syngelaki, RM; Liona C. Poon, MD; 63
9 Catalina de Paco Matallana, MD; Walter Plasencia, MD; Francisca S. Molina, MD; 64
10 Gemma Picciarelli, MD; Natasa Tul, MD; Ebru Celic, MD; Tze Kin Lau, MD; 65
11 Roberto Conturso, MD 66
12 67
13 68
14 OBJECTIVE: Preterm birth is the leading cause of neonatal death and transfusion). Analysis was by intention to treat. This trial is registered in 69
15 handicap in survivors. Although twins are found in 1.5% of preg- the ISRCTN registry, number 01096902. 70
16 nancies they account for about 25% of preterm births. Randomized 71
17 RESULTS: A total of 1180 (56.0%) of the 2107 eligible women agreed to 72
controlled trials in singleton pregnancies reported that the prophy-
18 take part in the trial; 590 received cervical pessary and 590 had 73
lactic use of progestogens, cervical cerclage, and cervical pessary
19 expectant management. Two of the former and 1 of the latter were lost to 74
reduce significantly the rate of early preterm birth. In twin preg-
20 follow-up. There were no significant differences between the pessary 75
nancies, progestogens and cervical cerclage have been shown to be
21 and control groups in rates of spontaneous birth <34 weeks (13.6% vs 76
ineffective in reducing preterm birth. The objective of this study was
22 12.9%; relative risk [RR], 1.054; 95% confidence interval [CI], 77
to test the hypothesis that the insertion of a cervical pessary in twin
23 0.787e1.413; P ¼ .722), perinatal death (2.5% vs 2.7%; RR, 0.908; 78
pregnancies would reduce the rate of spontaneous early preterm
24 95% CI, 0.553e1.491; P ¼ .702), adverse neonatal outcome (10.0 vs 79
birth.
25 9.2%; RR, 1.094; 95% CI, 0.851e1.407; P ¼ .524), or neonatal 80
26 STUDY DESIGN: This was a multicenter, randomized controlled trial in therapy (17.9% vs 17.2%; RR, 1.040; 95% CI, 0.871e1.242; 81
27 unselected twin pregnancies of cervical pessary placement from P ¼ .701). A post hoc subgroup analysis of 214 women with short cervix 82
þ0 þ6
28 20 -24 weeks’ gestation until elective removal or delivery vs (25 mm) showed no benefit from the insertion of a cervical pessary. 83
29 expectant management. Primary outcome was spontaneous birth 84
CONCLUSION: In women with twin pregnancy, routine treatment with
30 <34 weeks. Secondary outcomes included perinatal death and a 85
cervical pessary does not reduce the rate of spontaneous early preterm
31 composite of adverse neonatal outcomes (intraventricular hemor- 86
birth.
32 rhage, respiratory distress syndrome, retinopathy of prematurity, or 87
33 necrotizing enterocolitis) or need for neonatal therapy (ventilation, Key words: Arabin pessary, cervical length, neonatal morbidity, 88
34 phototherapy, treatment for proven or suspected sepsis, or blood prematurity, preterm birth, sonographic short cervix, twins 89
35 90
36 91
37
38 P reterm birth is responsible for of 1.5% of pregnancies,3 account for controlled trials (RCT) in singleton
>70% of all neonatal and infant about 25% of preterm births.1 Mortality pregnancies with short cervical length
deaths.1 Additionally, children born and morbidity are inversely related to reported that the prophylactic use of
92
93
39 94
preterm, compared to those born at gestational age at delivery and are progesterone reduces significantly the 95
40
term, have a 10-fold increase in risk of therefore more common in cases with rate of preterm birth and neonatal 96
41
cerebral palsy.2 Twins, with a prevalence early preterm birth.1,4,5 Randomized morbidity.6-9 Cervical cerclage in 97
42
43 98
44 99
Q2 From the Harris Birthright Research Center for Fetal Medicine, King’s College, London, United Kingdom (Drs Nicolaides, Poon, and Celic, and Ms
45 100
Syngelaki); Department of Obstetrics and Gynecology, Hospital Universitario Virgen de La Arrixaca, Murcia, Spain (Dr Paco Matallana); Maternal Fetal
46 Medicine Unit, Hospital Universitario Materno Infantil de Canarias, Las Palmas de Gran Canaria, Canary Islands, Spain (Dr Plasencia); Department of
101
47 Obstetrics and Gynecology, University Hospital of Granada (CHUG), Granada, Spain (Dr Molina); Department of Fetal Medicine, Medway Maritime 102
48 Hospital, Gillingham, United Kingdom (Dr Picciarelli); Department of Perinatology, Division of Obstetrics and Gynecology, University Medical Center 103
49 Ljubljana, Slovenia (Dr Tul); Department of Obstetrics and Gynecology, Chinese University of Hong Kong, Hong Kong, China (Dr Lau); and Department of 104
Obstetrics and Gynecology, Ospedale Valduce, Como, Italy (Dr Conturso). 105
50
51 Received Aug. 15, 2015; revised Aug. 23, 2015; accepted Aug. 24, 2015. 106
52 Q10 The study was supported by a grant from the Fetal Medicine Foundation (United Kingdom charity no. 1037116). 107
53 The authors report no conflict of interest. 108
54 Corresponding author: Kypros H. Nicolaides, MD. kypros@fetalmedicine.com 109
55 0002-9378/$36.00  ª 2015 Elsevier Inc. All rights reserved.  http://dx.doi.org/10.1016/j.ajog.2015.08.051 110

MONTH 2015 American Journal of Obstetrics & Gynecology 1

FLA 5.4.0 DTD
YMOB10615_proof
3 September 2015
8:14 pm
ce
 Research Obstetrics ajog.org

111 167
112 Kingdom. The trial was registered in the 168
FIGURE 1 ISRCTN registry, number N01096902.
113 169
114 170
Randomization
115 171
116 Eligible women were randomized in a 172
117 1:1 ratio to either cervical pessary or 173
118 expectant management, using a World 174
119 Wide Webebased application with a 175
120 computer-generated random-number 176
121 list. In the random-sequence generation 177
122 there were no restrictions, such as block 178
123 size or stratification by site. At each 179
124 center the patients agreeing to partici- 180
125 pate in the study were registered with a 181
126 central computer that then instructed 182
127 the operator as to whether the patient 183
128 should receive a cervical pessary or be 184
Trial profile. managed expectantly. Consequently,
129 185
130
Nicolaides. RCT of cervical pessary in twin gestations. Am J Obstet Gynecol 2015. there was no way for study personnel to 186
131 know or guess the group assignment 187
132 prior to allocation. 188
133 singleton pregnancies with short cervix management, would reduce the rate of 189
134 is beneficial only in the subgroup with spontaneous birth at <34 weeks’ Procedures
190
135 history of preterm birth.10,11 In twin gestation. Gestational age was determined from the 191
136 pregnancies, progestogens and cervical menstrual history and confirmed from 192
137 cerclage have been shown to be ineffec- the measurement of the crownerump 193
tive in reducing preterm birth.11-15 M ATERIALS AND M ETHODS length of the bigger fetus at 11-13 weeks’
138 194
139 An alternative approach for preven- Study design and participants gestation.21 At the same scan chorio- 195
140 tion of preterm birth is transvaginal This was an open-label randomized nicity was determine from examination 196
141 placement of a silicone pessary around study of cervical pessary vs expectant of the junction between the intertwin 197
142 the cervix; this is thought to support the management in twin pregnancies in 23 membrane and the placenta.22 198
143 cervix and change its direction toward maternity hospitals in the United Cervical length was measured by Q3 199
144 the sacrum, thereby reducing the direct Kingdom, Spain, Germany, Austria, transvaginal ultrasound examination at 200
145 pressure from the uterine contents on Slovenia, Portugal, Italy, Belgium, 20-24 weeks with women, who had 201
146 the cervical cana.16,17 Two RCTs, pub- Albania, China, Brazil, and Chile. emptied the bladder, placed in the dorsal 202
147 lished after the start of this study, in All women with twin pregnancies lithotomy position as previously 203
148 singleton pregnancies with short cervix undergoing routine ultrasound exami- described,23 by operators with certifica- 204
149 provided contradictory results on the nation at 20þ0-24þ6 weeks’ gestation for tion of competence in the technique 205
150 effect of cervical pessary on the rate of assessment of fetal anatomy and mea- (Fetal Medicine Foundation Certificate 206
151 spontaneous birth at <34 weeks; in 1 surement of cervical length were eligible of Competence in Cervical Assessment). 207
152 study, the pessary reduced the rate from for the study. Exclusion criteria were Cervical pessaries (CE0482, MED/ 208
153 27-6%,18 but in the second study of 108 maternal age <16 years, fetal death, CERT ISO 9003/EN 46003), which Q4 209
154 pregnancies there was no significant ef- major fetal defect, severe twin-to-twin consist of flexible silicone, were pur- 210
155 fect (5.5% vs 9.4%).19 A RCT in 813 transfusion syndrome or selective fetal chased from the manufacturer (Dr Ara- 211
156 unselected multiple pregnancies, pub- growth restriction, cervical cerclage bin GmbH & Co, Witten, Germany). 212
157 lished after the start of this study, re- in situ, painful regular uterine Speculum examination was carried out 213
158 ported that cervical pessary did not contractions, and history of rup- to inspect the cervix for any pathology 214
159 reduce significantly the rate of birth at tured membranes diagnosed before and obtain a high vaginal swab for 215
160 <32 weeks (12% vs 10%), but in an randomization. bacteriological examination. If there was 216
161 unplanned subgroup analysis of 133 Women agreeing to participate in the offensive vaginal discharge antibiotic 217
162 women with cervical length <38 mm the study gave written informed consent. therapy was given and insertion of the 218
163 rate was reduced (29% vs 14%).20 The study was approved by the National pessary was delayed until the discharge 219
164 The objective of this multicenter RCT Research Ethics Committee in the subsided. The pessary was inserted 220
165 was to test the hypothesis that the United Kingdom, as well as the local through the vagina with the woman in 221
166 insertion of a cervical pessary in twin ethics committees of the participating the recumbent position and placed up- 222
pregnancies, compared to expectant hospitals outside of the United ward around the cervix.16,18 The

2 American Journal of Obstetrics & Gynecology MONTH 2015

FLA 5.4.0 DTD
YMOB10615_proof
3 September 2015
8:14 pm
ce
 ajog.org Obstetrics Research
223 279
224 research team members introducing the 280
cervical pessaries received instruction on TABLE 1
225 281
selecting the appropriate size and intro- Characteristics of study participants
226 282
227 ducing the device. Pessary group, Control group,
283
Women in the control group received Characteristics n [ 590 n [ 590
228 284
229 the same obstetrical care as those in the Age, median (IQR), y 33.1 (29.5e36.7) 33.2 (29.1e36.6) 285
230 pessary group. Follow-up visits for ul- Weight, median (IQR), kg 67.0 (60.0e76.3) 68.0 (60.0e79.0) 286
231 trasound assessment of fetal growth and 287
Height, median (IQR), cm 165 (160e170) 164 (160e169)
232 cervical length were carried out every 4 288
weeks until 34 weeks’ gestation. If after Racial origin, n (%)
233 289
234 26 weeks the cervical length was <10 Caucasian 497 (84.2) 483 (81.9) 290
235 mm, steroids were administered for fetal Afro-Caribbean 43 (7.3) 54 (9.2) 291
236 lung maturation. At the time of 292
randomization, the participants were South Asian 19 (3.2) 20 (3.4)
237 293
informed that a symptom related to the East Asian 19 (3.2) 22 (3.7)
238 294
239 insertion of the pessary could include Mixed 12 (2.0) 11 (1.9) 295
240 increased vaginal discharge. At each 296
Obstetric history, n (%)
241 follow-up visit we asked the participants 297
in both arms of the study and recorded Nulliparous 363 (61.5) 360 (61.0)
242 298
243 their answer as to whether they had Parous 227 (38.5) 230 (39.0) 299
244 noted an increase in severity or fre- Delivery at 24e33 wk 9 (4.0) 15 (6.5) 300
245 quency of this symptom and whether 301
they had developed any new symptoms Delivery at 34e36 wk 11 (4.8) 18 (7.8)
246 302
since the beginning of treatment. Delivery at 37 wk 207 (91.2) 197 (85.7)
247 303
248 Women reporting increased vaginal Conception, n (%) 304
249 discharge were examined by a doctor for 305
Spontaneous 373 (63.2) 366 (62.0)
250 evidence of infection; bacterial swabs 306
were taken and antibiotic therapy was Ovulation drugs 21 (3.6) 20 (3.4)
251 307
252 given without removal of the pessary. In vitro fertilization 196 (33.2) 204 (34.6) 308
253 The cervical pessary was removed by a 309
Cigarette smoking during pregnancy, n (%) 45 (7.6) 53 (9.0)
254 simple vaginal examination at 37 weeks’ 310
gestation in asymptomatic patients. Previous cervical surgery, n (%)
255 311
256 Earlier removal of the pessary was un- Loop excision of transformation of zone 14 (2.4) 17 (2.9) 312
257 dertaken if: firstly, there was medically Cone biopsy 5 (0.8) 4 (0.7) 313
258 indicated induction of labor or elective 314
cesarean delivery; secondly, there was Chorionicity, n (%)
259 315
260 preterm labor not responding to toco- Dichorionic 479 (81.2) 479 (81.2) 316
261 lytic therapy or preterm prelabor Monochorionic 111 (18.8) 111 (18.8) 317
262 rupture of the membranes or active 318
vaginal bleeding; and thirdly, at patient GA at randomization, median (IQR), wk 22.6 (21.4e23.9) 22.7 (21.4e23.9)
263 319
request because of discomfort. GA at pessary insertion, median (IQR), wk 22.7 (21.7e23.9) e
264 320
265 Quality control of screening, handling Cervical length at randomization 321
266 of data, and verification of adherence to 322
Median (IQR), mm 32.0 (27.0e36.0) 32.0 (27.0e37.0)
267 protocols at the different centers were 323
performed on a regular basis by the trial 25 mm, n (%) 107 (18.1) 108 (18.3)
268 324
269 coordinators. Data on pregnancy out- GA, gestational age; IQR, interquartile range.
325
270 comes were obtained from hospital ma- Nicolaides. RCT of cervical pessary in twin gestations. Am J Obstet Gynecol 2015.
326
271 ternity records or the patients’ general 327
272 medical practitioners. The records of all 328
273 patients delivering at <34 weeks were 329
274 examined to determine whether the Outcome measures death, composite of major adverse 330
275 birth was medically indicated or spon- The primary outcome was spontaneous events for the neonate before discharge 331
276 taneous. Spontaneous births included birth from randomization to <34 weeks from the hospital (intraventricular 332
277 those with spontaneous onset of labor (237 days) of gestation. The secondary hemorrhage, respiratory distress syn- 333
278 and those with rupture of membranes outcome measures were: birthweight drome, retinopathy of prematurity, or 334
before labor. (mean <2.5 kg and <1.5 kg), perinatal necrotizing enterocolitis), composite of

MONTH 2015 American Journal of Obstetrics & Gynecology 3

FLA 5.4.0 DTD
YMOB10615_proof
3 September 2015
8:14 pm
ce
 Research Obstetrics ajog.org

335 391
336 TABLE 2 392
337 Q9 Outcomes according to study group 393
338 Pregnancy level Fetal/neonatal level 394
339 Pessary Control Pessary Control 395
340 group, group, group, group, 396
341 Outcome n [ 588 n [ 589 RR (95% CI) n [ 1176 n [ 1178 RR (95% CI) 397
342 Primary outcome 398
343 Spontaneous birth at 80 (13.6) 76 (12.9) 1.054 (0.787e1.413) e e e 399
344 <34 wk, n (%) 400
345 Other outcome 401
346 measures 402
347 403
Spontaneous birth at e e e
348 <34 wk, n (%) 404
349 405
Dichorionic twins, 62/477 (13.0) 62/478 (13.0) 1.002 (0.722e1.392) e e e
350 n/n (%) 406
351 407
Monochorionic 18/111 (16.2) 14/111 (12.6) 1.286 (0.673e2.455) e e e
352 408
twins, n/n (%)
353 409
354 Gestational age at 36.6 (34.9e37.9) 36.7 (35.0e37.9) e e e e 410
355 birth, median (IQR) 411
356 Any birth at <34 wk, 98 (16.7) 92 (15.6) 1.067 (0.822e1.385) e e e 412
357 n (%) 413
358 Any birth at <32 wk, 52 (8.8) 53 (9.0) 0.983 (0.682e1.416) e e e 414
359 n (%) 415
360 Any birth at <30 wk, 32 (5.4) 26 (4.4) 1.233 (0.744e2.042) e e e 416
361 n (%) 417
362 Any birth at <28 wk, 19 (3.2) 15 (2.5) 1.269 (0.651e2.473) e e e 418
363 n (%) 419
364 Secondary outcomes 420
365 421
366 Birthweight 422
367 Mean (IQR), g e e e 2331 2353 423
368 (2020e2740) (2050e2732) 424
369 <2500 g, n (%) 395 (67.2) 407 (69.1) 0.972 (0.899e1.051) 664 (56.5) 670 (56.9) 0.993 (0.925e1.065) 425
370 <1500 g, n (%) 60 (10.2) 65 (11.0) 0.925 (0.664e1.288) 100 (8.5) 96 (8.1) 1.043 (0.798e1.364) 426
371 427
372 Perinatal death, n (%) 20 (3.4) 22 (3.7) 0.911 (0.502e1.651) 29 (2.5) 32 (2.7) 0.908 (0.553e1.491) 428
373 Fetal death, n (%) 7 (1.2) 14 (2.4) 0.501 (0.204e1.232) 12 (1.0) 18 (1.5) 0.668 (0.323e1.380) 429
374 Neonatal death, 13 (2.2) 9 (1.5) 1.447 (0.623e3.359) 17 (1.4) 14 (1.2) 1.216 (0.602e2.456) 430
375 n (%) 431
376 432
Secondary outcomes n ¼ 579 n ¼ 579 n ¼ 1147 n ¼ 1146
377 in survivors, n (%) 433
378 434
Adverse neonatal 88 (15.2) 69 (11.9) 1.275 (0.951e1.710) 115 (10.0) 105 (9.2) 1.094 (0.851e1.407)
379 event
435
380 436
381 Intraventricular 16 (2.8) 12 (2.1) 1.333 (0.636e2.793) 18 (1.6) 15 (1.3) 1.199 (0.607e2.367) 437
hemorrhage
382 438
383 Respiratory 84 (14.5) 67 (11.6) 1.254 (0.929e1.692) 109 (9.5) 100 (8.7) 1.089 (0.841e1.411) 439
distress syndrome
384 440
385 Retinopathy of 8 (1.4) 3 (0.5) 2.667 (0.711e10.001) 12 (1.0) 3 (0.3) 3.997 (1.131e14.125) 441
386 prematurity 442
387 Necrotizing 6 (1.0) 6 (1.0) 1.000 (0.324e3.082) 8 (0.7) 6 (0.5) 1.332 (0.464e3.827) 443
388 enterocolitis 444
389 445
Nicolaides. RCT of cervical pessary in twin gestations. Am J Obstet Gynecol 2015. (continued)
390 446

4 American Journal of Obstetrics & Gynecology MONTH 2015

FLA 5.4.0 DTD
YMOB10615_proof
3 September 2015
8:14 pm
ce
 ajog.org Obstetrics Research
447 503
448 504
TABLE 2
449 505
Outcomes according to study group (continued)
450 506
451 Pregnancy level Fetal/neonatal level 507
452 Pessary Control Pessary Control 508
453 group, group, group, group, 509
Outcome n [ 588 n [ 589 RR (95% CI) n [ 1176 n [ 1178 RR (95% CI)
454 510
455 Neonatal therapy 137 (23.7) 127 (21.9) 1.079 (0.873e1.334) 205 (17.9) 197 (17.2) 1.040 (0.871e1.242) 511
456 Ventilation 80 (13.8) 64 (11.1) 1.250 (0.919e1.701) 114 (9.9) 97 (8.5) 1.174 (0.907e1.520) 512
457 Phototherapy 86 (14.9) 80 (13.8) 1.075 (0.811e1.425) 111 (9.7) 116 (10.1) 0.956 (0.747e1.224) 513
458 514
Treatment for 41 (7.1) 45 (7.8) 0.911 (0.606e1.369) 66 (5.8) 66 (5.8) 0.999 (0.717e1.392)
459 515
sepsis
460 516
461 Blood transfusion 26 (4.5) 25 (4.3) 1.040 (0.608e1.779) 36 (3.1) 36 (3.1) 0.999 (0.634e1.574) 517
462 Percentages for major adverse neonatal events and neonatal therapy were calculated after excluding cases of perinatal deaths. 518
463 CI, confidence interval; IQR, interquartile range; RR, relative risk. 519
464 Nicolaides. RCT of cervical pessary in twin gestations. Am J Obstet Gynecol 2015. 520
465 521
466 522
467 neonatal therapy (ventilation, photo- significance. The risk of spontaneous <38 mm the rate of early preterm birth 523
468 therapy, treatment for proven or sus- preterm birth <34 weeks was quantified was halved. The cut-off of 38 mm was 20 524
469 pected sepsis, or blood transfusion), and by the relative risk and 95% confidence selected because only 1% of the patients 525
470 major maternal complication attribut- interval (CI). The risk of spontaneous had cervical length <25 mm, which was 526
471 able to the pessary. birth from randomization until 34 weeks the cut-off selected for a preplanned 527
472 was assessed using KaplaneMeier anal- subgroup analysis. However, in our 528
473 ysis, where gestational age was the time study the cervical length was <25 mm in
25 529
Statistical analysis
474 530
475 The sample size calculation was based on scale, spontaneous birth was the event, 18% of cases. 531
476 detecting a treatment effect that pro- and elective deliveries were treated as 532
477 duces a one-third reduction in the inci- censored. For the purposes of this anal- Role of the funding source 533
478 dence of spontaneous delivery between ysis, all pregnancies were considered to Funding for the study was provided by 534
479 randomization and 33þ6 weeks from an be no longer at risk for the event at the the Fetal Medicine Foundation (United 535
480 anticipated 13% in the expectant man- start of the 34th week. Hazard ratios Kingdom charity no. 1037116), which 536
481 agement group. In the computer simu- were estimated with the use of the Cox had no role in study design, data 537
482 lations it was assumed that the proportional hazards model, with a collection, data analysis, data interpre- 538
483 distribution of cervical lengths and risks formal test of the proportional hazards tation, or the writing of the report. The
25,26 539
484 in the expectant group were the same as assumption. Odds ratios were con- corresponding author had full access to
540
previously reported in our population.24 verted to relative risk with the use of the all the data in the study and had final
485 27 541
486 Using logistic regression analysis, with method of Zhang and Yu. Results on responsibility for the decision to submit 542
487 adjustment for cervical length, a total perinatal and neonatal outcome were for publication. 543
488 sample of 1180 patients has 85% power examined both at the pregnancy and 544
489 of detecting this difference at a (2-tailed) fetal/neonatal level because of the po- R ESULTS 545
significance level of 5%. tential of nonindependence of outcomes Study population
490 546
491 Statistical analyses were by intention from the 2 twins arising from the same A total of 1180 of the 2107 eligible 547
492 to treat and no interim analyses were pregnancy. pregnant women agreed to take part in 548
493 performed. Baseline data for the cervical the trial (Figure 1). The participants ½F1 549
494 pessary and expectant groups were Post hoc analysis were recruited from August 2008 550
495 summarized by the median and the We conducted a post hoc analysis to through May 2011. There were 600 551
496 interquartile range. Comparisons be- examine the effect of cervical pessary in women from England, 391 from Spain, 552
497 tween groups were performed with the women with short cervix (25 mm). 145 from other European countries, and 553
498 use of the ManneWhitney U test. Uni- The reason for undertaking this analysis 44 from non-European countries. There 554
499 variate comparisons of dichotomous is that a recent RCT in multiple preg- were no important differences in base- 555
500 data were performed with the use of nancies reported that although the pes- line characteristics between the pessary 556
501 Fisher exact test. The P values for all sary was not beneficial in the total and the expectant groups (Table 1). Two ½T1 557
502 hypothesis tests were 2-sided and P <.05 population, in an unplanned subgroup pregnancies in the pessary group and 1 558
was considered to indicate statistical analysis of those with cervical length in the controls were lost to follow-up.

MONTH 2015 American Journal of Obstetrics & Gynecology 5

FLA 5.4.0 DTD
YMOB10615_proof
3 September 2015
8:14 pm
ce
 Research Obstetrics ajog.org

559 615
560 Two of the patients in the control group 1.061; 95% CI, 0.776e1.453; P ¼ .709) 616
FIGURE 2 were treated with vaginal progesterone (Figure 2). ½F2
561 617
562 from 26 and 28 weeks’ gestation, There were no cases of maternal death 618
web 4C=FPO

563 respectively, because of cervical short- or serious vaginal trauma either during 619
564 ening; in both cases delivery was >34 insertion or removal of the pessary. 620
565 weeks. There was 1 case where the pessary was 621
566 associated with cervical edema requiring 622
567 Outcomes removal under general anesthesia. There 623
568 There was no significant difference be- were 4 cases of chorioamnionitis, 3 in the 624
569 tween the cervical pessary and control pessary, and 1 in the control group, 625
570 groups in rates of spontaneous birth at including 2 in women with miscarriage 626
571 <34 weeks, perinatal death, adverse and 2 in those with preterm prelabor 627
572 neonatal event, or neonatal therapy rupture of membranes. 628
573 (Table 2). Logistic regression analysis, ½T2 629
with adjustment for cervical length, Post hoc subgroup analysis
574 630
575 demonstrated no significant effect of the The median cervical length at randomi- 631
Q8 Kaplan-Meier plot of proportion of continued
576 cervical pessary in the rate of sponta- zation was 32 mm in both the pessary 632
pregnancy without delivery in cervical pessary
577 neous birth at <34 weeks (odds ratio, group and controls and in both groups 633
and control groups.
578 1.058; 95% CI, 0.740e1.511; P ¼.7584). there was an inverse correlation between 634
Nicolaides. RCT of cervical pessary in twin gestations. Am J
579 Obstet Gynecol 2015. The cumulative percentage of women cervical length and rate of spontaneous 635
580 who did not give birth spontaneously at birth at <34 weeks, which was not 636
581 <34 weeks was not significantly different significantly different between the 2 637
between the 2 groups (hazard ratio, groups (Figure 3). ½F3
582 638
583 Post hoc subgroup analysis of 214 639
584 women with short cervix showed no 640
585 FIGURE 3 benefit from the insertion of a cervical 641
586 pessary (Table 3). The cumulative per- ½T3 642
587 centage of women who did not give birth 643
web 4C=FPO

588 spontaneously at <34 weeks was not 644

589 significantly different between the 2 645
590 treatment groups in either those with 646
591 cervical length 25 mm (hazard ratio, 647
592 1.256; 95% CI, 0.760e2.074; P ¼ .374) 648
593 or those with length >25 mm (hazard 649
594 ratio, 0.975; 95% CI, 0.652e1.458; 650
595 P ¼ .902) (Figure 4). ½F4
651
596 652
Adverse events
597 653
598 At recruitment to the trial, in the cervical 654
599 pessary compared to control group, 655
600 there was no significant difference in 656
601 reported vaginal discharge (10.9% vs 657
602 10.2%, P ¼ .705) or pelvic discomfort 658
603 (1.2% vs 1.5%, P ¼ .802). In any one of 659
604 the follow-up visits, cervical pessary was 660
605 associated with significantly higher rate 661
606 of vaginal discharge (42.1% vs 20.4%, P 662
607 < .0001), but not pelvic discomfort 663
608 (5.8% vs 5.1%, P ¼ .695). 664
609 In the cervical pessary group, vaginal 665
610 swabs demonstrated an infection, most 666
611 Association between cervical length at randomization and rate of spontaneous birth at <34 weeks in commonly with Candida albicans, group 667
612 pessary (red bars, red interrupted regression curve) and control (white bars, black regression curve) B streptococcus, or Gardnerella vaginalis, 668
613 groups. in 14.0% (82/585) of cases at recruit- 669
614 Nicolaides. RCT of cervical pessary in twin gestations. Am J Obstet Gynecol 2015. ment to the trial and in 20.9% (116/555) 670
in any one of the follow-up visits. The

6 American Journal of Obstetrics & Gynecology MONTH 2015

FLA 5.4.0 DTD
YMOB10615_proof
3 September 2015
8:14 pm
ce
 ajog.org Obstetrics Research
671 727
672 728
TABLE 3
673 729
Outcomes according to cervical length at randomization £25 mm and >25 mm
674 730
Pregnancy level Fetal/neonatal level
675 731
676 Pessary Control Pessary Control 732
677 Outcome group group RR (95% CI) group group RR (95% CI) 733
678 Cervical length 25 mm (n ¼ 214) n ¼ 106 n ¼ 108 n ¼ 212 n ¼ 216 734
679 Primary outcome, n (%) 735
680 736
Spontaneous birth at <34 wk 33 (31.1) 28 (25.9) 1.201 (0.784e1.839) e e e
681 737
682 Secondary outcomes, n (%) 738
683 Birthweight <2500 g 82 (77.4) 89 (82.4) 0.939 (0.820e1.074) 149 (70.3) 150 (69.4) 1.012 (0.894e1.146) 739
684 Birthweight <1500 g 24 (22.6) 21 (19.4) 1.164 (0.692e1.960) 45 (21.2) 36 (16.7) 1.274 (0.858e1.891) 740
685 741
Perinatal death 13 (12.3) 6 (5.6) 2.208 (0.872e5.592) 20 (9.4) 12 (5.6) 1.698 (0.852e3.386)
686 742
687 Secondary outcomes in survivors, n (%) n ¼ 99 n ¼ 102 n ¼ 192 n ¼ 204 743
688 Adverse neonatal event 23 (23.2) 20 (19.6) 1.185 (0.696e2.016) 34 (17.7) 30 (14.7) 1.204 (0.768e1.888) 744
689 Neonatal therapy 36 (36.4) 31 (30.4) 1.197 (0.808e1.772) 56 (29.2) 52 (25.5) 1.144 (0.829e1.579) 745
690 746
Cervical length >25 mm (n ¼ 963) n ¼ 482 n ¼ 481 n ¼ 964 n ¼ 962
691 747
692 Primary outcome, n (%) 748
693 Spontaneous birth at <34 wk 47 (9.8) 48 (10.0) 0.977 (0.667e1.432) e e e 749
694 Secondary outcomes, n (%)
750
695 751
696 Birthweight <2500 g 313 (64.9) 318 (66.1) 0.982 (0.896e1.077) 515 (53.4) 520 (54.1) 0.988 (0.910e1.074) 752
697 Birthweight <1500 g 36 (7.5) 44 (9.1) 0.817 (0.535e1.245) 55 (5.7) 60 (6.2) 0.915 (0.642e1.304) 753
698 Perinatal death 7 (1.5) 16 (3.3) 0.437 (0.181e1.052) 9 (0.9) 20 (2.1) 0.449 (0.206e0.981)a 754
699 755
Secondary outcomes in survivors, n (%) n ¼ 480 n ¼ 477 n ¼ 955 n ¼ 942
700 756
701 Adverse neonatal event 65 (13.5) 49 (10.3) 1.318 (0.930e1.868) 81 (8.5) 75 (8.0) 1.065 (0.789e1.439) 757
702 Neonatal therapy 101 (21.0) 96 (20.1) 1.046 (0.815e1.341) 149 (15.6) 146 (15.5) 1.007 (0.816e1.242) 758
703 Percentages for major adverse neonatal events and neonatal therapy were calculated after excluding cases of perinatal deaths. 759
704 CI, confidence intervals; RR, relative risk. 760
705 a
Significant P value < .05. 761
706 Nicolaides. RCT of cervical pessary in twin gestations. Am J Obstet Gynecol 2015. 762
707 763
708 764
709 respective values in the control group patient request group. In the latter The cervical pessary was well tolerated 765
710 were 13.4% (76/567) and 16.8% (86/ group, the rate of spontaneous birth at by most women and only 5% requested 766
711 511) and these were not significantly <34 weeks was not significantly higher removal. The pessary doubled the rate of 767
712 different from the pessary group to that in the total group treated with vaginal discharge but did not increase 768
713 (P ¼ .797 and P ¼ .100). pessary placement (16.1% vs 13.6%, the rate of cervicovaginal infection. 769
714 P ¼ .589). In this randomized twins cohort, the Q5 770
715 771
Removal of the pessary at <34 median cervical length at 20-24 weeks’
716 772
weeks’ gestation C OMMENT gestation, the overall rate of spontaneous
717 birth at <34 weeks was 13% and this rate 773
718 The cervical pessary was removed at <34 Main findings
774
weeks in 22.3% (131/588) of pregnan- The findings of this trial demonstrate was inversely related to cervical length.
719 These findings are consistent with our 775
720 cies, including 18 for iatrogenic delivery, that in unselected twin pregnancies, or 776
34 for preterm labor, 48 for preterm in the subgroup with cervical length previous study involving 1163 twin
721 pregnancies, which was the basis for the 777
722 prelabor rupture of membranes, and 31 25 mm, placements of a cervical 778
for patient request. Subsequently, there pessary at 20-24 weeks’ gestation does power calculations of this trial.24
723 779
724 was birth at <34 weeks in 94.4% (17/18) not reduce the rate of spontaneous early 780
of the iatrogenic group, 90.2% (74/82) of preterm birth, perinatal death, adverse Strengths and limitations
725 781
726 those with preterm labor or rupture of neonatal outcome, or need for neonatal The strengths of the study are: first, 782
membranes, and 22.7% (7/31) of the therapy. RCT with central randomization and

MONTH 2015 American Journal of Obstetrics & Gynecology 7

FLA 5.4.0 DTD
YMOB10615_proof
3 September 2015
8:14 pm
ce
 Research Obstetrics ajog.org

783 839
784 preterm birth may require further 840
FIGURE 4 investigation. However, before such
785 841
786 study is undertaken it is important that 842
web 4C=FPO

787 the technique for measuring cervical 843

788 length is standardized and the operators 844
789 demonstrate their competence in un- 845
dertaking such measurements. -
790 846
791 847
792 ACKNOWLEDGMENT 848
793 Professor Zarko Alfirevic, University of Liverpool, 849
794 United Kingdom, and Professor Steve Thornton, 850
University of Exeter, United Kingdom, were
795 members of the trial steering committee and
851
796 Professor David Wright, Institute of Health 852
797 Research, University of Exeter, United Kingdom, 853
798 provided the sample size calculation and was 854
799 responsible for trial data monitoring. Help in the 855
recruitment of patients was provided by the
800 following doctors who did not receive funding or
856
801 compensation for this study: Antonio Leal, 857
Kaplan-Meier plot of proportion of continued pregnancy without delivery in cervical pessary and
802 Evdoxia Skyfta, Anna Paula Mosconi, Jesus 858
control groups in women with cervical length randomization 25 mm (left) and >25 mm (right).
803 Rodriguez Calvo, Mara Mitrea, Esperanza 859
Nicolaides. RCT of cervical pessary in twin gestations. Am J Obstet Gynecol 2015. Gonzalez, and Paula Vargas employed by King’s
804 860
College Hospital, London, United Kingdom;
805 Mauro Parra-Corderro employed by Hospital Q6
861
806 Clinico Universidad de Chile; James Anderson 862
recruitment of the desired number of management, did not reduce signifi-
807 and Susana Pereira employed by Hospital de 863
patients with nearly complete follow-up; cantly the rate of poor perinatal outcome
808 Sao Teotonio, Viseu, Portugal; Chineze Otigbah 864
second, there were no changes to the (13% vs 14%) or birth at <32 weeks employed by Queen’s Hospital, London, United
809 865
protocol after commencement of the (10% vs 12%).20 However, in a subgroup Kingdom; Christof Worda employed by the
810 Medical University of Vienna, Austria; Adolfo Liao
866
trial, no outcomes were selectively of 133 women with cervical length
811 employed by Hospital das Clinicas da Uni- 867
dropped post hoc, and the person who <25th percentile (<38 mm), a pessary
812 versidade de Sao Paulo, Brazil; Matthias Scheier 868
performed the statistical analysis was group (n ¼ 78) compared to controls
813 employed by the University Hospital Innsbruck, 869
blinded to the allocated interventions; (n ¼ 55) significantly reduced the rate of Austria; Kastriot Dallaku employed by Spitali
814 870
third, measurement of cervical length by both poor perinatal outcome (12% vs Universitar Obstetrik Gjinekologjik, Tirane,
815 871
appropriately trained sonographers; and 29%) and birth at <32 weeks (14% vs Albania; Lars Hellmeyer employed by Philipps
816 University Hospital, Marburg, Germany; Ilka 872
fourth, the rate of spontaneous birth at 29%).20 In this trial the median cervical
817 Fuchs and Wiebke Schulze employed by the 873
<34 weeks was the same as the one length was 44 mm and it was <25 mm,
818 CharitéeUniversity Hospital Berlin, Germany; 874
estimated for the power calculations. which was the originally planned cut-off Natasa Vrhkar and Marina Jakimovska
819 875
A potential limitation of the study is for the subgroup analysis, in only 1% of employed by University Medical Center Ljubl-
820 jana, Slovenia; Shuk Yi Hui and Lai Wah Law
876
that many research team doctors were cases. The respective values in our study,
821 employed by Chinese University of Hong Kong, 877
involved in the insertion of the pessary in which all measurements of cervical
822 China; and Frederic Chantraine employed by 878
and, unlike measurement of cervical length were carried out by doctors with
823 Citadelle Hospital, Liège, Belgium. 879
length, they did not receive supervised extensive experience in the technique,
824 880
training in doing so. It is therefore not were 32 mm and 18%, respectively.
825 881
possible to be certain that there was REFERENCES
826 882
appropriate insertion in all cases. Conclusions and implications 1. Office for National Statistics. Gestation-specific
827 883
Another potential limitation arises from Twin pregnancies are at substantially infant mortality, 2012. Available at: http://www.
828 884
the inevitable open-label nature of the higher risk of early preterm birth than ons.gov.uk/ons/rel/child-health/gestation-specific-
829 885
trial that could have affected medical singleton pregnancies and this risk is infant-mortality-in-england-and-wales/2012/stb-
830 gestation-specific-infant-mortality-2012.html. 886
decision making. inversely related to sonographically
831 Accessed Feb. 8, 2015. 887
832 measured cervical length at 20-24 weeks’ 2. Kodjebacheva GD, Sabo T. Influence of pre- 888
Comparison with results of previous gestation. Insertion of cervical pessary at mature birth on the health conditions, receipt of
833 889
studies around 22 weeks in both unselected special education and sport participation of
834 890
A multicenter RCT in 813 unselected twins and in those with short cervix does children aged 6-17 years in the USA. J Public
835 Health (Oxf) 2015;pii:fdv098. Q7 891
836 multiple pregnancies, including 795 not reduce the rate of spontaneous early 892
3. Office for National Statistics. Births in England
837 with twins, reported that cervical pessary preterm birth. The extent to which cer- and Wales by characteristics of birth, 2013. Avail- 893
838 inserted at a median gestational age vical pessary inserted <20 weeks in twins able at: http://www.ons.gov.uk/ons/rel/vsob1/ 894
of 19 weeks, compared to expectant with short cervix reduces the rate of early characteristics-of-birth-2-englandeandewales/

8 American Journal of Obstetrics & Gynecology MONTH 2015

FLA 5.4.0 DTD
YMOB10615_proof
3 September 2015
8:14 pm
ce
 ajog.org Obstetrics Research
895 951
2013/sb-characteristics-of-birth-2.html. Accessed gestations and previous preterm birth: a meta- cervix (PECEP): an open-label randomized
896 Feb. 8, 2015. analysis. Obstet Gynecol 2011;117:663-71. controlled trial. Lancet 2012;379:1800-6. 952
897 4. Saigal S, Doyle LW. An overview of mortality 11. Berghella V, Odibo AO, To MS, Rust OA, 19. Hui SY, Chor CM, Lau TK, Lao TT, 953
898 and sequelae of preterm birth from infancy to Althuisius SM. Cerclage for short cervix on ul- Leung TY. Cerclage pessary for preventing 954
899 adulthood. Lancet 2008;371:261-9. trasonography, meta-analysis of trials using in- preterm birth in women with a singleton preg- 955
5. D’Onofrio BM, Class QA, Rickert ME, dividual patient-level data. Obstet Gynecol nancy and a short cervix at 20 to 24 weeks: a
900 956
Larsson H, Långström N, Lichtenstein P. Pre- 2005;106:181-9. randomized controlled trial. Am J Perinatol
901 term birth and mortality and morbidity: a 12. Rouse DJ, Caritis SN, Peaceman AM, et al. 2013;30:283-8. 957
902 population-based quasi-experimental study. A trial of 17 alpha-hydroxyprogesterone cap- 20. Liem S, Schuit E, Hegeman M, et al. Cervical 958
903 JAMA Psychiatry 2013;70:1231-40. roate to prevent prematurity in twins. N Engl J pessaries for prevention of preterm birth in 959
904 6. Fonseca EB, Celik E, Parra M, Singh M, Med 2007;357:454-61. women with a multiple pregnancy (ProTWIN): a 960
Nicolaides KH. Progesterone and the risk of 13. Norman JE, Mackenzie F, Owen P, et al. multicenter, open-label randomized controlled
905 961
preterm birth among women with a short cervix. Progesterone for the prevention of preterm birth trial. Lancet 2013;382:1341-9.
906 N Engl J Med 2007;357:462-9. in twin pregnancy (STOPPIT): a randomized, 21. Robinson HP, Fleming JE. A critical evalua- 962
907 7. Defranco EA, O’Brien JM, Adair CD, et al. double-blind, placebo-controlled study and tion of sonar “crown-rump length” measure- 963
908 Vaginal progesterone is associated with a meta-analysis. Lancet 2009;373:2034-40. ments. Br J Obstet Gynaecol 1975;82:702-10. 964
909 decrease in risk for early preterm birth and 14. Rode L, Klein K, Nicolaides KH, Krampl- 22. Sepulveda W, Sebire NJ, Hughes K, 965
improved neonatal outcome in women with a Bettelheim E, Tabor A; for the PREDICT Group. Odibo A, Nicolaides KH. The lambda sign at 10-
910 966
short cervix: a secondary analysis from a Prevention of preterm delivery in twin gestations 14 weeks of gestation as a predictor of chorio-
911 randomized, double-blind, placebo-controlled (PREDICT): a multicenter, randomized, placebo- nicity in twin pregnancies. Ultrasound Obstet 967
912 trial. Ultrasound Obstet Gynecol 2007;30: controlled trial on the effect of vaginal micronized Gynecol 1996;7:421-3. 968
913 697-705. progesterone. Ultrasound Obstet Gynecol 23. Sonek J, Shellhaas C. Cervical sonography: a 969
914 8. Hassan SS, Romero R, Vidyadhari D, et al. 2011;38:272-80. review. Ultrasound Obstet Gynecol 1998;11:71-8. 970
Vaginal progesterone reduces the rate of pre- 15. Schuit E, Stock S, Rode L, et al. Effective- 24. To MS, Fonseca EB, Molina FS, Cacho AM,
915 971
term birth in women with a sonographic short ness of progestogens to improve perinatal Nicolaides KH. Maternal characteristics and
916 cervix: a multicenter, randomized, double-blind, outcome in twin pregnancies: an individual cervical length in the prediction of spontaneous 972
917 placebo-controlled trial. Ultrasound Obstet participant data meta-analysis. BJOG early preterm delivery in twins. Am J Obstet 973
918 Gynecol 2011;38:18-31. 2015;122:27-37. Gynecol 2006;194:1360-5. 974
919 9. Romero R, Nicolaides K, Conde-Agudelo A, 16. Arabin B, Halbesma JR, Vork F, Hubener M, 25. Hosmer DWLS. Applied survival analysis: 975
et al. Vaginal progesterone in women with an van Eyck J. Is treatment with vaginal pessaries regression modeling of time to event data. New
920 976
asymptomatic sonographic short cervix in the an option in patients with a sonographically York (NY): John Wiley; 1999.
921 midtrimester decreases preterm delivery and detected short cervix? J Perinat Med 2003;31: 26. Grambsch PMTT. Proportional hazards 977
922 neonatal morbidity: a systematic review and 122-33. tests and diagnostics based on weighted re- 978
923 metaanalysis of individual patient data. Am J 17. Dharan VB, Ludmir J. Alternative treatment siduals. Biometrika 1994;81:515-26. 979
924 Obstet Gynecol 2012;206:124.e1-19. for a short cervix: the cervical pessary. Semin 27. Zhang J, Yu KF. What’s the relative risk? A 980
10. Berghella V, Rafael TJ, Szychowski JM, Perinatol 2009;33:338-42. method of correcting the odds ratio in cohort
925 981
Rust OA, Owen J. Cerclage for short cervix on 18. Goya M, Pratcorona L, Merced C, et al. studies of common outcomes. JAMA 1998;280:
926 ultrasonography in women with singleton Cervical pessary in pregnant women with a short 1690-1. 982
927 983
928 984
929 985
930 986
931 987
932 988
933 989
934 990
935 991
936 992
937 993
938 994
939 995
940 996
941 997
942 998
943 999
944 1000
945 1001
946 1002
947 1003
948 1004
949 1005
950 1006

MONTH 2015 American Journal of Obstetrics & Gynecology 9

FLA 5.4.0 DTD
YMOB10615_proof
3 September 2015
8:14 pm
ce