Contraception 89 (2014) 434 – 439

Original research article

Can ultrasound predict IUD expulsion after medical abortion?☆,☆☆,☆☆☆,★ Noa’a Shimoni a,⁎, Anne Davis b , Carolyn Westhoff b a

Rutgers New Jersey Medical School Columbia University Medical Center Received 6 August 2013; revised 24 December 2013; accepted 3 January 2014 b

Abstract Objectives: Our randomized trial compared early and delayed intrauterine device (IUD) insertion following medical abortion. In this planned substudy, we explore if endometrial thickness and initial IUD position were associated with IUD expulsion. We also describe IUD movement within the uterus during the 6 months after insertion. Study design: We recruited women undergoing medical abortion and choosing the copper IUD for contraception (n = 156). Participants were randomly assigned to early insertion 1 week after mifepristone or delayed insertion 4–6 weeks later. We measured endometrial thickness by transvaginal sonogram 1 week after abortion and IUD distance from the fundal aspect of the endometrial cavity three times: at insertion, 6–8 weeks later and at 6 months. Results: We analyzed endometrial thickness in 113 women, baseline IUD position in 114 women and IUD movement in 65 women. Women who expelled IUDs (n = 15) had slightly thicker endometria (p = .007) and slightly lower baseline IUD positions (p = .03) than those who retained IUDs, but no clear cutoffs emerged in the receiver operating characteristic curve analysis. Retained IUDs commonly moved up and down throughout the 6 months (from 14 mm towards the fundus to 32 mm towards the cervix). Overall, retained IUDs moved a median of 2 mm towards the cervix between insertion and exit (p b .0001). Conclusions: After medical abortion, the risk of IUD expulsion increases with thicker endometria and lower baseline position. Since no clear cutoffs emerged in the analysis and expulsion remained uncommon even with thicker endometria, we do not recommend restricting IUD insertion based on ultrasound data. Implication: Copper T IUDs often move within the uterus without expelling. Expulsion is uncommon, and we do not recommend restricting IUD insertion based on ultrasound data. © 2014 Elsevier Inc. All rights reserved. Keywords: Postabortion contraception; Postabortion IUD; Copper IUD; Medical abortion; Postabortion sonogram

1. Introduction

☆

Funding: This research was funded by the Society of Family Planning. ☆☆ CuT380A intrauterine devices were donated by DuraMed (currently Teva Women’s Health, Inc.). ☆☆☆ Potential conflict of interest statement: Dr. Shimoni does not have any potential conflicts of interest to report. Dr. Davis receives research support from Bayer Healthcare. Dr. Westhoff serves on a Scientific Advisory Board of Agile Therapeutics, and Data Safety and Monitoring Boards of several studies supported by Bayer Healthcare and Merck. Columbia University Medical Center receives support from Bayer Healthcare and Medicine 360 for research. ★ Clinical trials registration number: NCT00737178. ⁎ Corresponding author. Rutgers New Jersey Medical School, BHSB E-1562, Newark, NJ 07103, USA. E-mail address: [email protected] (N. Shimoni). 0010-7824/$ – see front matter © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.contraception.2014.01.006

Intrauterine device (IUD) insertion immediately after early surgical abortion is increasingly common. This practice is safe and results in earlier use, greater uptake and better long-term contraceptive protection [1–3]. Studies of IUD insertion after medical abortion are few. Our recent randomized trial (n = 156) compared CuT380A IUD insertion 1 week after medical abortion with insertion 4–6 weeks after medical abortion and found a greater uptake of IUDs in women randomized to early insertion and no excess bleeding, expulsion or infection during 6 months of follow-up [4]. An observational study of IUDs inserted 8–9 days after medical abortion (n = 118) also demonstrated low rates of expulsion (n = 4/97, 4.1%) within 3 months [5].

N. Shimoni et al. / Contraception 89 (2014) 434–439

The goals of this planned substudy were to explore whether sonographic measurements of endometrial thickness and IUD position at insertion were associated with IUD expulsion or removal. We hypothesized that expulsions or removals would be more frequent among women with greater tissue remaining in the uterus or with IUDs farther from the fundus immediately after insertion. Additionally, we sought to describe IUD movement within the uterus during the 6 months in women who retained their IUDs.

2. Materials and methods This study was conducted at Columbia University Medical Center with approval from the local institutional review board. We recruited women undergoing abortion with 200 mg mifepristone and 800 mcg buccal misoprostol up to 63 days of gestation. After contraceptive counseling and prior to treatment, patients who requested the CuT380A (ParaGard, DuraMed Pharmaceuticals, Pomona, NY, USA, now Teva Women’s Health, Inc.) as their postabortion contraceptive and provided study consent were enrolled [4]. We randomized participants at the medical abortion follow-up visit 1 week later, prior to assessing abortion completion by ultrasound. Women randomized to immediate insertion underwent IUD placement during that visit, and women randomized to usual insertion received an appointment to return 4–6 weeks after the abortion for IUD insertion. In both groups, IUD insertion was delayed in cases of continuing pregnancy or retained tissue (defined as intrauterine sac remnants or intrauterine tissue with continued pregnancy symptoms or heavy bleeding). Such participants underwent IUD insertion after aspiration or additional misoprostol. Participants routinely underwent a transvaginal sonogram at the abortion follow-up visit using a G.E. Logit 2 ultrasound machine with a 7.5-mHz vaginal probe. We measured endometrial thickness prior to IUD insertion by recording the thickest anterior posterior measurement in the sagittal plane. We measured IUD distance from the most fundal aspect of the endometrial cavity to the top of the IUD three times: immediately after insertion, at the 6–8 week IUD check and at the 6-month exit visit. One study investigator (N.S.) inserted 93% (n = 125/134) of IUDs and performed 79% (n = 278/352) of sonograms. Four other study providers inserted the remaining 5% of IUDs, and 2% of IUDs were placed outside the study. Nine study providers performed the remaining 21% of sonograms. An IUD was considered expelled in cases of complete expulsion (IUD completely outside the uterus or extruding from the external os) and in cases of sonographic expulsion (IUD stem in the cervical canal by ultrasound examination but above the external os and thus not visible on speculum exam). We used SAS 9.2 (SAS Institute Inc., Cary, NC, USA) for statistical analyses. Continuous variables were compared

435

using the Wilcoxon–Mann–Whitney U test and the Wilcoxon signed rank sum test. We constructed two receiver operating characteristic (ROC) curves to evaluate if IUD expulsion could be predicted by endometrial thickness or baseline IUD position. 3. Results In this substudy, we analyzed data of the 134 participants who had IUDs placed. The average participant age was 26 years (± 6). Participants were predominantly Hispanic, 85% were parous, and the mean gestational age was 49 days (± 7). Of the 134 women who had IUDs placed, 15 (11%) experienced an expulsion (9 complete and 6 sonographic), and 13 (10%) requested IUD removal during the study for reasons other than expulsion. We excluded 19 of 134 participants from these analyses due to the following: further intervention required to complete the abortion (n = 8), levonorgestrel IUD insertion outside the study (n = 1) and lack of follow-up after IUD placement (n = 10). We excluded two participants from the endometrial thickness analysis due to missing baseline endometrial thickness data and one participant from the baseline position analysis due to missing baseline IUD position data. Thus, we included 113 women in the endometrial thickness analysis and 114 women in the baseline IUD position analysis. Fig. 1 details the flow of participants through the trial. In this subanalysis, we analyzed the randomization groups together and also stratified by IUD insertion timing. 3.1. Endometrial thickness analysis We measured endometrial thickness a median of 7 days (range 4–28) after mifepristone administration (Fig. 2A) and inserted IUDs a median of 7 days (range 4–28) after mifepristone administration in the early group and 29 days (range 25–113) in the delayed insertion group. At the medical abortion follow-up visit, the 15 women who subsequently expelled the IUD had an endometrial stripe that measured a median of 2 mm thicker than those (n = 98) who retained the device (median 12 mm versus 10 mm, p = .007). When stratified by randomization group, the results were similar; women in either the immediate or delayed group with thicker endometria were more likely to expel their IUDs (data not shown). Analyses limited to parous participants (n = 96) yielded similar results (p = .02). Participant requests for IUD removal (n = 13) were not associated with endometrial thickness (median 11 mm vs 10 mm, p = .29). The ROC curve that evaluated sequential endometrial thicknesses as predictors for expulsion produced an area under the curve of 0.72 [95% confidence interval (CI) 0.62– 0.83]. The ROC curve and corresponding sensitivities, specificities, positive and negative predictive values, and positive likelihood ratios are presented in Fig. 2B and

436

N. Shimoni et al. / Contraception 89 (2014) 434–439 Enrolled n=204 Excluded from study (n=48): Did not return for randomization visit 27 Chlamydia 4 Anemic 1 Declined IUD 16 Randomized n=156 Excluded from sono analysis (n=22): Declined IUD 18 Lost to follow-up 4 IUD placed n=134

Excluded from sono analysis (EMT: n=21, Position: n=20): Intervention required 8 Lost to follow-up 10 LNG-IUS placed 1 Missing EMT sono data 2 Missing baseline position data 1

Baseline EMT data available n=113 Basline IUD position available n=114

IUD expelled n=15

IUD removed n=13

IUD retained at least 6 months n=85 3 position sonos available n=65

Fig. 1. Participant flow. EMT, endometrial thickness.

Table 1. This ROC curve demonstrates no endometrial thickness cutoff that clearly predicts IUD expulsion. No expulsions occurred when endometrial thickness was less than 8 mm; however, most women had thicker endometria than this. Using an endometrial thickness cutoff of 8 mm, the sensitivity of detecting expulsion was 100% since all expulsions occurred in women with endometrial thickness greater than 8 mm. However, the specificity at 8 mm was only 36%, as the majority of women with greater endometrial thicknesses did not expel their IUDs. 3.2. Baseline IUD position analysis The distribution of baseline IUD position is shown in Fig. 3. Initial IUD position (immediately after insertion) was weakly associated with expulsion. Expelled and retained IUDs were a median of 4 mm and 2 mm from the most fundal aspect of the endometrial cavity, respectively. IUDs positioned lower in the uterus were more likely to expel (p = .03). The immediate group baseline IUD position (n = 61) was a median of 4.5 mm from the fundus in those who expelled versus 2 mm in those who did not (p = .008). Delayed group position (n = 53) was 3 mm for both expelled and retained IUDs (p = .89). The ROC curve evaluating initial IUD position after insertion as a predictor for expulsion did not reveal a cutoff where expulsion was likely. The area under the curve was 0.68 (95% CI 0.53–0.83). Requests for voluntary IUD

removal (n = 13) were not associated with baseline IUD position (p = .89). 3.3. IUD movement analysis We studied changes in IUD position over the 6-month study. Of the 114 women with baseline position measurements, we excluded 13 for IUD removal and 15 for expulsion. IUD position was available for 86, 67 and 82 women at baseline, follow-up and exit, respectively. Of these, 65 women had all three position measurements recorded; their data were used for the IUD movement analysis. We saw IUD movement both towards the fundus and towards the cervix throughout the 6 months, with 31 of 65 women reversing the direction of movement. While the majority of IUDs moved a few millimeters, we recorded IUD movement up to 43 mm towards the cervix without expulsion and IUD movement of up to 22 mm towards the fundus between follow-up and exit. We graphed the movement of the subset of women (n = 31/65) whose IUDs moved more than 2 mm during the study (Fig. 4). Overall, IUDs moved slightly towards the cervix during the 6 months. The median IUD positions at insertion, at the 6–8 week IUD follow-up and at the 6-month exit visit were 2 mm [interquartile range (IQR) 2–4], 3 mm (IQR 2–5) and 4 mm (IQR 3–6), respectively, from the most fundal aspect of the endometrial cavity; thus, IUDs moved a median of

N. Shimoni et al. / Contraception 89 (2014) 434–439

A

437

Endometrial thickness one week after mifepristone by expulsion status

Number of Participants

18 16 Expelled

14

Not expelled

12 12

10 8 6

11

4 6

2 0

9

9

8

6 4

5

5 5

5

5

1

3 1

1

2

3

1

4

5

6

7

8

1

1

1

1

1

2

1

2

3

2

1 0

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

Endometrial thickness (mm)

B

ROC curve: Sensitivity and 1-specificity of EMT as a predictor for expulsion (AUC=.72) 1 0.9

Sensitivity

0.8 0.7 0.6 0.5

Diagonal

0.4

ROC curve

0.3 0.2 0.1 0 0

0.2

0.4

0.6

0.8

1

1-specificity Fig. 2. (A) Endometrial thickness by expulsion status. (B) ROC curve demonstrating sensitivity and 1−specificity for each endometrial thickness, AUC = 0.72 (95% CI 0.61–0.83).

2 mm towards the cervix between insertion and 6 months later (p b .0001). When stratified by randomization group, IUDs in both groups moved towards the cervix over the 6month study period.

Table 1 Sensitivity, specificity, positive and negative predictive values (PPV, NPV), and likelihood ratio of a positive test (LR +) at each endometrial thickness to predict expulsion. EMT (mm)

Sensitivity

Specificity

PPV

NPV

LR +

3 5 7 9 11 13 15 17 19 21 23 25 27 29

1.00 1.00 1.00 1.00 0.87 0.47 0.33 0.13 0.13 0.00 0.00 0.00 0.00 0.00

0.01 0.05 0.16 0.36 0.54 0.70 0.82 0.92 0.95 0.99 0.99 0.99 0.99 1.00

0.13 0.14 0.15 0.19 0.22 0.19 0.22 0.20 0.29 0.00 0.00 0.00 0.00 *

1.00 1.00 1.00 1.00 0.96 0.90 0.89 0.87 0.88 0.87 0.87 0.87 0.87 0.87

1.01 1.05 1.20 1.56 1.89 1.58 1.81 1.63 2.61 0.00 0.00 0.00 0.00 *

4. Discussion Our study found that women with thicker endometria 1 week later after medical abortion were slightly more likely to expel the copper IUD and that expulsions were unlikely with thin endometria. The endometrial thicknesses overlapped considerably between the women who retained or expelled the IUD, and our analysis did not find a clear cutoff when expulsion became predictable. Most women did not expel the IUD, even with thicker endometria. We found it compelling that a participant with an endometrial thickness of 27 mm retained her IUD. We know from the literature that at least a third of women do not return for IUD insertion after abortion. Thus, denying women early IUD insertion decreases IUD use. We conducted a simple sensitivity analysis to look at 6-month IUD use if we were to delay insertion for the 78 women with endometrial thicknesses greater than 8 mm. Based on the literature and our data, we assume a 7% expulsion rate and that a third of women turned away will not return for insertion (n = 26). Of the remaining 52 IUD insertions, 7% of women (n = 4) will expel the IUD, resulting in 48 women using the IUD at 6 months. In comparison, if we insert IUDs

438

N. Shimoni et al. / Contraception 89 (2014) 434–439

Number of Participants

45 40 35

Expelled

Not expelled

30 25 39

20 15 10

15

12

2

3

2

3

3

4

5

6

9

5 5

0 0

7

1

3

1

2

7

1

1

8

9

1

1

1

1

1

1 1

2

1

10 11 12 13 14 15 16 17 18 19 20

Distance from most fundal aspect of endometrium (mm) Fig. 3. Initial IUD position by expulsion status.

Distance from IUD to endometrial fundus (mm)

for all 78 women, 5 will expel the IUD, and 73 women will be using the IUD at 6 months. Thus, 25 additional women would be using IUDs at 6 months if offered early insertion. We weighed together the slightly increased expulsion rate, the low specificity for endometrial thickness to predict expulsion and the high likelihood women will not return for IUD insertion and conclude that a thicker endometrium post medical abortion should not preclude IUD insertion at the follow-up visit. Retained IUDs moved throughout the study. Although the median change in position was small, many individuals experienced large changes in IUD position. The 2-mm median movement towards the cervix we observed was consistent with another interval IUD position study [6]. In that study of interval copper IUD insertion (n = 214), the 90th percentile IUD-endometrial measurements (from the most fundal aspect of the endometrium to the top of the IUD) were 5, 6 and 8 mm immediately after insertion, at 30 days and at 90 days, respectively. In that study, only one IUD farther than the 90th percentile cutoff from the fundus postinsertion expelled by 90 days.

These results are also consistent with ultrasound data published by de Kroon et al. (n = 195), who showed that five of seven initially “misplaced” IUDs were subsequently “correctly placed” 6 weeks later (defining “correct” as less than 5 mm from the most fundal aspect of the endometrium) [7]. A cohort study by Morales-Rosello followed 32 women with interval IUDs placed more than 2 cm from the fundus [8]. Within 2–3 months, 31 of 32 moved towards the fundus, and 1 IUD moved toward the cervix without expelling. Together, these studies of IUD movement after interval insertion show that IUDs measured farther from the fundus immediately after insertion are not destined to expel and may move closer to the fundus with time [6,7]. We conclude from these studies that a small displacement from the uterine fundus is not an indication for IUD removal. Further, we infer that avoiding routine sonography may minimize unnecessary removals. Our study had several limitations. We studied the CuT380A only; our findings may not be generalizable to the levonorgestrel intrauterine system. Our study’s expulsion rate (in both the early and late insertion groups) was higher

Movement greater than 2 mm of retained IUDs over time (n=31/65) 48 25

20

15

10

5

0 At insertion

At follow-up

At exit

Fig. 4. IUD movement greater than 2 mm throughout the study. IUD movement in the 34 of 65 women whose IUDs moved 2 mm or less is not graphed.

N. Shimoni et al. / Contraception 89 (2014) 434–439

than commonly cited after suction abortion, which may, in part, be secondary to our definition of expulsion (complete expulsions out of uterus and any part of the IUD in the cervix as detected by research ultrasound surveillance). Our expulsion rate decreased to 6.7% (n = 9/134) when ultrasound-detected expulsions were omitted. This rate is consistent with the published literature [9]. Overall, copper IUD expulsion after medical abortion is uncommon. The risk of expulsion should be weighed against the risk of not returning for insertion and subsequent unintended pregnancy when method initiation is delayed. In our data, no clear cutoffs emerged to predict expulsion using ultrasound measurements; therefore, we do not recommend restricting IUD insertion based on ultrasound findings.

[3]

[4]

[5]

[6]

[7]

References

[8]

[1] Bednarek PH, Creinin MD, Reeves MF, Cwiak C, Espey E, Jensen JT. Immediate versus delayed IUD insertion after uterine aspiration. New England Journal of Medicine 2011;364(23):2208–17. [2] Fox MC, Oat-Judge J, Severson K, Jamshidi RM, Singh RH, McDonald-Mosley R, et al. Immediate placement of intrauterine

[9]

439

devices after first and second trimester pregnancy termination. Contraception 2010;83(1):34–40. Grimes DA, Lopez LM, Schulz KF, Stanwood NL. Immediate postabortal insertion of intrauterine devices. Cochrane Database of Systematic Reviews. 2010 (6):CD001777. PubMed PMID: 20556754. Shimoni N, Davis A, Ramos ME, Rosario L, Westhoff C. Timing of copper intrauterine device insertion after medical abortion: a randomized controlled trial. Obstetrics & Gynecology. 2011 Sep;118(3):623–8. PubMed PMID: 21860292. Betstadt SJ, Turok DK, Kapp N, Feng K-T, Borgatta L. Intrauterine device insertion after medical abortion. Contraception 83(6):517–21. Faúndes D, Perdigão A, Faúndes A, Bahamondes L, Petta CA. T-shaped IUDs accommodate in their position during the first 3 months after insertion. Contraception 2000;62(4):165–8. de Kroon CD, van Houwelingen JC, Trimbos JB, Jansen FW. The value of transvaginal ultrasound to monitor the position of an intrauterine device after insertion. A technology assessment study. Human Reproduction. 2003 Nov;18(11):2323–7. PubMed PMID: 14585882. Morales-Roselló J. Spontaneous upward movement of lowly placed Tshaped IUDs. Contraception 2005;72(6):430–1. Stanwood NL, Grimes DA, Schulz KF. Insertion of an intrauterine contraceptive device after induced or spontaneous abortion: a review of the evidence. BJOG: An International Journal of Obstetrics & Gynaecology. 2001 Nov;108(11):1168–73. PubMed PMID: 11762657.