Medical management of early pregnancy failure with misoprostol is an effective and safe alternative to surgical management by dilation and suction curettage (D&C).1 Ultrasonography is commonly used to determine if expulsion of the gestational sac has occurred after misoprostol treatment. Some investigators have used endometrial thickness to guide clinical management.2–4 However, endometrial thickness criteria for intervention vary among studies,5 and multiple investigators have questioned whether such measurement is clinically useful.6–8
Endometrial thickness has been shown to be a useful tool to predict intrauterine pathology in postmenopausal women.9 In the setting of both elective medical abortion and spontaneous abortion, a variety of endometrial thickness values have been proposed as cutoffs that indicate the need for surgical intervention by vacuum aspiration.10,11 However, the normal appearance of the endometrial cavity after spontaneous abortion is variable and poorly described.12
Prospectively collected endometrial thickness data from women having a D&C and medical treatment is lacking. Accordingly, we planned this secondary analysis as part of a multicenter randomized trial comparing the efficacy and safety of misoprostol to D&C for the treatment of early pregnancy failure.1 We sought to compare endometrial thickness after misoprostol to D&C and to evaluate the predictive value of endometrial thickness after misoprostol for subsequent surgical intervention.
MATERIALS AND METHODS
The trial was approved by the institutional review boards of the National Institute of Child Health and Human Development, Columbia University, the University of Miami, the University of Pennsylvania, the University of Pittsburgh, and the Clinical Trials and Surveys Corporation. The study protocol, population demographics, and treatment outcomes have been described previously.1 Briefly, women with an anembryonic gestation, embryonic or fetal demise, incomplete abortion, or inevitable abortion at less than 12 weeks of gestation were eligible for inclusion.
Women were randomized in a 3:1 ratio to medical or surgical management. Surgical management consisted of manual or electric dilation and suction curettage (D&C) performed or supervised by a study investigator. Women allocated to medical treatment had four 200-mcg tablets (800 mcg) of misoprostol inserted into the posterior fornix of the vagina through a speculum. These participants returned 2 days after misoprostol treatment (range 2–5 days) for a follow-up evaluation including transvaginal ultrasonography. If the ultrasound examination demonstrated a gestational sac or an endometrial lining more than 30 mm, a second 800-mcg dose of vaginal misoprostol was administered in the same fashion. Participants receiving a second dose of misoprostol returned 3 to 6 days later for repeat evaluation. If the expulsion of products of conception was still incomplete, dilation and suction curettage was offered.
All participants returned for follow-up visit 14 days after treatment (range 12–17 days). A telephone interview was conducted 29 days after treatment (range 24–34 days) to determine whether any participant underwent additional treatment. Women with symptoms potentially related to the study treatment were followed until resolution.
For this analysis, we examined women whose endometrial thickness was assessed during follow-up visits. We defined endometrial thickness as the maximal dimension of the endometrial cavity in the anterior-posterior plane of the uterus. From the misoprostol group, we included women who had an endometrial thickness measured 2 days or 14 days after treatment. From the D&C group, we included women who had an endometrial thickness measured 14 days after treatment.
We excluded women who had a gestational sac identified on the day of the ultrasonographic examination and women who underwent a D&C (or repeat D&C for the D&C group) before the ultrasonogram. For the evaluation 2 days after medical treatment, we excluded women if the interval between treatment and the ultrasonogram was less than 2 days or more than 4 days. For the evaluation 14 days after medical or surgical treatment, we excluded women if the interval between treatment and the ultrasonogram was less than 11 days or more than 19 days.
Demographic and outcomes variables were compared using Student t test for continuous variables and χ2 or Fisher exact tests, as appropriate, for proportions. We performed comparisons of the endometrial thickness on study day 15 between women receiving initial treatment with misoprostol and those undergoing D&C, using both Student t test and Mann-Whitney rank sum test.
We then examined the predictive value of endometrial thickness for subsequent surgical management among women in the misoprostol group. We constructed receiver operating characteristic (ROC) curves for the utility of endometrial thickness in predicting the need for subsequent D&C. We then fit maximum likelihood ROC models using a binormal distribution.13 Additionally, endometrial thickness was dichotomized at each 5-mm increment from 10 mm to 30 mm. We then examined each endometrial thickness threshold value (from 10 mm to 30mm in 5mm increments) as predictors of D&C and calculated the sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), and ROC curve area.14 All statistical analyses were performed using Stata 9 (StataCorp, College Station, TX).
A total of 652 women were enrolled between March 2002 and March 2004, of whom 491 were assigned to receive misoprostol and 161 to vacuum aspiration. Of women receiving misoprostol, 343 met the criteria for this secondary analysis at 2 days after treatment and 375 at 14 days after treatment. Of women undergoing vacuum aspiration, 137 met criteria for analysis at 14 days after treatment. Reasons for exclusion are shown in Table 1. The demographics of the women included in this analysis did not differ from the study as a whole (data not shown).
The endometrial thickness values at 14 days after treatment ranged widely for women in both groups (Fig. 1). The mean endometrial thickness was 9.0 mm for the misoprostol group and 6.9 mm for the D&C group. On average, the endometrial thickness was 2.1 mm greater (95% CI 1.0–3.2 mm) in the misoprostol group than the D&C group (P<.001).
Among the 343 women in the misoprostol group with endometrial thickness assessed at 2 days (range 2–4 days) after treatment, 13 women (3.8%) subsequently required surgical intervention. The D&C procedures were performed a median of 14 days after this evaluation (range 2–45 days). The mean endometrial thickness was 19.2 mm and 14.0 mm for women who did and did not require vacuum aspiration, respectively. The difference was 5.2 mm (95% CI 1.6–8.8 mm). The area under the fitted binormal-regression ROC curve was 0.713 (95% CI 0.577–0.849), shown in Figure 2. The test characteristics of endometrial thickness at 5-mm intervals are shown in Table 2.
With increasing time after misoprostol over the interval from 2 days to 4 days after treatment, the endometrial thickness decreased significantly. In women who did not need surgical intervention, the mean endometrial thickness was 15.6, 13.3, and 11.5 mm after a 2-day (n=160), 3-day (n=93), and 4-day (n=77) interval, respectively (P<.01). The decrease in endometrial thickness for women who required surgical intervention was similar. However, with only 13 women in this group, the numbers are too small to provide stable estimates of the mean. The number of women needing surgical intervention at each time point was 6 (3.6%), 3 (3.1%), and 4 (4.9%) after a 2-day, 3-day, and 4-day interval, respectively (P>.2).
Among the 375 women in the misoprostol group with endometrial thickness assessed at 14 days (range 11–19 days) after treatment, 12 (3.2%) women subsequently required surgical intervention. The D&C procedures were performed a median of 8 days after this evaluation (range 0–94 days). The mean endometrial thickness was 14.3 mm and 8.8 mm for women who did and did not require vacuum aspiration, respectively. The difference was 5.5 mm (95% CI 2.3–8.8 mm). The area under the fitted binormal-regression ROC curve was 0.727 (95% CI 0.583–0.872), shown in Figure 3. The test characteristics of endometrial thickness at 5-mm intervals are shown in Table 3. Endometrial thickness did not vary significantly by time interval between medical treatment and ultrasonography over the range of 11 days to 19 days between treatment and endometrial thickness assessment (P>.05).
Ultrasonography is commonly performed after medical management of early pregnancy failure to confirm expulsion of the gestational sac; however, it is not routinely performed after vacuum aspiration. Previous, smaller studies have shown that endometrial thickness varies widely after elective medical abortion and medical management of early pregnancy failure.5–7,15 Data are limited on the normal appearance of the uterine lining after vacuum aspiration.12,16 We found that the endometrial thickness for women undergoing medical management was similar to the endometrial thickness after surgical management.
Although endometrial thickness after medical management was significantly greater in women who subsequently underwent surgical intervention, these data show that endometrial thickness is not a useful test for predicting subsequent surgical intervention. Although the number of women needing D&C after the endometrial thickness assessment was small, this was more than that in prior studies of early pregnancy failure. Despite relatively few events, the width of the 95% confidence intervals around sensitivity, specificity, NPV, and PPV (Tables 2 and 3) excludes the possibility that endometrial thickness is a good test for predicting subsequent D&C.
Receiver operating characteristic curves are a useful tool for quantitatively summarizing the usefulness of a test with a continuous outcome such as endometrial thickness.17 The ROC curve compares the sensitivity to the false positive rate (1–specificity) over all possible threshold values. A perfect test would have an area under the ROC curve of 1.0, whereas a test equivalent to flipping a coin would have an area of 0.5. In the case presented here, the area under the ROC curve for both 2 days and 14 days after medical management is approximately 0.72. Thus endometrial thickness is better than chance. However, the upper limit of the 95% confidence intervals are both below 0.90, precluding the possibility that this test is good at discriminating between those women who will and will not need a D&C.
These findings corroborate the results of previous studies showing that endometrial thickness is not associated with a need for subsequent surgical intervention.6,7 Importantly, the inability of endometrial thickness to predict subsequent surgical intervention did not change significantly with increased threshold values, as shown in Tables 2 and 3. As expected, dichotomous thresholds did not perform as well as endometrial thickness as a continuous variable. For all threshold values, the area under the ROC curve was lower that the area under the ROC curve for endometrial thickness as a continuous predictor. For all endometrial thickness cutoff values from 10 mm through 30 mm, the sensitivity and specificity exclude the possibility that any cutoff makes a good test to detect those women who will need a D&C.
An endometrial thickness of less than 15 mm has been used to define a completely evacuated uterus after spontaneous abortion.2 This definition has been used in clinical trials of medical management of early pregnancy failure as an indication for surgical intervention.3,4,18 Although clinical presentation was used primarily to guide management, an endometrial thickness of 30 mm was used as a threshold for offering D&C in this study. However, although D&C was recommended to six participants meeting or exceeding this threshold, five participants declined and did not later have an indication for D&C. Consequently, even an endometrial thickness above 30 mm was not a reliable predictor of D&C.
Notably, two of the women underwent D&C because a gestational sac was identified at 14 days after treatment after the initial ultrasonogram at 2 days after treatment failed to identify a gestational sac. The charts for these participants were reviewed, and the ultrasonogram at 2 days after treatment revealed heterogeneous material within the uterus with endometrial thickness measurements of 13 mm and 17 mm. A third participant had a gestational sac identified at 14 days after treatment but was expectantly managed. The data for these three women was included in analysis for 2 days after treatment because future events were unknown at the time of that evaluation. These three cases show that ultrasonography soon after medical management can falsely confirm expulsion. However, these events seem to be rare.
For women choosing medical management of early pregnancy failure, ultrasonography is a clinically useful means of confirming expulsion of the gestational sac. However, endometrial thickness is not a useful predictor of subsequent surgical intervention when gestational sac expulsion has been confirmed. Therefore, in the management of early pregnancy failure, clinical presentation, not endometrial thickness, should be used to define the need for dilation and curettage.
1. Zhang J, Gilles JM, Barnhart K, Creinin MD, Westhoff C, Frederick MM, et al. A comparison of medical management with misoprostol and surgical management for early pregnancy failure. N Engl J Med 2005;353:761–9.
2. Nielsen S, Hahlin M. Expectant management of first-trimester spontaneous abortion. Lancet 1995;345:84–6.
3. Ayres-de-Campos D, Teixeira-da-Silva J, Campos I, Patricio B. Vaginal misoprostol in the management of first-trimester missed abortions. Int J Gynaecol Obstet 2000;71:53–7.
4. Graziosi GC, Mol BW, Reuwer PJ, Drogtrop A, Bruinse HW. Misoprostol versus curettage in women with early pregnancy failure after initial expectant management: a randomized trial. Hum Reprod 2004;19:1894–9.
5. Harwood B, Meckstroth KR, Mishell DR, Jain JK. Serum beta-human chorionic gonadotropin levels and endometrial thickness after medical abortion. Contraception 2001;63:255–6.
6. Creinin MD, Harwood B, Guido RS, Fox MC, Zhang J, et al. Endometrial thickness after misoprostol use for early pregnancy failure. Int J Gynaecol Obstet 2004;86:22–6.
7. Reynolds A, Ayres-de-Campos D, Costa MA, Montenegro N. How should success be defined when attempting medical resolution of first-trimester missed abortion? Eur J Obstet Gynecol Reprod Biol 2005;118:71–6.
8. Luise C, Jermy K, May C, Costello G, Collins WP, Bourne TH. Outcome of expectant management of spontaneous first trimester miscarriage: observational study. BMJ 2002;324:873–5.
9. Smith-Bindman R, Kerlikowske K, Feldstein VA, Subak L, Scheidler J, Segal M, et al. Endovaginal ultrasound to exclude endometrial cancer and other endometrial abnormalities. JAMA 1998;280:1510–7.
10. Durfee SM, Frates MC, Luong A, Benson CB. The sonographic and color doppler features of retained products of conception. J Ultrasound Med 2005;24:1181–6.
11. Cetin A, Cetin M. Diagnostic and therapeutic decision-making with transvaginal sonography for first trimester spontaneous abortion, clinically thought to be incomplete or complete. Contraception 1998;57:393–7.
12. Bar-Hava I, Aschkenazi S, Orvieto R, Perri T, Shalev J, Dicker D, et al. Spectrum of normal intrauterine cavity sonographic findings after first-trimester abortion. J Ultrasound Med 2001;20:1277–81.
13. Dorfman DD, Alf E Jr. Maximum likelihood estimation of parameters of signal detection theory—a direct solution. Psychometrika 1968;33:117–24.
14. Dorfman DD, Alf E Jr. Maximum likelihood estimation of parameters of signal detection theory and determination of confidence intervals-rating method data. J Math Psychol 1969;6:487–96.
15. Cowett AA, Cohen LS, Lichtenberg ES, Stika CS. Ultrasound evaluation of the endometrium after medical termination of pregnancy. Obstet Gynecol 2004;103:871–5.
16. Dillon EH, Case CQ, Ramos IM, Holland CK, Taylor KJ. Endovaginal US and Doppler findings after first-trimester abortion. Radiology 1993;186:87–91.
17. Obuchowski NA. Receiver operating characteristic curves and their use in radiology. Radiology 2003;229:3–8.
18. Nielsen S, Hahlin M, Platz-Christensen Randomised trial comparing expectant with medical management for first trimester miscarriages. Br J Obstet Gynaecol 1999;106: 804–7.
© 2008 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
Figure. No caption available.