- Email: [email protected]
The accuracy of using last menstrual period to determine gestational age for first trimester medication abortion: a systematic review
Contraception xx (2014) xxx – xxx
Review Article
The accuracy of using last menstrual period to determine gestational age for first trimester medication abortion: a systematic review☆,☆☆ Dana Schonberg a,⁎, Lin-Fan Wang a , Ariana H. Bennett b , Marji Gold a , Emily Jackson c a
Montefiore Medical Center/Albert Einstein College of Medicine, Bronx NY b Albert Einstein College of Medicine, Bronx NY c Planned Parenthood of Orange and San Bernardino Counties, Orange, CA Received 20 September 2013; revised 11 July 2014; accepted 12 July 2014
Abstract Objective: We sought to evaluate the accuracy of assessing gestational age (GA) prior to first trimester medication abortion using last menstrual period (LMP) compared to ultrasound (U/S). Study Design: We searched Medline, Embase and Cochrane databases through October 2013 for peer-reviewed articles comparing LMP to U/S for GA dating in abortion care. Two teams of investigators independently evaluated data using standard abstraction forms. The US Preventive Services Task Force and Quality Assessment of Diagnostic Accuracy Studies guidelines were used to assess quality. Results: Of 318 articles identified, 5 met inclusion criteria. Three studies reported that 2.5–11.8% of women were eligible for medication abortion by LMP and ineligible by U/S. The number of women who underestimated GA using LMP compared to U/S ranged from 1.8 to 14.8%, with lower rates found when the sample was limited to a GA b 63 days. Most women (90.5–99.1%) knew their LMP, 70.8–90.5% with certainty. Conclusion: Our results support that LMP can be used to assess GA prior to medication abortion at GA b 63 days. Further research looking at patient outcomes and identifying women eligible for medication abortion by LMP but ineligible by U/S is needed to confirm the safety and effectiveness of providing medication abortion using LMP alone to determine GA. © 2014 Elsevier Inc. All rights reserved. Keywords: Pregnancy dating; Menstruation; Medication abortion; Gestational age
1. Introduction While the number of abortions worldwide has remained stable in recent years, the proportion of unsafe abortions has increased [1]. Each year, it is estimated that 47,000 women die, and over 8 million women experience complications from unsafe abortion [2,3]. Medical methods of abortion hold promise to improve access to care in areas with limited resources, reducing maternal morbidity and mortality. Prior research has greatly simplified the process of mifepristone– misoprostol abortion. Alternative methods of follow-up [4–9] and self-administration of misoprostol [10–12] have allowed first trimester medication abortion (MAB) to be provided with
fewer visits to a health care facility. Further demedicalization of the process will allow more women to access safe abortion care in areas with limited resources. Previous research has called into question the need for routine ultrasound (U/S) for gestational dating prior to MAB and has demonstrated the feasibility of simplifying the abortion process for women by using ultrasonography only when needed [13,14]. We undertook this systematic review of published literature to evaluate the accuracy of using last menstrual period (LMP) compared to U/S for assessing gestational age (GA) prior to first trimester MAB.
2. Methods ☆
Funding: Society of Family Planning. The authors have no financial or other conflicts of interest to declare. ⁎ Corresponding author. E-mail address: [email protected] (D. Schonberg).
☆☆
http://dx.doi.org/10.1016/j.contraception.2014.07.004 0010-7824/© 2014 Elsevier Inc. All rights reserved.
We searched the Pubmed, Ovid Medline, Embase and Cochrane databases for peer-reviewed articles in all languages published from database inception through
2
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
October 2013, including evidence on the use of LMP for GA dating in abortion care (Appendix A). Reference lists from included articles identified by the search as well as key review articles were hand searched to identify additional articles. We did not attempt to identify unpublished articles or abstracts from scientific conferences. We included articles comparing GA dating by LMP to dating by U/S. We included only articles related to abortion care and not those related to obstetrical care. Articles were excluded if they lacked the necessary comparative data or if they did not pertain to abortion care at a GA less than 63 days. We also excluded case reports, letters and review articles. Our main outcome of interest was the accuracy of LMP to determine GA for MAB at GA less than 63 days. We focused on the number of women who underestimated their GA as well as the proportion of women eligible for a MAB by LMP but ineligible by U/S examination. We also report data on ectopic pregnancies, missed abortions and the clinical implications of inaccurate menstrual dating if available in articles included in the review. All five investigators independently reviewed all article titles and abstracts for possible study inclusion. After the first round of exclusion, the investigators divided into two teams (DS, LFW and EJ and MG, AB and EJ); each team was responsible for reviewing half of the remaining articles. Using standardized data abstraction forms, each investigator independently extracted and summarized data from the articles assigned to their team. The senior investigator (EJ) abstracted all articles and ensured consistency between the two teams. The two teams met regularly to review the abstraction process. At all stages, differences were resolved by consensus. We sent an email to article authors to request additional data if an article mentioned collecting both LMP and U/S data but did not present those data in the manuscript. Data acquired via author request were reviewed independently by each member of the research team, and consensus on the quality of data and appropriateness for inclusion was reached through discussion. We used the system for grading evidence developed by the US Preventive Services Task Force (USPSTF) [15] and the Quality Assessment of Diagnostic Accuracy Studies (QUADAS)-2 system [16]. Following the USPSTF guidelines, the internal validity of each study was judged based upon use of a relevant screening test (LMP) reliably administered to an appropriate sample of adequate size and compared to a credible reference standard (U/S). The risk of bias and applicability concerns identified in the QUADAS system were used to inform a decision on the overall quality of each study. Of note, we assessed data quality only as it applied to our specific research question; therefore, grading is not a reflection on the article as a whole.
remained. An additional 29 studies were identified after reviewing the citations of key articles, for a total of 318 articles for review. After excluding case reports, reviews, letters, editorials and articles not relevant to the topic question, 86 articles remained, and the full text of each article was reviewed (see Fig. 1). Five articles met inclusion criteria (Table 1); we were able to assess four articles as diagnostic accuracy studies [17–20] and one secondary analysis of a clinical trial [21]. Two of the included articles [18,21] cite additional publications for details on methodology [22,23]. Heterogeneity of data prevented meta-analysis, and lack of data presented in each article precluded the calculation of statistical measures including sensitivity and specificity. We attempted to contact the authors of 28 articles for additional data; 11 authors responded and three provided data. After review, it was agreed by all investigators that the data were not appropriate for inclusion in this review. 3.1. LMP compared to U/S Of the five articles included in the review, three articles [17–19] studied the number of women considered eligible for a MAB based upon LMP who were found to be ineligible based upon U/S. Bracken [17], in an article rated to be of “fair” quality, enrolled women seeking MAB at 10 clinics in the US. Bracken found that only 2.5% (76/3041) of women certain of their LMP would have been offered a MAB based upon LMP despite subsequent transvaginal U/S showing a GA above 63 days. When including women uncertain of their LMP, the number increased to 3.3% (142/4257) [17]. Blanchard, in an article rated to be of fair quality, studied women seeking first trimester abortion in three provinces in South Africa. The provider conducting the U/S was blinded to LMP. Blanchard found a higher rate of women, 11.8% (72/608), who were b 56 days by LMP but N 56 days by U/S [18]. In a study rated to be of “poor” quality, McGalliard
3. Results A total of 462 articles were retrieved via database searches; after adjusting for duplicates, 289 unique articles
Fig. 1. Flowchart of the systematic review.
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
found that 7.2% (17/237) of women referred to a UK hospital clinic for an abortion would have been offered a MAB based upon reported LMP despite having a GA by U/S greater than 63 days. McGalliard used transvaginal U/S unless the abdominal examination was larger than expected by LMP in which case transabdominal U/S was used. The authors collected LMP data and conducted the U/S, so this study was not blinded. Two articles reported the proportion of women who underestimated their GA using LMP compared with U/S [19,21]. McGalliard found that 14.8% of women (35/237) underestimated their GA by more than 1 week [19]. Of note, some of the women included in McGalliard's calculations had a GA above 13 weeks, unlike Harper who only included women with a GA less than 63 days in this calculation. Harper, in an article rated to be of poor quality, conducted a secondary analysis of a clinical trial to assess the safety and efficacy of MAB with less clinician oversight. Harper found that 1.8% of women (38/2121) with a GA less than 63 days by LMP underestimated their GA by more than 2 weeks compared to a physician's estimate based upon history, exam and transvaginal U/S [21,23]. Two articles reported on the strength of correlation between GA by LMP to that by U/S [19,20]. McGalliard found moderate agreement, r= 0.44 [19]. Harkness' study, which only studied correlation coefficients and was rated to be of poor quality, studied women b 63 days who had been referred. Measurements were taken by transabdominal U/S 48 h after mifepristone administration and were compared to GA by LMP as well as direct measurements of tissue. Harkness reported similar values of r= 0.48 between LMP and U/S measurement of gestational sac and r= 0.57 between LMP and U/S measurement of crown rump length [20]. Blanchard measured agreement and noted that 32% of LMP estimates of GA were within 1 week of U/S but did not report raw data or strength of correlation [18]. Two articles [18,21] assessed the impact of demographic variables on the accuracy of GA estimate by LMP. As demographic variables were studied as secondary outcomes, these studies may not have been powered to find a difference if one existed. Blanchard found that women who were certain of their LMP dates did not differ from others by age, gravidity, parity, education or employment status [18]. On the other hand, in multivariate analysis, Harper showed that Hispanic and African American women in the US were more likely to underestimate their GA by more than a week using LMP when compared to White women [21]. 3.2. Additional findings Four studies [17‐19,21] included the number of women able to report their LMP. Almost all women (90.5–99.1%) were able to recall their LMP. Young age was associated with inability to report LMP [21]. The number of women who stated that they were able to report their LMP with certainty varied but was still high (70.8–90.5%) [17,19].
3
Although Bracken reported that 70.8% of women could give a date with certainty, raw numbers were not given. Two studies [17,19] reported data on ectopic pregnancy, abnormal pregnancies or possible clinical implications of foregoing routine U/S. Bracken found that of 4484 women, screening with history and physical exam identified 32 as possibly having an ectopic or molar pregnancy. Of these 32 women, 9 were diagnosed with an ectopic pregnancy and 2 diagnosed with molar pregnancy using U/S or clinical and/or laboratory examinations. There were no ectopic pregnancies found in McGalliard's study population, but one woman (of 300) was diagnosed with an ectopic pregnancy when she presented emergently the day prior to her scheduled enrollment. The use of U/S by McGalliard added additional clinical information for 24/283 (8.5%) women: 4 were found not to be pregnant, 17 had an early pregnancy loss, and 3 had a twin pregnancy.
4. Discussion Our review found five articles that compared GA by LMP to that by U/S; all were deemed to be of fair or poor quality for the question under review. Rates of women presenting for care who were eligible for a MAB by LMP but ineligible by U/S were reported by three studies and ranged between 2.5 and 11.8%. Two studies reported rates of underestimation of GA by LMP of 1.8%–14.8%. Of particular interest is the proportion of women who underestimate their GA by LMP, as this pertains to the safety and effectiveness of abortion care. Harper showed low rates of underestimation (1.8%) for women at GA b 63 days. McGalliard showed higher rates but used a sample that included women beyond the first trimester. These results agree with previous research by Norman who found that LMP becomes less accurate at assessing GA as GA increases [24]. Similar results were found by Ellertson who reported low rates (1.4%) of women underestimating their GA when limiting the sample to GA b 56 days [25] and higher rates (10.4–12.4%) when including the entire first trimester [26]. Our review found rates of women presenting for care who were eligible for a MAB by LMP and ineligible by U/S of 2.5–11.8% [17–19]. Higher rates (7.2–11.8%) were found by Blanchard and McGalliard, but these studies used smaller sample sizes. These findings are consistent with prior work by Ellertson, which showed that 9.8–10% of women had a GA by LMP b 56 days but N 56 days by provider's estimate, with rates lowered to 4.6–5.5% when using a cutoff of 63 days [26]. When clinician exam was used in addition to LMP, Bracken found that only 1.6% (63/3874) of women were found to be b 63 days by LMP and N 63 days by U/S [17]. Prior research has shown that using LMP plus clinician exam accurately assesses GA and results in high success rates when used in MAB services [13]. Further research comparing LMP to U/S with and without the use of clinician exam could be of value.
4
Table 1 Summary of evidence for accuracy of reported LMP to determine GA. Study country funding
Study design
Population
Results
Strengths
Weaknesses
Quality of evidence
Bracken, 2011 [17]
Diagnostic accuracy
4484 women seeking early medication abortion at 10 US clinics in various cities
4447/4484 (99.1%) gave date for LMP; 70.8% gave date for LMP with certainty
Prospective data collection
Denominator for some calculations not clear
Fair
Accuracy of “certain” LMP vs. U/S: 76/3041 (2.5%) women would have been offered medication abprtion MAB ≥ 63 d by U/S (13/3041 [0.4%] ≥ 78 days)
Large N, diverse sample
Accuracy of ‘certain’ and ‘estimated’ LMP vs. U/S: 142/4257 (3.3%) women would have been offered MAB ≥ 63 days by U/S (26/4257 [0.6%] ≥ 78 days)
Methods used to measure GA clearly described
USA
Accuracy of LMP + clinician exam vs. U/S: 63/3874 (1.6%) women would have been offered MAB ≥ 63 d by U/S (1/3874 [0.02%] ≥ 78 days) Blanchard, 2007 [18,22]
Diagnostic accuracy
South Africa
673 women seeking first trimester abortion at 8 clinics
UK No funding reported
Women b 56 days GA by LMP, but N 56 days by U/S: 72/608 (11.8%)
Unknown if ultrasonographer blinded to LMP
Separate analysis of ‘certain’ and ‘uncertain’ LMP
Prospective data collection Large N
15th assigned as date of LMP for women who could only remember the month
Fair
Women referred to clinics may have had previous gestational age assessment
Poor
Multiple sites in 3 provinces
Anonymous donor
McGalliard, 2004 [19]
640/673 women (95.1%) reported LMP
Multicenter trial
Women referred to clinics may have had previous gestational age assessment
Diagnostic accuracy
262 women referred for abortion
LMP estimates within 1 week of U/S: 32% LMP estimates within 2 weeks of U/S: 51% LMP estimates within 3 weeks of U/S: 65%
U/S data collected separately and blinded
Level of agreement between methods: U/S vs. LMP estimate r= 0.36
Used standardized data collection forms
237/262 (90.5%) of women sure of their LMP
Prospective data collection
Women who were b 63 days by LMP, but N63 days by U/S: 17/237 (7.2%)
Large N
Mean difference: U/S mean 1 day b LMP (95% CI −36 to +34 days)
U/S data not gathered independently or
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
David & Lucille Packard Foundation and anonymous donor
Harper, 2002 [21,23] USA
2121 women b 63 days GA seeking abortion at 17 centers
blinded
For certain LMP: Discrepancy of N 7 days = 90/237 (38.0%) LMP underestimated GA by more than 1 week: 35/237 (14.8%) LMP overestimated GA by more than 1 week: 55/237 (23.2%)
Women with uncertain LMP excluded
2098/2121 women (98.9%) knew LMP LMP underestimated GA by more than 2 weeks: 38/2121 (1.8%) 259/2121 (12.2%) thought they had more time before reaching 63 d limit
Single center
Multi-center and geographically diverse
Retrospective, secondary data analysis
Prospective data collection
Women N63 days GA by clinical assessment excluded
Poor
Large N Statistical methods used to analyze GA data not reported Unknown if ultrasonographer blinded to LMP
Harkness, 1996 [20] UK No funding reported
Diagnostic accuracy study
Unreported number of women b63 days GA referred for MAB
Correlation coefficient (r) LMP and gest sac from scan: r= 0.48 (pb0.001, n=112) LMP and CRL from scan: r=0.57 (pb0.001, n=99)
Prospective data collection Single investigator gathered all U/S data
Women referred to clinics may have had previous gestational age assessment Women b17 years old excluded U/S performed after mifepristone administration Inconsistent N and small N reported for some measurements
Poor
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
Population Council
Secondary analysis of clinical trial assessing safety and efficacy of MAB
Correlation coefficient r=0.44
Unknown if ultrasonographer blinded to LMP
5
6
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
These studies show that most women seeking first trimester abortion can report a date for their LMP, and the majority can do so with confidence. As some of the studies in our review included women who were referred, they may have had prior knowledge of their GA, which could possibly bias these results. Still, our finding is consistent with earlier work by Wegienka and Baird [27], who established the validity of women's reported LMP by comparing prospectively recorded menstrual diaries to women's response when queried 4 weeks later. They found that 56% of women precisely recalled their LMP, 74% within 1 day, 81% within 2 days and 84% within 3 days. This conclusion remains limited, as women asked to prospectively record their menstrual histories may be more likely to accurately recall their LMP. Utilizing women's reported LMP to determine GA has the potential to make MAB services more accessible, particularly in areas where U/S is not readily available. Although using LMP to determine GA may result in a small number of women being offered a MAB after 63 days GA, research demonstrates that mifepristone and misoprostol are effective for terminating pregnancies beyond 63 days GA, with low rates of complications [28,29]. Therefore, these women would likely still have a safe MAB, albeit with a gradual decrease in efficacy as GA increases. Although Winikoff et al. noted no significant differences when comparing women undergoing MAB at 57–63 days GA with women 64–70 days GA, women at higher GA might experience a longer process of expelling the pregnancy and should be counseled appropriately [28]. A perceived benefit of routine use of U/S is the ability to identify women with abnormal pregnancies, such as nonviable, ectopic or molar pregnancies. While U/S provided some additional information, it did not provide further data for identifying ectopics over history and physical exam in the few articles that provided this information [17,19]. Studies that did report data about abnormal pregnancies [17,19] found low rates of ectopic pregnancies, consistent with population-based prevalence of around 2% [30–33]. U/S is a limited diagnostic test for ectopic, with sensitivity of only 25% for identifying ectopic pregnancy at β-hCG levels below 1500 mIU/mL, and 55% among all women regardless of β-hCG level [34]. Mol et al. showed that it is not cost effective to screen women for ectopic pregnancy using U/S unless a woman has already been identified for concern by history and physical exam [35]. Although women with spontaneous abortions and nonviable pregnancies, who would be eligible for additional treatment modalities such as expectant management, could be missed without routine use of U/S, both uterine aspiration and the medication regimen used in MAB are safe and effective treatments for spontaneous abortion. As we did not search systematically for data regarding ectopic pregnancies and missed abortions, limited conclusions can be drawn, and this remains an area for further research. There are some limitations to these data. Very few articles met our criteria for inclusion. Three studies included women
who were referred and, therefore, may have had knowledge of GA prior to enrollment. Some studies excluded women above a specific measured GA, potentially skewing the number of women who underestimated their GA. In addition, it is not known whether women seeking MABs and women seeking abortion procedures differ in their ability to assess their GA by LMP. If a difference exists, our results are further limited. Another possible limitation is the use of U/S as a reference standard. U/S has a margin of error that varies depending on the formula used to calculate GA as well as the GA of the pregnancy. Using U/S to measure GA between 7 and 10 weeks, Goldstein predicted a 95% confidence interval (CI) of ± 3 days [36]. This margin of error could affect the results of the articles included in the review. Most articles received fair or poor quality ratings due to the strict criteria used for judging quality. One strength of this review is that four of the included studies were specifically designed to assess the accuracy of GA by LMP compared to U/S. In summary, our review demonstrates that among patients seeking abortion at GA b 63 days, women's report of LMP can be used to determine GA. Our results support using LMP to assess GA prior to MAB. MAB would be more widely accessible if reported LMP could be relied upon to accurately assess GA. Research comparing LMP to U/S at GA above 63 days will be needed as the eligibility for MAB expands. Acknowledgement We would like to thank Dr. Jennifer Amico and Dr. Jason Fletcher for their comments and review of this paper. Appendix A. Search strategies Pubmed ((((("Gestational Age"[Mesh] OR "gestational age" OR "pregnancy duration" OR "pregnancy length" OR "pregnancy dating")) AND ("Menstruation"[Mesh] OR "Menstrual Cycle"[Mesh] OR menstrua* OR "last menstrual period" OR "LMP")) AND ("Abortion, Induced"[Mesh] OR "abortion") NOT "Abortion, Spontaneous"[Mesh])) Filters: Publication date to 2013/10/31 Ovid, 1946 through Oct 31 2013 1. abortion.mp. or Abortion, Induced/ 2. Abortion, Spontaneous/ 3. 1 not 2 4. gestational age.mp. or Gestational Age/ 5. pregnancy duration.mp. 6. pregnancy dating.mp. 7. pregnancy length.mp. 8. 4 or 5 or 6 or 7 9. menstrual cycle.mp. or Menstrual Cycle/ 10. menstruation.mp. or Menstruation/ 11. last menstrual period.mp.
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
12. LMP.mp. 13. menstrua*.mp. 14. 9 or 10 or 11 or 12 or 13 15. 3 and 8 and 14 Embase 'abortion'/exp OR 'abortion' OR 'abortion induced'/exp OR 'abortion induced' NOT ('abortion spontaneous'/exp OR 'abortion spontaneous') AND ('gestational age'/exp OR 'gestational age' OR 'pregnancy duration' OR 'pregnancy length' OR 'pregnancy dating') AND ('menstruation'/exp OR 'menstruation' OR 'menstrual cycle'/exp OR 'menstrual cycle' OR 'last menstrual period' OR 'lmp' OR menstrua*) AND ([article]/lim OR [article in press]/lim) AND [humans]/ lim AND [embase]/lim NOT [31-10-2013]/sd Cochrane #1 MeSH descriptor: [Abortion, Induced] explode all trees #2 MeSH descriptor: [Abortion, Spontaneous] explode all trees #3 abortion #4 #1 or #3 #5 #4 not #2 #6 MeSH descriptor: [Gestational Age] explode all trees #7 "gestational age" #8 "pregnancy duration" #9 "pregnancy length" #10 "pregnancy dating" #11 #6 or #7 or #8 or #9 or #10 #12 MeSH descriptor: [Menstruation] explode all trees #13 menstruation #14 MeSH descriptor: [Menstrual Cycle] explode all trees #15“menstrual cycle” #16 menstrua* #17 "last menstrual period" #18 LMP #19 #12 or #13 or #14 or #15 or #16 or #17 or #18 #18 #5 and #11 and #19 limited to reviews published before 10.31.13 References [1] Sedgh G, Singh S, Shah IH, Ahman E, Henshaw SK, Bankole A. Induced abortion: incidence and trends worldwide from 1995 to 2008. Lancet 2012;379(9816):625–32. [2] Unsafe abortion: global and regional estimates of the incidence of unsafe abortion and associated mortality in 2008. Geneva: World Health Organization; 2011. [3] Singh S, Wulf D, Hussain R, Bankole A, Sedgh G. Abortion worldwide: a decade of uneven progress. New York: Guttmacher Institute; 2009. [4] Clark W, Panton T, Hann L, Gold M. Medication abortion employing routine sequential measurements of serum hCG and sonography only when indicated. Contraception 2007;75(2):131–5. [5] Clark W, Bracken H, Tanenhaus J, Schweikert S, Lichtenberg ES, Winikoff B. Alternatives to a routine follow-up visit for early medical abortion. Obstet Gynecol 2010;115(2 Pt 1):264–72. [6] Perriera LK, Reeves MF, Chen BA, Hohmann HL, Hayes J, Creinin MD. Feasibility of telephone follow-up after medical abortion. Contraception 2010;81(2):143–9. [7] Fiala C, Safar P, Bygdeman M, Gemzell-Danielsson K. Verifying the effectiveness of medical abortion; ultrasound versus hCG testing. Eur J Obstet Gynecol Reprod Biol 2003;109(2):190–5.
7
[8] Ngoc NN, Bracken H, Blum J, Nga NT, Minh NH, von Nhang N, et al. Acceptability and feasibility of phone follow-up after early medical abortion in Vietnam. Obstet Gynecol 2014;23(1):88–95. [9] Cameron ST, Glasier A, Dewart H, Johnstone A, Burnside A. Telephone follow-up and self-performed urine pregnancy testing after early medical abortion: a service evaluation. Contraception 2012;86(1):67–73. [10] Schaff EA, Stadalius LS, Eisinger SH, Franks P. Vaginal misoprostol administered at home after mifepristone (RU486) for abortion. J Fam Pract 1997;44(4):353–60. [11] Elul B, Hajri S, Ngoc NN, Ellertson C, Slama CB, Pearlman E, et al. Can women in less-developed countries use a simplified medical abortion regimen? Lancet 2001;357(9266):1402–5. [12] Guengant JP, Bangou J, Elul B, Ellertson C. Mifepristone-misoprostol medical abortion: home administration of misoprostol in Guadeloupe. Contraception 1999;60(3):167–72. [13] Kaneshiro B, Edelman A, Sneeringer RK, Ponce de Leon RG. Expanding medical abortion: Can medical abortion be effectively provided without the routine use of ultrasound? Contraception 2011;83 (3):194–201. [14] Clark WH, Gold M, Grossman D, Winikoff B. Can mifepristone medical abortion be simplified? A review of the evidence and questions for future research. Contraception 2007;75(4):245–50. [15] Harris RP, Helfand M, Woolf SH, Lohr KN, Mulrow CD, Teutsch SM, et al. Current methods of the US Preventive Services Task Force: a review of the process. Am J Prev Med 2001;20(3 Suppl):21–35. [16] Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 2011;155(8):529–36. [17] Bracken H, Clark W, Lichtenberg ES, Schweikert SM, Tanenhaus J, Barajas A, et al. Alternatives to routine ultrasound for eligibility assessment prior to early termination of pregnancy with mifepristonemisoprostol. BJOG 2011;118(1):17–23. [18] Blanchard K, Cooper D, Dickson K, Cullingworth L, Mavimbela N, von Mollendorf C, et al. A comparison of women's, providers' and ultrasound assessments of pregnancy duration among termination of pregnancy clients in South Africa. BJOG 2007;114(5):569–75. [19] McGalliard C, Gaudoin M. Routine ultrasound for pregnancy termination requests increases women's choice and reduces inappropriate treatments. BJOG 2004;111(1):79–82. [20] Harkness LM, Rodger M, Baird DT. Morphological and molecular characteristics of living human fetuses between Carnegie stages 7 and 23: ultrasound scanning and direct measurements. Hum Reprod Update 1996;3(1):25–33. [21] Harper C, Ellertson C, Winikoff B. Could American women use mifepristone-misoprostol pills safely with less medical supervision? Contraception 2002;65(2):133–42. [22] Cooper D, Dickson K, Blanchard K, Cullingworth L, Mavimbela N, von Mollendorf C, et al. Medical abortion: the possibilities for introduction in the public sector in South Africa. Reprod Health Matters 2005;13(26):35–43. [23] Spitz IM, Bardin CW, Benton L, Robbins A. Early pregnancy termination with mifepristone and misoprostol in the United States. N Engl J Med 1998;338(18):1241–7. [24] Norman WV, Bergunder J, Eccles L. Accuracy of gestational age estimated by menstrual dating in women seeking abortion beyond nine weeks. J Obstet Gynaecol Can 2011;33(3):252–7. [25] Ellertson C, Elul B, Winikoff B. Can women use medical abortion without medical supervision? Reprod Health Matters 1997;5(9):149–61. [26] Ellertson C, Elul B, Ambardekar S, Wood L, Carroll J, Coyaji K. Accuracy of assessment of pregnancy duration by women seeking early abortions. Lancet 2000;355(9207):877–81. [27] Wegienka G, Baird DD. A comparison of recalled date of last menstrual period with prospectively recorded dates. J Womens Health 2005;14(3):248–52. [28] Winikoff B, Dzuba IG, Chong E, Goldberg AB, Lichtenberg ES, Ball C, et al. Extending outpatient medical abortion services through 70 days of gestational age. Obstet Gynecol 2012;120(5):1070–6.
8
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
[29] Gouk EV, Lincoln K, Khair A, Haslock J, Knight J, Cruickshank DJ. Medical termination of pregnancy at 63 to 83 days gestation. Br J Obstet Gynaecol 1999;106(6):535–9. [30] Centers for Disease C. Prevention. Ectopic pregnancy–United States, 1990-1992, 44(3). MMWR Morbidity and mortality weekly report; 1995, pp. 46–8. [31] Van Den Eeden SK, Shan J, Bruce C, Glasser M. Ectopic pregnancy rate and treatment utilization in a large managed care organization. Obstet Gynecol 2005;105(5 Pt 1):1052–7. [32] Hoover KW, Tao G, Kent CK. Trends in the diagnosis and treatment of ectopic pregnancy in the United States. Obstet Gynecol 2010;115 (3):495–502.
[33] Stulberg DB, Cain LR, Dahlquist I, Lauderdale DS. Ectopic pregnancy rates in the Medicaid population. Am J Obstet Gynecol 2013;208 (4):274.e1–7. [34] Barnhart KT, Simhan H, Kamelle SA. Diagnostic accuracy of ultrasound above and below the beta-hCG discriminatory zone. Obstet Gynecol 1999;94(4):583–7. [35] Mol BW, van der Veen F, Bossuyt PM. Symptom-free women at increased risk of ectopic pregnancy: should we screen? Acta Obstet Gynecol Scand 2002;81(7):661–72. [36] Goldstein SR, Wolfson R. Endovaginal ultrasonographic measurement of early embryonic size as a means of assessing gestational age. J Ultrasound Med 1994;13(1):27–31.
Review Article
The accuracy of using last menstrual period to determine gestational age for first trimester medication abortion: a systematic review☆,☆☆ Dana Schonberg a,⁎, Lin-Fan Wang a , Ariana H. Bennett b , Marji Gold a , Emily Jackson c a
Montefiore Medical Center/Albert Einstein College of Medicine, Bronx NY b Albert Einstein College of Medicine, Bronx NY c Planned Parenthood of Orange and San Bernardino Counties, Orange, CA Received 20 September 2013; revised 11 July 2014; accepted 12 July 2014
Abstract Objective: We sought to evaluate the accuracy of assessing gestational age (GA) prior to first trimester medication abortion using last menstrual period (LMP) compared to ultrasound (U/S). Study Design: We searched Medline, Embase and Cochrane databases through October 2013 for peer-reviewed articles comparing LMP to U/S for GA dating in abortion care. Two teams of investigators independently evaluated data using standard abstraction forms. The US Preventive Services Task Force and Quality Assessment of Diagnostic Accuracy Studies guidelines were used to assess quality. Results: Of 318 articles identified, 5 met inclusion criteria. Three studies reported that 2.5–11.8% of women were eligible for medication abortion by LMP and ineligible by U/S. The number of women who underestimated GA using LMP compared to U/S ranged from 1.8 to 14.8%, with lower rates found when the sample was limited to a GA b 63 days. Most women (90.5–99.1%) knew their LMP, 70.8–90.5% with certainty. Conclusion: Our results support that LMP can be used to assess GA prior to medication abortion at GA b 63 days. Further research looking at patient outcomes and identifying women eligible for medication abortion by LMP but ineligible by U/S is needed to confirm the safety and effectiveness of providing medication abortion using LMP alone to determine GA. © 2014 Elsevier Inc. All rights reserved. Keywords: Pregnancy dating; Menstruation; Medication abortion; Gestational age
1. Introduction While the number of abortions worldwide has remained stable in recent years, the proportion of unsafe abortions has increased [1]. Each year, it is estimated that 47,000 women die, and over 8 million women experience complications from unsafe abortion [2,3]. Medical methods of abortion hold promise to improve access to care in areas with limited resources, reducing maternal morbidity and mortality. Prior research has greatly simplified the process of mifepristone– misoprostol abortion. Alternative methods of follow-up [4–9] and self-administration of misoprostol [10–12] have allowed first trimester medication abortion (MAB) to be provided with
fewer visits to a health care facility. Further demedicalization of the process will allow more women to access safe abortion care in areas with limited resources. Previous research has called into question the need for routine ultrasound (U/S) for gestational dating prior to MAB and has demonstrated the feasibility of simplifying the abortion process for women by using ultrasonography only when needed [13,14]. We undertook this systematic review of published literature to evaluate the accuracy of using last menstrual period (LMP) compared to U/S for assessing gestational age (GA) prior to first trimester MAB.
2. Methods ☆
Funding: Society of Family Planning. The authors have no financial or other conflicts of interest to declare. ⁎ Corresponding author. E-mail address: [email protected] (D. Schonberg).
☆☆
http://dx.doi.org/10.1016/j.contraception.2014.07.004 0010-7824/© 2014 Elsevier Inc. All rights reserved.
We searched the Pubmed, Ovid Medline, Embase and Cochrane databases for peer-reviewed articles in all languages published from database inception through
2
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
October 2013, including evidence on the use of LMP for GA dating in abortion care (Appendix A). Reference lists from included articles identified by the search as well as key review articles were hand searched to identify additional articles. We did not attempt to identify unpublished articles or abstracts from scientific conferences. We included articles comparing GA dating by LMP to dating by U/S. We included only articles related to abortion care and not those related to obstetrical care. Articles were excluded if they lacked the necessary comparative data or if they did not pertain to abortion care at a GA less than 63 days. We also excluded case reports, letters and review articles. Our main outcome of interest was the accuracy of LMP to determine GA for MAB at GA less than 63 days. We focused on the number of women who underestimated their GA as well as the proportion of women eligible for a MAB by LMP but ineligible by U/S examination. We also report data on ectopic pregnancies, missed abortions and the clinical implications of inaccurate menstrual dating if available in articles included in the review. All five investigators independently reviewed all article titles and abstracts for possible study inclusion. After the first round of exclusion, the investigators divided into two teams (DS, LFW and EJ and MG, AB and EJ); each team was responsible for reviewing half of the remaining articles. Using standardized data abstraction forms, each investigator independently extracted and summarized data from the articles assigned to their team. The senior investigator (EJ) abstracted all articles and ensured consistency between the two teams. The two teams met regularly to review the abstraction process. At all stages, differences were resolved by consensus. We sent an email to article authors to request additional data if an article mentioned collecting both LMP and U/S data but did not present those data in the manuscript. Data acquired via author request were reviewed independently by each member of the research team, and consensus on the quality of data and appropriateness for inclusion was reached through discussion. We used the system for grading evidence developed by the US Preventive Services Task Force (USPSTF) [15] and the Quality Assessment of Diagnostic Accuracy Studies (QUADAS)-2 system [16]. Following the USPSTF guidelines, the internal validity of each study was judged based upon use of a relevant screening test (LMP) reliably administered to an appropriate sample of adequate size and compared to a credible reference standard (U/S). The risk of bias and applicability concerns identified in the QUADAS system were used to inform a decision on the overall quality of each study. Of note, we assessed data quality only as it applied to our specific research question; therefore, grading is not a reflection on the article as a whole.
remained. An additional 29 studies were identified after reviewing the citations of key articles, for a total of 318 articles for review. After excluding case reports, reviews, letters, editorials and articles not relevant to the topic question, 86 articles remained, and the full text of each article was reviewed (see Fig. 1). Five articles met inclusion criteria (Table 1); we were able to assess four articles as diagnostic accuracy studies [17–20] and one secondary analysis of a clinical trial [21]. Two of the included articles [18,21] cite additional publications for details on methodology [22,23]. Heterogeneity of data prevented meta-analysis, and lack of data presented in each article precluded the calculation of statistical measures including sensitivity and specificity. We attempted to contact the authors of 28 articles for additional data; 11 authors responded and three provided data. After review, it was agreed by all investigators that the data were not appropriate for inclusion in this review. 3.1. LMP compared to U/S Of the five articles included in the review, three articles [17–19] studied the number of women considered eligible for a MAB based upon LMP who were found to be ineligible based upon U/S. Bracken [17], in an article rated to be of “fair” quality, enrolled women seeking MAB at 10 clinics in the US. Bracken found that only 2.5% (76/3041) of women certain of their LMP would have been offered a MAB based upon LMP despite subsequent transvaginal U/S showing a GA above 63 days. When including women uncertain of their LMP, the number increased to 3.3% (142/4257) [17]. Blanchard, in an article rated to be of fair quality, studied women seeking first trimester abortion in three provinces in South Africa. The provider conducting the U/S was blinded to LMP. Blanchard found a higher rate of women, 11.8% (72/608), who were b 56 days by LMP but N 56 days by U/S [18]. In a study rated to be of “poor” quality, McGalliard
3. Results A total of 462 articles were retrieved via database searches; after adjusting for duplicates, 289 unique articles
Fig. 1. Flowchart of the systematic review.
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
found that 7.2% (17/237) of women referred to a UK hospital clinic for an abortion would have been offered a MAB based upon reported LMP despite having a GA by U/S greater than 63 days. McGalliard used transvaginal U/S unless the abdominal examination was larger than expected by LMP in which case transabdominal U/S was used. The authors collected LMP data and conducted the U/S, so this study was not blinded. Two articles reported the proportion of women who underestimated their GA using LMP compared with U/S [19,21]. McGalliard found that 14.8% of women (35/237) underestimated their GA by more than 1 week [19]. Of note, some of the women included in McGalliard's calculations had a GA above 13 weeks, unlike Harper who only included women with a GA less than 63 days in this calculation. Harper, in an article rated to be of poor quality, conducted a secondary analysis of a clinical trial to assess the safety and efficacy of MAB with less clinician oversight. Harper found that 1.8% of women (38/2121) with a GA less than 63 days by LMP underestimated their GA by more than 2 weeks compared to a physician's estimate based upon history, exam and transvaginal U/S [21,23]. Two articles reported on the strength of correlation between GA by LMP to that by U/S [19,20]. McGalliard found moderate agreement, r= 0.44 [19]. Harkness' study, which only studied correlation coefficients and was rated to be of poor quality, studied women b 63 days who had been referred. Measurements were taken by transabdominal U/S 48 h after mifepristone administration and were compared to GA by LMP as well as direct measurements of tissue. Harkness reported similar values of r= 0.48 between LMP and U/S measurement of gestational sac and r= 0.57 between LMP and U/S measurement of crown rump length [20]. Blanchard measured agreement and noted that 32% of LMP estimates of GA were within 1 week of U/S but did not report raw data or strength of correlation [18]. Two articles [18,21] assessed the impact of demographic variables on the accuracy of GA estimate by LMP. As demographic variables were studied as secondary outcomes, these studies may not have been powered to find a difference if one existed. Blanchard found that women who were certain of their LMP dates did not differ from others by age, gravidity, parity, education or employment status [18]. On the other hand, in multivariate analysis, Harper showed that Hispanic and African American women in the US were more likely to underestimate their GA by more than a week using LMP when compared to White women [21]. 3.2. Additional findings Four studies [17‐19,21] included the number of women able to report their LMP. Almost all women (90.5–99.1%) were able to recall their LMP. Young age was associated with inability to report LMP [21]. The number of women who stated that they were able to report their LMP with certainty varied but was still high (70.8–90.5%) [17,19].
3
Although Bracken reported that 70.8% of women could give a date with certainty, raw numbers were not given. Two studies [17,19] reported data on ectopic pregnancy, abnormal pregnancies or possible clinical implications of foregoing routine U/S. Bracken found that of 4484 women, screening with history and physical exam identified 32 as possibly having an ectopic or molar pregnancy. Of these 32 women, 9 were diagnosed with an ectopic pregnancy and 2 diagnosed with molar pregnancy using U/S or clinical and/or laboratory examinations. There were no ectopic pregnancies found in McGalliard's study population, but one woman (of 300) was diagnosed with an ectopic pregnancy when she presented emergently the day prior to her scheduled enrollment. The use of U/S by McGalliard added additional clinical information for 24/283 (8.5%) women: 4 were found not to be pregnant, 17 had an early pregnancy loss, and 3 had a twin pregnancy.
4. Discussion Our review found five articles that compared GA by LMP to that by U/S; all were deemed to be of fair or poor quality for the question under review. Rates of women presenting for care who were eligible for a MAB by LMP but ineligible by U/S were reported by three studies and ranged between 2.5 and 11.8%. Two studies reported rates of underestimation of GA by LMP of 1.8%–14.8%. Of particular interest is the proportion of women who underestimate their GA by LMP, as this pertains to the safety and effectiveness of abortion care. Harper showed low rates of underestimation (1.8%) for women at GA b 63 days. McGalliard showed higher rates but used a sample that included women beyond the first trimester. These results agree with previous research by Norman who found that LMP becomes less accurate at assessing GA as GA increases [24]. Similar results were found by Ellertson who reported low rates (1.4%) of women underestimating their GA when limiting the sample to GA b 56 days [25] and higher rates (10.4–12.4%) when including the entire first trimester [26]. Our review found rates of women presenting for care who were eligible for a MAB by LMP and ineligible by U/S of 2.5–11.8% [17–19]. Higher rates (7.2–11.8%) were found by Blanchard and McGalliard, but these studies used smaller sample sizes. These findings are consistent with prior work by Ellertson, which showed that 9.8–10% of women had a GA by LMP b 56 days but N 56 days by provider's estimate, with rates lowered to 4.6–5.5% when using a cutoff of 63 days [26]. When clinician exam was used in addition to LMP, Bracken found that only 1.6% (63/3874) of women were found to be b 63 days by LMP and N 63 days by U/S [17]. Prior research has shown that using LMP plus clinician exam accurately assesses GA and results in high success rates when used in MAB services [13]. Further research comparing LMP to U/S with and without the use of clinician exam could be of value.
4
Table 1 Summary of evidence for accuracy of reported LMP to determine GA. Study country funding
Study design
Population
Results
Strengths
Weaknesses
Quality of evidence
Bracken, 2011 [17]
Diagnostic accuracy
4484 women seeking early medication abortion at 10 US clinics in various cities
4447/4484 (99.1%) gave date for LMP; 70.8% gave date for LMP with certainty
Prospective data collection
Denominator for some calculations not clear
Fair
Accuracy of “certain” LMP vs. U/S: 76/3041 (2.5%) women would have been offered medication abprtion MAB ≥ 63 d by U/S (13/3041 [0.4%] ≥ 78 days)
Large N, diverse sample
Accuracy of ‘certain’ and ‘estimated’ LMP vs. U/S: 142/4257 (3.3%) women would have been offered MAB ≥ 63 days by U/S (26/4257 [0.6%] ≥ 78 days)
Methods used to measure GA clearly described
USA
Accuracy of LMP + clinician exam vs. U/S: 63/3874 (1.6%) women would have been offered MAB ≥ 63 d by U/S (1/3874 [0.02%] ≥ 78 days) Blanchard, 2007 [18,22]
Diagnostic accuracy
South Africa
673 women seeking first trimester abortion at 8 clinics
UK No funding reported
Women b 56 days GA by LMP, but N 56 days by U/S: 72/608 (11.8%)
Unknown if ultrasonographer blinded to LMP
Separate analysis of ‘certain’ and ‘uncertain’ LMP
Prospective data collection Large N
15th assigned as date of LMP for women who could only remember the month
Fair
Women referred to clinics may have had previous gestational age assessment
Poor
Multiple sites in 3 provinces
Anonymous donor
McGalliard, 2004 [19]
640/673 women (95.1%) reported LMP
Multicenter trial
Women referred to clinics may have had previous gestational age assessment
Diagnostic accuracy
262 women referred for abortion
LMP estimates within 1 week of U/S: 32% LMP estimates within 2 weeks of U/S: 51% LMP estimates within 3 weeks of U/S: 65%
U/S data collected separately and blinded
Level of agreement between methods: U/S vs. LMP estimate r= 0.36
Used standardized data collection forms
237/262 (90.5%) of women sure of their LMP
Prospective data collection
Women who were b 63 days by LMP, but N63 days by U/S: 17/237 (7.2%)
Large N
Mean difference: U/S mean 1 day b LMP (95% CI −36 to +34 days)
U/S data not gathered independently or
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
David & Lucille Packard Foundation and anonymous donor
Harper, 2002 [21,23] USA
2121 women b 63 days GA seeking abortion at 17 centers
blinded
For certain LMP: Discrepancy of N 7 days = 90/237 (38.0%) LMP underestimated GA by more than 1 week: 35/237 (14.8%) LMP overestimated GA by more than 1 week: 55/237 (23.2%)
Women with uncertain LMP excluded
2098/2121 women (98.9%) knew LMP LMP underestimated GA by more than 2 weeks: 38/2121 (1.8%) 259/2121 (12.2%) thought they had more time before reaching 63 d limit
Single center
Multi-center and geographically diverse
Retrospective, secondary data analysis
Prospective data collection
Women N63 days GA by clinical assessment excluded
Poor
Large N Statistical methods used to analyze GA data not reported Unknown if ultrasonographer blinded to LMP
Harkness, 1996 [20] UK No funding reported
Diagnostic accuracy study
Unreported number of women b63 days GA referred for MAB
Correlation coefficient (r) LMP and gest sac from scan: r= 0.48 (pb0.001, n=112) LMP and CRL from scan: r=0.57 (pb0.001, n=99)
Prospective data collection Single investigator gathered all U/S data
Women referred to clinics may have had previous gestational age assessment Women b17 years old excluded U/S performed after mifepristone administration Inconsistent N and small N reported for some measurements
Poor
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
Population Council
Secondary analysis of clinical trial assessing safety and efficacy of MAB
Correlation coefficient r=0.44
Unknown if ultrasonographer blinded to LMP
5
6
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
These studies show that most women seeking first trimester abortion can report a date for their LMP, and the majority can do so with confidence. As some of the studies in our review included women who were referred, they may have had prior knowledge of their GA, which could possibly bias these results. Still, our finding is consistent with earlier work by Wegienka and Baird [27], who established the validity of women's reported LMP by comparing prospectively recorded menstrual diaries to women's response when queried 4 weeks later. They found that 56% of women precisely recalled their LMP, 74% within 1 day, 81% within 2 days and 84% within 3 days. This conclusion remains limited, as women asked to prospectively record their menstrual histories may be more likely to accurately recall their LMP. Utilizing women's reported LMP to determine GA has the potential to make MAB services more accessible, particularly in areas where U/S is not readily available. Although using LMP to determine GA may result in a small number of women being offered a MAB after 63 days GA, research demonstrates that mifepristone and misoprostol are effective for terminating pregnancies beyond 63 days GA, with low rates of complications [28,29]. Therefore, these women would likely still have a safe MAB, albeit with a gradual decrease in efficacy as GA increases. Although Winikoff et al. noted no significant differences when comparing women undergoing MAB at 57–63 days GA with women 64–70 days GA, women at higher GA might experience a longer process of expelling the pregnancy and should be counseled appropriately [28]. A perceived benefit of routine use of U/S is the ability to identify women with abnormal pregnancies, such as nonviable, ectopic or molar pregnancies. While U/S provided some additional information, it did not provide further data for identifying ectopics over history and physical exam in the few articles that provided this information [17,19]. Studies that did report data about abnormal pregnancies [17,19] found low rates of ectopic pregnancies, consistent with population-based prevalence of around 2% [30–33]. U/S is a limited diagnostic test for ectopic, with sensitivity of only 25% for identifying ectopic pregnancy at β-hCG levels below 1500 mIU/mL, and 55% among all women regardless of β-hCG level [34]. Mol et al. showed that it is not cost effective to screen women for ectopic pregnancy using U/S unless a woman has already been identified for concern by history and physical exam [35]. Although women with spontaneous abortions and nonviable pregnancies, who would be eligible for additional treatment modalities such as expectant management, could be missed without routine use of U/S, both uterine aspiration and the medication regimen used in MAB are safe and effective treatments for spontaneous abortion. As we did not search systematically for data regarding ectopic pregnancies and missed abortions, limited conclusions can be drawn, and this remains an area for further research. There are some limitations to these data. Very few articles met our criteria for inclusion. Three studies included women
who were referred and, therefore, may have had knowledge of GA prior to enrollment. Some studies excluded women above a specific measured GA, potentially skewing the number of women who underestimated their GA. In addition, it is not known whether women seeking MABs and women seeking abortion procedures differ in their ability to assess their GA by LMP. If a difference exists, our results are further limited. Another possible limitation is the use of U/S as a reference standard. U/S has a margin of error that varies depending on the formula used to calculate GA as well as the GA of the pregnancy. Using U/S to measure GA between 7 and 10 weeks, Goldstein predicted a 95% confidence interval (CI) of ± 3 days [36]. This margin of error could affect the results of the articles included in the review. Most articles received fair or poor quality ratings due to the strict criteria used for judging quality. One strength of this review is that four of the included studies were specifically designed to assess the accuracy of GA by LMP compared to U/S. In summary, our review demonstrates that among patients seeking abortion at GA b 63 days, women's report of LMP can be used to determine GA. Our results support using LMP to assess GA prior to MAB. MAB would be more widely accessible if reported LMP could be relied upon to accurately assess GA. Research comparing LMP to U/S at GA above 63 days will be needed as the eligibility for MAB expands. Acknowledgement We would like to thank Dr. Jennifer Amico and Dr. Jason Fletcher for their comments and review of this paper. Appendix A. Search strategies Pubmed ((((("Gestational Age"[Mesh] OR "gestational age" OR "pregnancy duration" OR "pregnancy length" OR "pregnancy dating")) AND ("Menstruation"[Mesh] OR "Menstrual Cycle"[Mesh] OR menstrua* OR "last menstrual period" OR "LMP")) AND ("Abortion, Induced"[Mesh] OR "abortion") NOT "Abortion, Spontaneous"[Mesh])) Filters: Publication date to 2013/10/31 Ovid, 1946 through Oct 31 2013 1. abortion.mp. or Abortion, Induced/ 2. Abortion, Spontaneous/ 3. 1 not 2 4. gestational age.mp. or Gestational Age/ 5. pregnancy duration.mp. 6. pregnancy dating.mp. 7. pregnancy length.mp. 8. 4 or 5 or 6 or 7 9. menstrual cycle.mp. or Menstrual Cycle/ 10. menstruation.mp. or Menstruation/ 11. last menstrual period.mp.
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
12. LMP.mp. 13. menstrua*.mp. 14. 9 or 10 or 11 or 12 or 13 15. 3 and 8 and 14 Embase 'abortion'/exp OR 'abortion' OR 'abortion induced'/exp OR 'abortion induced' NOT ('abortion spontaneous'/exp OR 'abortion spontaneous') AND ('gestational age'/exp OR 'gestational age' OR 'pregnancy duration' OR 'pregnancy length' OR 'pregnancy dating') AND ('menstruation'/exp OR 'menstruation' OR 'menstrual cycle'/exp OR 'menstrual cycle' OR 'last menstrual period' OR 'lmp' OR menstrua*) AND ([article]/lim OR [article in press]/lim) AND [humans]/ lim AND [embase]/lim NOT [31-10-2013]/sd Cochrane #1 MeSH descriptor: [Abortion, Induced] explode all trees #2 MeSH descriptor: [Abortion, Spontaneous] explode all trees #3 abortion #4 #1 or #3 #5 #4 not #2 #6 MeSH descriptor: [Gestational Age] explode all trees #7 "gestational age" #8 "pregnancy duration" #9 "pregnancy length" #10 "pregnancy dating" #11 #6 or #7 or #8 or #9 or #10 #12 MeSH descriptor: [Menstruation] explode all trees #13 menstruation #14 MeSH descriptor: [Menstrual Cycle] explode all trees #15“menstrual cycle” #16 menstrua* #17 "last menstrual period" #18 LMP #19 #12 or #13 or #14 or #15 or #16 or #17 or #18 #18 #5 and #11 and #19 limited to reviews published before 10.31.13 References [1] Sedgh G, Singh S, Shah IH, Ahman E, Henshaw SK, Bankole A. Induced abortion: incidence and trends worldwide from 1995 to 2008. Lancet 2012;379(9816):625–32. [2] Unsafe abortion: global and regional estimates of the incidence of unsafe abortion and associated mortality in 2008. Geneva: World Health Organization; 2011. [3] Singh S, Wulf D, Hussain R, Bankole A, Sedgh G. Abortion worldwide: a decade of uneven progress. New York: Guttmacher Institute; 2009. [4] Clark W, Panton T, Hann L, Gold M. Medication abortion employing routine sequential measurements of serum hCG and sonography only when indicated. Contraception 2007;75(2):131–5. [5] Clark W, Bracken H, Tanenhaus J, Schweikert S, Lichtenberg ES, Winikoff B. Alternatives to a routine follow-up visit for early medical abortion. Obstet Gynecol 2010;115(2 Pt 1):264–72. [6] Perriera LK, Reeves MF, Chen BA, Hohmann HL, Hayes J, Creinin MD. Feasibility of telephone follow-up after medical abortion. Contraception 2010;81(2):143–9. [7] Fiala C, Safar P, Bygdeman M, Gemzell-Danielsson K. Verifying the effectiveness of medical abortion; ultrasound versus hCG testing. Eur J Obstet Gynecol Reprod Biol 2003;109(2):190–5.
7
[8] Ngoc NN, Bracken H, Blum J, Nga NT, Minh NH, von Nhang N, et al. Acceptability and feasibility of phone follow-up after early medical abortion in Vietnam. Obstet Gynecol 2014;23(1):88–95. [9] Cameron ST, Glasier A, Dewart H, Johnstone A, Burnside A. Telephone follow-up and self-performed urine pregnancy testing after early medical abortion: a service evaluation. Contraception 2012;86(1):67–73. [10] Schaff EA, Stadalius LS, Eisinger SH, Franks P. Vaginal misoprostol administered at home after mifepristone (RU486) for abortion. J Fam Pract 1997;44(4):353–60. [11] Elul B, Hajri S, Ngoc NN, Ellertson C, Slama CB, Pearlman E, et al. Can women in less-developed countries use a simplified medical abortion regimen? Lancet 2001;357(9266):1402–5. [12] Guengant JP, Bangou J, Elul B, Ellertson C. Mifepristone-misoprostol medical abortion: home administration of misoprostol in Guadeloupe. Contraception 1999;60(3):167–72. [13] Kaneshiro B, Edelman A, Sneeringer RK, Ponce de Leon RG. Expanding medical abortion: Can medical abortion be effectively provided without the routine use of ultrasound? Contraception 2011;83 (3):194–201. [14] Clark WH, Gold M, Grossman D, Winikoff B. Can mifepristone medical abortion be simplified? A review of the evidence and questions for future research. Contraception 2007;75(4):245–50. [15] Harris RP, Helfand M, Woolf SH, Lohr KN, Mulrow CD, Teutsch SM, et al. Current methods of the US Preventive Services Task Force: a review of the process. Am J Prev Med 2001;20(3 Suppl):21–35. [16] Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 2011;155(8):529–36. [17] Bracken H, Clark W, Lichtenberg ES, Schweikert SM, Tanenhaus J, Barajas A, et al. Alternatives to routine ultrasound for eligibility assessment prior to early termination of pregnancy with mifepristonemisoprostol. BJOG 2011;118(1):17–23. [18] Blanchard K, Cooper D, Dickson K, Cullingworth L, Mavimbela N, von Mollendorf C, et al. A comparison of women's, providers' and ultrasound assessments of pregnancy duration among termination of pregnancy clients in South Africa. BJOG 2007;114(5):569–75. [19] McGalliard C, Gaudoin M. Routine ultrasound for pregnancy termination requests increases women's choice and reduces inappropriate treatments. BJOG 2004;111(1):79–82. [20] Harkness LM, Rodger M, Baird DT. Morphological and molecular characteristics of living human fetuses between Carnegie stages 7 and 23: ultrasound scanning and direct measurements. Hum Reprod Update 1996;3(1):25–33. [21] Harper C, Ellertson C, Winikoff B. Could American women use mifepristone-misoprostol pills safely with less medical supervision? Contraception 2002;65(2):133–42. [22] Cooper D, Dickson K, Blanchard K, Cullingworth L, Mavimbela N, von Mollendorf C, et al. Medical abortion: the possibilities for introduction in the public sector in South Africa. Reprod Health Matters 2005;13(26):35–43. [23] Spitz IM, Bardin CW, Benton L, Robbins A. Early pregnancy termination with mifepristone and misoprostol in the United States. N Engl J Med 1998;338(18):1241–7. [24] Norman WV, Bergunder J, Eccles L. Accuracy of gestational age estimated by menstrual dating in women seeking abortion beyond nine weeks. J Obstet Gynaecol Can 2011;33(3):252–7. [25] Ellertson C, Elul B, Winikoff B. Can women use medical abortion without medical supervision? Reprod Health Matters 1997;5(9):149–61. [26] Ellertson C, Elul B, Ambardekar S, Wood L, Carroll J, Coyaji K. Accuracy of assessment of pregnancy duration by women seeking early abortions. Lancet 2000;355(9207):877–81. [27] Wegienka G, Baird DD. A comparison of recalled date of last menstrual period with prospectively recorded dates. J Womens Health 2005;14(3):248–52. [28] Winikoff B, Dzuba IG, Chong E, Goldberg AB, Lichtenberg ES, Ball C, et al. Extending outpatient medical abortion services through 70 days of gestational age. Obstet Gynecol 2012;120(5):1070–6.
8
D. Schonberg et al. / Contraception xx (2014) xxx–xxx
[29] Gouk EV, Lincoln K, Khair A, Haslock J, Knight J, Cruickshank DJ. Medical termination of pregnancy at 63 to 83 days gestation. Br J Obstet Gynaecol 1999;106(6):535–9. [30] Centers for Disease C. Prevention. Ectopic pregnancy–United States, 1990-1992, 44(3). MMWR Morbidity and mortality weekly report; 1995, pp. 46–8. [31] Van Den Eeden SK, Shan J, Bruce C, Glasser M. Ectopic pregnancy rate and treatment utilization in a large managed care organization. Obstet Gynecol 2005;105(5 Pt 1):1052–7. [32] Hoover KW, Tao G, Kent CK. Trends in the diagnosis and treatment of ectopic pregnancy in the United States. Obstet Gynecol 2010;115 (3):495–502.
[33] Stulberg DB, Cain LR, Dahlquist I, Lauderdale DS. Ectopic pregnancy rates in the Medicaid population. Am J Obstet Gynecol 2013;208 (4):274.e1–7. [34] Barnhart KT, Simhan H, Kamelle SA. Diagnostic accuracy of ultrasound above and below the beta-hCG discriminatory zone. Obstet Gynecol 1999;94(4):583–7. [35] Mol BW, van der Veen F, Bossuyt PM. Symptom-free women at increased risk of ectopic pregnancy: should we screen? Acta Obstet Gynecol Scand 2002;81(7):661–72. [36] Goldstein SR, Wolfson R. Endovaginal ultrasonographic measurement of early embryonic size as a means of assessing gestational age. J Ultrasound Med 1994;13(1):27–31.