Secondary Logo

Journal Logo

Diagnosing Ectopic Pregnancy: Decision Analysis Comparing Six Strategies



Objective To compare six published methods of diagnosing ectopic pregnancy.

Methods Decision analysis compared six diagnostic algorithms involving combinations of clinical examination, transvaginal ultrasound, serum progesterone, serum hCG, and D&C. The population was composed of hemodynamically stable women who presented to a tertiary care university emergency department with abdominal pain or bleeding in their first trimesters. Outcome measures included number of missed ectopic pregnancies, potentially interrupted intra-uterine pregnancies, surgical and diagnostic procedures, time until diagnosis, and cost.

Results Ultrasound followed by serum hCG in women with nondiagnostic scans yielded the most favorable outcomes; no ectopic pregnancy was missed, only 1% of all potential intrauterine pregnancies were interrupted, and time to diagnosis averaged 1.46 days. Quantitative hCG measurement followed by ultrasound only in women with hCG levels above the discriminatory zone was optimal if sensitivity of ultrasound to diagnose intrauterine pregnancy was less than 93%. Serum progesterone measurement was not favored because it was associated with missed ectopic pregnancies (2.6%).

Conclusion Given the current accuracy of tests for diagnosing ectopic pregnancy, algorithms using a combination of ultrasound and hCG resulted in the best outcomes. Ultrasound as the first step was the most efficient and accurate method of diagnosing ectopic pregnancies.

The most efficient method of diagnosing ectopic pregnancy is ultrasound followed by quantitative hCG measurement in cases in which ultrasound is nondiagnostic.

Department of Obstetrics and Gynecology, the Center for Reproductive Medicine and Surgery, and the Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania Medical Center, Philadelphia, Pennsylvania.

Kurt Barnhart, MD, University of Pennsylvania Medical Center, Division of Human Reproduction, 106 Dulles, 3400 Spruce Street, Philadelphia, PA 10104; E-mail:

Funding for this study was provided from National Institutes of Health grant RO1-HD-36455-01A1.

Received July 10, 2000. Received in revised form October 16, 2000. Accepted October 26, 2000.

Ectopic pregnancy is a major cause of morbidity and mortality in reproductive-aged women, accounting for 9% of pregnancy-related deaths in the first trimester.1,2 The number of ectopic pregnancies in the United States has increased from 17,800 in 1970 to 108,000 in 1992, a sixfold rise.3 Despite increasing incidence, the mortality rate associated with ectopic pregnancy has fallen dramatically thanks to improved diagnostic methods. Early detection before tubal rupture allows for outpatient treatment, reducing the risk of major complications and future infertility.4–6 Delay in diagnosis might lead to rupture, resulting in intra-abdominal hemorrhage, need for laparotomy, blood transfusion, and death.

Abdominal pain or vaginal bleeding in the first trimester are the only clinical signs and symptoms of ectopic pregnancy, but they are not sensitive or specific.7–9 Most women with such symptoms have normal intrauterine pregnancies; therefore, diagnosis of ectopic pregnancy must be accurate and timely, but should use methods that do not interrupt viable intrauterine pregnancies. Several published protocols for diagnosing women at risk for ectopic pregnancy use clinical examination, transvaginal ultrasonography, serum quantitative hCG, and serum progesterone9–15 selectively and in various permutations, but no approach has been established to be superior to others.

Identification of ectopic pregnancy can be difficult by ultrasound, so the most efficient way to rule out ectopic pregnancy is to diagnose intrauterine pregnancy. The sensitivity of ultrasound for diagnosing intrauterine pregnancy approaches 100% in gestations greater than 5.5 weeks.16,17 Often, exact gestational age is not known at evaluation and serum hCG is used as a surrogate marker for it.10 The discriminatory zone is defined as the serum hCG level above which ultrasound is expected to detect a viable intrauterine pregnancy.10,17,18 When an intrauterine pregnancy is not visualized on ultrasound with hCG above the discriminatory zone, we presume the gestation is nonviable or ectopic. In those circumstances, a D&C is done to confirm intra-uterine pregnancy. Without endometrial curettings with chorionic villi, ectopic pregnancy is presumed. Transvaginal ultrasound and quantitative hCGs are routinely used in diagnostic strategies, but there is debate about which should be done first.10

There has been much literature suggesting serum progesterone might aid diagnosis of ectopic pregnancy. A single serum progesterone level above 25 ng/mL is strongly associated with a normal intrauterine pregnancy, whereas levels below 5 ng/mL are associated with abnormal gestations (ectopic or abnormal intra-uterine pregnancy).9,11–13,19,20 Other authors have advocated transvaginal ultrasound alone, followed by hospitalization and repeat ultrasound, if indicated, as the best way to diagnose ectopic pregnancy and prevent rupture.14 Hospitalization for women with nondiagnostic initial ultrasounds ensures 100% compliance with follow-up. We did a decision analysis to find the optimal protocol for diagnosing ectopic pregnancy21 and compared it with diagnosis by clinical examination alone.

Back to Top | Article Outline

Materials and Methods


Our six diagnostic algorithms were analyzed for clinically stable women with pain or bleeding and positive pregnancy tests. Those with signs of acute intra-abdominal hemorrhage were not candidates for those diagnostic protocols. Figure 1 displays a simplified schematic of the first strategy.

Figure 1

Figure 1

Back to Top | Article Outline

Ultrasound followed by quantitative hCG.

In this protocol, a transvaginal ultrasound is the initial diagnostic test for women at risk of ectopic pregnancy. If a viable intrauterine pregnancy is diagnosed, the woman is discharged with appropriate follow-up obstetric care. If an ectopic pregnancy is found, she is referred for appropriate treatment. In this case, we assume management includes laparoscopic confirmation of diagnoses with surgical treatment. If a nonviable intrauterine pregnancy is found, then a D&C is done. Curettings are evaluated microscopically to confirm chorionic villi.

In the case of a nondiagnostic ultrasound, quantitative hCG is measured. A D&C is done to document chorionic villi if the level is greater than the established discriminatory zone. When chorionic villi are absent, laparoscopy is done for presumed ectopic pregnancy. If quantitative hCG is below the discriminatory zone, the woman is discharged and observed with serial quantitative hCG measurements. When quantitative hCG rises appropriately above the discriminatory zone, ultrasound is repeated. The woman is treated based on sonographic findings described above. If quantitative hCG rises or falls abnormally, a D&C is done to confirm nonviable intrauterine pregnancy. If no chorionic villi are collected by D&C, the woman is assumed to have an ectopic pregnancy. No additional ultrasound is done and there is no surgical intervention as long as hCG values decline to undetectable levels.10,14,15

Back to Top | Article Outline

Quantitative hCG followed by ultrasound.

In this case, serum quantitative hCG is the first step in the evaluation. Transvaginal ultrasound is done only when hCG concentration is greater than the discriminatory zone. If an intrauterine pregnancy, ectopic pregnancy, or non-viable intrauterine pregnancy is found on ultrasound, the described steps are taken. However, when ultrasound is nondiagnostic, a D&C is done. Endometrial curettings without documented chorionic villi indicate ectopic pregnancy and warrant surgery.

When initial quantitative hCG is below the discriminatory zone, women are discharged without preliminary ultrasound and serial quantitative hCG measurements are made with described follow-up.10

Back to Top | Article Outline

Progesterone followed by ultrasound and quantitative hCG.

Serum progesterone is measured first in this algorithm. Women with levels of at least 25 ng/mL are presumed to have viable intrauterine pregnancies and are discharged with follow-up prenatal care and routine ultrasound later. Those with levels less than 5 ng/mL are presumed to have nonviable intrauterine pregnancies and have D&C. Endometrial curettings without chorionic villi indicate ectopic pregnancy and warrant laparoscopic treatment. Women with progesterone levels between 5 and 25 ng/mL are further evaluated according to the first strategy with ultrasound and quantitative hCG measurement if ultrasound is nondiagnostic.9,12,13

Back to Top | Article Outline

Progesterone followed by quantitative hCG measurement and ultrasound.

As in the third protocol, serum progesterone is measured initially and women are assigned to three categories depending on progesterone values. Those with levels of at least 25 ng/mL or less than 5 ng/mL are treated as described in the progesterone–ultrasound–hCG algorithm. However, those with levels between 5 and 25 ng/mL are evaluated according to the second protocol: quantitative hCG measurement followed by ultrasound, if indicated.9,12,13

Back to Top | Article Outline

Ultrasound followed by repeat ultrasound.

Transvaginal ultrasound is done initially and appropriate measures are taken depending on findings. Women with nondiagnostic ultrasound are hospitalized and ultrasound is repeated 24 hours later. Those with definitive repeat ultrasound are treated appropriately. However, those with nondiagnostic repeat ultrasound have D&C. As described, curettings without chorionic villi indicate surgery for presumed ectopic pregnancy.16

Back to Top | Article Outline

Clinical examination.

Physical examination alone is used to diagnose ectopic pregnancy without serum hCG, serum progesterone, or ultrasound.10

We constructed a decision tree to compare six strategies for diagnosing ectopic pregnancy in hemodynamically stable women who presented to the emergency department with abdominal pain or bleeding in their first trimesters. Those models were constructed and analyzed with Data 3-0 for Windows (Treeage Software, Williamstown, MA). A sample schematic of the first strategy is shown in Figure 1.

The women in this study represent women diagnosed and treated after presentation to an inner-city emergency department. The assumed prevalences of ectopic pregnancy, intrauterine pregnancy, and nonviable intrauterine pregnancy were 9.4%, 61.1%, and 29.9%, respectively, based on weighted averages from three studies that represented populations from similar urban teaching hospitals.10,14,15

Probabilities used in this analysis are outlined in Table 2 with references. Probabilities were drawn primarily from published prospective and retrospective cohort studies. Much data were taken from two studies conducted at our own institution (Hospital of the University of Pennsylvania).10,15 All probabilities are in the form of sensitivities from the literature. Focus on sensitivity (instead of specificity) was chosen to maximize diagnoses of women with ectopic pregnancies.

Table 2

Table 2

Back to Top | Article Outline


Using a hypothetical cohort of 10,000 women with spontaneous pregnancies who had experienced pain or bleeding and calculating pathway probabilities, several outcomes were assessed. The primary outcome was number of missed ectopic pregnancies per 10,000 women. That was considered the most important outcome because failure to identify an ectopic pregnancy can lead to substantial maternal morbidity and even death. Secondary outcomes included number of potentially interrupted intrauterine pregnancies, performed D&Cs or laparoscopies done, ultrasounds, blood collection, admissions, days until diagnosis, and hospital charges. Ultrasound is not 100% sensitive for detecting intrauterine pregnancy, so some potential intrauterine pregnancies will be interrupted by D&Cs done to document chorionic villi. The actual number of interrupted intrauterine pregnancies is probably significantly less than the number calculated in this model because many early intrauterine pregnancies will not result in viable pregnancies because of miscarriage or desired termination.

The time for definitive diagnosis in subjects observed with outpatient serial hCG values is 4 days.10 That time frame was used to compare average time until diagnosis per patient and number of blood draws. To provide a reference for comparing diagnostic algorithms, the price of each surgical and diagnostic procedure was estimated according to average charges at the Hospital of the University of Pennsylvania: laparoscopy = $5000, D&C = $2500, ultrasound = $784, blood draw (progesterone or hCG) = $75, admission for 24 hours = $1500.

One- and two-way sensitivity analyses were done using probability ranges shown in Table 1 to find whether changing specific probabilities would change outcomes. Outcomes assessed in the sensitivity analysis were missed ectopic pregnancy and interrupted intra-uterine pregnancy.

Table 1

Table 1

Several assumptions were made in this study. D&C and laparoscopy were assumed 100% diagnostic, meaning that no pregnancy was missed when D&C was done to document chorionic villi and no ectopic pregnancy was missed by laparoscopy. Also, although ultrasound is not 100% sensitive, the assumption was that ultrasound diagnosis is always accurate. Therefore, ultrasound might not detect an intrauterine pregnancy and fall into the nondiagnostic category, but we assume that intrauterine pregnancy cannot be mistaken for an ectopic pregnancy and vice versa. The possibility of a heterotopic pregnancy was not accounted for in the model.

Back to Top | Article Outline


Table 2 shows calculated outcomes for each strategy. No ectopic pregnancies were missed with strategies involving only ultrasound and hCG. Among those strategies, transvaginal ultrasound as the first step led to the fewest interrupted intrauterine pregnancies. Those two strategies were similar in number of procedures and days until diagnosis.

The progesterone algorithms missed more women with ectopic pregnancies and required more surgeries than those without progesterone and had the fewest number of interrupted intrauterine pregnancies.

The strategy of ultrasound followed by ultrasound was associated with no missed ectopic pregnancies and the shortest time until diagnosis. Clinical examination alone resulted in no incremental costs, no procedures, and no interrupted pregnancies; however, all ectopic pregnancies were missed.

One- and two-way sensitivity analyses using probability ranges are shown in Table 1. The strategy of ultrasound followed by quantitative hCG was no longer the optimal protocol (fewest interrupted intrauterine pregnancies and missed ectopics) if overall sensitivity of ultrasound fell below 93%, odds ratio if the sensitivity of ultrasound to detect intrauterine pregnancies in women with hCG measurements greater than the discriminatory zone exceeded 99%. In those cases, the strategy of hCG followed by ultrasound had the fewest negative outcomes (Figure 2). The strategy with ultrasound and repeat ultrasound had the most favorable results only if the sensitivity of repeat ultrasound exceeded 80%.

Figure 2

Figure 2

The progesterone strategies were optimal only if an ectopic pregnancy was never associated with progesterone concentration greater than or equal to 25 ng/mL (thereby missing no ectopic pregnancies).

Back to Top | Article Outline


Early diagnosis of unruptured ectopic pregnancy is paramount for eliminating mortality, reducing morbidity, and preserving fertility. If one suspects ectopic pregnancy, sensitive and accurate diagnostic tests exist to confirm it or rule it out. However, controversy exists regarding which to use, and in what order, to most efficiently diagnose ectopic pregnancy. We found that algorithms using transvaginal ultrasound and quantitative hCG values were the best diagnostic strategies because no potential ectopic pregnancy was missed. Specifically, the best strategy was transvaginal ultrasound on all women who presented with abdominal pain or vaginal bleeding in their first trimesters. If ultrasound is nondiagnostic, hCG should be measured and the woman treated according to level. Using that algorithm, no ectopic pregnancies were missed, few potential intrauterine pregnancies were interrupted, timely diagnoses were achieved, and with few surgical procedures. This strategy is equally effective in high-and low-risk populations.

The second best diagnostic strategy was to start with a quantitative hCG and do ultrasound only on women with levels above the established discriminatory zone. That algorithm resulted in 1% more pregnancies that were potentially interrupted and a slightly longer time until diagnosis compared with the strategy of ultrasound and quantitative hCG. However, that strategy is recommended if diagnostic accuracy of ultrasound differs from the assumptions in our analysis. For example, if the overall sensitivity of ultrasound for detecting intrauterine pregnancy is less than 93% or greater than 99% in women with quantitative hCGs above the discriminatory zone, then hCG followed by ultrasound was superior (resulting in fewer interrupted intrauterine pregnancies). Factors that might decrease sensitivity of ultrasound include inexperienced ultrasonographer, maternal obesity, uterine myomas, and patient intolerance of examination. In those situations, ultrasound only in women with a high hCG would decrease the number of false-negative and false-positive ultrasound diagnoses. The sensitivity of ultrasound for detecting intrauterine pregnancy above the discriminatory zone was dependent on quantitative hCG level chosen as the discriminatory zone. Raising the level of the discriminatory zone improved sensitivity of ultrasound for detecting intrauterine pregnancy, making hCG followed by ultrasound favorable by limiting the number of interrupted intrauterine pregnancies.

The algorithms using progesterone are problematic because they were associated with missed ectopic pregnancies, which can lead to devastating consequences. In algorithms using progesterone, as prevalence of ectopic pregnancy rose, more ectopic pregnancies were missed and more intrauterine pregnancies interrupted. As noted by advocates of progesterone, those protocols were not appropriate for diagnosing women at high risk of ectopic pregnancy, such as those with histories of tubal surgery, pelvic infections, ectopic pregnancy or those undergoing ovulation induction. The addition of progesterone to algorithms using ultrasound and hCG prolongs the time spent in the emergency department for a large subset of women with indeterminate progesterone levels (5–25 ng/mL). If serum progesterone is to be considered in the diagnosis of ectopic pregnancy, the optimal threshold must be validated by each institution. Cutoffs provided in this study might not be generalizable to every institution.20,22

Ultrasound followed by hospitalization and repeat ultrasound did not compare favorably with other algorithms because of the increased intrauterine pregnancies that were interrupted. However, that protocol might be useful for treating women for whom compliance with follow-up is a concern.14 That strategy is optimal when sensitivity of repeat ultrasound is increased from 50% to 80%, which might be achieved if ultrasound is done several days after initial evaluation rather than 24 hours after admission. That interval would allow further development of an intrauterine pregnancy, increasing the chance that it would be seen with ultrasound. However, that approach would be expected to increase the length of hospitalization and delay diagnosis of ectopic pregnancy, resulting in greater risk of rupture.

Used alone, clinical examination does not aid in diagnosis of ectopic pregnancy in a stable woman. It is not possible to distinguish ectopic pregnancy from normal or abnormal intrauterine pregnancy using only history and physical examination. If clinical examination is used exclusively, all ectopic pregnancies would be missed and only 13% of all intrauterine pregnancies would be diagnosed accurately.

The assumptions used in conducting this decision analysis were unlikely to alter our main outcomes. For instance, we assumed that D&C and laparoscopy are 100% diagnostic and that there are no false-positive diagnoses by ultrasound. In reality, those situations can occur, although rarely.23 For example, not detecting chorionic villi that were actually present from D&C would result in unnecessary laparoscopy but would not be expected to change the number of ectopic pregnancies missed. Assumption that no spontaneous heterotopic pregnancies existed in our population did not significantly alter the data because that entity occurs in approximately one of 30,000 pregnancies.24 Prevalence of ectopic pregnancy after assisted reproductive technology was not evaluated in this study.

Back to Top | Article Outline


1. Centers for Disease Control and Prevention. Ectopic pregnancy—United States, 1988–1989. MMWR 1992;41:591–4.
2. Centers for Disease Control and Prevention. Ectopic pregnancy—United States, 1990–1992. MMWR 1995;44:46–8.
3. NCHS. Advanced report of final mortality statistics, 1992. Monthly vital statistics report vol. 43, no. 6, suppl. Hyattsville, MD: US Department of Health and Human Services, Public Health Service, CDC, 1994.
4. Carson SA, Buster JE. Ectopic pregnancy. N Engl J Med 1993;329:1174–81.
5. Stovall TG, Ling FW. Ectopic pregnancy: Diagnostic and therapeutic algorithms minimizing surgical intervention. J Reprod Med 1993;38:807–12.
6. Stovall TG, Ling FW, Buster JE. Outpatient chemotherapy for unruptured ectopic pregnancy. Fertil Steril 1989;51:435–8.
7. Dorfman SF, Grimes DA, Cates W, Binkin NJ, Kafrissen ME, O'Reilly KR. Ectopic pregnancy mortality, United States 1979–1980: Clinical aspects. Obstet Gynecol 1984:64:386–9.
8. Abbot JT, Emmans LS, Lowenstein ST. Ectopic pregnancy: Ten common pitfalls in diagnosis. Am J Emerg Med 1990;8:5515–22.
9. Stovall RG, Kellerman AL, Ling FW, Buster JE. Emergency department diagnosis of ectopic pregnancy. Ann Emerg Med 1993;19:1098–103.
10. Barnhart K, Mennuti MT, Benjamin I, Jacobson S, Goodman D, Coutifaris C. Prompt diagnosis of ectopic pregnancy in an emergency department setting. Obstet Gynecol 1994;84:1010–5.
11. Stovall TG, Ling FW, Carson SA, Buster JE. Serum progesterone and uterine curettage in differential diagnosis of ectopic pregnancy. Fertil Steril 1992;57:456–8.
12. Stovall TG, Ling FW, Cope BJ, Buster JE. Preventing ruptured ectopic pregnancy with a single serum progesterone. Am J Obstet Gynecol 1989;160:1425–31.
13. McCord ML, Arheart KL, Muram DM, Stovall TG, Buster JE, Carson SA. Single serum progesterone as a screen for ectopic pregnancy: Exchanging specificity and sensitivity to obtain optimal test performance. Fertil Steril 1996;66:513–6.
14. Kaplan BC, Dart RG, Moskos M, Kuligowska E, Chun B, Adel Hamid M, et al. Ectopic pregnancy: Prospective study with improved diagnostic accuracy. Ann Emerg Med 1996;28:10–7.
15. Barnhart KT, Kamelle SA, Simhan H. Diagnostic accuracy of ultrasound, above and below the beta-hCG discriminatory zone. Obstet Gynecol 1999;94:583–7.
16. Goldstein SR, Snyder JR, Watson C, Danon M. Very early pregnancy detection with endovaginal ultrasound. Obstet Gynecol 1988;72:200–4.
17. Timor-Tritsch IE, Yeh MN, Peisner DB, Lesser KB, Salvik BS. The use of transvaginal ultrasound in the diagnosis of ectopic pregnancy. Am J Obstet Gynecol 1988;161:157–61.
18. Fleischer AC, Pennell RG, McKee MS, Worell JA, Keefe B, Herbert CM, et al. Ectopic pregnancy: Features at transvaginal sonography. Radiology 1990;174:375–8.
19. Stovall TG, Ling FW, Andersen RN, Buster JE. Improved sensitivity and specificity of a single measurement of serum progesterone over serial quantitative β-human chorionic gonadotropin in screening for ectopic pregnancy. Hum Reprod 1992;7:723–5.
20. Mol BW, Lijmer TG, Ankum WM, Van Der Veen F, Bossuyt PM. The accuracy of single serum progesterone measurement in the diagnosis of ectopic pregnancy: A meta-analysis. Hum Reprod 1998;13:3220–7.
21. Richardson WS, Allan SD. Users' guide to the medical literature: How to use a clinical decision analysis. JAMA 1995;273:1292–5.
22. Anonymous. Serum progesterone in the diagnosis of ectopic pregnancy (editorial). Lancet 1992;340:583.
23. Spandorfer SD, Menzin AW, Barnhart KT, LiVolsi VA, Pfeifer SM. Efficacy of frozen-section evaluation of uterine curettings in the diagnosis of ectopic pregnancy. Am J Obstet Gynecol 1996;175:603–5.
24. Richards SR, Stempel LS, Carlton BD. Heterotopic pregnancy: Reappraisal of incidence. Am J Obstet Gynecol 1982:142:928–30.

Cited By:

This article has been cited 1 time(s).

Obstetrics & Gynecology
Prediction of Location of a Symptomatic Early Gestation Based Solely on Clinical Presentation
Barnhart, KT; Casanova, B; Sammel, MD; Timbers, K; Chung, K; Kulp, JL
Obstetrics & Gynecology, 112(6): 1319-1326.
PDF (251) | CrossRef
Back to Top | Article Outline
© 2001 The American College of Obstetricians and Gynecologists