Barnhart, Kurt T. MD, MSCE; Fay, Courtney A.; Suescum, Maria MD; Sammel, Mary D. ScD; Appleby, Dina MS; Shaunik, Alka MD; Dean, Anthony J. MD
Vaginal bleeding, abdominal pain, or both in the first trimester of pregnancy are common reasons for women to seek emergency department care.1,2 These symptoms occur in as many as 15–25% of pregnancies1; and although 50–70% of these women go on to have normal gestations, the remaining women have either abnormal or ectopic pregnancies, many of whom require urgent or emergent gynecologic intervention.1–5
Ultrasonography is the one of the most important tools in the diagnosis of these patients.1,3,6–11 However, transvaginal ultrasonography is a subjective test and does not always produce a definitive yes or no answer. When an intrauterine gestation is identified with ultrasonography, the patient can be discharged for outpatient follow-up with an obstetrician.1,4,5 Systematic reviews demonstrate that ultrasound scans performed in the emergency department have a sensitivity of 90% and specificity of 98% in the detection of an intrauterine pregnancy.9
Identification of women without an intrauterine pregnancy is a strategy to identify a small higher-risk group who need urgent evaluation for possible treatment or outpatient surveillance.1,4,5,8–10 The diagnostic considerations of patients with nondiagnostic initial ultrasonography include early intrauterine pregnancy (before development of a yolk sac), spontaneous miscarriage, or ectopic pregnancy. In stable patients, the differentiation of these possibilities will require gynecologic outpatient observation over the following days and weeks until a definitive diagnosis can be made. The extent and acuity of follow-up is based on clinical suspicion and is greatly influenced by the findings on an initial ultrasound scan.1
Often, although not diagnostic, a probable impression of intrauterine pregnancy or ectopic pregnancy is reported. If the initial ultrasound impression was incorrect, a woman is at risk of under treatment and morbidity resulting from a missed ectopic treatment, overtreatment of a spontaneous miscarriage, or interruption of a potentially viable intrauterine pregnancy. The objective of this study was to estimate the use of pelvic ultrasonography in an undifferentiated cohort of women who presented to an emergency department with first-trimester vaginal bleeding, abdominal pain, or both who did not have an uncomplicated intrauterine pregnancy on initial ultrasonography. The study sought to determine the diagnostic accuracy of an ultrasound scan in this group and to investigate how clinical factors such as initial quantitative human chorionic gonadotropin (hCG) value in women presenting with symptoms (pain, bleeding, or both) affected the accuracy of the initial ultrasound scan.
MATERIALS AND METHODS
Approval to conduct this study was obtained from the Institutional Review Board of the University of Pennsylvania. The majority (approximately 70%) of women presenting to the emergency department with first-trimester pain, bleeding, or both are diagnosed by ultrasonography to have an uncomplicated intrauterine pregnancy. These patients are discharged without a gynecologic consult and are therefore not included in this cohort. All patients who are not discharged according to these criteria are consulted to the gynecology service, which determines the need for acute gynecologic intervention or outpatient follow-up. Thus, the cohort for this study is comprised of all women presenting to our emergency department with first-trimester pain, bleeding, or both and one or more of the following: 1) an indeterminate ultrasonography (no definite intrauterine pregnancy or ectopic pregnancy); 2) an abnormal intrauterine pregnancy; 3) an ectopic pregnancy that was not immediately admitted for operative management; and 4) an intrauterine pregnancy requiring gynecologic evaluation.
A database is continually maintained by the gynecology service of all women meeting these criteria. It includes each patient's initial findings along with details of their continued evaluation until definitive diagnosis. Patient records from August 1999 to September 2007 are included in this study.
All patients in this series received a transvaginal ultrasonography that was reviewed and interpreted by a board-certified radiologist. The ultrasound findings were communicated to the emergency department attending before gynecology consultation. Based on previous consensus recommendations,12 initial ultrasound diagnoses were classified in one of the following six categories: 1) definite intrauterine pregnancy (visualization of a gestational sac with a yolk sac, embryo, or both); 2) probable intrauterine pregnancy (intrauterine echogenic sac-like structure without visualization of a yolk sac or embryo); 3) definite ectopic pregnancy (extrauterine gestational sac with yolk sac, embryo or both); 4) probable ectopic pregnancy (inhomogeneous adnexal mass or extrauterine sac-like structure without identification of a yolk sac or embryo); 5) nondiagnostic or pregnancy of unknown location (no evidence of either ectopic pregnancy or intrauterine pregnancy); and 6) nonviable intrauterine pregnancy (ultrasound evidence of a fetal death, anembryonic gestation, or retained products of conception).
Women were followed by the gynecology service until a definitive diagnosis was made or the patient was lost to follow-up. Based on previous consensus recommendations, final clinical diagnoses were classified into one of the three following categories12: 1) visualized intrauterine pregnancy: intrauterine gestational sac with yolk sac or embryo; 2) ectopic pregnancy: visualized extrauterine gestational sac with yolk sac or embryo or nonvisualized ectopic pregnancy: no products of conception on uterine evacuation or confirmed with surgical pathologic specimens and a rise in postoperative quantitative hCG concentration); or 3) spontaneous miscarriage: identification of products of conception on uterine evacuation or complete resolution of hCG from the serum.
The primary analysis included the comparison of the emergency department ultrasound diagnosis on the initial visit (ultrasound diagnosis) with a final clinical diagnosis of ectopic pregnancy, intrauterine pregnancy, or spontaneous miscarriage (final diagnosis). Sensitivity, specificity, positive predictive value, and negative predictive values were calculated for each ultrasound diagnosis other than nondiagnostic. For the of probable ectopic pregnancy, ultrasound diagnoses of definite ectopic pregnancy were excluded from the calculations; similarly, calculations for the probable intrauterine pregnancy ultrasound diagnosis excluded patients with an ultrasound diagnosis of definite intrauterine pregnancy.
For further analyses, patients with an ultrasound diagnosis of definite ectopic pregnancy and probable ectopic pregnancy were combined under a general ultrasound diagnostic group of ectopic pregnancy. Similarly, patients with an ultrasound diagnosis of definite intrauterine pregnancy and probable intrauterine pregnancy were combined under a general ultrasound diagnostic group of intrauterine pregnancy. Sensitivity, specificity, positive predictive value, and negative predictive value were then calculated comparing these two initial ultrasound diagnostic groups to the corresponding final clinical diagnosis. Overall test accuracy was calculated by dividing the number of patients for whom the initial ultrasound diagnostic groups (combined probable and definite) as well as the nonviable intrauterine pregnancy ultrasound diagnosis corresponded with the final clinical diagnosis by the total number of patients who received diagnostic ultrasound examinations.
Sensitivity, specificity, positive predictive value, and negative predictive value of the definite ectopic pregnancy ultrasound diagnosis were compared with the corresponding test statistics for the probable ectopic pregnancy ultrasound diagnosis. Likewise, the same four test statistics for the definite intrauterine pregnancy diagnosis were compared with those for the probable intrauterine pregnancy diagnoses. Analyses were then stratified by three different clinical variables: initial serum hCG levels above or below the discriminatory zone of 2,000 milli-international units/mL, the presence or absence of bleeding, and the presence or absence of pain at presentation. Sensitivity, specificity, positive predictive value, and negative predictive value for each of the ultrasound diagnostic groups described as well as overall test accuracy were calculated for each clinical variable and compared using tests of independent proportions through Pearson chi square tests. A P<.05 was considered significant.
The cohort consisted of 2,058 women over a 7-year period. A total of 178 (8.6%) were lost to follow-up resulting in 1,880 evaluable women. The mean age of the women in the study was 26 years (range, 13–48 years), and the mean parity was 1.3 (range, 0–9). The initial interpretation of the ultrasound diagnosis was considered definite in 277 (14.7%) women, probable in 462 (24.6%) women, and nondiagnostic in 1,141 (60.7%) of women. Serum hCG was above 2,000 or greater in 535 women (28.5%) and less than 2,000 in 1,345 (71.5%) women. The distribution of final clinical outcomes stratified by initial ultrasound diagnosis is shown in Table 1.
A comparison of specific ultrasound diagnoses to final clinical diagnosis is also detailed in Table 1. Overall, the accuracy of diagnostic ultrasound examination in predicting final outcome was 78.0%. Ectopic pregnancy was ultimately diagnosed in 380 of 1,880 (20.2%) women. Initial ultrasonography (using criteria of definite or probable) identified ectopic pregnancy in 189 of 380 of women for a sensitivity rate of 49.7%. Of the 219 women with definite or probable ectopic pregnancy on initial ultrasound diagnosis, 189 were correctly diagnosed for a positive predictive value of 86.3%.
A final diagnosis of a visualized intrauterine pregnancy was made in 501 of 1,880 (26.6%) women. Initial ultrasonography identified an intrauterine pregnancy in 193 of the 501 women for a sensitivity of 38.5%. Of the 317 women in whom a definite or probable intrauterine pregnancy was noted on initial scan, 193 were correctly diagnosed for a positive predictive value of (60.9%). Spontaneous miscarriage ultimately was diagnosed in 999 of 1,880 (53.1%) women.
Accuracy, sensitivity, specificity, and predictive values of initial ultrasonography are also presented in Table 1. An ultrasound diagnosis of a definite ectopic pregnancy has 99.9% specificity with a high positive predictive value (98%) in detecting women eventually diagnosed with ectopic pregnancy. When the ultrasound diagnosis was probable for ectopic pregnancy, sensitivity significantly higher (42.1–13.2% P<.001), whereas the specificity remained high (98%); but the positive predictive value was significantly lower (82.7% compared with 98.0%; P<.005).
An ultrasound diagnosis of a definite intrauterine pregnancy was almost 100% specific. When the ultrasound diagnosis was probable but not definite, sensitively significantly increased from 4.0% to 36% (P<.001), whereas specificity (99.8% compared with 92.2% P<.001) and positive predicted value (87.0% compared with 58.8% P=.073) decreased.
Of the 739 women identified as having an ultrasound diagnosis in any one of the five categories other than indeterminate, there were 163 (22.1%) cases in which the ultrasound diagnosis did not match the final diagnosis. These diagnostic errors are also detailed in Table 2. Four of 163 (2.5%) cases were given a wrong definite diagnosis of either an intrauterine pregnancy or ectopic pregnancy. One (0.6%) was a false-negative scan for ectopic pregnancy. A total 117 of 163 (71.8%) women had an initial ultrasound diagnosis of probable intrauterine pregnancy but ultimately had a spontaneous miscarriage, representing the most common diagnostic error.
When examining the data by final outcome, it is noted that among patients not initially in the nondiagnostic ultrasound category, six of the 195 (3.1%) patients with a final diagnosis of ectopic pregnancy and 14 of the 207 (6.8%) patients with a final diagnosis of intrauterine pregnancy received an incorrect initial ultrasound diagnosis. Four of the six women in this group with ectopic pregnancy were initially classified as probable intrauterine pregnancy and two were classified as having a nonviable intrauterine pregnancy. Of the 14 misdiagnosed women with a final diagnosis of intrauterine pregnancy, seven were initially classified as probable ectopic pregnancy and seven were initially classified as a nonviable intrauterine pregnancy.
Table 2 presents the sensitivity, specificity, positive predictive value, and negative predictive value of ultrasound results for the final diagnoses of ectopic pregnancy, intrauterine pregnancy, and spontaneous miscarriage stratified by initial hCG concentration. The sensitivity, negative predictive value, and overall accuracy of the initial ultrasonography was significantly higher for all final diagnoses when the serum hCG level was 2,000 milli-international units/mL or greater.
Table 3 presents the results of the analysis of ultrasound detection of final outcome stratified by presence compared with absence of bleeding. A total of 71.8% of the study population had symptoms of vaginal bleeding at the time of initial presentation. The accuracy of the initial ultrasonography was significantly higher when bleeding was not a presenting symptom (P<.001). In addition, the sensitivity in detecting an ectopic pregnancy and the positive predictive value in detecting an intrauterine pregnancy were statistically significantly higher in the absence of bleeding (P=.02 and P<.001, respectively).
Table 4 presents the results of the analyses of ultrasound detection of final outcome stratified by presence compared with absence of pain. Approximately 56.5% of study participants reported abdominal pain during their initial visit. Accuracy of ultrasonography in detecting final outcome did not differ based on by presence or absence of pain.
The cohort studied pregnant patients in whom ultrasound findings are expected to affect follow-up and treatment. Previous studies suggest that approximately 20% of women with symptomatic early pregnancy presenting to the emergency department will not receive a definite diagnosis on the initial emergency department visit.4,5,10 This cohort, in which 20% of the women ultimately were diagnosed with ectopic pregnancy and 27% with ongoing intrauterine pregnancy, is similar to previously described populations with nondiagnostic initial ultrasound examinations.4,5
This study differs from previous studies in several ways. First, our cohort is larger. Second, we investigated the diagnostic impact of radiologists' ultrasound interpretations that are qualified as probable or definite. Third, the size of our cohort afforded us the opportunity to do subset analysis based on patients' cardinal clinical findings.
In this clinically important population, ultrasound findings resulted in a definite impression for almost 15% of the population at initial presentation and was probable in an additional 25%, leading to low sensitivity of initial ultrasound scan for the diagnosis of all three outcomes. In this clinical situation, if the clinician has a sense of the pretest probability of an outcome in the patient he or she is examining, the predictive value and accuracy of an initial ultrasonography may be of greater use than sensitivity and specificity.
The importance of risk stratification in this patient group arises from the potential clinical and psychologic costs of either over- or underdiagnosis. If a women is erroneously reassured that she has an intrauterine pregnancy and not an ectopic pregnancy, outpatient evaluation may not be continued, may not be complied with, or may not be intensive enough to diagnose an unruptured ectopic pregnancy. Conversely, if a woman is mistakenly diagnosed with an ectopic pregnancy or nonviable gestation, she may undergo unnecessary surgical or medical therapy that may interrupt a desired potentially viable gestation. It should be noted that compliance with outpatient surveillance of women at risk for ectopic pregnancy is low.13
Our data demonstrate that when criteria used to make a diagnosis of an extrauterine gestation with ultrasonography include visualization of a yolk sac or embryo (classified as definite),12 there are very few incorrect diagnoses and ultrasonography has almost 100% specificity with a 98% positive predictive value in identifying an ectopic pregnancy. We noted only one case of a false-positive diagnosis of an ectopic pregnancy in this group. However, when criteria are liberalized to include an extrauterine mass without a yolk sac or embryo (defined as probable),12 the number of diagnostic errors is significantly higher. Although the sensitivity of initial ultrasound scan result among patients with probable ectopic increases, the positive predictive value diminishes to a level at which approximately one in five diagnoses is falsely positive. When the diagnostic criteria for an intrauterine pregnancy are not limited to those intrauterine sacs that contain a yolk sac or embryo, a similar increase in sensitivity at the expense of predictive value is noted with one in four diagnoses of probable intrauterine pregnancy being falsely positive.
The most common error noted in this study, accounting for approximately 70% of the incorrect diagnoses, was the misclassification of miscarriage as an ongoing intrauterine pregnancy. It is acknowledged that this error does not have severe clinical consequences, but misdiagnosis of viability may lead to delays in therapeutic intervention and psychologic effects.14 These findings reconfirm the limited ability of a single ultrasonography to predict viability of an intrauterine pregnancy.14
Several clinically important errors in initial ultrasound diagnoses were noted, including four cases of false-positive diagnosis of intrauterine pregnancy and two cases of false-positive diagnosis of miscarriage in which the patient ultimately had an ectopic pregnancy. These errors represent a false-negative rate for an ectopic pregnancy of six of 195 (3%). There were also false-positive diagnoses in seven of 203 (3.4%) of a miscarriage in which intrauterine pregnancy was the final diagnosis. If clinical management had been undertaken based solely on the ultrasound report, severe clinical consequences might have occurred. A total of two-thirds of women whose ectopic pregnancy was misdiagnosed later presented with signs of tubal rupture.
Our data confirm that there are clinical situations that significantly affect the test characteristics of initial ultrasonography in this population. It has been previously noted that ultrasonography is less sensitive and accurate when the hCG value is less than 1,500 milli-international units/mL with sensitivity and positive predictive value for the diagnosis of ectopic pregnancy of 25% and 60%, respectively.15 Our finding of a sensitivity of 34% with a positive predictive value of 80% in women with an hCG 2,000 milli-international units/mL or less is similar. A positive predictive value of 80% for an ectopic pregnancy and 58% for an intrauterine pregnancy represents a clinically important misdiagnosis rate in this group.
Clinical symptoms also affected accuracy of initial ultrasound results in our study. The presence of bleeding resulted in a significant 12% decline in overall accuracy. When bleeding was a complaint, there was a significant decline in sensitivity for the diagnosis of ectopic pregnancy as well as a significant decline in predictive value for the diagnosis of an intrauterine pregnancy and a spontaneous miscarriage. Conversely, the presence of pain as a chief complaint has little effect on accuracy of ultrasonography.
In conclusion, transvaginal ultrasonography remains a mainstay in the management of symptomatic first-trimester pregnancy and is an essential tool for detecting an ectopic pregnancy. However, clinicians need to be aware of the limitations of sonography with significant numbers of patients whose final diagnosis differs from the initial impression. Our results provide the gynecologist consulted on such emergency department patients a sense of the degree of risk or reassurance that should be conveyed to patients when weighing the alternatives of immediate surgical or medical intervention compared with watchful waiting. There are significantly more false negative and false-positive diagnoses when the criteria used to identify a gestational sac in the uterus or adnexa do not include the presence of a yolk sac or an embryo. Additionally, accuracy of the initial ultrasonography is significantly lower when the hCG value is low and when the patient reports bleeding. In these clinical situations, extra care should be taken in the interpretation of ultrasound findings and subsequent plan of observation, intervention pending a definitive diagnosis, or both.
1. Barnhart KT. Clinical practice. Ectopic pregnancy. N Engl J Med 2009;361:379–87.
2. Wittels KA, Pelletier AJ, Brown DF, Camargo CA Jr. United States emergency department visits for vaginal bleeding during early pregnancy, 1993–2003. Am J Obstet Gynecol 2008;198:523.el–6.
3. Goldstein SR, Snyder JR, Watson C, Danon M. Very early pregnancy detection with endovaginal ultrasound. Obstet Gynecol 1988;72:200–4.
4. 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.
5. Tayal VS, Cohen H, Norton HJ. Outcome of patients with an indeterminate emergency department first-trimester pelvic ultrasound to rule out ectopic pregnancy. Acad Emerg Med 2004;11:912–7.
6. Kirk E, Papageorghiou AT, Condous G, Tan L, Bora S, Bourne T. The diagnostic effectiveness of an initial transvaginal scan in detecting ectopic pregnancy. Hum Reprod 2007;22:2824–8.
7. Condous G, Okaro E, Khalid A, Lu C, Van Huffel S, Timmerman D, et al. The accuracy of transvaginal ultrasonography for the diagnosis of ectopic pregnancy prior to surgery. Hum Reprod 2005;20:1404–9.
8. Peisner DB, Timor-Tritsch IE. The discriminatory zone of beta-hCG for vaginal probes. J Clin Ultrasound 1990;18:280–5.
9. McRae A, Edmonds M, Murray H. Diagnostic accuracy and clinical utility of emergency department targeted ultrasonography in the evaluation of first-trimester pelvic pain and bleeding: a systematic review. CJEM 2009;11:355–64.
10. Kaplan BC, Dart RG, Muskos 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.
11. Mateer JR, Aiman EJ, Brown MH, Olson DW. Ultrasonographic examination by emergency physicians of patients at risk for ectopic pregnancy. Acad Emerg Med 1995;2:867–73.
12. Barnhart K, van Mello NM, Bourne T, Kirk E, Van Calster B, Bottomley C, et al. Pregnancy of unknown location: a consensus statement of nomenclature, definitions, and outcome. Fertil Steril 2010 Oct 12 [Epub ahead of print].
13. Jaspan D, Giraldo-Isaza M, Dandolu V, Cohen AW. Compliance with methotrexate therapy for presumed ectopic pregnancy in an inner-city population. Fertil Steril 2010;94:1122–4.
14. Lok IH, Neugebauer R. Psychological morbidity following miscarriage. Best Pract Res Clin Obstet Gynaecol 2007;21:229–47.
15. Barnhart KT, Simhan H, Kamelle SA. Diagnostic accuracy of ultrasound above and below the beta-hCG discriminatory zone. Obstet Gynecol 1999;94:583–7.
© 2011 by The American College of Obstetricians and Gynecologists.