Connolly, AnnaMarie MD; Ryan, David H. MD; Stuebe, Alison M. MD; Wolfe, Honor M. MD
Many women experience vaginal bleeding and pain in the first trimester of pregnancy. Estimates for first-trimester bleeding range from 15% to 25%, with roughly half of these pregnancies ending in miscarriage by 20 weeks of gestation.1,2 A woman presenting with these symptoms may have a normal intrauterine pregnancy, embryonic death, or ectopic pregnancy. The increased availability of diagnostic technology has led to evaluation of pregnancies at early gestational ages to better assess these symptoms.3–13 This technology includes the measurement of β-hCG and the use of ultrasonography.
Although the discriminatory level describes the serum β-hCG level at which ultrasonographic findings should be detected, the threshold level for β-hCG refers to the lowest serum β-hCG level at which an ultrasonographic finding such as a gestational sac or fetal pole can be detected. The concept of the discriminatory “zone” was first described by Kadar and colleagues3 in relation to visualization of a gestational sac with the use of transabdominal ultrasonography at a β-hCG level of 6,500 milli-international units/mL. With the development of transvaginal ultrasonography, progressively lower discriminatory levels of β-hCG were found to be associated with normally developing intrauterine pregnancies. These levels ranged from 1,000 milli-international units/mL to 2,000 milli-international units/mL.4–14 A discriminatory β-hCG zone commonly used in clinical practice, 1,500–2,000 milli-international units/mL, is derived from the work of Barnhart et al.8 These authors followed 68 consecutive pregnancies in women with reliable menstrual histories who ovulated spontaneously. Repeat ultrasonographic studies and serum β-hCG levels were performed serially until a gestational sac was visualized. The phase of this study during which this discriminatory zone was established (“Phase I”) did not evaluate women presenting with pain or bleeding, and thus may not be generalizable to the clinical settings where ultrasonography is commonly used.8 Of note, serum β-hCG levels have been measured over the years using the First International Reference Preparation, the Second International Standard, the Third International Standard, and the Fourth International Standard. The Second International Standard produces values that measure approximately half the value of the First International Reference Preparation, the Third International Standard, and Fourth International Standard. All β-hCG levels reported in this article are referenced against the First International Reference Preparation, the Third International Standard, or the Fourth International Standard. Changes in ultrasonography technology over the past 15 years have resulted in improved diagnostic capabilities. We hypothesized that these improvements would result in both lower discriminatory and threshold β-hCG values in the first trimester of pregnancy. We performed a retrospective cohort study to evaluate the effect of these changes on discriminatory and threshold β-hCG levels in the first trimester of pregnancy.
MATERIALS AND METHODS
After receiving approval by the Office of Human Research Ethics at the University of North Carolina, we performed a retrospective cohort study of all women presenting to the Emergency Department at the University of North Carolina with vaginal bleeding, pelvic pain, or both in the first trimester of pregnancy from August 1, 2007, through July 31 2009. Women in whom transvaginal ultrasonography and serum β-hCG levels (Third International Standard) were performed within 6 hours of each other were included in the analysis. Those with an ectopic pregnancy, multiple gestation, suboptimal scanning conditions (eg, uterine leiomyomas), or unknown pregnancy outcome were excluded from further analysis. For women for whom data from multiple visits were available, only the earliest visit was used in the analysis.
Viable intrauterine gestations were defined by the presence of fetal cardiac activity at the time of the initial or follow-up ultrasonograms or the delivery of a live neonate or stillbirth whose gestational age corresponded with that predicted by the original ultrasonogram. Nonviable intrauterine pregnancies were defined by any of the following: 1) the presence of a gestational sac with a mean sac diameter 20 mm or greater without ultrasonographic visualization of a fetal pole; 2) a crown rump length of greater than or equal to 5 mm without ultrasonographic documentation of cardiac activity; 2) a falling serum β-hCG level within 1 week of the initial β-hCG; 4) a negative serum βhCGβ-hCG or urine pregnancy test within 3 months; or 5) documentation of a “failed pregnancy” in an operative report. Operative reports, rather than pathology reports, were used because these were uniformly available for chart review, which was not the case for pathology reports. All surgical procedures were performed within the University of North Carolina health care system.
Transvaginal ultrasonography examinations were performed by registered diagnostic medical ultrasonographers from the department of radiology or the prenatal diagnostic ultrasound unit in the department of obstetrics and gynecology at the University of North Carolina. All studies were reviewed by a board-certified maternal-fetal medicine physician with experience in the interpretation of first-trimester obstetric ultrasonograms who made the determination of the presence of a gestational sac, yolk sac, and fetal pole. Ultrasonographic examinations were performed using a 8- to 9-MHz vaginal transducer on a GE Logic E8, Acuson Sequoia, Siemens S2000, or Phillips IU22 ultrasonography machine.
Serum β-hCG levels were reported according to the Third International Reference Preparation and all samples were analyzed in the same laboratory.
To determine the threshold for detection of a gestational sac, yolk sac, and fetal pole, we analyzed all viable pregnancies and identified the lowest serum β-hCG level at which these structures were identified. To determine discriminatory levels, we used logistic regression to model the association between β-hCG value and visualization of a gestational sac, yolk sac, and fetal pole using fractional polynomials.15 Fractional polynomials allow for a nonlinear association between a continuous parameter and an outcome of interest. We selected the parameterization with the best fit based on the log likelihood test. We used the resulting models to identify the β-hCG value at which there would be a 50%, 90%, 95%, and 99% probability of visualizing each structure in a viable pregnancy. We defined confidence intervals (CIs) as the β-hCG value at which the lower and upper 95% CIs for visualizing each structure was 50%, 90%, 95%, or 99%. SAS 9.2 was used for all analyses.
During the study period, 1,015 women presenting with vaginal bleeding, pain, or both in the first trimester of pregnancy had an ultrasonogram and serum β-hCG level performed within 6 hours of each other. In this series, 295 cases were excluded (160 pregnancies with unknown outcomes, 92 ectopic pregnancies, and 43 pregnancies excluded for other reasons such as multiple gestation and suboptimal scanning conditions). For women for whom data from multiple visits were available (n=69), only the earliest visit was used in the analysis. As such, a total of 651 pregnancies (366 viable, 285 nonviable) were analyzed.
The majority of nonviable pregnancies had low levels of β-hCG. However, there were nonviable pregnancies seen with β-hCG levels as high as 122,000 milli-international units/mL. As expected, higher β-hCG levels were associated with a greater proportion of ultimately viable pregnancies.
Prior work has defined the discriminatory level as the highest β-hCG value where a given structure should be visualized on ultrasonography. In our population, the highest observed values at which structures were not visualized in ultimately viable pregnancies were 2,317 milli-international units/mL, 9,975 milli-international units/mL, and 35,486 milli-international units/mL for gestational sac, yolk sac, and fetal pole, respectively.
We further used logistic regression with fractional polynomials to model the association between serum β-hCG level and visualization of a gestational sac, yolk sac, and fetal pole. Using these models, we identified the β-hCG level for which the predicted probability of visualizing each structure was 50%, 90%, 95%, and 99% (Fig. 1; Table 1). We defined the modeled discriminatory β-hCG value as the value at which there was a 99% probability of visualizing these structures in a viable pregnancy. In our sample, the modeled discriminatory β-hCG values were 3,510 milli-international units/mL, 17,716 milli-international units/mL, and 47,685 milli-international units/mL for gestational sac, yolk sac, and fetal pole.
Using a serum β-hCG level of 1,500 milli-international units/mL, the predicted probability of seeing a gestational sac in a viable pregnancy was 80.4% (95% CI 56.5–92.8%). Using the previously reported β-hCG threshold of 2,000 milli-international units/mL, the predicted probability reached 91.2% (95% CI 77.6–99.4%). All discriminatory values were higher than those previously reported (Table 2).
The lowest β-hCG value at which a gestational sac was visualized in a viable pregnancy, the threshold value, was 390 milli-international units/mL. The threshold values for yolk sac and fetal pole visualization were 1,094 milli-international units/mL and 1,394 milli-international units/mL, respectively, in viable pregnancies (Table 2).
This is the first study of which we are aware using cross-sectional data to specifically establish discriminatory and threshold serum β-hCG levels in a large population of patients presenting with pain, bleeding, or both using current ultrasonographic technology. Lower threshold values than previously reported were found for visualization of a gestational sac, yolk sac, and fetal pole, reflecting the use of higher frequency ultrasonographic probes. However, the discriminatory levels were higher than previously reported for all three structures, reflecting the limitations of ultrasonography in clinical practice rather than a controlled research environment. In addition, we reported 50%, 90%, 95%, and 99% probabilities for visualization of three early first-trimester ultrasonographic findings. These values may allow clinicians to better counsel patients about the likelihood of an ultimately viable gestation, which could affect clinical care positively. Although previous work by Doubilet and Benson examined 202 patients over an 11-year period and found that all nine of the women with β-hCG levels 2,000 milli-international units/mL or higher had ultrasonographically confirmed intrauterine pregnancies, this work did not examine threshold and discriminatory levels for each of the three early gestational structures seen ultrasonographically.16
As expected, improvements in ultrasonography technology over the past 10–15 years, including the routine use of 8- to 12-MHz probes, have resulted in lower threshold values for the detection of early intrauterine pregnancy findings. The threshold value of 390 milli-international units/mL for gestational sac, the earliest ultrasonographically visible pregnancy structure, was lower than previously reported. Values for yolk sac (1,094 milli-international units/mL) and fetal pole (1,394 milli-international units/mL) were markedly lower than in previous studies (Table 2).4–13 Unexpectedly, however, we found that the discriminatory β-hCG levels for visualization of early gestational structures were higher than previously reported.4–14 Particularly, our cross-sectional study found that, in our population, the β-hCG level associated with a 99% predicted probability of detecting an intrauterine gestational sac was 3,510 milli-international units/mL. Previously reported original evidence, much of which was obtained through longitudinally conducted work, had placed the β-hCG discriminatory zone at a lower level of 1,000–2,000 milli-international units/mL.4–6,8–14 Potential explanations for these findings include the limitations of actual, cross-sectional, clinical practice; patient discomfort; patient anatomy; duration of the examination; and delayed rather than real-time visualization of ultrasonographic images by physicians.
Previous reports that established currently used discriminatory β-hCG values4–6,8–12,14 were characterized by small sample size (10–74 pregnancies) and were performed in the late 1980s and early 1990s using 3.5- to 7.5-MHz probes.4–6,8–12,14 This prior work was often longitudinal in study design.5,6,8,12,14 Patients were followed serially with transvaginal ultrasonographic examinations and serum β-hCG levels in controlled research settings until the ultrasonographic findings under study were detected. Previous work by Barnhart et al13 examining 333 consecutive pregnant women who presented to the emergency department investigated the diagnostic accuracy of a discriminatory level commonly used in clinical practice, 1,500 milli-international units/mL; however, they did not seek to report on individual discriminatory levels for early gestational structures.
The serum β-hCG level is widely used in clinical practice to inform medical decision-making regarding the location or viability of a pregnancy on the basis of a single clinical encounter. Many have reported the limitations of such use of a single β-hCG level and ultrasonographic examination.13,16–19 This underlines the importance of discriminatory β-hCG levels that are associated with high probabilities of detecting early gestational structures. We found that the serum β-hCG level associated with 99% probability of detection of a gestational sac was 3,510 milli-international units/mL, higher than the current discriminatory level. Use of the previously reported β-hCG level of 1,500 milli-international units/mL as the discriminatory value for patients in this study would have detected only 80% of the ultimately viable pregnancies with the potential mismanagement of up to 20% of viable pregnancies as nonviable gestations. Even at the higher β-hCG value of 2,000 milli-international units/mL, the predicted probability reached only 91.2%. Serum β-hCG levels associated with 99% probability of detection of a yolk sac and a fetal pole were 17,716 milli-international units/mL and 47,685 milli-international units/mL, respectively.
This study has several limitations, including variation in ultrasonographer ability and ultrasonographic equipment, although all ultrasonographers were registered diagnostic medical ultrasonographers and all ultrasonographic studies were performed with a minimum of an 8-mHz transvaginal probe. Although patients with suboptimal scanning conditions were not included in this study, there were variations in image quality based on maternal factors such as body habitus and discomfort. All of these limitations are inherent to clinical practice. Given strict imaging criteria and the high level of training of ultrasonographers and physicians, the discriminatory levels described in this work are likely the lowest that should be used in general clinical practice.
In conclusion, improvements in ultrasonographic technology have lowered threshold values for ultrasonographic visualization of a gestational sac, yolk sac, and fetal pole in the first trimester. However, currently used discriminatory serum β-hCG levels are too low for use in clinical practice and may result in the underdiagnosis of ultimately viable gestations.
1. Everett C. Incidence and outcome of bleeding before the 20th week of pregnancy: prospective study from general practice. BMJ 1997;315:32–4.
2. Hasan R, Baird DD, Herring AH, Jonsson Funk ML, Hartmann KE. Association between first-trimester vaginal bleeding and miscarriage. Obstet Gynecol 2009;114:860–7.
3. Kadar N, DeVore G, Romero R. Discriminatory hCG zone: its use in the sonographic evaluation for ectopic pregnancy. Obstet Gynecol 1981;58:156–61.
4. Goldstein SR, Snyder JR, Watson C, Danon M. Very early pregnancy detection with endovaginal ultrasound. Obstet Gynecol 1988;72:200–4.
5. Aleem FA, DeFazio M, GIntautas J. Endovaginal sonography for the early diagnosis of intrauterine and ectopic pregnancies. Hum Reprod 1990;5:755–8.
6. Fossum GT, Davaian V, Kletzky OA. Early detection of pregnancy with transvaginal ultrasound. Fertil Steril 1988;49:788–91.
7. Cacciatore B, Stenman UH, Ulostalo P. Diagnosis of ectopic pregnancy by vaginal ultrasonography in combination with a discriminatory serum hCG level of 1000IU/I (IRP). Br J Obstet Gynaecol 1990;97:904–8.
8. 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.
9. Bernaschek G, Rudelstorfer R, Csaicsich P. Vaginal sonography versus serum human chorionic gonadotropin in early detection of pregnancy. Am J Obstet Gynecol 1988;158:608–12.
10. Nyberg DA, Mack LA, Laing FC, Jeffrey RB. Early pregnancy complications: endovaginal sonographic findings correlated with human chorionic gonadotropin levels. Radiology 1988;167:619–22.
11. Bree RL, Edwards M, Bohm-Vélez M, Beyler S, Roberts J, Mendelson EB. Transvaginal sonography in the evaluation of normal early pregnancy: correlation with HCG level. AJR Am J Roentgenol 1989;153:75–9.
12. Daya S, Woods S, Lappalainen R, Caco C. Transvaginal ultrasound scanning in early pregnancy and correlation with human chorionic gonadotropin levels. J Clin Ultrasound 1991;19:139–42.
13. Barnhart KT, Simhan H, Kamelle SA. Diagnostic accuracy of ultrasound above and below the beta-hCG discriminatory zone. Obstet Gynecol 1999;94:583–7.
14. Bateman BG, Nunley WC, Kolp LA, Kitchin JD, Felder R. Vaginal sonography findings and hCG dynamics of early intrauterine and tubal pregnancies. Obstet Gynecol 1990;75:421–7.
15. Royston F, Altman DG. Regression using fractional polynomials of continuous covariates: parsimonious parametric modeling (with discussion). Appl Stat 1994;43:429–67.
16. Doubilet PM, Benson CB. Further evidence against the reliability of the human chorionic gonatropin discriminatory level. J Ultrasound Med 2011;30:1637–42.
17. Seeber BE, Sammel MD, Guo W, Zhou L, Hummel A, Barnhart KT. Application of redefined human chorionic gonadotropin curves for the diagnosis of women at risk for ectopic pregnancy Fertil Steril 2006;86:454–9.
18. Gracia CR, Barnhart KT. Diagnosing ectopic pregnancy: decision analysis comparing six strategies. Obstet Gynecol 2001;97:464–70.
19. Barnhart KT, Fay CA, Suescum M, Sammel MD, Appleby D, Shaunik A, et al.. Clinical factors affecting the accuracy of ultrasonography in symptomatic first-trimester pregnancy. Obstet Gynecol 2011;117:299–306.
Figure. No available...Image Tools