Skip Navigation LinksHome > December 2008 - Volume 112 - Issue 6 > Prediction of Location of a Symptomatic Early Gestation Base...
Obstetrics & Gynecology:
doi: 10.1097/AOG.0b013e31818eddcf
Original Research

Prediction of Location of a Symptomatic Early Gestation Based Solely on Clinical Presentation

Barnhart, Kurt T. MD, MSCE1; Casanova, Bruno MD, MSCE1; Sammel, Mary D. ScD2; Timbers, Kelly CNP1; Chung, Karine MD, MSCE1; Kulp, J L. MD1

Free Access
Article Outline
Collapse Box

Author Information

From the 1Department of Obstetrics and Gynecology, Division of Reproductive Endocrinology, and 2Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania, Philadelphia, Pennsylvania.

Supported by the National Institutes of Health (RO1 HD036455 to Dr. Barnhart).

Corresponding author: Dr. Kurt T. Barnhart, Department of Obstetrics and Gynecology, Division of Reproductive Endocrinology, University of Pennsylvania. 3701 Market Street, 8th floor, Philadelphia, PA, 19104; e-mail: kbarnhart@obgyn.upenn.edu.

Financial Disclosure Dr. Barnhart has been an investigator on clinical trials for Boehringer Ingelheim (Ingelheim, Germany), Duramed (Cincinnati, OH), Johnson & Johnson (New Brunswick, NJ), MGI Pharma (Bloomington, MN), Organon (Roseland, NJ), Third Wave Technologies (Madison, WI), Wyeth-Ayerst (Madison, NJ), and Xanodyne (Newport, KY). He has served on the speakers bureau for Organon and has been a consultant to Novo Nordisk (Bagsværd, Denmark). He has also been a medical–legal consultant to Pfizer (New York, NY). The other authors have no potential conflicts of interest to disclose.

Collapse Box

Abstract

OBJECTIVE: To evaluate three strategies for diagnosis of women at risk for ectopic pregnancy based on information collected at initial presentation.

METHODS: Strength of association for risk factors, signs, and symptoms obtained at initial presentation of women with pain, bleeding, or both in a first-trimester pregnancy and a nondiagnostic ultrasound examination were calculated using a cohort of 2,026 women. Three models (logistic regression, a numeric scoring system, and a Classification and Regression Tree) were created to predict final outcome and tested on a second cohort of 1,634 women. Accuracy was assessed using 2×2 tables evaluating the decision of send home compared with do not send home (combination of monitor or intervene) and intervene compared with do not intervene (monitor or send home). Sensitivity, specificity, and predictive values were calculated.

RESULTS: The ultimate diagnosis of women in the test population was 304 (18.6%) patients with an ectopic pregnancy, 834 (51.0%) with miscarriage, and 494 (30.2%) with an ongoing intrauterine pregnancy. A total of 95.9% of patients with ectopic pregnancy or miscarriage were correctly assigned to the strategy to monitor or intervene upon based on the scoring system, and 97.6% based on the Classification and Regression Tree. The specificity of the decision to send a patient home with a likely intrauterine pregnancy was greater than 95% for all three methods. The sensitivity of all strategies in the decision to intervene for an ectopic pregnancy was greater than 98%.

CONCLUSION: A simplified scoring system based on five factors (age, ectopic history, bleeding, prior miscarriage, and human chorionic gonadotropin level) was as effective as a Classification and Regression Tree or logistic regression modes in predicting outcome of women at risk for ectopic pregnancy. Prediction of location of a symptomatic first-trimester pregnancy based on clinical symptoms and risk factors is possible, but must be used in conjunction with outpatient surveillance.

LEVEL OF EVIDENCE: II

Ectopic pregnancy is the leading pregnancy-related cause of death in the first trimester of pregnancy and a major contributor to maternal morbidity.1,2 As a tubal pregnancy progresses, it erodes into blood vessels and can cause massive intraabdominal bleeding. There are limitations in the strategies currently used to diagnose ectopic pregnancy. Even with the use of diagnostic algorithms that systematically evaluate all women at risk for an ectopic pregnancy, only 50% of women with an ectopic pregnancy can be diagnosed upon presentation. Diagnosis in the remaining 50% represents a clinical conundrum and can take up to 6 weeks.3 If the diagnosis of ectopic pregnancy is delayed, rupture of the fallopian tube, resulting in greater risks of morbidity and mortality, can occur. Moreover, an ectopic pregnancy of large size is not amenable to medical therapy and may require major surgery (laparotomy) instead of laparoscopy, which can cause greater damage to the fallopian tube (and greater impairment of fertility), even if treated before rupture.

The presence of abdominal pain and vaginal bleeding in a woman known or suspected to be pregnant should be evaluated urgently to investigate the possibility of an ectopic pregnancy or spontaneous abortion.3,4 Ultrasonography is the mainstay in the diagnosis of women at risk for ectopic pregnancy, but 8–25% of women at risk will have a nondiagnostic ultrasound examination at presentation3,4 resulting in a pregnancy of unknown location.5–7 A solitary human chorionic gonadotropin (hCG) value is not sufficient to definitively diagnose a woman at risk for ectopic pregnancy.3,5,8 Thus, a combination of serial ultrasonography and hCG levels is the most efficient method for surveillance and ultimate diagnosis of women at risk.8,9 However, this diagnostic strategy is cumbersome and requires a high index of clinical suspicion.9,10 Additionally, up to 30% of all women with an ectopic pregnancy present to anemergency department after rupture has occurred or, as we have demonstrated, rupture occurs during the lengthy diagnostic process, which can take weeks.3

Because current strategies of diagnosis and treatment of women at risk for ectopic pregnancy have decreased morbidity, future strategies should focus on minimizing the potential for interrupting a desired viable intrauterine pregnancy and reducing the number of costly and invasive procedures used to confirm a diagnosis. A strategy assessing women at high and low risk for ectopic pregnancy, at initial presentation, would allow a more efficient use of resources and may improve accuracy of diagnosis. The goal of this study was to assess three different strategies to develop a simple, accurate, noninvasive method of diagnosis of ectopic pregnancy based on information available at presentation for care in a woman whose initial ultrasonography is nondiagnostic (does not demonstrate an intrauterine or extrauterine pregnancy) and who is in need of further outpatient surveillance to make a definitive diagnosis.

Back to Top | Article Outline

MATERIALS AND METHODS

This analysis is based on two retrospective cohorts conducted at the University of Pennsylvania Medical Center. Information was obtained from a data set, currently maintained at this Institution, of all women who present with complaints of bleeding and/or pain during the first trimester of pregnancy (by positive pregnancy test or a history of a missed period). The potential risk factors used in our analysis were obtained from the history, clinical presentation, and diagnostic tests gathered by clinicians at initial presentation of a woman at risk for ectopic pregnancy. Approval to conduct the study was obtained from the Institutional Review Board of the University of Pennsylvania.

Risk factors included age, gravidity, number of live births, number of prior cesarean deliveries, number of spontaneous abortions, number of elective abortions, number of previous ectopic pregnancies, pelvic inflammatory disease (defined as inpatient treatment), previous outpatient treatment of gonorrhea and/or Chlamydia, history of pelvic surgery other than cesarean deliveries, history of intrauterine device use, and clinical findings on admission such as pain, hCG level, bleeding, and current gonorrhea/Chlamydia infection. Women were followed until a definite diagnosis was made of spontaneous abortion, ectopic pregnancy, or intrauterine pregnancy. A normal intrauterine pregnancy was confirmed by current progression of the pregnancy by ultrasonography with visualization of an intrauterine yolk sac, fetal pole, or fetal heartbeat. Ectopic pregnancy was diagnosed by the presence of chorionic villi in the fallopian tube after surgical treatment, visualization of an extrauterine gestational sac (with yolk sac and/or embryonic cardiac activity) on ultrasonography for those treated medically or by a rise or stall of the hCG level, after dilatation and evacuation (and no evidence of chorionic villi in the endometrial samples obtained). Spontaneous abortion was confirmed by either histopathologic confirmation of the presence of products of conception from endometrial curettings obtained after dilatation and curettage or by spontaneous decline of hCG level to 5 milli-international units/mL or less. Pain and bleeding were both defined as present if self-reported as presenting symptom for care.

We have previously demonstrated that when using risk factors as predictors of ectopic pregnancy, it is optimal to divide the three potential outcomes (ectopic pregnancy, spontaneous abortion, and ongoing intrauterine pregnancy) into a dichotomous outcome of intrauterine pregnancy compared with nonviable gestation (spontaneous abortion and ectopic pregnancy).11 This approach allows better classification (greater area under a receiver-operating characteristic curve) than the dichotomization into extrauterine pregnancy compared with intrauterine pregnancy (ectopic pregnancy compared with intrauterine pregnancy plus spontaneous abortion).11 Thus our initial step was to divide our sample between those patients diagnosed with intrauterine pregnancy compared with those diagnosed with other types of outcomes (spontaneous abortion and ectopic pregnancy).

Analyses were performed on the first cohort. Bivariable associations were evaluated using Student t test for continuous variables, whereas categorical variables were evaluated by Pearson χ2 test of association. Odds ratios and Wald statistics were constructed to estimate the associations and test for statistical significance between the categorical risk factors and diagnosis. Stratified analyses were then performed to look for effect modifiers and confounding variables. In the case of categorical values, one of the categories was chosen as the reference group. Three methods of creating a predictive model were explored with a focus on discrimination: 1) A logistic regression; 2) a numeric scoring system, and 3) a Classification and Regression Tree (CART, Salford Systems, San Diego, CA) algorithm.

A logistic regression model was created using backward selection methods. At each step, the largest P value variable was removed from the model, and this process was repeated until all variables had a P≤.05 or were identified as confounding variables. Potential confounders were kept in the model if they affected the coefficient estimates of the other variables by 15% or more, giving the final multivariable model.

A second method to predict outcome was then generated in the form of a numeric scoring system. For this system a numeric value (raging from a possible –2 to +14) was assigned to each risk factor that was in the final model based on its strength of associations estimated from the logistic regression model. For example, an odds ratio (OR) of approximately 1.3–1.5 was given a score of 1, an OR of 1.5–2.0 was given a score of 2, an OR of 3.0 or more was given a score of 3, and an OR of 0.5 was given a score of –1. A total score of –2 to –1 was correlated with a low risk for nonviable gestation (ectopic pregnancy or spontaneous abortion); a total score of 0–4 corresponded to an intermediate risk for nonviable gestation, and a total score of 5 or more showed a high risk for a nonviable gestation.

The third method used was a CART (Salford Systems), which was also created based on the variables identified by screening them with logistic regression. Classification and Regression Tree was used to allow for further consideration of combinations of patient characteristics that may further discriminate a potentially viable compared with nonviable gestation. Classification and Regression Tree uses a computer-based technique of cross-validation, CART dividing the sample into 10 roughly equal parts, each containing a similar distribution for the dependent variable. Classification and Regression Tree takes the first nine parts of the data, and uses the remaining 1/10 of the data to obtain initial estimates of the error rate of selected trees. The results of the 10 minitest samples are then combined to form error rates for trees of each possible size; these error rates are applied to the tree based on the entire sample. Decision tables were then generated for the scoring system and CART, and each of them was applied independently to our cohort. Different alternatives of management were then used based on the prediction from each of the modeling strategies for the diagnosis of a potentially viable compared with nonviable gestation.

The developed models were then applied to a second, or test, cohort using logistic regression to assess accuracy of classification. A three-tiered clinical action plan or decision rule was developed based on the outcome from these clinical prediction rules. The three possible action plans were to 1) “send home,” 2) “intervene,” or 3) “monitor.” The three clinical prediction strategies (described above) were applied to the data set retrospectively. Each model was applied to these data to obtain recommendations. In the case of the scoring system, patients with a total score compatible with a low risk (–2 to –1) for ectopic pregnancy or spontaneous abortion (or likely viable intrauterine pregnancy) were assigned the treatment “send home” with less intensive evaluation. Patients at intermediate risk (0–4) were assigned standard clinical evaluation with serial outpatient surveillance, whereas women at high risk for ectopic pregnancy or spontaneous abortion (5 or more) were recommended for intervention to distinguish a miscarriage from an ectopic pregnancy. The recommendations for each of these methods were then compared with the actual definitive diagnosis to assess accuracy. Thus we were able to calculate and compare the sensitivity, specificity, and predictive value for each model.

For the calculation of test characteristic, the three outcomes and the three decision rules were collapsed to 2×2 tables evaluating the decision of send home compared with do not sent home (combination of monitor or intervene) and intervene compared with do not intervene (monitor or send home). Importantly, we focused on each of the three methods’ specificity to send home only patients with a high suspicion of an intrauterine pregnancy and their predictive value for intervening only on those patients with a low suspicion for an intrauterine pregnancy. All the statistical analysis were performed using SAS .9 (SAS Institute Inc., Cary, NC), Stata 9 (StataCorp LP, College Station, TX) and CART (Salford Systems).

Back to Top | Article Outline

RESULTS

The original cohort, consisting of 2,026 patients and whose initial ultrasonography was nondiagnostic and needed outpatient surveillance to obtain a definitive diagnosis, was used in this analysis. These patients presented to the University of Pennsylvania Medical Center with complaints of pain and/or bleeding in the first trimester, collected over a 9-year period (January 1990 through July 1999). A total of 367 (18.11%) patients were diagnosed with ectopic pregnancy, 1,192 (58.84%) patients were diagnosed with a spontaneous abortion, and 467 (23.05%) had an ongoing intrauterine pregnancy. Overall, the median gravidity of our total population was 2 gestations, and the median parity was 1, whereas the mean serum hCG level found was 4,269.7 milli-international units/mL. For women ultimately diagnosed with an intrauterine pregnancy, mean gravidity was 2.3, mean parity was 0.86, and mean hCG was 6,488.9 milli-international units/mL. In the ectopic pregnancy plus spontaneous abortion group mean gravidity of 2.40, mean parity of 0.91, and mean hCG of 3,697 milli-international units/mL, were found. A description of our cohort is presented in Table 1.

Table 1
Table 1
Image Tools

Each prediction model was then tested on a new cohort of women (1,634) collected at the same institution under the same circumstances for 5 years (July 1999 through July 2004). The final diagnosis of this cohort was intrauterine pregnancy 494 (30.2%), spontaneous abortion 834 (51.0%), and ectopic pregnancy 304 (18.6%).

The final multivariable logistic regression model differentiating intrauterine pregnancy from nonviable gestation included the following variables: age, number of ectopic pregnancies, parity, vaginal bleeding on presentation, previous live births, spontaneous abortions, and serum hCG level (Table 2). A scoring system was then developed based on the values of odds ratios obtained from the logistic regression model. This scoring system is presented in Table 3. A CART tree was also developed as the third method of evaluation. After testing several configurations of CART options, it was found that CART with a node size of 40, under which no more partitions were to be made, brought the best results. The optimal CART tree obtained is presented in Figure 1.

Table 2
Table 2
Image Tools
Table 3
Table 3
Image Tools
Fig. 1
Fig. 1
Image Tools

Information regarding three different treatment options for sending the patient home, continue monitoring or intervene, for both CART and the scoring system, are presented in Table 4. We found that 95.9% of patients with ectopic pregnancy or spontaneous abortion were correctly monitored or intervened based on the scoring system, whereas the same number was 97.6% based on CART.

Table 4
Table 4
Image Tools

A summary of the collapsed 2×2 decision tables (send home/do not send home), (intervene/do not intervene) for the three methods under evaluation are presented in Tables 5 and 6. In these two tables sensitivity, specificity, negative predictive values and positive predictive values were presented for each of the methods under evaluation.

Table 5
Table 5
Image Tools
Table 6
Table 6
Image Tools
Back to Top | Article Outline

DISCUSSION

The current standard of care when evaluating a woman with a potentially complicated first-trimester pregnancy is to take a detailed history of risk factors and ascertain the clinical course. The classic risk factors for ectopic pregnancy and miscarriage have been well-characterized in a normal population12 as well as in the population presenting with pain and bleeding.13,14 Knowledge of risk factors is particularly valuable if they can inform clinical care. The use of risk factors and presenting information as part of a clinical decision rule can be helpful for clinicians in cases when diagnosis is not definitive upon initial presentation for care. A clinical prediction rule enables clinicians to use information-based risk factors to assess the probability of developing it in the future.15 This type of strategy has been used to predict children at risk for dehydration,16 interpretation of testing to assess for pulmonary embolism,17 management of cytology for women with abnormal Pap tests,18 and in the prediction of iron deficiency anemia in pregnant women.19 In the situation of women at risk for an ectopic pregnancy, a potential clinical decision rule would be used for a woman with a “pregnancy of unknown location” or when the initial ultrasonography was nondiagnostic and further tests are required to confirm a diagnosis.

In this study we used three different methods to assess the predictive ability of signs and symptoms at presentation to discriminate an intrauterine pregnancy from the presence of a nonviable gestation in a first trimester pregnancy, each with its own advantages and disadvantages. Logistic regression is a generalized linear statistical model used to predict a specific outcome and is widely used in medical statistics today. This method is very precise, but requires the use of a computer to estimate risk. The second method, a scoring system, uses the variables included in the final logistic model in a simplified form by assigning scores to each based on the weight of the odds ratios. This method is very easy to use, and it does not require a computer system. However it may not be as precise as a full logistic regression model, because risk was “rounded” to obtain a numeric score. We also considered CART, an automated data-analysis tool that searches for important patterns and relationships to generate accurate and reliable predictive models.20 CART uses tree-building algorithm to determine a set of if-then logical (split) conditions that permit accurate prediction of a specific outcome. The advantages of using CART are the simplicity of the results and that there is no implicit assumption about the underlying relationships between the predictor variables and the dependent variable.

The diagnosis of women at risk for ectopic pregnancy is complicated by the fact that there are three possible outcomes (ectopic pregnancy, spontaneous abortion, and ongoing intrauterine pregnancy) that dictate very different treatment strategies and consequences of missed diagnosis. Our three-tiered decision model was designed to mimic potential clinical care. “Send home” suggested a strategy of discharging the patient from the hospital with minimal further evaluation planned other than referral for potential care as there was high suspicion of intrauterine pregnancy. “Intervene” suggested a strategy for a patient with a nonviable pregnancy at high risk for ectopic pregnancy. “Intervention” does not indicate the need for surgical procedure, rather the need for increased level of care or counseling. The “increased care,” recommending further urgent evaluation, is suggested to diagnose or treat (such as laparoscopy, uterine curettage, expectant management with heightened suspicion, and/or surveillance). “Monitor” suggests a strategy when there is insufficient information to distinguish among ectopic pregnancy, spontaneous abortion, and intrauterine pregnancy. In this instance the clinical prediction rule was not informative and current standard care of serial hCG values and ultrasonography should be used to make the ultimate definitive diagnosis in a woman at risk for ectopic pregnancy.

Our strategy was to assess the capacity of each method to identify patients with an intrauterine pregnancy, for which no further treatment was necessary (send home/do not send home) and also to identify patients with a potential ectopic pregnancy who require immediate treatment (intervene/do not intervene). The advantage of our tiered treatment/surveillance strategy was that sensitivity and specificity could be maximized at different decision points in an effort to minimize false-positive and false-negative diagnoses. For example, we attempted to maximize specificity (at the expense of sensitivity) when confronted with the decision to send home, reducing the possibility of sending women with an ectopic pregnancy home with no follow-up. Conversely, we maximized sensitivity and negative predictive value (at the expense of specificity) when confronted with the decision to intervene. In this instance, intervention (to distinguish a miscarriage from an ectopic pregnancy) was only entertained when nonviable gestation was confirmed, thus the chance of interrupting a viable pregnancy was low.

The test characteristics of each of our three strategies, in these two clinical situations, demonstrated a remarkable equivalence. The specificity of diagnosis in the decision to send home was greater than 95% for all three methods, with negative predictive value approximately 81%. In the case of the decision to intervene, all strategies had a sensitivity greater than 98%, with all three having a negative predictive value of greater than 93%. Based on these data we concluded that the use of a simple numeric scoring system can be as accurate as a more complicated computer model derived using logistic regression or CART. Therefore, data obtained on the initial presentation of a woman at risk for ectopic pregnancy can be used to inform a clinician as to risk of ectopic pregnancy and possibly aid in the decision on the frequency of follow-up. Of note, race was not part of our model, because it was unavailable in the majority of cases in our data set.

There are two important limitations of the diagnostic strategy described. The decision to send home or intervene based on risk factors is only made in a small subset of the population. In the case of the scoring system, 70% of those ultimately diagnosed with an intrauterine pregnancy, and 81% of those diagnosed with a nonviable gestation are recommended to be treated with the usual standard of care, including surveillance with serial hCG values and/or ultrasonography.9,10,21–23 Moreover, the accuracy of the rule is not 100%. Of the approximately 30% of women with an intrauterine pregnancy, 1.8% would have been falsely diagnosed as a nonviable gestation. Therefore, this classification system cannot be used in isolation without continued outpatient surveillance.

In summary, the strategy of surveillance of a woman who presents with pain and bleeding in the first trimester in those not initially definitely diagnosed with an ongoing intrauterine pregnancy, spontaneous abortion, or ectopic pregnancy is cumbersome. Others have attempted to simply follow up examining hCG values. 24–26 However, we have demonstrated that a simple clinical decision rule can be used to assess the chance of a woman having an ongoing intrauterine pregnancy despite the presence of vaginal bleeding/pain. This allows clinicians to potentially individualize a patient to outpatient follow-up or to intervention to distinguish an ongoing intrauterine pregnancy from a miscarriage and ectopic pregnancy. The simple rule based on five factors has good diagnostic accuracy, and distinguishes approximately 30% of the population as high or low risk. Like other clinical decision rules, this simple scoring system should not be used solely, but in combination with current strategies to definitively diagnose women at risk for ectopic pregnancy.

Back to Top | Article Outline

REFERENCES

1. Tenore JL. Ectopic pregnancy. Am Fam Physician 2000;61:1080–8.

2. Atrash HK, Hughes JM, Hogue CJ. Ectopic pregnancy in the United States, 1970-1983. MMWR CDC Surveill Summ 1986;35:29SS–37SS.

3. 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.

4. Mol BM, Hajenius PJ, Ankum WM, Bossuyt PM, van der Veen F. Screening for ectopic pregnancy in symptom-free women at increased risk. Obstet Gynecol 1997;89:704–7.

5. Condous G, Lu C, Van Huffel SV, Timmerman D, Bourne T. Human chorionic gonadotrophin and progesterone levels in pregnancies of unknown location. Int J Gynaecol Obstet 2004;86:351–7.

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. Barnhart KT, Katz I, Hummel A, Gracia CR. Presumed diagnosis of ectopic pregnancy. Obstet Gynecol 2002;100:505–10.

8. Gracia CR, Barnhart KT. Diagnosing ectopic pregnancy: decision analysis comparing six strategies. Obstet Gynecol 2001;97:464–70.

9. 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.

10. American College of Obstetricians and Gynecologists. ACOG Practice Bulletin No. 94: medical management of ectopic pregnancy. Obstet Gynecol 2008;111:1479–85.

11. Casanova BC, Sammel MD, Chittams J, Timbers K, Kulp JL, Barnhart KT. Prediction of outcome for women with symptomatic first trimester pregnancy: the focus should be on intrauterine rather than ectopic gestation. J Women’s Health, in press.

12. Ankum WM, Mol BW, Van der Veen F, Bossuyt PM. Risk factors for ectopic pregnancy: a meta-analysis. Fertil Steril 1996;65:1093–9.

13. Barnhart KT, Sammel MD, Gracia CR, Chittams J, Hummel AC, Shaunik A. Risk factors for ectopic pregnancy in women with symptomatic first-trimester pregnancies. Fertil Steril 2006;86:36–43.

14. Gracia CR, Sammel MD, Chittams J, Hummel AC, Shaunik A, Barnhart KT. Risk factors for spontaneous abortion in early symptomatic first-trimester pregnancies. Obstet Gynecol. 2005;106:993–9.

15. Baron JA. The clinical utility of risk factor data. J Clin Epidemiol 1989;42:1013–20.

16. Gorelick MH, Shaw KN, Murphy KO. Validity and reliability of clinical signs in the diagnosis of dehydration in children. Pediatrics 1997;99:E6.

17. Value of the ventilation/perfusion scan in acute pulmonary embolism. Results of the prospective investigation of pulmonary embolism diagnosis (PIOPED). The PIOPED Investigators. JAMA 1990;263:2753–9.

18. ASCUS-LSIL Triage Study (ALTS) Group. Results of a randomized trial on the management of cytology interpretations of atypical squamous cells of undetermined significance. Am J Obstet Gynecol 2003;188:1383–92.

19. Casanova BF, Sammel MD, Macones GA. Development of a clinical prediction rule for iron deficiency anemia in pregnancy. Am J Obstet Gynecol 2005;193:460–6.

20. Lemon SC, Roy J, Clark MA, Friedmann PD, Rakowski W. Classification and regression tree analysis in public health: methodological review and comparison with logistic regression. Ann Behav Med 2003;26:172–81.

21. Barnhart K, Sammel MD, Chung K, Zhou L, Hummel AC, Guo W. Decline of serum human chorionic gonadotropin and spontaneous complete abortion: defining the normal curve. Obstet Gynecol 2004;104:975–81.

22. Barnhart KT, Sammel MD, Rinaudo PF, Zhou L, Hummel AC, Guo W. Symptomatic patients with an early viable intrauterine pregnancy: HCG curves redefined. Obstet Gynecol 2004;104:50–5.

23. Silva C, Sammel MD, Zhou L, Gracia C, Guo W, Hummel AC, Barnhart KT. Human chorionic gonadotropin profile for women with an ectopic pregnancy. Obstet Gynecol 2006;107:605–10.

24. Condous G, Okaro E, Khalid A, Timmerman D, Lu C, Zhou Y, et al. The use of a new logistic regression model for predicting the outcome of pregnancies of unknown location. Hum Reprod 2004;19:1900–10.

25. Condous G, Van Calster B, Kirk E, Haider Z, Timmerman D, Van Huffel S, et al. Prediction of ectopic pregnancy in women with a pregnancy of unknown location. Ultrasound Obstet Gynecol 2007;29:680–7.

26. Condous G, Van Calster B, Kirk E, Timmerman D, Van Huffel S, Bourne T. Prospective cross-validation of three methods of predicting failing pregnancies of unknown location. Hum Reprod 2007;22:1156–60.

Cited By:

This article has been cited 4 time(s).

Fertility and Sterility
Validation of a clinical risk scoring system, based solely on clinical presentation, for the management of pregnancy of unknown location
Barnhart, KT; Sammel, MD; Takacs, P; Chung, K; Morse, CB; O'Brien, KO; Allen-Taylor, L; Shaunik, A
Fertility and Sterility, 99(1): 193-198.
10.1016/j.fertnstert.2012.09.012
CrossRef
Journal of Ultrasound in Medicine
Efficient Use of Early Obstetric Sonography in the Emergency Department Setting
Goodman, LR; Eichelberger, KY; Wolfe, H; Steiner, AZ
Journal of Ultrasound in Medicine, 31(): 1917-1923.

New England Journal of Medicine
Ectopic Pregnancy
Barnhart, KT
New England Journal of Medicine, 361(4): 379-387.

Clinical Obstetrics and Gynecology
Ectopic Pregnancy:: Role of Laparoscopic Treatment
EHRENBERG-BUCHNER, S; SANDADI, S; MOAWAD, NS; PINKERTON, JS; HURD, WW
Clinical Obstetrics and Gynecology, 52(3): 372-379.
10.1097/GRF.0b013e3181b0be24
PDF (131) | CrossRef
Back to Top | Article Outline

© 2008 The American College of Obstetricians and Gynecologists

Login

Article Tools

Images

Share