Accurate prediction of pregnancy viability by means of a simple scoring system

Human Reproduction, Vol.28, No.1 pp. 68– 76, 2013 Advanced Access publication on October 30, 2012 doi:10.1093/humrep/des352 ORIGINAL ARTICLE Early pregnancy Accurate prediction of pregnancy viability by means of a simple scoring system Cecilia Bottomley 1,*, Vanya Van Belle 2, Emma Kirk3, Sabine Van Huffel 2, Dirk Timmerman 4, and Tom Bourne 3,4 1 Department of Obstetrics and Gynaecology, Chelsea and Westminster Hospital, 369 Fulham Road, London SW10 9NH, UK 2IBBT Future Health Department, ESAT, SCD-SISTA, Kasteelpark Arenberg 10, Box 2446, 3001 Leuven, Belgium 3Imperial College London, Hammersmith Campus, Du Cane Road, London, UK 4Department of Obstetrics and Gynaecology, University Hospitals KU Leuven, Leuven, Belgium *Correspondence address. Tel: +44-7985-937833; Fax: +44-0208-746-5074; E-mail: Submitted on January 27, 2012; resubmitted on July 26, 2012; accepted on September 3, 2012 study question: What is the performance of a simple scoring system to predict whether women will have an ongoing viable intrauterine pregnancy beyond the first trimester? summary answer: A simple scoring system using demographic and initial ultrasound variables accurately predicts pregnancy viability beyond the first trimester with an area under the curve (AUC) in a receiver operating characteristic curve of 0.924 [95% confidence interval (CI) 0.900–0.947] on an independent test set. what is known already: Individual demographic and ultrasound factors, such as maternal age, vaginal bleeding and gestational sac size, are strong predictors of miscarriage. Previous mathematical models have combined individual risk factors with reasonable performance. A simple scoring system derived from a mathematical model that can be easily implemented in clinical practice has not previously been described for the prediction of ongoing viability. study design, size and duration: This was a prospective observational study in a single early pregnancy assessment centre during a 9-month period. participants/materials, setting and methods: A cohort of 1881 consecutive women undergoing transvaginal ultrasound scan at a gestational age ,84 days were included. Women were excluded if the first trimester outcome was not known. Demographic features, symptoms and ultrasound variables were tested for their influence on ongoing viability. Logistic regression was used to determine the influence on first trimester viability from demographics and symptoms alone, ultrasound findings alone and then from all the variables combined. Each model was developed on a training data set, and a simple scoring system was derived from this. This scoring system was tested on an independent test data set. main results and the role of chance: The final outcome based on a total of 1435 participants was an ongoing viable pregnancy in 885 (61.7%) and early pregnancy loss in 550 (38.3%) women. The scoring system using significant demographic variables alone (maternal age and amount of bleeding) to predict ongoing viability gave an AUC of 0.724 (95% CI ¼ 0.692–0.756) in the training set and 0.729 (95% CI ¼ 0.684–0.774) in the test set. The scoring system using significant ultrasound variables alone (mean gestation sac diameter, mean yolk sac diameter and the presence of fetal heart beat) gave an AUC of 0.873 (95% CI ¼ 0.850– 0.897) and 0.900 (95% CI ¼ 0.871– 0.928) in the training and the test sets, respectively. The final scoring system using demographic and ultrasound variables together gave an AUC of 0.901 (95% CI ¼ 0.881 –0.920) and 0.924 (CI ¼ 0.900–0.947) in the training and the test sets, respectively. After defining the cut-off at which the sensitivity is 0.90 on the training set, this model performed with a sensitivity of 0.92, specificity of 0.73, positive predictive value of 84.7% and negative predictive value of 85.4% in the test set. limitations, reasons for caution: BMI and smoking variables were a potential omission in the data collection and might further improve the model performance if included. A further limitation is the absence of information on either bleeding or pain in 18% of women. Caution should be exercised before implementation of this scoring system prior to further external validation studies & The Author 2012. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. For Permissions, please email: 69 Early pregnancy viability score wider implications of the findings: This simple scoring system incorporates readily available data that are routinely collected in clinical practice and does not rely on complex data entry. As such it could, unlike most mathematical models, be easily incorporated into normal early pregnancy care, where women may appreciate an individualized calculation of the likelihood of ongoing pregnancy viability. study funding/competing interest(s): Research by V.V.B. supported by Research Council KUL: GOA MaNet, PFV/10/ 002 (OPTEC), several PhD/postdoc & fellow grants; IWT: TBM070706-IOTA3, PhD Grants; IBBT; Belgian Federal Science Policy Office: IUAP P7/(DYSCO, ‘Dynamical systems, control and optimization’, 2012– 2017). T.B. is supported by the Imperial Healthcare NHS Trust NIHR Biomedical Research Centre. trial registration number: Not applicable. Key words: miscarriage / first trimester / pregnancy / transvaginal ultrasound / viability Introduction Miscarriage [defined as the loss of pregnancy before 24 completed weeks’ of gestation (RCOG, 2006)] is the most common serious pregnancy complication. It is not generally associated with serious physical morbidity or mortality (Trinder et al., 2006), but very often has a significant social and psychological impact on the mother (Lok and Neugebauer, 2007). The physician has an important role in making an accurate and certain diagnosis of pregnancy failure and relaying this information in a sensitive and professional manner. Evidence-based practice should always be employed to ensure that an accurate diagnosis is made in order to avoid inadvertent termination of a potentially viable pregnancy; but the physician must also avoid giving false assurance regarding a pregnancy that is in fact likely to fail. The exact incidence of miscarriage is difficult to assess. A commonly quoted figure for the rate of clinical miscarriage is around one in five recognized pregnancies (Savitz et al., 2002). Prospective observational studies suggest a rate of 31–38% when the high number of ‘biochemical’ pregnancies is included (Ellish et al., 1996; Zinaman et al., 1996). It is likely that the rate of miscarriage of clinically recognized pregnancies is increasing as a result of earlier recognition of pregnancy with commercially available, highly sensitive, urine-based, home pregnancy tests and the easy availability of transvaginal ultrasonography (TVS). The number of women who suffer a miscarriage at home without any medical involvement is unknown but the miscarriage rate among women attending an early pr (...truncated)