Predicting Outcomes in Pregnancies of Unknown Location

Emma Kirk; Tom Bourne


Women's Health. 2008;4(5):491-499. 

In This Article

Mathematical Models

Recently, mathematical models using logistic-regression and Bayesian networks have been developed in order to predict the outcome of PULs ( Table 1 ). Two logistic-regression models (M1 and M4) have been developed to predict the outcomes of PULs, based on the results of two serum hCG levels taken at 0 and 48h, respectively, ( Table 1 ).[21,22,23] Model M1 is based on the hCG ratio and its performance was evaluated using receiver operator characteristic (ROC) analysis. It gave an area under the ROC curve (AUC) of 0.975 for failing PUL, 0.966 for IUP and 0.885 for ectopic pregnancy. For the detection of ectopic pregnancy, it had a sensitivity of 91.7%, a specificity of 84.2%, a positive likelihood ratio of 5.8, a positive predictive value of 27.5% and a negative predictive value of 99.4%.[21] When tested in the clinical setting, this model was found to compare favorably with subjective assessment by experienced nurse operators.[24] M4 is based on the hCG average ([hCG 0h + hCG 48h]/2), the hCG ratio and its quadratic effect.[22] During development on a set of 376 PULs, M4 gave an AUC of 0.978 (95% CI:0.954-1.000) for the prediction of failing PUL, 0.974 (95% CI:0.954-0.994) for the prediction of IUP and 0.900 (95% CI:0.812-0.988) for ectopic pregnancy.[22] This model appears superior to M1 when comparing AUCs for the prediction of ectopic pregnancy in a PUL population but, in real terms, it did not result in many more pregnancies being correctly identified as ectopic pregnancies.[22] The obvious advantage of these models is that their use is independent of clinical experience and they do not require any understanding of clinical biochemistry in early pregnancy. In a recent prospective study investigating 363 PULs, it was demonstrated that such a model can be used as a basis on which to rationalize the management of PULs, as it can successfully minimize their follow-up by reducing the number of visits, scans and blood tests, as well as intervention rates.[7] Although misclassification of the final outcome resulting in rupture of ectopic pregnancy is a possibility with the use of such mathematical models, in this prospective study there was no serious morbidity in any of the women and, specifically, no ruptured ectopic pregnancies.[7] These models may, therefore, have the potential to de-skill the interpretation of serum hCG levels and lead to more standardized management protocols. However, they need to be tested prospectively and in other patient populations.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: