Background: Women having pregnancies of unknown location (PUL) can be defined as those having positive pregnancy test when no pregnancy is visualized on transvaginal ultrasound (TVS).
Aim: To identify diagnostic parameters which are predictive of ectopic pregnancies in women with early pregnancies of unknown location.
Methods: We undertook a prospective observational study of pregnant women with suspected early pregnancy complications. Ninety-four patients were classified as having a pregnancy of unknown location (PUL) by transvaginal ultrasound; blood sample was taken on presentation to measure the serum human chorionic gonadotrophin (,-HCG) and progesterone levels. All collected data were tested by univariate analysis and then analyzed in a stepwise procedure to form a logistic model for predicting ectopic pregnancy.
Results: A total of 2675 women were referred for suspected early pregnancy complications. In 94 (4%) patients the location of the pregnancy was unknown. Three parameters were found to be statistically significant for predicting ectopic pregnancy:progesterone level, vaginal bleeding associated with pain and the presence of free fluid in the pouch of Douglas. The overall model described by these variables offer a sensitivity of 79 %and a specificity of 59% in the prediction of ectopic pregnancy.
Conclusion: Logistic regression model can help in the clinical decision-making in women with pregnancy of unknown location.