Clinical use of a model to predict the viability of early intrauterine pregnancies when no embryo is visible on ultrasound.

BACKGROUND

When a small gestational sac with no visible embryo is seen at an early pregnancy ultrasound scan, the clinician cannot distinguish a viable from a non-viable pregnancy. A test for the prediction of early pregnancy viability at the initial visit was developed in 2003. Maternal age, gestational sac diameter (GSD) and serum progesterone levels were used in a logistic regression model to create an algorithm for estimation of the probability of a viable pregnancy. The objective of this study was to assess how well the test performed in routine clinical practice.

METHODS

This is a retrospective observational study of women who had the test performed in our Early Pregnancy Unit over a 6-year period. Inclusion criteria were a spontaneous conception, gestational sac of <20 mm mean diameter, no visible embryo on transvaginal ultrasound scan and outcome data regarding the viability of the pregnancy.

RESULTS

Of 5163 potentially eligible women, 472 had the test performed (9.1%) and 400 met the inclusion criteria for the study. Women who were older or with vaginal bleeding, a more advanced gestational age or a history of previous first trimester miscarriages were more likely to have the test performed. At follow-up, 199/400 (49.8%) women had a viable intrauterine pregnancy, and 201/400 (50.2%) had a non-viable pregnancy. The logistic regression model performed better than serum progesterone, β-hCG, mean GSD or maternal age alone as single parameters to differentiate between viable and non-viable pregnancies, but the area under the curve was lower than in the 2003 study [0.85 (standard error 0.021) versus 0.97 (standard error 0.011)].

CONCLUSIONS

Although less effective than in the original study, the logistic regression model was able to predict pregnancy viability with reasonable accuracy when applied in clinical practice. The test appears to be under utilized and further prospective studies are needed to establish if the test is of clinical benefit, for example, in reducing patient anxiety.