Predictive value of plasma human chorionic gonadotrophin following assisted conception treatment

Abstract

A total of 429 pregnancies after assisted conception treatment was analysed, using receiver operator characteristic curves. The best balance between sensitivity and specificity for predicting viable (single and multiple births) and non-viable (fetal heart positive abortions, ectopic and biochemical pregnancies) outcomes was human chorionic gonadotrophin (HCG) 50 IU/l on day 14 and 200 IU/l on day 21 after treatment. Utilizing these indices all pregnancies could be classified into one of four groups. In group A (day 14 HCG <50 IU/l and day 21 <200 IU/l), the probability of a birth was 0%, pregnancy loss 72% and ectopic pregnancy 28%. Conversely for group D (day 14 HCG >50 IU/l and day 21 >1000 IU/l), the likelihood of a birth was 90%, pregnancy loss 8% and ectopic pregnancy only 1%. Between groups A and D there was, as expected, a gradually shifting balance in favour of a reduction in ectopic (28, 13, 3, 1%) and biochemical pregnancies (70, 36, 33, 2%) and an increase in fetal heart positive pregnancy losses (2, 6, 13, 7%) and births (0, 44, 50, 90%). The majority of multiple pregnancies (98%) occurred in group D. Two accurately linked HCG measurements allowed a greater predictive accuracy of pregnancy outcome than could be obtained using either alone.