O–122 Developing a robust model for predicting the outcome of pregnancies complicated by threatened miscarriage in the first trimester

Introduction: Approximately 250,000 miscarriages occur in the United Kingdom each year with significant psychological distress to affected families. 20% of all pregnancies are complicated by threatened miscarriage, presenting with vaginal bleeding while the pregnancies remain viable. These pregnancies have a poorer outcome than pregnancies in which there is no bleeding and about 20% miscarry. We currently have no way of predicting which threatened miscarriages will result in pregnancy demise and so are unable to intervene by way of counseling or attempts at rescuing the pregnancy. This has led to wasteful and potentially harmful interventions including unnecessary blood tests, ultrasound scans, hospital admission for bed rest, sexual abstinence, low dose aspirin and progesterone supplementation. Materials and Methods: This was a prospective longitudinal single site study. Women with threatened miscarriages between 6 and 10 weeks gestation were recruited at onset of vaginal bleeding. Each participant had a viability ultrasound scan on recruitment, weekly charting of pain and bleeding scores for five weeks, weekly assays of serum progesterone and hCG for five weeks, and an exit scan on study completion (four weeks after recruitment). The primary outcome measure was detection of fetal demise within five weeks of recruitment and the secondary outcome measure was interval to fetal demise. Logistic regression was used to determine the study parameters that impacted most on the risk of miscarriage and these were used to derive a Pregnancy Viability Index (PVI). Results: 117 women with threatened miscarriage were recruited of which 15 (12.8%) dropped out of the study; their outcome data were unknown so they were excluded from further analysis. Outcome data for 102 women who concluded the study were analysed. At study conclusion, 80 pregnancies (78.4%) were ongoing while 22 (21.6%) had miscarried. Interval to miscarriage could be accurately ascertained in 18 pregnancies of which 10 (56%) occurred within 7 days, 13 (72%) within 14 days and 16 (89%) within 21 days of onset of vaginal bleeding. Logistic regression determined that the parameters that impacted most on the risk of miscarriage were serum progesterone and hCG levels, fetal crown-rump-length, bleeding score and maternal age; these were used to derive the Pregnancy Viability Index (PVI). PVI averaged 808 and 106 for ongoing and demised pregnancies respectively (p = <0.001). Setting a PVI discriminatory value of 145 accurately predicted the outcomes of 95/102 pregnancies (93%) consisting of 75/80 (94%) ongoing pregnancies and 20/22 (91%) demised pregnancies. Therefore a PVI of 145 had a 97% positive predictive value and a 75% negative predictive value of pregnancy viability five weeks after threatened miscarriage. Conclusions: This study has demonstrated the feasibility of predicting the outcome of pregnancies complicated by threatened miscarriage. Clinical impact § we can offer targeted psychological counseling and support to affected women/couples hoping to reduce anxiety levels and improve the overall experience § we can eliminate unnecessary, wasteful and potentially harmful clinical practices for affected women/couples § this model lends itself to widespread adoption because it is simple, inexpensive and reproducible without the need for sophisticated equipment or gadgets Scientific impact § we have for the first time demonstrated a robust model for predicting the outcome of pregnancies complicated by threatened miscarriage in the first trimester § this will facilitate further investigation of the approximately 21% of such pregnancies likely to miscarry to gain a better understanding of the mechanisms underlying the miscarriage process particularly with regard to the role of genetic abnormalities in threatened miscarriage § it will facilitate investigations of interventions to rescue such pregnancies